It is currently Thu Aug 28, 2014 8:09 am




Forum locked This topic is locked, you cannot edit posts or make further replies.  [ 1 post ] 
 Trouble-shooting a P9700S 
Author Message
User avatar

Joined: Sun Nov 09, 2008 2:49 pm
Posts: 575
Post Trouble-shooting a P9700S
giulio wrote:

Hi I've jst ordered and payed with paypal a 9700 kit. I just wondered: does it include patch cables? how many of them? if they're not enough for my needs may I ask you to put some more in I'd pay the difference, of course) thanks

Giulio Moro

-0-0-0-0-0-0-0-0-

Scott Lee ha scritto:

Hello Giulio.

Thank you for your order! Yes, a patch cord kit is included. It is a collection of connectors and wire to make patch cords for the system. With several short ones and a few longer ones, you'll find you can make some useful and interesting patches. Here is a tip that you won't see in the instructions for making the cords: For the Gate-trigger output, use one connector with two or three wires/plugs so it can split to multiple points (each module has a gate-trigger input (G)). You will notice there is a US-type 120vac to 12vac (1 Amp) wall-mount transformer supplied with the kit and this is the only type we have. You may use this with a travel adapter, or substitute one with a similar output to match your wall-outlet and voltage (it wires direct so any connector on the end is not a concern). I have attached some extra information I have prepared about the P9700S which you might find useful.

Thank you. Sincerely, Scott Lee

-0-0-0-0-0-0-

*From*: giulio <oromoiluig@yahoo.it>
*Sent*: Thursday, January 01, 2009 8:16 PM
*To*: Scott Lee <scott@paia.com>
*Subject*: SPAM-MED: Re: Thank you for ordering from PAiA.

Hi I've finally managed to build my Paia 9700.
I must say that a few parts were missing: all of the 1k resistors, 4 33k resistors, and all of the
15pf caps in the VCO(in place of them there were 7 470pf). I managed to get these parts from my
local dealer. Furthermore(that was not a problem), in both the VCO and VCA frmount kit there was one
of the L brackets which had both of the holes threated.
So far I've begun the testing of the VCO. The problem I found is that the pulse width amount (as set
by the pulse width pot + the modulator) slightly affects the pitch of all of the OSC B outputs. For
instance turning the pulse width pot from 0 to maximum makes the pitch slightly rise and then fall
again.
At a first glance wiring looks ok. Maybe should I check some shorts on the PCB?

Just one more info-question: on the midi2CV PCB I cannot understand if the G and SG points are
shorted together. The two points are very close to each other and it looks like there is a
connection between them. Anyway, presently they are connected together on my midi2cv as I couldn't
manage not to short them. Is it right or wrong? I noticed that when the shield of the cable that
goes from the output of the VCO to my mixer+amp touches the metal frame of my table the volume goes
to 0 .

Thanks
Giulio Moro

-0-0-0-0-0-0-0-0-


Scott Lee ha scritto:

Hello Giulio.

Congratulations.  That is quite a project!  I apologize for the missing resistors and mis-supplied capacitors.  Thanks too for the report on the L-Brackets with a thread on both holes.  It is possible a new order has a manufacturing defect and was poured in with the others in the bin for theh parts.

The interaction you are noticing in the VCOB between pw and pitch can be corrected with a replacement of the 91k resistor at R59 with a 27k.

Yes, G and SG join at the MIDI2CV8.  Being the power source for the other modules.  The two grounds do start as the same point, but as the ground extends to the modules over wires and pcb traces, There are separate circuits for G as power ground and SG as signal ground.
The vco output going away when it touches metal might be the result of the metal being at ground potential and the wave on the cable being in contact with both the tip and sleeve terminals of the phone jack at the module or the phone plug of the cable.  I wonder if this occurs when the cable is in other waveform outputs, or, from the noise source?

Thank you! Sincerely,  Scott Lee
-0-0-0-0-0-0-0-0-0-0-0-0-
giulio wrote:


Thank you very much Scott, substituting the resistor solved the issue, in a first moment. Now there are some more problems, but I cannot tell if these have begun just after the resistor substition(and thus could be derived by it): as I substituted the resistor I verifyed that the pulse width modulation did not affect pitch anymore, then I moved around the synth and now there are some problems.
I must tell you that I hadn't tried to connect the midi2cv before today, and, as I'll
explain later, it doesn't work, so I'm not putting some CV into the VCO, but just turn the pots around.
The VCO now features the following issues:
- from time to time it doesn't power on(no output and no red led), but then I'll turn the switch
power on the midi2cv off and on a few times without touching anything and then in a few attempts the
VCO powers up(this used to happen even before the resistor substitution)
- The sounds I hear from the sine,triangle and ramp output now sound about the same to me, and I think they are all ramps. I'm sure there is no sine output.
- the normal-lfo switch seems to switch between two different lfo's,instead that betweeen a normal and a lfo, one with a lower freq of about 10Hz(I hear about ten "tock" in a second) and one with a much longer period(longer than a minute)
- the ar/cycle switch on the modulators VCO alters the pitch of both osc's(and oscA in both LFO and normal modes)  the pitch is slightly higher when in ar mode and lower in cycle mode(I also tried to plug a plug into the modulation out to disable normalization,but it did not have any effect,as expected).
- I've just realized that now the pulse width modulation affects pitch again!

As for the midi2cv, first of all a question: is it normal that the dip switch is INSIDE the panel?
How am I suppose to move the switches? I use a small flat head screwdriver
Now with the issue: when I power it up with the switches set as suggested in the testing section of
the manual the outputs go to about 11.7 V DC and keep that value.
I tried different random settings for the switches and the result was the same...

Sorry for the troubles I'm causing...
Thanks,
Giulio
-0-0-0-0-0-0-0-0-
Scott Lee ha scritto:

Hello Giulio.

Thank you for this message.  I see too you have sent another with files too, but right away, there are a couple of possibilities that come to mind.

The ac transformer:  It must not attach with an added connector unless it is a plastic type and the two circuits or transformer wires are isolated from the chassis, or it is in-line and not mounted on the chassis.  Otherwise, the ac voltage wire could be in contact with the chassis, or, the neutral wire could be in contact with the chassis, and this could cause the voltage to be unreliable or damage the transformer or make hum noise in the ground.

The MIDI2CV8 power switch:  These can fail or be intermittent if the solder terminals misalign or become loose or wobbly due to heat from soldering softening the plastic body.  A replacement switch might be needed if you see crooked terminals or they wiggle.

You asked about the grounds joining at the MIDI2CV8 before, and yes, they do join.  But from the MIDI2CV8 out to the other modules, they are separate wires G and SG.  If there is not a complete circuit on G or SG to the VCO and the other modules which chain from one to the other, they can behave very strange, if they operate at all. Take a second look at the soldering for the G and SG wires at the printed circuit board and at the terminal that is in the four-circuit power supply connector cover.  A catch can be pressed through the tiny opening to release the terminal from the cover for inspection.  It is possible the solder did not flow on the terminal, or, it has deformed (smashed) so it isn't contacting the mating pin of the four-pin pcb mounted header.
About the DIP Switch for the MIDI2CV8 MIDI Channel and Operating Mode, it is not unusual for a small screwdriver or other stylus or pointed object to be needed to get at the tiny levers, but the part should face out the opening, at a right angle to the pcb.  It should not be on the same plane as the pcb.  Did you get the wrong type?

I have attached some extra notes about the P9700S which you might find useful.

Thank you. Sincerely,  Scott Lee
-0-0-0-0-0-0-
giulio wrote:


Hi Scott. Two wires go from the AC transformer directly to the midi2cv PCBoard, so no shorting on case. As for the DIP switch: I'm attaching some photos to this message to show you what I have.
I forgot to mention that the other modules do power up even when the VCO does not. The modules are chained in the following way as for the power supply: MIDI2CV, VCO, VCF, VCA , so VCF and VCA are powered by the VCO, and even though it does not power on, the VCF and the VCA do. So I don't think it's a matter of the power switch, but I suppose it is something inside the VCO.
Furthermore, with the VCO-MIDI2CV power supply disconnected(and the others connected) I checked on the VCO if G and SG shorted and they were not. I also checked that there is a complete circuit on G or SG to the VCO and the other modules which chain from one to the other.
I tested the VCO board for the voltages specified in the attachments you sent me.
+12V reads 11.80 on IC2, -12V reads -11.87 on IC1. Voltage across r1 is 0.88 , across r2 is 2.19.
The only other values that did not match were:
R86/87: I read -11.76
R87/95: I read 4.90
IC8: I read pin5 0.38 - pin6 0.37 - pin7 2.32
I haven't checked voltages on the MIDI2CV yet. If you got more documents ready with reference voltages please send them to me.
Be sure to check the email with attachments I sent you yesterday, I think it is important.
Thanks
Giulio
-0-0-0-0-0-0-0-0-0-0-
Scott Lee ha scritto:

Hello Giulio

Thanks for these updates.  I did see and hear your files, after looking over your measurements.  I like your scope images!  The recordings sound good too!  Now if only the VCO sounded good!
It is surprising that it would have suddenly changed they way it did, but the voltage reading at R86/87 points to a possible mix-up of the 1k and 10k trims for the Scale and Sym locations.  A way the sudden difference could have occurred is if one or both of the mini-toggle switches failed due to heat from soldering causing the terminals to misalign or become loose.

