Midi 2CV8 Linearity Issue
Page 1 of 2

Author:  lucidsounds [ Mon Apr 01, 2013 3:31 pm ]
Post subject:  Midi 2CV8 Linearity Issue

I have a 9700 series which has a Midi2CV8. Whilst attempting to tune the oscillators I noticed a strange problem where the note voltage will suddenly jump between two particular notes - not excessively - but enough to raise the pitch of the successive notes by 20~30 cents.

First of all, I used a Kenton Midi CV and Fluke DVM to check my figures - each note increments by 1/12 or approx. 0.0833333V. So using this Kenton, I measured...

C1: 1.000
Eb1: 1.250
F#1: 1.500
A1: 1.751
C2: 2.001

...and so on, all the way up, which makes sense. However, when using the Midi2CV8 I see this:

C1: 1.001
Eb1: 1.246
F#1: 1.502 (notice how that is slightly non-linear)
A1: 1.748
C2: 2.002
Eb2: 2.247
F#: 2.500
A2: 2.769 (big jump there)
C3: 3.022
Eb3: 3.266
F#: 3.521
A3: 3.766
C4: 4.022
Eb4: 4.266
F#4: 4.519
A4: 4.766

and so on...the 'jump' occurs between G2 and Ab2, where if G2 is in tune, Ab2 will be 20~30 cents or so sharp, as will the remaining notes above that point.

So what could be causing that? The self-test routine appears to run fine. The DAC tune trimmer is between midway and 1 o'clock. The outputs appear to be free of noise, as does the 5V regulator.

Is it possible the DAC is somehow non-linear?

Many thanks in advance,


Author:  PAiA-Scott [ Tue Apr 02, 2013 8:57 am ]
Post subject:  scale adjust, 9700k MIDI2CV8

Check out the procedure listed in this P9700S overview:


It might make the difference going for 1v changes for octave spaced notes other than ones corresponding to C.

Was this for a recent order of a P9700S or one that's been around for a while? We updated the power supply connections to the 4051 demultiplexer IC so it's more tolerant to the variations in supply voltages which can occur as the load changes (all modules in the system connected or just some, or just the MIDI2CV8). At first it was just a cut printed-circuit-trace between pin 8 and the negative supply and a jump added between this pin and pin 7 (0v), then we changed the boards to be like this. When the voltage to the part gets to be too high, cross-talk between channels will begin to occur.


Author:  lucidsounds [ Tue Apr 02, 2013 1:39 pm ]
Post subject:  Re: Midi 2CV8 Linearity Issue

Hi Scott,

I think this unit is about 5 years old. The 4051 Pin 7 is connected to the zero volt rail on the PCB (no jumper).

Even using the revised procedure I still get a jump in the voltage between G2 and Ab2. I'm pretty sure it's not a stuck or missing bit, because all the notes increment as you go up the keyboard - it's just that the interval between some of them seems wrong.

I notice the DAC is a MC1408L6 which has an accuracy of +/- 0.78%. Would it be a good idea to swap this for a DAC08HP which has an accuracy of +/- 0.1%?

Many thanks,


Author:  PAiA-Scott [ Tue Apr 02, 2013 3:03 pm ]
Post subject:  9700 MIDI2CV8, IC13

Maybe pin 7 of IC 13 is floating??? The boards were changed to have a printed-circuit-trace joining pins 7 and 8 of IC3 about April of 2011 or a couple of years ago.

If you don't see a jumper joining pin 7 with pin 8, adding it might make the difference.

Starting about five years ago, we were cutting the printed circuit trace carrying the negative supply to pin 7 (right near the pin on the bottom side of the board), and adding a piece of bare 30 guage wire through the holes and linking pins 7 and 8.

Maybe the wire fell out or was cleared from the board before the kit was built?

Maybe the kit is even older than five years and pin 7 is still joined by the printed circuit to the negative DC supply, and, the power source to the kit is unipolar DC instead of the usual AC, or, dual-polarity DC.

Usually, MIDI2CV8 is wired to a wall-mounted 12VAC transformer; however, it can be powered by dual-polarity DC supplies.

Powering the MIDI2CV8

If the trouble isn't power (or power connections to IC13), look over the troubleshooting tips/suggestions below in case we're overlooking something we haven't considered yet.


If this is all good, then it might be worth trying the "tighter" part as a substitute for the IC5 D/A Converter.

Author:  Kyhotay [ Tue Apr 02, 2013 10:28 pm ]
Post subject:  Re: Midi 2CV8 Linearity Issue

Scott, is that a new picture? You lost the nerd look and went for cool? Nice...

Author:  PAiA-Scott [ Wed Apr 03, 2013 10:07 am ]
Post subject:  photo

Aaahh... Thanks Brian. It's a part of the five year plan. Now, back to the nerdy talk.

Author:  RoyGwinn [ Wed Apr 03, 2013 12:28 pm ]
Post subject:  Re: Midi 2CV8 Linearity Issue

G2 is 31 semitones up from the bottom, and G# is 32. That's a major rollover point for the DAC (from 00111110 to 01000000), which means it's likely that DAC nonlinearity is the issue. DAC0802LCN is the best part that fits.


Author:  PAiA-Scott [ Wed Apr 03, 2013 12:49 pm ]
Post subject:  bit change

Thanks Roy!

I wondered about that and counted-up to 22 from the low C on the keyboard. The Low-Key transpose feature must be asserted since C1 is reported as being 1V. So the count would be considering C0 as 1.

Author:  lucidsounds [ Wed Apr 03, 2013 1:40 pm ]
Post subject:  Re: Midi 2CV8 Linearity Issue

Hello again,

OK, first of all I made the mistake of confusing Pins 7 and 8 on the 4051. Pin 8 was grounded, and Pin 7 went to the negative rail. I have now modified this, as you suggested, by breaking the track to Pin 7 and linking Pin 7 to Pin 8, so both pins are now at Ground potential. I checked the linearity again, and the problem is still exactly the same.

By the way the date on the PCB is 1996 so it's a little older than I first thought.

A local supplier can get a Texas Instruments DAC0800LCN to me tomorrow cheaply, so I plan to try that in place of the MC1408L6 - the non-linearity is rated at +/- 0.19% which is not quite as good as the DAC0802 but considerably better than the 1408, so it will prove the issue.

Many thanks for the continued help!


Author:  Kyhotay [ Wed Apr 03, 2013 2:34 pm ]
Post subject:  Re: Midi 2CV8 Linearity Issue

Scott, keep the nerd talk flowing along with the cool look!

I've often wondered about the linearity of the MIDI>CV myself. I attributed it more to not getting the VCOs in tune. But last weekend I was driving 2 9720 VCOs A&B for four voices tuned to unison with the CV out from my Moog Voyager XL along with the 3 Moog VCOs in unison. Except for some tweaking on one of the PAiA VCOs all 7 (ok, really 6) Oscillators track from the Moog. When I try it from the MIDI-CV I get some off pitched notes that I need to pay better attention to but range is more like 3 octaves compared to running 5-6 octaves off the Moog CV. It's the high end that is off.

I'm going to try this once more with the Blacet MIDI>CV and see what happens. I also might try the ARP 2600 keyboard that outputs 1V/Oct.

I read that diatribe a few months back about the 8 bit resolution not being accurate enough. Maybe so, but there's not a lot out there beyond 12 bits that I have noticed. Still, is a 12 bit DAC fit/form& function identical to the 8 bit or is it a total redesign?


Page 1 of 2 All times are UTC - 6 hours [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
Not able to open ./cache/data_global.php