Sunday, August 11, 2013

Polysix - Modifying the MG Delay Circuit

As I've been discussing in the last few posts, the MG Delay circuit on my Korg Polysix doesn't work quite right -- even with an MG Delay of zero, it still suppresses my MG signal at the start of each new note.  In my most recent post, I found that my MG Delay control voltage (CV) from the Polysix DAC is still a tiny bit too high, even when set to zero.  Since I can't do much about the DAC, I chose instead to modify the keypress signal that is compared to the MG Delay. The result is that my MG Delay now works as I think that it should -- when set to zero, the MG signal is applied smoothly across all notes, with no MG transients at the start of the notes.  Below is an audio demo with the MG Delay modified (top) compared to how it sounded on my Polysix prior to modification (bottom).  In addition to the MG Delay being set to zero in both cases, all of the other settings are the same, too.



Find the MG Delay Circuit:  A schematic of the relevant part of the Polysix circuitry is shown below.  The MG Delay functionality is effected by IC14, which compares the MG Delay CV produced by the DAC sample-and-hold (arriving at Pin 2 of IC14) to a voltage signal that pulses which each new keypress (arriving at Pin 3 of IC 14).  Whenever Pin 3 goes lower than Pin 2, the MG is suppressed.  I'd like modify the circuit so that, when the MG Delay is set to zero, the voltage at Pin 3 does not drop below whatever voltage is being delivered by the DAC to Pin 2.


Swap R98 to Control the Voltage Drop at Pin 3:  The voltage at Pin 3 drops with each new keypress because a keypress causes Q5 to conduct, which allows the charge stored in C33 to drain out via R98. Because Q5 only conducts for a short amount of time, we can limit how low C33 gets by constricting the flow of charge out of C33.  The easiest way to do that is to simply swap R98 from its default 4.7K value to a higher value.  At first, I tried 20K, but decided that 10K was better.

Removing R98.
New R98.  I first tried 20K.  I settled on 10K.
Viewing the Impact of R98:  To visually confirming that I correctly affected the voltage at Pin 3, I used my oscilloscope to view the voltage at Pin 3 (as altered by my modification of R98) and the voltage at Pin 2 (the MG Delay CV).  As shown in the picture below, changing R98 now keeps the voltage at Pin 3 from dropping below the MG Delay CV that is being applied to Pin 2.  As a result, the output of IC14 now stays high, which means that the MG signal is not suppressed by the start of the note.  Success!


Ensuring I Did Not Over-Correct:  It would be very easy to use too large a value for R98.  How do you know that you have an OK value?  If the resistor were too large, then the MG Delay functionality would be defeated for values other than zero.  I'm not trying to completely defeat the MG Delay, I just wanted zero to be zero.  So, to make sure that my MG Delay still worked well, I turned the MG Delay knob a little bit until the DAC put out a higher MG Delay CV.  It turns out that there is a pretty big dead zone on my MG Delay knob (that is unrelated at all to changes to R98)...I had to turn my knob all the way up to "1" in order to see the DAC output tick up one notch.  Once I saw it tick up, I retested the voltage at Pin 3 to ensure that it dropped low enough to trigger the MG suppression.  With my initial change to an R98 of 20K, Pin 3 did not drop low enough.  By swapping R98 to 10K, I got the result shown in the picture below.  As you can see, Pin 3 does indeed drop briefly below Pin 2, as desired.  Success!


So, with this modification, I have achieved my goal of making the MG Delay work as I want.  When turned to zero, the MG Delay is defeated.  When turned slightly above zero, the MG Delay works as before.  I'm pleased.

12 comments:

  1. So all that's needed to accomplish this mod is a 10k resistor and that's it? I'm somewhat new to electronics but comfortable with modding and learning.

    ReplyDelete
    Replies
    1. Yes, it was just one resistor..BUT, remember that I was trying to fix a problem exhibited by *my* Polysix. Your MG Delay circuit on your Polysix probably works just fine (mine did when I first got it). If your Polysix doesn't exhibit the same problem, there's no need for this mod.

      There are lots of other cool mods to the Polysix that I've talked about in my pages. So, if you're interesting in modding, I'd check them out instead.

      Delete
  2. There is a delay in mine too. About the same as yours had. As a simple example.. If I have settings set (delay zero, rate highest, level highest) when I simply press a note there is a delay of about that much time before any modulation occurs. I would love for it to be immediate.
    In your original example you play some separate notes and one can hear the delay, but in the corrected version you play a sequence where I can't tell if the notes are being re attacked (with release long and mono mode) or if they are all just being held down long enough to make it smooth. Because when I make a sequence like that... If I reattack each note, there is a delay before it goes to proper modulation setting, but of I hold notes down overlapping, then it is smooth.
    I believe I have the same need though. It would be terrific to fix this.

    ReplyDelete
    Replies
    1. You're describing it very well. Sounds like you have the same problem! You can try the same resistor value that I tried...but, if you re-read my text, you'll note that I tried a couple different resistors until I got one that was just right.

      Because it's easy to fool oneself with testing modifications only with your ear, I used an oscilloscope to check the response of my synth with each of my different resistor values. If you don't have an o-scope, you can probably use your ear and try different resistors until it sounds right.

      Good luck!

      Delete
    2. Thanks so much! So.. A total newbie question... When I buy a resistor such as this one, is there anything else about it that I need to know and select? Like is it a certain material? Or are there any other factors to know except that it's 10k?

      Delete
    3. Resistors are pretty simple. In addition to the resistance value, options include: (1) package size, (2) tolerance, (3) power rating, and (4) material.

      For this application, the only thing that matters is package size...there are "through-hole" resistors and "surface-mount" resistors (which are tiny). You want through hole, like I show in my pictures.

      For tolerance, any tolerance is fine. Mine happen to be 1%, but it doesn't matter here. For power rating, mine are standard 1/4W resistors, but even the skinny 1/6W guys would be fine here. For material, it doesn't matter. The standard ones are metal film (?), and they're totally fine.

      For basic 10K resistors, you could buy some real cheap guys from Sparkfun:
      https://www.sparkfun.com/products/11508

      Good luck!

      Chip

      Delete
    4. Or, if you're feeling bold, you can buy any value you want from a "serious" electronics store like Digikey. Digikey takes some practice, though. I wrote a post about it:

      http://synthhacker.blogspot.com/2012/12/buying-parts-and-using-digikey.html

      Delete
    5. Perfect. I thank you kindly!
      I don't know if this is the same Chip that is on the Polysix forums and if you've already read this, but I could use a quick opinion or two about accessing/creating a few CV patch points as jacks to make the synth more modular with some hard patch ability. Do you think this is possible? Are the CV's accessible on the boards? I'd love to gain access to things like the MG LFO input/output, the PWM LFO input/output, etc..

      Delete
    6. Yeah, I'm the same guy. I'm happy to continue the conversation over in that forum...

      Delete
    7. Radical. Here's my last message. Would be awesome to hear your thoughts. http://groups.yahoo.com/neo/groups/PolySix/conversations/messages/3918

      Delete
  3. This comment has been removed by the author.

    ReplyDelete
  4. I did this mod today! Thanks Chip.
    I went ahead and decided to just chance it and put in a 10K and assume our problems were similar in value. It seems they were very close at least.. The result sounds great! Nothing noticeably funny, and the new notes sure are not having any momentary non-effected segment. I might someday hook up my scope and spend the time but i have so many things i'm trying to do and hey what sounds good sounds good! If i find i hear something funny later on then i'll hook up my scope and a decade resistor and get to work switching values up and down.

    ReplyDelete