The Sym trim is the waveform symmetry (shape) and the Scale Trim is a setting related to frequency.  With power switched off, you should get comparable resistance measurements on the two sets of trims in VCOA and VCOB (the red and black probe placement should be consistent from A to B; ie, red to R85/86 and black to ground, and, red to R4/R5 and black to ground).  The marking for these trims is on the underside of these parts, I think.

For the images of the DIP Switch for Channel & Mode selection, I see that you have the proper part.  Yes, it is not so convenient, but it is something that does not have to be changed often.
Thank you. Sincerely,  Scott Lee
-0-0-0-0-0-0-0-0-
giulio wrote:


Hi scott
maybe I was wrong: in the preceding measurements I measured voltage across R86/87 and R87/95 with the black probe on one end of the first resistor and the red probe on one end(the one that is not connected to the other resistor)of the second resistor.
What about the other voltages I measured that were wrong?

The trims are in the correct position. I've fixed one thing: I realized that I had moved the sym trim of OSC A to one of the ends, now I adjusted it for the "most mellow tone". I'm attaching a JPG of the sine wave I get. Am I supposed to get any better of this?
Well, so the sine is fixed, at last!

Now the biggest problem is the instability of the oscillators. I've just realized that(even with modulator output set to 0) the pitch of OscA seems to follow the brightness of the LED.
I've also realized that WHILE I connect other modules to the power supply of the VCO the pitch varies, but I suppose it's pretty normal.

I'll get a couple of new switches to substitute the existing ones.

Best regards,
Giulio
-0-0-0-0-0-0-0-0-
Hello Giulio.

This is good news, the trim setting fixing the VCO wave.  I wonder now if the remaining trouble is only due to the voltage from the MIDI2CV8 being a bit low for the other modules.  Is it a 12vac 1000mA (or 1A) transformer that you are using.  If so, maybe you are using the US type with a travel converter and there is some loss.  the dc voltage going in to each module should be about 15 to 16v.  If it is less, then the modules are not getting enough power from the transformer and a 13v 1A is the answer, or 12v with more mA.

Thank you. --Scott
-0-0-0-0-0-0-0-0-0-
giulio wrote:
I'm now using a 100VA 220/12 AC transformer(I had it at home for another project), so it is not underdimensioned...
The pitch is NOT stable even though the VCO is the only module connected and if I'm not doing anything.

Thanks
Giulio

-0-0-0-0-0-0-0-0-0-0-

Hello Giulio

I apologize for this delayed response, but I had to be away from emails a couple of days.
Your power supply certainly does have a powerful rating, but I wonder if it might be too good, in that the voltage might be a more constant 12v ac than would be available from a 12vac wall-mount type similar to the type we provide with the kit.  For these, the resulting rectified dc outputs on the 9700K MIDI2CV8 are are +18v and -18v dc without the other modules connected, then, when they're connected and presenting more of a load to the supply, the dc is +16 and -16.

The voltage regulators on the VCO, VCF, and VCA modules must be getting a minimum of 15v dc for them to provide the needed, stabilized 12v dc supplies. Since we covered the possibility of the 9720 VCO module not getting a good G or SG circuit, it just made me think that maybe it is the voltage dipping down a bit with the modulator section action.

If the dc to the module is 15v or more then maybe there is a connection where the solder didn't flow on the panel control and connector ground wiring.  Wire C connects the pcb SG circuit up to the panel at R203-1.  It goes two ways from here to panel ground terminals including the two mini toggle switches, but if a joint where multiple wires join did not flow, the parts beyond would be referencing each other instead of the zero potential SG (signal ground/circuit-common).  See figs 3 and 4 of the assembly manual illustrations for these wires and connecting points.

Thank you. Sincerely,  Scott Lee

-0-0-0-0-0-0-0-0-0-0-0-0-0-0-0-0-0-0-0-0-

giulio wrote:

Hello Scott

I've discovered a strange thing: after verifying the AC input(it reads about 12V AC at the poles of the power on switch in the midi2cv), I tested the DC output of the module, and it reads.... 31V DC
with other modules disconnected and 30V DC with other modules connected. I don't know how these results come to happen. The transformer I'm using is in fact a 2x12V AC with three outlets marked 12 0 12  and I'm connecting one wire to the first 12 and the other to 0.
Tomorrow I'll try to substitute the transformer. Can I use a wall-mount transformer with the following charachteristics?
INPUT: AC100-240V,50/60Hz,0.6A
OUTPUT: DC 12V/2A
I also have some other wall mount DC transformers giving an output of up to 15V / 700mA, are they suitable?
Or should I get something else?
Where should I connect the 2 wires coming from the wall mount DC transfoemer(I can't understand exactly what is said in the "Custom apps" chapter of the manual)?
Thanks
Giulio

...then...

Ah ah! the 2x12V AC  transformer I was using had two primaries: one for 230V and one for 400V . I plugged 230V into the 400V winding and thus I got about 13V on the output(where I should get the 24V with "ordinary wiring"). And...now(almost)everything works perfectly in the VCO: it now retains a constant pitch over time!
I'm getting 16.5 VDC out of the rectified output of the midi2cv now, with all the other modules
connected. Great!(I must say I was stupid  to measure the voltage between the (+) and (-), instead as you may have noticed from my last email. Anyway, with the old "ordinary wiring" I got about 15V DC out of the midi2cv, and, as you stated, they weren't enough). Great! The LFO switch still gives troubles, I'm going to replace it. Now we can focus on the not working midi2cv. With the switches set on selftest mode, I power the unit up and the output of each channel suddenly goes to about 13V and then keeps that value. In the 9700k tss.rtf file you sent me is written in the first lines: "Double-check to be sure the 'conditional' steps weren't overlooked on page 11 (jumpers  at Out-In and OctC-G, and, 5600 @ R28)." what is this referrin to? can you send me more documentation if available? I've done a quick test as described in that file for voltages on the 28 pins. I had no scope available at the time, so I only veryfied high and low voltages. Only pin 14 is 0V, while 20,25 and 26 are about 1.5 V. Also pin 27(which should be high) is 1.5V.

Thanks Scott, we'll manage to have it working!
Giulio

-0-0-0-0-0-0-0-0-
Hello Giulio.

This is good news!

Yes, OK, for the conditional steps.  These are the ones on page 12 of the MIDI2CV8 manual.  If you are using the 9700VHZ option for the alternative cv scaling, then you are to skip ahead from page 11; however, the P9700S does not use the vhz option so the assembly is completed on page 11 with two jumper wires:  a) Out to In, and, b) OctC to G.  Then, a 5600ohm green-blue-red-gold is to be installed at the R28 location.
These steps must be performed in order for the CVandTrigger outputs to occur.
It is possible the 1.5v readings you got for high/low tests were due to pulsing as the micro-controller attempted to address out of the normal program range.  This could occur as a result of pins being loaded by the tester.  If trouble still persists, look for possible weak links between the micro-controller and the EPROM firmware IC (with the MIDI2CV8 label)--a solder that did not flow or a pin that bent under the part and didn't make it into the socket, etc.

To be sure about the Channel and Mode switch settings, for the setting as in the image you sent IMG_2962 with the sections On Off On On _ Off On Off Off you have selected Channel Three and Mode Six as shown on page 24 of the manual.

Thank you.

Sincerely,  Scott Lee
-0-0-0-0-0-0-0-0-0-0-0-0-0-0-
giulio wrote:

About what I've just written:"Furthermore, I noticed that even with the ADSR of the VCA set with attack,decay and release to 0 and sustain to maximum, the volume of the A output slightly decreases as the note is sustained, until it reaches a final value slightly softer than the starting one. "
This happens with the gate and trigger outputs of the midi2cv connected to the respective inputs of the VCA
...then...
giulio wrote:

Hi Scott, I was stupid and I did not follow those conditional steps.
Now I've done that and it works, as for the self test and midi IN test. Just one thing about this: I happen not to know what "status on basic channel" and the other similiar phrases do mean in MIDI term. I just want to let you know that in MIDI IN test the led after every signal I send(note on-off,program change or pitch wheel) blinks regularly(50% duty cicle).
Anyway, I finally connected all the things together, and now as I press keys on the keyboard the VCA sends out a sound, as in a true synth ;-)
Some notes about tuning: I spent many minutes following the tuning procedures(mode1,channel1) as written in the manual, and I couldn't manage to get in tune. At last I noticed that the keyboards I have do send velocity data and these velocity data do slightly(a quarter of tone) affect the voltage sent out of output 1(pitch control).
How can I get rid of this?
Another thing about the midi2cv is that when I release the key on my keyboards the pitch of the oscillator changes, thus making impossible to use Release values on the VCA ADSR different than 0.
Furthermore, I noticed that even with the ADSR of the VCA set with attack,decay and release to 0 and sustain to maximum, the volume of the A output slightly decreases as the note is sustained, until it reaches a final value slightly softer than the starting one.

