sprocket

AOUT_NG VCO Tracking

11 posts in this topic

Building A SEQV4 with a bunch of Midi outs (Via IIC>Midi) and CV outs (Via AOUT_NG), but a concern has come up via this description regarding the AOUT_NG on the uCApps website...

The featured precision is perfect for controlling analog filters, VCAs etc. Despite the slightly inferior specs of the DAC, some real world tests confirmed that the precision is even sufficient to drive VCOs. So as long as you don't require extraordinary precision and saving a few bucks isn't your central aim, the NG will be the right choice.

I may being nitpicky, but what does this exactly mean? How will the lack of precision manifest itself? Pitch inaccuracy and zippering? To what degree? Has anyone noticed any problems using it?

Thanks for the help!

Ian

Share this post


Link to post
Share on other sites

Building A SEQV4 with a bunch of Midi outs (Via IIC>Midi) and CV outs (Via AOUT_NG), but a concern has come up via this description regarding the AOUT_NG on the uCApps website...

The featured precision is perfect for controlling analog filters, VCAs etc. Despite the slightly inferior specs of the DAC, some real world tests confirmed that the precision is even sufficient to drive VCOs. So as long as you don't require extraordinary precision and saving a few bucks isn't your central aim, the NG will be the right choice.

I may being nitpicky, but what does this exactly mean? How will the lack of precision manifest itself? Pitch inaccuracy and zippering? To what degree? Has anyone noticed any problems using it?

well it more or less means what it says :tongue:

From the datasheet specs, the NG DAC is somewhat inferior, like e.g. a typical INL (integral nonlinearity) of +/- 2 LSB instead of +/- 0.5 LSB etc. As 12 bits of resolution are distributed over 10.67 Volts, this means we use a resolution of 384 values per octave, or 32 values per semitone. so one LSB (minimum value change) is a deviation of about 3 cents. So when you test different values/pitches, you can expect the output of the AOUT_NG to be off from the ideal value not more than +/- 6 cents while the old AOUT will be off by not more than +/- 1.5 cents.

But these numbers are the worst case deviation of a typical example IC over the full output range, so as the full output is a range of more than 10 octaves, you won't ever need this full range in musical use of a VCO, you'll probably use something like the middle 5 octaves. The diagram in the NG DAC's datasheet shows that in the middle range you can expect an imprecision of about +/- 1 LSB or +/- 3 cents, so if you calibrate the outputs to be precise in the usual range, these +/- 3 cents won't really be noticeable by the average listener.

Next question is: 3 cents is about 2.6 mV, so a) do you really have the measuring equipment to even reliably calibrate an output to that precision or even better? A 20$ multimeter isn't up to that task. and b) how linear and precise is your VCO?

So from a theoretical point of view: The NG will give you somewhat higher pitch inaccuracy. Will anyone notice? If you've got well trained ears, maybe, maybe not. As to zippering/glitching, I don't really know. The NG does glitch a bit when values change, so do most other DACs, too. Does that matter when driving VCOs? Probably not. I'd say, if someone is a perfectionist, he'll probably be unhappy by the pure thought of knowing that he's not using the very best available technology. On the other hand, an AOUT_NG is less than half the price of an AOUT and all parts are easily available, and these were the main reasons for introducing it.

I don't have access to a VCO (or even an old AOUT for comparison) so I can't tell you first hand experience in this use case. I can't remember someone else posting experiences with NG and VCOs, except TK and Polykobol who tried the NG on pk's RSF PolyKobol, and they were quite happy with the performance.

As you said you're already in the process of building NGs, and obviously want to use it for driving VCOs, why not just try it out and post your experiences after testing? Actually I'd be very much interested in all first hand experience opinions on that topic :)

Seppoman

Share this post


Link to post
Share on other sites

If zipper noise should be a problem, perhaps an idea would be to add some vactrols (optocouplers) after D/A ?

Share this post


Link to post
Share on other sites

If zipper noise should be a problem, perhaps an idea would be to add some vactrols (optocouplers) after D/A ?

Don't you introduce some nonlinearity by using vactrols? I guess with a vactrol, all discussion about a few millivolts of DAC INL would be unneccessary.

What I did on my SSM2044 module, that VCF is quite touchy when it comes to steppy CV (regardless of glitch noise issues, the output of a 12 bit DAC has sharp edges - the DAC is fast, the update rate on e.g. MBSID V2 is only every 2 ms, so there are hard jumps in the CV), I added a simple low pass filter on the CV input to smoothe the voltage a bit.

