Page 1 of 1

FATMAN notes dither

Posted: Mon Oct 29, 2012 11:36 am
by Bachmanm
On certian notes, at random times, my Fatman'd notes "dither". By this I mean the pitch trilss between the actula note and some other descrete step. Sometimes fast (5 hz or so),, sometimes slow (1hz).

Only certian keys do this, like B4.

It is not pitchbend (none on my keyboard) and it is not software (to random for that)

I think there is a DAC bit wavering? bad solder joint?


BTW, I am in the middle of a big FATMAN upgrade and coming up with some brand new circuits.. so stay posted.. that is if I can fix this issue..

Re: FATMAN notes dither

Posted: Tue Oct 30, 2012 11:20 am
by Bachmanm
If I just touch IC4, the data latch for the DAC, with my finger, I can get the dithering to start or stop. It seems not to be pressure (indicating a bad solder joint) but seems to be the noise casued by my body (we are all coursing wi th 60Hz hum and such from all the RF around us).

If i look at the Data coming out of the latch and going to the DAC, I see the dithering on bit 3 clearly.

I will try a shield... but it makes me wonder.. don't other people have this problem?

Using a scope on the address lines from the Micro to the latch, they are noisy and not purely ON/Off.. meaning there are some intermediate levels in there.

I think this is an artifact of a single layer PCB with no ground plane

Re: FATMAN notes dither

Posted: Tue Nov 06, 2012 3:42 pm
by PAiA-Scott
An input to IC4 might be floating due to a jumper wire not reaching all the way through into the joint at each end, or, a joint that didn't flow to the wire and pad, or, a broken solder-pad/circuit-trace interface and touching the IC causes enough movement to cause the change.

Check closely in the area where there is a concentration of jumper wires between ICs 2 and 4. There's an image of this area in the trouble-shooting tips/suggestions pdf here:

Look at the connections for and related to IC8 sections A and B. They link pitch and velocity demultiplexing control from the microcontroller to electronic switch sections IC11A and IC11B. Pins 4 and 6 of IC8 should have a steady 2.5v so 0-5 levels from the uC are shifted to 0-8 levels for IC11 (voltages on resistors R29 and R30 are 0v or 8v), pins 5 and 13.

Viewed on a scope, the demulitiplexing control looks like a stream of pulses. On the data lines to and from the DAC Latch IC4, they aren't so regular since they change according to data/address/Pitch/Velocity. It can help to compare the "look" of the scope image for lines that originate at the uC and should be getting to the latch inputs. A noticeable difference is probably pointing to an open circuit. This can occur on high order address bits to the IC3 EPROM Firmware too, where a floating input is sympathetic to a neighbor and things are usually OK but if that neighbor changes, not.

Re: FATMAN notes dither

Posted: Tue Nov 06, 2012 4:17 pm
by Bachmanm

Thank you!!.

I should have closed out this post.. but for the record, you are right.. After I made my last post, I couldn't sleep until it was fixed. I looked at the bit lines to the DAC and data lines to the latch. One of the DAC bit lines was wavering.. it should have had a steady waveforem (depending on pitch and velocity, as you said).. so I looked at the data line associated with that bit and it was noisey and not pulling up all the way. So I thought I would add a pull-up resistor just to see if it helped.. when I did so, I noticed the REAL problem, the pad that goes to a jumper wire had broken free of the board.. barely.

I imagine the original builder (not me) soldered in the jumper wire and then thought to tidey them up and pressed down on it to straight the jumper out.. and broke the pad..

May I suggest Paia lay out these board with much wider traces. No reason that they could not be 2 or 3 times wider in most places. I am sure this is a fairly common issue around the jumper wires. Better yet, go to a 2 layer board. It is not that much more expensive and would save so many problems in assembly. If Paia wants to send me the electronic schematic and layout, I would do this for free.

On to more mods. I am working on an improved velocity mod for the VCA. I'll let you know how it works when it is all done.. the basic idea is to use the velocity to vary the ADSR trip point..


Re: FATMAN notes dither

Posted: Mon Nov 19, 2012 1:28 pm
by PAiA-Scott
Good news! and thanks for your kind offer.

FatMan was one of the last boards done single-sided with jumper wires.

We have boards to last for a while though, thanks.

The jumper wires do provide a practice stage before getting to the electronic components, but yes, the pads and traces associated with them are a bit too small for beginners -- especially with the lead-free solder and the extra heat it needs.

For kits returned for service, I'm seeing that it might be a good idea for us to stress the importance of seeking solder such as Kester "44" in a 60/40 alloy and about 0.031" diameter.

That's a neat idea with the velocity cv setting the VCA ADSR peak level, but the sustain phase of the envelope would still be according to the panel control setting.

It might be more effective just connecting an opto-isolator with a resistive element inside connected in a voltage divider with the Velocity CV varying the LED and the ADSR amount into the VCA.

Re: FATMAN notes dither

Posted: Mon Nov 19, 2012 2:09 pm
by Bachmanm

Glad to see the end of jumpers! There was a day when a 2 layer was a significant increase in cost (though plateing etc) back in 1978, I was etching my own boards, and the single layer boards were much easier to make.

On the Vleoity mod idea. I tested several different varients before deciding on this Attack peak/Vleocity modification idea.

The velocity sets the reference level for the ADSR peak comparitor (IC8:D). The existing Velocity pot is modified to produce a voltage that is diode or'd with the velocity voltage and thus provides a minimum peak setting. If you turn up the Velocity pot to max,the ADSR acts just as it does normally. You need a minium setting becasue Velocity voltage goes to 0 at note off. If I had an EEPROM burner, I would change that code.

Since velocity is limiting the Attack peak , if the peak is below the sustain level, the resulting volume is controlled by the velocity alone and not the sustain level as diode D6 is blocking this voltage. When the velocity voltage is above the sustian setting, then you do get some peaking (decay and sustian). This actually works very well and sounds nices. Very soft notes only have an attack and release. Loud notes have all 4 stages.

The tick is that the ADSR cap needs to be reset to 0 quickly at the start of a new note (gate on) or the relaese stage can block the next attack stage.