Thanks
Giulio
-0-0-0-0-0-0-0-
Hello Giulio.

Good news again!
The phrases are related to the MIDI messages.  MIDI messages contain multiple bytes or data words made from data bits.  A "status" byte is sent to indicate the type of "data" bytes that will follow.  A status byte can be sent with each message, but to save on the number of bytes that are sent, it can just be sent once.  If your keyboard operates with
"running status" like this, it will send the Note On Status byte once, then the other messages for key-presses and key-releases can be Note-number Data byte, Velocity Data byte (with velocity at some greater than zero value for a 'note-on' and equal to zero for a note-off).
So, for the MIDI In test, if the first MIDI byte that is sent in is a Status byte on the selected MIDI Channel the MIDI LED is operated on most of the time with repeated flashes off.  If the first message received by MIDI2CV8 is Status but not on the selected MIDI Channel, the MIDI LED is operated in equal on and off times.  This is what you saw, so the MIDI Channel for the keyboard and the MIDI2CV8 were not the same.  Then, for the third condition, the first message seen by the MIDI2CV8 is a Data byte.  This is likely an indication your controller sends the MIDI under Running Status and that the cable was connected to MIDI2CV8 after the status byte had been sent (changing status, ie pressing a program change button or rolling a pitch or mod wheel sends a new status (Program Change, Pitch Wheel, Control Change)).  Note; however, MIDI2CV8 must be reset (power off, then on) for the MIDI Test to stop indicating the test result and begin testing again.

About the pitch changing when keys are released, you are sending only the pitch cv to the vco, right?  Velocity should not be sent to the VCO or affecting the pitch.  Is the voltage from the MIDI2CV8 to the other three modules holding at more than 15v?

When measuring the pitch cv for 1v changes for octave-spaced key-presses on the MIDI controller, watch to see if there is a difference in the voltage for the key-pressed and the key-released.  It should hold steady.  I am including tips and suggestions for getting the tuning set below.

Also, the Sustain phase of the VCA ADSR envelope should hold for the duration of the Gate-trigger being active (for a key being held down).  Look at page 22 of the manual for the 9710 VCA for the circuit sections associated with the VCA ADSR.  There are many things that go on at once, so it is simplest to just look for possible mistakes or errors or soldering trouble with the parts shown here.
First set the MIDI2CV8 to be outputting 0V for the low key pressed on the controller, by powering the 9700K while holding the low key on the controller so that releasing the key will cause the first message input to the MIDI2CV8 to be a note-off (it interprets this as a signal to output zero volts for this note number).

A dvm can now be used to check that octave spaced key presses are outputting a 1v pitch cv change (down to a hundredth of a volt or so) as set by the trim on the MIDI2CV8. The most accurate setting will be obtained if you press octaves that start a few notes above the lowest key, ie if your low key is a C, press D or higher for the lowest octave reading.

The usual tuned setting for the VCO scale trims is about a 1:00 setting for the pointer of there disk which covers a cw range from about 7:00 (ccw) to 5:00 (cw). Start with the trim at 1 o'clock.

After having been powered for a minute or two patch the low-key transposed pitch CV over to the VCOA P2 input. Set the two pitch controls to they're both in-tune at unison at about a mid-rotation setting for the low key pressed on the controller. Play an octave higher and the adjust scaleA for an octave relationship with B. Go back to the low note and adjust the A panel pitch control for unison and then, again, press an octave higher and adjust the scaleA for an octave relationship with B. After going back and forth like this another time or two, you should find the scale is about as close as it can get.

Then, do this all again pressing some note a few keys higher than the lowest key (with VCOB adjusted to match at this new low pitch) and confirm/tweak the scale to match for higher pressed octave relationships with B (using the A panel pitch to realign A with B after adjusting the scale). When the scaleA adjustment is complete, move the pitch control over to VCOB P2 and make the panel Pitch and scaleB trim adjustments to get unison and octaves with the VCOA pitch (adjusted to match B for the low key pressed and pitch CV sent to it).

After the scaleB is set, move the Pitch CV over to the VCOA P1 input and listen to the two vcos tracking across the keyboard range (again with the low-key transpose asserted) after aligning their pitch while a higher note is pressed, say the third or fourth octave. Confirm they track for a higher octave pressed and hold as you press keys going down towards the low key. You should be able to turn up the sound of the controller or some other MIDI Sound Module and hear that the tones made from the VCOs can be aligned using the panel pitch controls and follow as you play different ranges on the keyboard (make the alignment between the two when a key in the middle C to A 440 range is pressed--its easier to hear the beating slow as they are tuned).

Thanks. --Scott

-0-0-0-0-0-0-0-0-0-0-0-

giulio wrote:


Hi Scott

I have two keyboards capable of sending midi data: a Roland VK-7 and a Kawai K3,  they are both set to send midi data on channel one. With the midicv2 is in "midi in self test"mode and the channel is set to 1, the led always blinks as I described(50% duty cycle). It starts blinking both if I turn on the midi2cv with the keyboard connected already on and if I turn on the keyboard connected with the midi2cv already on. I think this is quite strange.
It's strange also because if I then turn it into mode 1 and reset the midi2cv, it works(that is it outputs pitch CV's, triggers and gates regarding data received on channel 1, though pitch CV is not stable as I described).
Of course I was sending only the pitchCV to the VCO.
With a voltmeter I managed to set the scale of the MIDI2CV, so it now tracks correctly(as for the C's, which voltages are integers, but I suppose the other keys work well too) a 5 octave range, with keys pressed with the same velocity(low velocity).
Here I am reporting some values I measured with the voltmeter(I tracked notes on the first 2octaves as in that range my voltmeter is capable of reading millivolts).

C
minimum velocity: 1.016
after release: 1.003
maximum velocity: 1.025
after release: 1.011
A
minimum velocity: 1.771
after release: 1.755
maximum velocity: 1.783
after release: 1.765

I also verified that output 2 follows velocity: from about 0 for minimum velocity to about 10.30 V for maximum velocity.

With all modules connected the voltage on the red wires holds on 16.5V

A good news: I've replaced the LFO-normal switch on OSCA and now it seems to work correctly.
Another(sigh!) bad news: the VCO still does not power on each time I turn the power on the midi2cv, though other modules connected to it do power on.
In those cases I have to turn power off and then on again, and most of the times it works.

As for the VCA, I'll check it in the days to come.

Best regards,
Giulio
-0-0-0-0-0-0-0-0-0-0-
Hello Giulio.

Thank you for these updates.  Your mention of the blinking starting as soon as the keyboard is connected or the MIDI2CV8 being powered, points to the controller sending a System Real-time Message such as Active Sensing or MIDI Timing Clock (just a status byte but without any channel information).  Since the MIDI2CV8 does respond to Channel Messages, I believe it is just a matter of me not recognizing this as being a possible condition for MIDI In test :-[

About the velocity cv affecting the pitch cv, this is not normal.  Take a close look at the soldering and parts associated with transistors Q4- Q7 (R34-R37, R23, R24, R25, R27), IC13, and IC11A, IC11B (C10, C11) .  These parts are all related to the CVs being distributed to their specific outputs.  Inspect the soldering for these circuit sections. R34-R37 should be 100k brown-black-yellow-gold and it is easy to get the 47ohm (yellow-violet-black-gold) put in place of one of these parts and vice-versa.  Maybe there is too much ripple voltage on the dc power to the MIDI2CV8 and this is getting on the voltage that goes out as the cvs.  Capacitors C19 and C21 are 2200uF to help smooth the dc that is V+ and V- (about 16 to 18 vdc positive and negative).  If a connector has been added on the chassis to bring in the ac power to the MIDI2CV8, it could be making ground noise that could be contributing to the variation.  Insulating washers can prevent this contact (or a plastic connector which isolates the transformer wires from the chassis, or, a the connector wired in-line and not touching the chassis).  If there is more than a usual amount of ripple due to the load on the power source, then the trouble might go away if you try just the MIDI2CV8 alone and without the other modules connected.  This could explain the inconsistency with the VCO power-up too, and maybe the VCA ADSR changing.