S

Share this post


Link to post
Share on other sites

probably true, just saw some schematic the other day (I think by Buchla or Wiard) that used a bargraph LED driver and some vactrols as an optical mixer, the vactrols were used to deal with the zipper noise.

Share this post


Link to post
Share on other sites

Don't you introduce some nonlinearity by using vactrols? I guess with a vactrol, all discussion about a few millivolts of DAC INL would be unneccessary.

What I did on my SSM2044 module, that VCF is quite touchy when it comes to steppy CV (regardless of glitch noise issues, the output of a 12 bit DAC has sharp edges - the DAC is fast, the update rate on e.g. MBSID V2 is only every 2 ms, so there are hard jumps in the CV), I added a simple low pass filter on the CV input to smoothe the voltage a bit.

S

these linearity errors are code dependent, hence, if you notice it (?) try giving it an offset in digital realm and compensate for it in analog realm.

Share this post


Link to post
Share on other sites

Building A SEQV4 with a bunch of Midi outs (Via IIC>Midi) and CV outs (Via AOUT_NG)

FYI I'm doing the same w/ my MB-SEQ. I'm (slowly) working on a case design for this as well based on an off-the-shelf Hammond case (http://www.hammondmfg.com/pdf/1456RH3.pdf). I'll let you know if I come across anything interesting in my build. I also plan to have an internal power supply for both the MB-SEQ and the AOUT_NG based on the design used in the MB-808.

Share this post


Link to post
Share on other sites

FYI I'm doing the same w/ my MB-SEQ. I'm (slowly) working on a case design for this as well based on an off-the-shelf Hammond case (http://www.hammondmfg.com/pdf/1456RH3.pdf). I'll let you know if I come across anything interesting in my build. I also plan to have an internal power supply for both the MB-SEQ and the AOUT_NG based on the design used in the MB-808.

very cool. I'm doing something very similar. Internal power supply with an iec connector, but I'm going to (hopefully, if i get around to it), design a case in turbocad and send it to a shop to be made (with all of the cutouts and standoffs). I need to see if I can grab the files in dxf or something alike so I can work around the correct board dimensions.

Share this post


Link to post
Share on other sites

Wow, Seppoman explanation was really educating.

I can only talk from a user's point of view. I have been using the AOUT_NG to control analog gear directly from the MB-SEQ V4. I have been driving VCOs from its outputs and I think the resolution is good enough for my needs.

I am using the unipolar mode. I have not tried the bipolar mode yet.

No Lab precision here. I just don't have the tools to measure 1V/oct signals with high accuraccy. I tune my VCOs and calibrate the AOUT_NG by using a regular chromatic tuner on the audio output of the VCO.

I have been using the Kenton Pro-2000 for years. Compared to the NG:

The Pro-2000 is highly configurable. The AOUT_NG inside a MB-SEQ is not configurable.

I wanted to have the AOUT_NG inside the SEQ even that I knew I would loose all the configuration options on the MIDIbox-CV stand alone unit. The "problems" I have found are more related to the features that are not configurable or aren't implemented in the SEQ.

Btw: I also mounted the PSU inside the case. For the AOUT_NG PSU I used a Velleman K8042 Kit:

http://www.velleman.eu/distributor/products/view/?id=353668

I can buy these at my local store. Costs something like 10 EUR and the only optional thing, not included, are the heatsinks.

Just tell if you want me to test anything special or to make a deeper comparison of the Kenton vs AOUT.

Share this post


Link to post
Share on other sites

Just tell if you want me to test anything special or to make a deeper comparison of the Kenton vs AOUT.

That would be nice indeed, since I was assuming, that via the mixer-page one could address the AOUT_NG (so that - for instance - LFO's could be generated via CC-Messages). Like it says in the SEQ-manual:

Values are sent directly over the selected MIDI port - all ports are selectable, even Loopback and AOUT!
Edited by oshidash

Share this post


Link to post
Share on other sites

That would be nice indeed, since I was assuming, that via the mixer-page one could address the AOUT_NG (so that - for instance - LFO's could be generated via CC-Messages).

Sorry, maybe I didn't explain myself correctly in my last post. Not being a native English speaker I often find it more difficult that I first expect to explain "technical" things like this in a short-simple-clear way. I'll try my best. :)

Yes, you can use MIDI CC to control the NG CV outs, but each CV output is fixed to one specific CC. This is what I meant by "not configurable". In the same way, notes on each MIDI channel are linked to fixed CV/Gate outs. So the "configuration" is done by sending the appropriate messages on the appropriate MIDI/AOUT channels.