Thank you. Sincerely,  Scott Lee
-0-0-0-0-0-0-
giulio wrote:

Hello Scott,
I checked all the components indicated, and I saw no shortcircuits or wrong part in the wrong place. I also re-melted the solder on eache of those soldering, with no changes in the behaviour of the module.
I measured AC voltage across pin "+" and pin "-" of the midi2cv, with no other module connected and it read about 370mV.
I have no metallic connector for the power supply on the chassis : I'm still "working in progress", so there are two wires from the power supply transformer that go directly to the power switch on the front panel of the midi2cv.
I know little about electronics and nothing about digital electronics, but I noticed that pitch isn't affected only by velocity, but also by releasing the key, as I told you. This could mean that it is affected in some way by the gate(that changes when key is released) and/or by the "note off" signal, beside the velocity...
I tried to connect only the VCO to the midi2cv and to power up and down the whole thing several times: it still happens from time to time that the vco does not power up.

Thank you
Giulio
-0-0-0-0-0-0-0-0-
Hello Giulio.

Thank you for this reply.  The power does seem seem to be OK.
I wonder if the trouble might be with the IC13 part, a 4051, on the MIDI2CV8.  Sometimes these can be more sensitive to their power supply range than is usual and when this is the situation, they will tend to feel warm to hot to the touch when in operation.  We have recently started modifying the printed-circuit-boards at this IC13 position by cutting the printed circuit for the run to pin 7 of the part and include a wire link to join pins 7 and 8.  This puts pin 7 on the ground circuit instead of the negative dc (V-) circuit better suiting these more sensitive parts.  If your board does not have this change, I think that making the change might be the key to stopping the interaction between the outputs.
Use a knife to scrape and open the printed circuit trace leading to pin 7 of IC13 on the bottom side of the board (see fig. 1b of the MIDI2CV8 assembly manual illustrations).  This trace can be seen looking through the invisible board just to the right of the letter A by the DAC Tune marking on the board.   Then,  solder a short piece of wire (a leftover clipping from a resistor or capacitor) to link pins 7 and 8.  The pins of the IC count up in a counter clockwise direction around the part from the notch as viewed from the top.

For the VCO, it seems that something doesn't have a good connection or there is something awry that is resulting in a teetering condition.  The connections are easy to check visually, and, that the proper parts are in the specified spots. So we can go over these first, then, make some tests to try and zero in on the trouble.    If you take the assembly illustration pages loose from the manual, it makes it easier to trace points on the board and panel wiring and connections as shown on figs. 3, 4, and 5.  Fig 6 shows the circuit sections in a block form.  The Glide circuit links the pitch cv to both VCO sections A and B.  The modulator section links with the VCO B pulse width.

Inspect the board soldering under some good light (a sunny window?) looking for joints that have a tubular look resting on the pad with the wire or component lead passing through.  This joint might or might not make the connection.  Any like this should be reheated, flowing in a tiny bit more fresh solder.  The solder should look 'one' with the solder pad and the wire/component-lead.    Check the wire links soldered to the panel controls and connectors as shown on fig. 3.  If moving any wires shows movement in the joint, it would be good to flow the solder at the joint again, feeding in a bit more solder.  Take a look at all the connectors for any sign of a bulging.  Heat during soldering can soften the plastic allowing the contact in the part to misalign and the closed-circuit connectors with the X terminal must have good contact for the normally closed circuit.

Wires to R200-3 and R201-3, the Attack and Release controls are joined with the #2 terminals via wire links.  These controls each have a connection with R207 violet-brown-brown-gold 750 ohms and wire back to the board at point AE.  When wire N gets connected with ground via the modulator AR/Cycle switch, it goes to AR mode (light is off, waiting for Gate-trigger or Pulse-trigger).
Wire A to R206-3 carries the +12v power supply up to the Pitch and PW controls.  It jumpers over to R204-3, then to R203-3.

Wires J and Y and the bare wire link on R-205 complete the circuit through R205 the Glide control for the pitch cv to both vco sections.

Look for any stray wires or connector terminals in the area where the board meets the panel and the panel parts are near the board parts.
Of course, don't overlook the possibility of a patch cord having a solder that did not flow or with a frayed wire in it that is sometimes shorting the tip and sleeve circuits of the plug.
For testing the voltages, measure the +12v and -12v supplies to see if one of them not going to the full voltage is related to the trouble.  Then, the pitch cv that is on pin 14 of IC7D (also to J5-X on the panel and to R14 on the board).  For these measurements in the modulator section, set the Attack and Release controls at minimum.  AR/Cycle can be either way.  The 'flip-flop' circuit made by the parts including transistors Q5 and Q6 should have normal power-up voltages of about 11v on Q5-c/R38 and about 1v on Q6-c/R37.  Resistors R96 and R103 make a voltage reference for IC8C, pin 9 and the dc voltage here should be about 5v.

Here's to success!

Thank you. Sincerely,  Scott Lee
-0-0-0-0-
giulio wrote:

I've also realized that measuring with an ohmeter the resistance between o1 and the other o spots on the pcb something strange happens: the ohmeter reads about 900k when set to 20M scale, but when set to 2M it reads "1" (that is out of range).
Desperately yours,
Giulio

....then...

Furthermore, pitch is still affected in both mode 2 and 3 by the gate(that is: on release of the key, pitch lowers a bit). I'd say that every pitch is affected by big changes in the other outputs, as if the signal is not buffered enough
Giulio

....then....

Hello Scott. I did what you told me on the midi2cv, but it did not work. In mode 2 pitch1 is affected by velocity1, and pitch2 by velocity2. In mode 3 pitch is not affected by velocity.
So I think it could "succesfully" be used in mode 3  - multi , but at present I am not able to send a program 0 MIDI message.
Thanks
Giulio
-0-0-0-0-0-0-
Hello Giulio.

Thank you for these updates.  I thought we might have solved it by now.  But there are some other possibilities.

Maybe there was a mix-up on the ICs installed at the IC10,  IC11, and IC12 spots.  They are all quad op-amps with the same pin-outs, but the TL-084 ones for IC11 and IC12 have higher input impedance so as to present a negligible load to the capacitors which hold the charge of voltage sent through IC13.  IC10 is a LM324.  Check to be sure these didn't get put in each others place.

The eight capacitors which store the charge are 0.1uF value and these are grouped there at the corner of the board near ICs 11 and 12.  You should see markings on these that is to eliminate the decimal point by listing the value in pF (decimal moves six places to the left for uF), 104, for 10 0000 pF (the same as 0.1uF).

IC3 the micro-controller operates IC13 via transistors Q4, Q5, Q6 and Q7.  Transistor Q4 is on the line to IC13, INH, which is stops the input to output connections for this part.  It is supposed to prevent the input signal from getting through to the outputs and the capacitors and op-amp stages in the times the output channels are changing.  I think we have already checked to be sure the 100k attached to its collector, R34 is a 100k (brown black yellow gold) and not the similar looking 47ohm (yellow violet black gold), and that the solders for this transistor, its base resistor, R27, IC3pin11, and IC13pin8 are good, but maybe there is something about the conditional steps we'd encountered before that still needs attention.

The special resistor #R32 should not be in place, and, the other special resistor *R28 should be a 5600 blue grey red gold.
Please check and double-check all these things and let me know what you find.  Our next step might be to try replacements at IC13 and Q4.

About the resistance tests you made.  It is probably the amount of voltage applied to the circuit for the test that resulted in the different readings you saw, but it does not tell about the trouble.
One more thought.  If you can switch off the extra messages sent by your controller, ie Active Sensing or MIDI Timing Clock (the ones we figured were causing the unexpected MIDI Test flashing), it might free up some processing time that is occurring and possibly making the voltage updates to the capacitors and op-amp stages at an expected rate.

Thanks again. Sincerely,  Scott Lee

-0-0-0-0-0-

Hello Scott!
The trouble has been shot!
The issue was Q4: I removed Q8 (which I think is useless if I don't use a trs connector) and on my multimeter its hfe was about 120 . I then removed Q4, and its hfe was about 50. I put the transistor that used to be in Q8 in Q4 , and now velocity and gate do not affect pitch anymore!
I've already managed to tune the midi2v8 and the VCO, so I'll be testing the synth in the next days,  and maybe I'll be back with some more bugs...
I still can't understand what the "rings" of the connectors are intended for...
Thanks
Giulio

-0-0-0-0-0-0-0-

Hello Giulio.

Good news!  But these transistors should not affect the voltage output.  They do share a connection with the output voltage through a base resistor, so if the transistor has trouble or a short-circuit, this might be a reason for the voltage output not being normal.
These transistors are Open Collector circuits.  A voltage load connected to them is switched to ground or zero-volts when the voltage corresponding to a gate or pulse trigger is active, or, they conduct an amount of current to ground from a voltage source for a varying amount of voltage at the output.  When interfacing with a Moog, for instance, the S-trigger connector can be wired to the Sleeve (ground) and the Ring (transistor o.c.) of the TRS (tip ring sleeve)  connector on MIDI2CV8 so a gate trigger or pulse trigger ' operates this Switch-trigger input (pulls the voltage present on the input to zero).  Or, as shown in the example in the MIDI2CV8, an LED connected through a resistor to a voltage source can have the current through it controlled by the o.c. transistor.

When only using the voltage outputs, you should only wire to the Tip and Sleeve circuits of the MIDI2CV8 trs connectors.  It is OK to use a mono TS tip sleeve plug in the MIDI2CV8 trs sockets.  This simply connects the ring with the sleeve and the o.c. transistor is joined with ground--it does nothing.

Thank you. Sincerely,  Scott Lee

-0-0-0-0-

giulio wrote:

Hi Scott!
Just a question about tuning: how should I proceed, if I have neither the MIDI2CV8 nor the VCO tuned? that is, in the manual is written the procedure to regulate the DAC , having a tuned VCO, or to tune the VCO having a precise voltage source, but I have not that!
I've tried to regulate the DAC reading 5V (when pressing last C on keyboard)out of output1 with my voltmeter, but I don't think it is precise enough: using it to read the "stabilized voltage" out of the 12V voltage regulators, it reads 11.88V...

Thanks
Giulio

-0-0-0-0-0-0-

Hello Giulio.

You may set the MIDI2CV8 to 1v/octave by connecting a dvm to the pitch cv output and adjusting the trim so the change for octave spaced key presses is 1v.  Then, connect the VCO and set its scale trims.

Here is some text I have for the procedure:

First set the MIDI2CV8 to be outputting 0V for the low key pressed on the controller, by powering the 9700K while holding the low key on the controller so that releasing the key will cause the first message input to the MIDI2CV8 to be a note-off (it interprets this as a signal to output zero volts for this note number).

A dvm can now be used to check that octave spaced key presses are outputting a 1v pitch cv change (down to a hundredth of a volt or so) as set by the trim on the MIDI2CV8. The most accurate setting will be obtained if you press octaves that start a few notes above the lowest key, ie if your low key is a C, press D or higher for the lowest octave reading.

The usual tuned setting for the VCO scale trims is about a 1:00 setting for the pointer of there disk which covers a cw range from about 7:00 (ccw) to 5:00 (cw). Start with the trim at 1 o'clock.

After having been powered for a minute or two patch the low-key transposed pitch CV over to the VCOA P2 input. Set the two pitch controls to they're both in-tune at unison at about a mid-rotation setting for the low key pressed on the controller. Play an octave higher and the adjust scaleA for an octave relationship with B. Go back to the low note and adjust the A panel pitch control for unison and then, again, press an octave higher and adjust the scaleA for an octave relationship with B. After going back and forth like this another time or two, you should find the scale is about as close as it can get.

Then, do this all again pressing some note a few keys higher than the lowest key (with VCOB adjusted to match at this new low pitch) and confirm/tweak the scale to match for higher pressed octave relationships with B (using the A panel pitch to realign A with B after adjusting the scale). When the scaleA adjustment is complete, move the pitch control over to VCOB P2 and make the panel Pitch and scaleB trim adjustments to get unison and octaves with the VCOA pitch (adjusted to match B for the low key pressed and pitch CV sent to it).

After the scaleB is set, move the Pitch CV over to the VCOA P1 input and listen to the two vcos tracking across the keyboard range (again with the low-key transpose asserted) after aligning their pitch while a higher note is pressed, say the third or fourth octave. Confirm they track for a higher octave pressed and hold as you press keys going down towards the low key. You should be able to turn up the sound of the controller or some other MIDI Sound Module and hear that the tones made from the VCOs can be aligned using the panel pitch controls and follow as you play different ranges on the keyboard (make the alignment between the two when a key in the middle C to A 440 range is pressed--its easier to hear the beating slow as they are tuned).


Thank you. Sincerely,  Scott Lee

ps  It helps to keep the VCO Modulation control at minimum so there isn't any possible pw modulation which can make the VCOB pitch change slightly (often, a change of R59 from 91k to 27k prevents this interaction).

Thanks!  --Scott

-0-0-0-0-0-0-0-0-0-0-0-0-

giulio wrote:

Hi Scott, I've been playing around with the beast for a while and now I'm back with some old and some new issues:
first - You'll remember that I had a problem with the VCO: sometimes it did not power on when I turned on the power on the midi2cv. Then I had to turn the power off and on a few times to have the VCO power up, too. I found out that when it did not power up the 7912 would provide the correct voltages, but the 7812 would not. In fact I can get something out of it's outputs: frome the saw and tri out of oscb I get a sound similiar to a saw , whose pitch cannot be affected by either the pitch knob or by some voltage put into the P1, but the modulator led doesn't light at all and nothing comes out of osc1.
With the power turned on, I heated with the soldering iron the pins of the 7812, and it turned out that heating the ground pin would get the VCO to power on. I though this would stop the problems, and in fact I immediatley tried to turn the power off and on some 20 times, and it always worked. But after a few hours the issue came out again, and it could always be solved with the heating of the ground pin of the 7812. I thought it could be a problem of the copper circuit on the board, so I shorted the ground pin to the closest ground pin available with a piece of wire. It did not solve the problem, so I replaced the 7812, but still I've got the same issue: from time to time it will not power up, but heating the ground pin will get the osc working.
second - I just noticed that the modulators when gated would perform the attack and release phases in the time set by the "attack" and "release" knobs, even if the key is released before the end of the attack phase. In practice, if the key is held down for less than the attack time, the gate input seem seems behave like a trigger, but it works as expected if the key is held down longer. This happens to all of the three modulators(the 2 AR's and the ADSR), so I maybe this is the way it is expected to work, but it seems strange to me.
third - In the VCA module with no input connected I can hear part of the noise normalized to L input out of the R input, unless the panpot is set at about 2 o'clock or more. Furthermore, the A output seems very noisy to me: with a jack plugged into L input to disable the noise, when the adsr is gated I can hear quite a lot of noise coming out of the A output
fourth - I read the VCA manual and I cannot understand the difference between VCA(using the expo converter normalled between the adsr and the four quadrant multiplier) and balanced modulator(bypassing that expo converter)...

Thanks
Giulio
-0-0-0-0-0-0-0-0-
Hello Giulio.

I can't imagine that heating the joint for the 7812 would affect the circuit as you describe, but it does make sense that heating of the part could change the output voltage slightly enough that the trouble stops.  The ground to the 7812 is on the circuit labeled G, while the vco circuitry is all referencing a different ground circuit SG.  Maybe one or the other is not a reliable connection and the soldering operation is pulling a floating circuit back to zero potential???

The voltage from the MIDI2CV8 over the four-circuit dc connectors is usually about 16v positive and negative dc.  But if it is dipping down a bit lower than that, this could be reflected in the module operation, particularly the modulator section where there are transistors biased to the positive supply.  Please check to see that the dc on the MIDI2CV8 four-circuit tap is at +16 and -16 (up to 18 is OK).
Out on the end of the four-circuit power connector cables the terminals will release so the wire can be pulled out of the cover by putting a stylus, pick, or small knife-point into the rectangular opening to press down on the metal catch which springs into the void when the terminal is inserted.  Pull out the wires for the G and SG so you can see the solder flowed to the wire and the terminal, but also that the terminal has a arched shape and is not flattened and possibly not making a reliable contact with the header pins when the connectors mate.  If the metal catch isn't springing into the void, it could result in the terminal not meeting the pin when the connectors meet.  If any of the terminals look to have troubles, let me know and I will mail you some more (or a new power connector cable assembly).

Is your VCO connected as the first link in the chain from the MIDI2CV8 on through to the other modules on the four-circuit supply (ie MIDI2CV8 to VCO to VCF to VCA)?  If so, it makes it less likely for the G or SG to be open as the other modules would be having troubles too.  But if the order is with the VCO last, this is possible.  It is also possible the ground is being established via patching and the trouble isn't manifesting on them as it is on the VCO.

If there is not enough voltage from the MIDI2CV8 four-circuit dc supply, then just connecting the VCO to the MIDI2CV8 and leaving the VCF and VCA disconnected might allow the VCO to get more voltage.  If this corrects the trouble, there is a change that can be tried that might make the difference.  We have recently been modifying the MIDI2CV8 boards by cutting the circuit trace which carries the negative supply voltage (V-) to pin 7 of IC13 (on the bottom of the board near pin 7) and using a piece of wire to join pins 7 and 8 (ground/circuit-common).  This change prevents too much voltage from getting to this 4051 IC when the MIDI2CV8 does not have a load on it's four-circuit dc tap.  Maybe there is some power being wasted in your system that can be saved by making this change and it is an amount that would fix the trouble.
For the VCA, check the operation of the L input connector:  When a plug is inserted, the noise to the shunt of connector J1(terminal and contact X) should be moved away from contact with the tip terminal/contact of J1 (T), the input to the VCA A.  The noise should stop when a plug is inserted to move the tip terminal away from the shunt contact.  If it does stop, but can still be detected in the audio output, it might be coupling through tightly grouped wiring or maybe just a 'weaker' noise transistor is needed for the Q5 spot.  Wire P carries the noise source up to the L input connector and getting this wire away from others might help.
The operation you have described for the modulator is correct.  If a Gate deactivates before the attack is complete, it will be like a pulse-trigger, the middle events will be skipped and there will only be attack and release.  Once fired, the attack completes and the release follows.