I'll explain one of these "problems" I found while working with the NG.

Referring to the table on ucapps:

Note 9 Key Value -> CV Out #1, Velocity -> CV Out #2, Gate #1 and #2 triggered

Note 10 Key Value -> CV Out #3, Velocity -> CV Out #4, Gate #3 and #4 triggered

Note 11 Key Value -> CV Out #5, Velocity -> CV Out #6, Gate #5 and #6 triggered

Note 12 Key Value -> CV Out #7, Velocity -> CV Out #8, Gate #7 and #8 triggered

Note 13 Velocity -> CV Out #1, Key Value -> CV Out #2, Gate #1 and #2 triggered

Note 14 Velocity -> CV Out #3, Key Value -> CV Out #4, Gate #3 and #4 triggered

Note 15 Velocity -> CV Out #5, Key Value -> CV Out #6, Gate #5 and #6 triggered

But as fas as I see in V4, it is actually working like this:

Note 9 Key Value -> CV Out #1, Velocity -> CV Out #2, Gate #1 triggered

Note 10 Key Value -> CV Out #3, Velocity -> CV Out #4, Gate #2 triggered

Note 11 Key Value -> CV Out #5, Velocity -> CV Out #6, Gate #3 triggered

Note 12 Key Value -> CV Out #7, Velocity -> CV Out #8, Gate #4 triggered

Note 13 Velocity -> CV Out #1, Key Value -> CV Out #2, Gate #1 triggered

Note 14 Velocity -> CV Out #3, Key Value -> CV Out #4, Gate #2 triggered

Note 15 Velocity -> CV Out #5, Key Value -> CV Out #6, Gate #3 triggered

Which would be fine for me, but...

I was using a track on AOUT 9 and a normal VCO-VCF-VCA modular patch. CV Out #1 going to the VCO 1V/oct CV in. Gate #1 routed to the (VCA) ADSR Gate in. And CV Out #2 controlling the VCF Cutoff. So Velocity affected the Freq. of the filter. Then I added another module on the audio chain. And wanted to add a CC Layer to the seq. track and route it to another NG CV Out. The problem is that any CC routed to AOUT Chn 9 will output on CV#1 and will open Gate# 1.

Mode Note 10 any CC routed to AOUT Chn 10 will open Gate# 2.

Mode Note 11 any CC routed to AOUT Chn 11 will open Gate# 3.

Mode Note 12 any CC routed to AOUT Chn 12 will open Gate# 4.

So none of the 4 modes worked for this. I had to use Note 13 and switch the cables on CV Outs #1 and #2. Then any CC routed to AOUT Chn 13 will open Gate# 5 which is not ruining my ADSR Gates.

Then I realized I could not use another CC->CV on that Track and would have to use another Track (assigned to another AOUT Chn) to control another CV parameter on my modular synth patch.

To this type of "problems" I was referring before. But I didn't mean to compare the Kenton and the NG in terms of best/worst. I just wanted to point out that the NG inside the SEQ is missing some parameters/features found on most MIDI-CV converters, including the MIDIbox-CV stand alone unit. The things that I probably miss the most:

-Note Memory is only 1 note. This is fine while playing the SEQ tracks, but not that good when playing directly from a keyboard. 2 or 3 would be enough for me and then it would be nice to choose note priority between Lower/Higher/Newest and Multi or Single trigger Gates.

-On the Kenton, PitchBend, Portamento and Modulation (using one of the internal LFOs) are implemented by default at the same output as the notes, without having to use another CV output/assignment. I am so used to have these working without having to patch extra cables/modules (Yes, I am lazy, I know).

Finally, one great thing about SEQ V4 are the FX LFOs. By using the ExtraCC# they can be routed straight to a CV out.

I attached 2 recordings demonstrating the different resolutions of a triangle LFO controlling the whole cutoff frequency range of a self-oscillating VCF.

NG_LFO_04steps.mp3

4 steps LFO. 8 LFO cycles of each 48, 96, 192, 384 ppqn.

NG_LFO_08steps.mp3

8 steps LFO. 4 LFO cycles of each 48, 96, 192, 384 ppqn.

IMHO even at 48ppqn this is way better than LFOs created using CCs! :)

I hope some of this made any sense and sorry for the long post.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now