In the VCA section, the ADSR is connected such that is runs through an exponential converter circuit so the raising and lowering of the volume sounds correct operating as a VCA, but if a plug is inserted, the ADSR control is shunted and the control is linear so an audio signal for example can modulate another audio signal without the exponential function.

Thank you. Sincerely,  Scott Lee
-0-0-0-0-
giulio wrote:

As for the 7812, I haven't mentioned what follows:
when the VCO does not power on, the output voltage of the 7812 is about 0, while the input is about 18V . Heating that joint raises the output voltage from 0 to 11.82V. I've connected the ground of the 7812 to the ground pin of the 7912 with a piece of wire.

The VCO is connected as the first link in the chain on through to the other modules, and the issue appears even if it is the only module connected to the midi2cv. I've already done the modification on the MIDI2CV.

The noise source in the VCA behaves as it should: when a jack is plugged into the L input, I can not hear any noise out of either the L or R ouptut. The issue is that whatever signal is fed into the L input (the normalled noise, or an external signal), I can hear some of this signal coming out of the R output, unless the panpot is turned to three o'clock or further clockwise. This is not a big trouble, but I'm wondering if it is supposed to behave like that.
Furthermore, is the background noise coming out of the A output whenever it is gated or triggered, normal?it is much louder than the quiet background noise of the L and R outputs.

Sincerely,
Giulio
-0-0-0-0-0-0-
Hello Giulio.

Thank you for these updates.  It is still possible an open circuit exists between the ground pin of the 7812 IC2 and the power ground G.  The G circuit is to IC2 before it goes to the 7912 at IC1.  It runs from G to C1 and C2, from there to C3 and C5 and two ways with a branch to C6 and C4 and a branch to IC2 pin 2 and then to IC1 pin 1.  Did you check the terminals in the housing of the power connector cable to be sure the G is reliable over to the 9720 module from the 9700K MIDI2CV8?

For these three terminal voltage regulators, the output voltage should not exceed the input voltage.  This shouldn't happen under normal conditions on the modules of a P9700S, but if there was a wrong capacitor (and with greater than specified value) on the output of the regulator, this could work to hold the output voltage to be more than the input voltage when the power source is switched-off.  C6, at the output of the 7812 should be 4.7uF (4u7).

The L and R VCA sections are joined together by resistors for a passive mix to the L+R connector, so, yes it is normal that some of the L signal will get to the R output connector.  Note that the L output connector is a closed-circuit type so if a plug and cable is inserted here to tap the L VCA output, then no L signal should get to the L+R output, or the R output.  The L and R VCAs can be used independently with patches to the inputs and outputs of each, or, if only inputs are connected to them, they are mixed or combined through resistances to the L+R output, and on to the VCA A input.  Putting a plug into the VCA A input shunts the L+R signal away from the VCA A input so VCA A can be used independently.

The noise you hear at VCA A output might be due to DC shift.  It can make a brushing sound as it moves and the null trims should work to prevent dc voltage from developing at the output.  The Carrier Null should work to remove any dc offset from the ADSR, but notice there are many equal value resistances to the two inputs of IC4 and IC6 sections in VCA A.  These work to keep the dc from the output too so be sure these are all the ones specified and that they have good solder connections.

Thank you. Sincerely,  Scott Lee

-0-0-0-0-

Giulio Moro wrote:


Sorry to have to resend this, but I sent it last thursday and haven't received any answer yet, maybe you've trashed it unintentionally...


Hi Scott: I've got some news about the VCO: the power connector cable to the vco seems reliable. The 7912 on the VCO does work each time I turn the power on, even when the 7812 fails. I realized that there are not one but two ways to have the 7812 working, if it fails when I power the whole thing up: the first, as previously described, was to heat the ground pin. The second was to short the ground and out pins for a while: immediately after that the output voltage goes to 11.85 and holds that value until the power is turned off.
So I tried the following: I cut the PCB to open the circuit traces going from V+ to the rest of the
circuit(the one that goes to the A spot and the one that goes from C5 to IC4), thus leaving
connected to the 7812 output pin only C6 and C5.
I connected my voltmeter to the output pin of the 7812 and then turn the power off and on for more than 40 times. It never failed.
I then reestabilished with a jumper wire the trace going to the A spot and the rest of the circuit.
Again I turned the power off and on several times and it never failed.
I then removed that jumper wire and reestabilished the connection to IC4 only.
Again I turned the power off and on several times and it never failed.
Then, as to check what seemed  by then clear, I reconnected both traces. After a couple of powerups,
the 7912 failed again.
As to be precise, when the 7912 fails, the output pin voltage is -0.72V.
Input voltage to the 7912 is about 17.8V.
Hope this helps!

Giulio
-0-0-0-0-0-0-
Hello Giulio.

I apologize for the delayed response.  I took the day off Friday and when I returned Monday morning, my email was a bit overwhelming...

Maybe an IC that is linking the supplies internally.  I have seen this occur with the TL 084 Quad Op-Amps (there are two, one at IC3 and one at IC7).  Since you'd reported VCOB having unusual waves when the trouble with the V+ occurred, it could be IC3 with its four sections in VCOB.  If replacing IC3 doesn't do it, then IC7 would be the next to try.

It can be difficult replacing ICs on these boards with the plated-through holes.  To remove the ICs without stressing the board too much, clip each leg on one row of pins from the top side with diagonal cutters.  Lift the part and rock it back and forth a few times and the other row of pins will break.  Then grasp each pin stub with needle-nose or forceps or tweezers and remove them as the joint is melted.  Then, desolder each joint with a vacuum bulb or spring loaded desolder pump after melting a bit of new soldering onto each pad for some fresh flux in the joints.
Or, let me know if you would like to send the kit here for me to do this for you.

Thank you. Sincerely,  Scott Lee

-0-0-0-0-0-0-

Giulio Moro wrote:


Hi Scott, bad news from here:

I've replaced IC3 but this did not solve anything. I then replaced IC7, but still everything is exactly as it was before.

The only good news in all of this is that the VCO still behaves as 4 months ago(that is: I did not worsen nor break anything! ;-) )

I remind you that when the 7912 fails, its output pin measures about -0.70V.
What would you suggest to do next?

I considered putting a button in front of the vco to briefly short ground and out pins of the 7912 , which seems the only way to have it working when the 7912 fails, but that would be not fair to drill a new hole in the faceplate...

Thank-you
Giulio
-0-0-0-0-0-0-0-
Hello Giulio.

It is good to hear you got the ICs changed, but not so good that there is not a correction to the trouble.  It does now occur to me that you have twice mentioned the 7912 failing and needing to short its output before the voltage appears.  I thought it was the 7812 at first and then in the next message the mention of the 7912 was a typo.
It is possible your 9720 kit has two of the same voltage regulators at the IC1 and IC2 designations?

IC1 nearest the rear edge of the board is the 7912 negative regulator and IC2 is the 7812 positive regulator.  Also, a difference between 78xx positive and 79xx negative regulators is that the middle pin and metal tab on the 78xx is ground/circuit-common while on the 79xx the middle pin and tab is the negative input voltage:

7812:  1) Input   2) Ground    3) Output

7912  1) Ground   2) Input    3) Output

I just want to make sure we aren't looking beyond the root of the trouble....

Thank you. --Scott

-0-0-0-0-0-0-

Giulio Moro wrote:

Hi Scott, probably I mistyped : on my board i've got the IC's in the right place: 7912 at IC1 and 7812 at IC2

When the 7812 fails  its output is -0,70V and shorting the output and the ground makes its output go to 11.95V , and it keeps that value.

Thanks
Giulio
-0-0-0-0-0-0-
Hello Giulio.

Thank you for your patience with this mysterious trouble.

For the breaks you made in the V+ circuits, then jumpered, remove these jumpers and use low value resistors to make the links, 10ohms or 100ohms.  Then, a dc voltage measurement can be made with the probes touching the two ends of one resistor, then the other, to see the voltage that appears on power-up and after.  This will tell us about the current flow through each leg:  Divide the voltage measured, by the resistance (10) and we'll have the current.  It should be less than the amount that flows through R1, the 22 ohm which connects the regulator to the V+ circuit.  The usual mA through R1 is about 33.

The regulators have short-circuit protection and will shut down if too much current flows at the output.  Follow the wire from A up to the panel to be sure another wire has not melted into and is shorting with this wire.  At the panel destination, R206-3, inspect to be sure there isn't any contact/melting with other wires and that the circuits are to R204-3, then R203-3 (look for wires near these pot terminals too--the soldering might have melted their insulation allowing a short).

Also, if for some reason, the output voltage is exceeding the input voltage and resulting in the condition, then you could try soldering a diode from the output back to the input (cathode or banded end towards the input pin 1).  I'm looking at the Fig.1 in the manual showing the board traces and thinking, we have checked the power cable to be sure G is making a good connection (as the ground for the power supply parts), but maybe the circuit isn't complete from G, up between C1 and C2, to C3 and C5 and out to C6 and C4, and, IC2 and IC1.

A direct connection from the input 17v or so, over to the output 12v circuit (with the diode added to keep the output voltage from exceeding the input voltage) would surely work to heat a part that might have trouble on the inside (ie the other parts on the +12v circuit), pointing to it as the culprit.

Thank you. Sincerely,  Scott Lee

-0-0-0-0-0-0-

Giulio Moro wrote:


Hi Scott thank-you for your response, it gave me something to do this evening. I ended up with a 2 and half hours of work on the 9720.

I replaced the jumpers with 10ohm resistors. The one that goes to IC3 carried 6.18mA and the one going to the A spot and the rest of the circuit carried 22.2mA . Across R1 I measured a current of about 36mA (827mV in fact).

I checked the A wire and there seem to be no unwantend short. As to be sure I removed temporarily the wire going to the panel and the issue still happened, so this is not the problem.

I tried to put the diode from output to input pin1, nothing good happened. I left the diode in place , should I remove it?. Furthermore I checked with my tester that there is continuity between the Ground spot at the power supply connector and the Ground pin of IC1 and IC2, and it is OK.

I did not really understand the last part of your mail , where you say "A direct connection from the input 17v or so, over to the output 12v circuit (with the diode added to keep the output voltage from exceeding the input voltage) would surely work to heat a part that might have trouble on the inside (ie the other parts on the +12v circuit), pointing to it as the culprit. "
Should I try to short the 17V input and the output of  the 7812 and see if something becomes hot on the board?

Now some good news:
as nothing did seem to work out right, I tried, with a "trial and error" method, to find out what could be happening. So I cut each trace going from the output of the 7812 to the rest of circuit. I found out that the problem was in the one that goes to R81 and IC7. That is: breaking the trace from pin  4  of IC8  to R81 DOES solve the problem on the 7812(though this obviously doesn't produce any output from the oscillators , while the modulator still works) : I have turned the power off and on many many times and the voltage regulators never fail.
This lead me to think that the issue is in IC7 (it's strange, because I've just replaced it...)
Anyway, I made some measurements on the pins of IC7 both when the 7812 fails and when it works
when the output of the 7812 is 11.95, IC7 behaves like this:

pin1 -1.55
pin2 -6.00
pin3 -6.00
pin4 11.95(output of the 7812)
pin5 -5.90(touching this with the voltmeter does raise the pitch of oscA)
pin6 -6.00
pin7 -6.00
pin8 0.00
pin9 0.00
pin10 0.00
pin11 -11.90
pin12 0.00
pin13 0.00
pin14 0.00
when the output of the 7812 is -0.70, IC7 behaves like this:
Pin 1 -0.42
pin 2 -5.32
pin 3 -5.49
pin 4  -0.73 (output of the 7812)
pin 5 -5.62
pin 6 -7.64
pin 7 -7.82
pin 8 -1.37
pin 9 -1.40
pin 10  -0.30
pin 11 -11.90 (V-)
pin 12 0.00
pin 13  -0.30
pin 14 -0.30

I think we're towards the end, or at least we've circumscribed the problem.
Hope to hear from you soon
Thanks
Giulio

-0-0-0-0-0-0-0-

Hello Giulio.

Your list of voltages for IC7 helps.  If you look at the schematic and the pcb image, fig 7 and fig 1, you can see that IC7 pins 1 and 2 and 5 are all joined.  So, you should have similar readings on these.  I wonder if there might be a break in the printed circuit from pin 1 to pin 2???  But also, this circuit runs out to R19 so confirm this is indeed only connecting with R19 out on this branch.  Maybe capacitor C11 has a leg leaning against R19???

Thanks.  --Scott

-0-0-0-0-

Giulio wrote:

Just one more thing: I measured the voltages on pin 4 of the TL084's when it is disconnected from the output of IC2 : in IC3 pin4 is about-10.5V while in IC7 pin4 is about-9.5V
...
thanks
Giulio

-0-0-0-0-0-0-0-0-0-

Hello Giulio.

I disconnected pin 4 on the TL084 ICs in the VCO I have here and they each have 10.5v readings, so
the 9.5v you see on IC7 is probably a sign of trouble in this part.  Maybe it has had the same thing go wrong before and after replacing it.  Did you look at the capacitor C11 to be sure it's legs aren't touching another nearby part? Double-check the J5 wiring too.  It is the output of IC7:D through a the closed circuit shunt (X) to VCOB input and should open if an external voltage is connected.

Let me know if you would like me to mail a replacement for IC7.

Thank you. Sincerely,  Scott Lee

-0-0-0-0-0-0-0-

Giulio Moro wrote:

Hi Scott
I measured again pin4 of IC7 with pin4 disconnected, it is -10.5 not -9.5 as I previously stated.
After this I swapped the TL084's (a friend of mine desoldered them for me with a desoldering machine) and still the problem exists when pin4 of IC7 is connected.
I also replaced IC8 but this did not change the situation; I would already replace  IC6 but my dealer does not have any LM339 .
I don't know what to do anymore!
Think I'll put this button to push to briefly short to ground the output of the 7812 when it fails.

Now as for the tuning of the VCO's:
I've thermally coupled the temperature compensating transistors, should I do the same between the LM13700 and those transistors?
I cannot manage to have them tracking correctly more than 3 or 4 octaves:
I used a guitar tuner(Planet Waves TU-4).
I adjusted the pitch of a high note(the 4th C on the keyboard) with the pitch knob and then used the trimmer to match the VCO and a low note(lower C).
I did this a few times until the pitches where correct. I then played the notes between the two C's and they where a little out of tune(a little sharp) except for a few notes above and below the two C's , but they where not too terrible. Notes above the high C would instead be really flat (the 5th C would be half way between B and C).
If I repeated calibration with starting with the pitch knob set an octave above 4th C , scaling would be equally acceptable for 3-4 octaves(one octave higher than in the previous case).
Is this normal?
I think I'd better put a rotary switch to select between different ranges (each with a dedicated scale trimmer), and use each position for a maximum of 3 octaves, to have the thing tuned across the whole keyboard.

Thanks
Giulio

-0-0-0-0-0-0-

Scott Lee ha scritto:
Hello Giulio.

Oh no!  We must be getting closer to finding the solution.  I will send a LM339 (also the MC3302
is a similar part), and new capacitors for the power supply filtering.  I am wondering now if there might be something related to coupling between the supplies via the capacitors and the shut-down of the regulator is due to unusual current at the instant of power-up.  I will mail another 7812 too, as maybe there is a variation available from your supplier that is not tolerant to use in the 9720VCO.  So a parts pack is on the way to you.

...

For the tuning, no, it should not be necessary to thermally couple the transistors with the 13700 IC. But it is a good idea to wait a minute or two after power-up for the circuits to warm to 'operating'
temperature. Sometimes a tuner might be overloaded and harmonics are sensed.

Here's a way to get them by ear... 

First set the MIDI2CV8 to be outputting 0V for the low key pressed on the controller, by powering the 9700K while holding the low key on the controller so that releasing the key will cause the first message input to the MIDI2CV8 to be a note-off (it interprets this as a signal to output zero volts for this note number).

A dvm can now be used to check that octave spaced key presses are outputting a 1v pitch cv change (down to a hundredth of a volt or so) as set by the trim on the MIDI2CV8. The most accurate setting will be obtained if you press octaves that start a few notes above the lowest key, ie if your low key is a C, press D or higher for the lowest octave reading.

The usual tuned setting for the VCO scale trims is about a 1:00 setting for the pointer of there disk which covers a cw range from about 7:00 (ccw) to 5:00 (cw). Start with the trim at 1 o'clock.

After having been powered for a minute or two patch the low-key transposed pitch CV over to the VCOA P2 input. Set the two pitch controls to they're both in-tune at unison at about a mid-rotation setting for the low key pressed on the controller. Play an octave higher and the adjust scaleA for an octave relationship with B. Go back to the low note and adjust the A panel pitch control for unison and then, again, press an octave higher and adjust the scaleA for an octave relationship with B. After going back and forth like this another time or two, you should find the scale is about as close as it can get.

Then, do this all again pressing some note a few keys higher than the lowest key (with VCOB adjusted to match at this new low pitch) and confirm/tweak the scale to match for higher pressed octave relationships with B (using the A panel pitch to realign A with B after adjusting the scale). When the scaleA adjustment is complete, move the pitch control over to VCOB P2 and make the panel Pitch and scaleB trim adjustments to get unison and octaves with the VCOA pitch (adjusted to match B for the low key pressed and pitch CV sent to it).

After the scaleB is set, move the Pitch CV over to the VCOA P1 input and listen to the two vcos tracking across the keyboard range (again with the low-key transpose asserted) after aligning their pitch while a higher note is pressed, say the third or fourth octave. Confirm they track for a higher octave pressed and hold as you press keys going down towards the low key. You should be able to turn up the sound of the controller or some other MIDI Sound Module and hear that the tones made from the VCOs can be aligned using the panel pitch controls and follow as you play different ranges on the keyboard (make the alignment between the two when a key in the middle C to A 440 range is pressed--its easier to hear the beating slow as they are tuned).


Thank you. Sincerely,  Scott Lee

ps  It helps to keep the VCO Modulation control at minimum so there isn't any possible pw modulation which can make the VCOB pitch change slightly (often, a change of R59 from 91k to 27k prevents this interaction).

Thanks!  --Scott

-0-0-0-0-

Giulio Moro wrote:


Hi Scott, it's been a while since you shipped the package, but I haven't received it yet... do you have a tracking number for it or something ?

thanks
Giulio
-0-0-0-0-0-0-
Hello Giulio.

I apologize for this delayed response...just back from a vacation.

No, there will not have been a tracking number, but there is a customs declaration label on the packet describing the contents as missing parts for an 'electronic hobby kit' originating in the USA with a valuation of a few dollars.  It is usually about three weeks for something like this, so maybe you have the pieces now.  If not, it should be soon.

Thank you. Sincerely,  Scott Lee
-0-0-0-0-
Giulio Moro wrote:


Thanks. Yes, they arrived last week , I forgot to tell you.
I'll try to put them on as soon as I have some leisure time.

Thanks
Giulio
-0-0-0-0-0-
Giulio Moro wrote:

I forgot the attachment...

Hello Scott, hope you're enjoying your summer.
I've been on holiday but before and after it I've been working hard on
the 9720 VCO tryin' to fix it and to get it tuned.
I received the pack with the parts, and I used them to replace the ones
in my 7920 but I didn't get to solve it's main problem: it still doesn't
power on each time I turn the power on the midi2cv8, still the 7812
fails(that is : its output goes to about -0.7V instead of +12.00V, as
I've described many times so far).
Just to sum up, so far I've replaced all of the IC's on the 9720,
(including the 7912), except for the lm13700's. As for the 13700's I've
done what follows:I've had them carefully removed by a friend of mine,
and then I turned the power on without them on board, but still the
power supply failed, so I can say the cause of the trouble is not the 13700.
So I put a button in front of the synth, to push in case the 7812 fails.
This button briefly shorts the output of the 7812 to ground and this
causes the 7812 to power on properly.
I told you I was unsatisfied with the tuning of the osc's. It seemed
like notes in the middle of the tuning interval(wheter it was 2,3 or 4
octaves the impression was the same) were a little flat, and notes
outside this interval were flat.
Here's how I made the test :
Powered up the unit and let it warm for about 10minutes.
I set the midi2cv8 trimmer so that now it is precise enough(after my
voltmeter), almost in the first 5 octaves, and then I tuned and tested
both oscillators a few times.
I tuned each VCO multiple times, each time repeating the following
procedure:
I tweaked the scale trim and the frequency pot as described in the
manual, between notes A1 and A5.
When at last these two notes were in tune I played notes A,C and E
between A1 and A5, and then notes outside this range(that is: lower than
A1 and higher than A5) and measured with Chromatia Tuner(a software
tuner, http://www.fmjsoft.com/chromatia.html) the error of each note.
My impressions were confirmed by this test: notes between the tuning
interval are sharp, and notes outside the tuning interval are flat.
What is strange to me is that both VCO's seem to behave the same way.
In the attached pdf you'll find the average values I found after testing
OSCA 4 times and OSCB 3 times. You'll also find a graphical
representation of the behaviour of each OSC, and you'll see that they
seem to behave almost the same(wrong) way.

As a further consideration I can say that when tuning within two
octaves(C2  and C4, for instance) the thing is playable inside the
tuning range, but for greater tuning intervals it is almost unplayable.

Where do we go from here?

Sorry to keep bugging you...
Thanks
Giulio

-0-0-0-0-0-0-

Hello Giulio.

Ah...yes...summertime...

It's good hearing you've had a holiday and are back at it.  Thanks for this update and the chart you made.  Since the error is practically the same for the two VCO sections, it seems that the MIDI2CV8 might not be sending a linear control, or the tuner isn't getting the fundamental, but a harmonic.  And that, V+ trouble...crazy!?!   I wonder if you might need more capacitance on the MIDI2CV8, where the rectified dc is filtered before being sent out as the V+ and V- dc supplies to the modules in the P9700S?

The greatest value we have here is a 2200uF at the 25v rating, but I can make an order for larger ones and send them to you, or, today I could send a couple more 2200uF for you to use connected as paralleled pairs in place of the ones on your board.  Please let me know if you have a preference for one way or the other.

Thank you. Sincerely,  Scott Lee

-0-0-0-0-0-

giulio wrote:

As I haven't heard anything from you in the last week maybe you did not receive my email, so I'm sending it once more:


Hello Scott,
 I had a couple of 4700uF, 35V at home, so I replaced the
 existing 2200uF, but still the VCO won't power on,
 furthermore we must remember that the other modules do power
 on at anytime. I also tried in the past days to completely
 rebuild the power supply of the 9720 on another board with
 using new components, and connected it to the 9720 PCboard,
 but still the problem was not solved.
 Sigh!
 
 I checked the output of the midi2cv8 and this is what I
 read(using a good voltmeter):
 A1 0.75
 C2 1.00
 E2 1.33
 A2 1.75
 C3 2.00
 E3 2.33
 A3 1.75
 C4 3.00
 E4 3.34
 A4 3.75
 C5 4.01
 E5 4.34
 A5 4.75
these values are pretty much the same both with and without the output1 of the midi2cv8 connected to the input of the VCO. so these seem good, too! gulp!
As a further consideration I must say that I DO hear that those notes are out of tune, so it's not a tuner's issue!

So where do we go from here?
Thanks
Giulio

-0-0-0-0-0-0-0-

Scott Lee ha scritto:

Hello Giulio.

Thank you for this reminder.  I apologize for this delayed response.
Is it possible that you can send photos of the module showing the wires to the rear of the panel, the top side of the printed circuit board and the bottom side of the printed circuit board?  Maybe I can spot something we haven't considered.

The MIDI2CV8 voltages look good.

Did you get and use the replacement 9791 power connector cable I sent?

I am willing to have a look at it here too.  Attached is the service authorization form.  It has information about service, instructions for return, and areas to include your details.  Please print and complete the form and include it with the return.  Mark it and the package with the authorization number/code 919VCO.

Thank you. Sincerely,  Scott Lee

-0-0-0-0-

giulio wrote:

Hello Scott
I'll take the photos as soon as I can.
I did not receive the 9791 power connector cable you mentioned, and I have no idea of what it is...

Thanks
Giulio

-0-0-0-0-0-0-
Thank you Giulio.  I will mail another 9791 Power Connector Cable.  We now have them pre-assembled with the four wires attached to terminals and inserted in a four-circuit housing to mate with the four-circuit header.  --Scott
-0-0-0-0-0-
giulio wrote:
Hello Scott, just one more thing:
I just want to remind you that when the 7812 fails, heating with the soldering iron its output pin would suddenly make its output raise up to 12V.

This is strange, too...

Thanks
Giulio

-0-0-0-0-0-0-0-0-

Hello Giulio.

Yes, I remember that.  This makes me think the G circuit of the four-circuit dc power connection with the positive and negative dc supplies and the power and signal grounds, might be open, or partially open.  Maybe a terminal is squashed in the housing and when mated with the header isn't a reliable connection.  There could be a cold joint on the header pin associated with G.  Maybe in replacing a 7812, the printed-circuit-feedthrough, linking the top and bottom sides of the board, detached when the part was removed (soldering the replacement on the top and bottom sides would fix this, but don't overlook the possibility of this having occurred on another part replaced and on the G circuit).

The power-connector-cable I'm sending is pre-assembled, so it will just be a matter of removing the one in place on the 9720, clearing the holes of solder, and installing the new one. 

Thank you. Sincerely,  Scott Lee

Attachment:
P9700Snotes.pdf [51.48 KiB]
Downloaded 1318 times

_________________
Scott Lee
scott@paia.com


Thu Sep 03, 2009 1:04 pm
Profile
Display posts from previous:  Sort by  
Forum locked This topic is locked, you cannot edit posts or make further replies.  [ 1 post ] 


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron
Powered by phpBB © phpBB Group.
Theme designed by Vjacheslav Trushkin , adapted for PAIA by Illumimedia Design Studios