Korg Electribe ESX-1, continued / updated Nov. 2019

11/3/2019: OK, I have yet another update on this!

Unfortunately, about 2 years after I implemented the fix I describe below (debouncing capacitors for each pot, plus a shield) the issue came back!
I wanted to take a closer look through the circuitry with my oscilloscope but did not yet get a chance to because when I opened it up, the issue went away. So I just did my best guess of what might help and packed it up yet again.

I had a good idea of what might help, because thinking more about the symptoms had led me to a stronger idea of what might actually be the issue.

There are really two symptoms:
1. constant random jittering of pot values, which is what people usually focus on
2. the pots can NOT be turned down to 0,  the minimum value anything can reach at full counterclockwise position is 2 or 3 for me usually.

The second symptom may actually be a better indicator as to the source of the issue, in my unit at least. What the second symptom seems to point towards is an inadequate connection to 0V (ground) on the pot board. So the 0V on the pot board has a potential with respect to the 0V that is the reference of the ADC on the other board, which causes the ADC to never read any pot as having reached its minimum value; if this potential is constantly fluctuating due to changes in the small amount of resistance between the main board and the pot board, this will be read by the ADC as changes in output voltage.

In case the concept of an ADC is not familiar to you… an ADC (analog to digital converter) may be inside of or external to a microprocessor or microcontroller, and what it does is convert an incoming analog voltage or a stream of multiplexed analog voltages (in this case, the voltages that every pot is set to at a given moment) to a digital value (a number represented in binary) that the CPU/MCU can use and store. The ADC “decides” how to do this by seeing where each incoming voltage falls in the range between a maximum and minimum reference voltage that are fed into it. The voltage range between these two reference voltages is divided into some power-of-two’s worth of equally spaced “slices” depending on the resolution of the ADC (e.g. 2 bit, 8 bit, 16 bit, etc) and then each incoming voltage is turned into a binary number depending on which slice it fell in. A 2 bit ADC means you have two bits to represent numbers with, meaning it can represent four values and divides the range into four slices. An 8 bit ADC gives you 8 bits to represent values with, meaning you get 256 slices (2^8=256).

Anyway, all it takes is a little bit of voltage offset between the 0V feeding the pots and the 0V that the ADC is looking at, and no pot will ever be able to reach what the ADC interprets as 0V. And, if this voltage offset is fluctuating at all, a pot whose value is NOT changing will be read as changing by the ADC, because the changing relationship to the 0V reference recontextualizes it as falling in a different slice.

All this is to say that I surmised that a very likely cause of this symptom was something that is super common in everything we work on: excessive ground impedance, AKA a “ground loop!” So I desoldered, and resoldered the connections bringing ground across between the boards, because their lead free solder was looking a little crappy.

So far the problem is gone but it really wouldn’t surprise me if yet again, my theory for the cause of the issue is disproven by the problem coming back in a year or two. Then I’ll be trying to find another explanation, and another solution, again.

See below for the previous installment of the saga, in which I explain a solution I came up with based on a theory as to the cause of the issue that I no longer believe is correct ¯\_(ツ)_/¯

***

A few weeks ago I wrote about a “job” I did on one of my own pieces of gear, my Korg Electribe ESX-1. After replacing the pots (explained in previous post) and feeling so hopeful for my Electribe’s future, I later realized the biggest problem that causes the Electribe controls to jitter is NOT the pots themselves as it so often is with other pieces of gear, but rather induced electrical noise from the circuit board underneath affecting the voltages being sent to the A/D circuitry by the pots.

 I finally realized this a few hours before I had to perform and fixed it by doing two things. I did these two things at the same time because I had that time crunch before the show. I suspect, but have not verified, that the one that actually fixed it was probably the shield.

    1. I added a 1 uf capacitor between the output of each pot and 0v to filter out high-frequency noise there. This is done by putting it right across the lugs of the pot. There are already SMD filter capacitors there. The service manual does not disclose their value. I figured more couldn’t hurt.
      filter capacitors across pot outputs
      filter capacitors across pot outputs

      2. I shielded the entire back of the pot area with a copper sheet. This is adhesive-backed copper that I left the paper backing on so that it wouldn’t short against the panel board. This is probably not the most clean or economical way to do this but it was what I had and I had no time to source something better. It is just screwed in by the normal board screws and then I soldered a wire directly to the copper and to a ground point, that conducts any induction it collects to ground.

ESX-1 with pot area shielded from induction
ESX-1 with pot area shielded from induction

So now it is really and truly jitter-free.

12 thoughts on “Korg Electribe ESX-1, continued / updated Nov. 2019”

    1. Try making a shield like I did, and instead of soldering it, you might be able to get away with just screwing some of the PCB-fastening screws in through the top of it. Take note, the underside must remain insulated though so if you use a copper shielding sheet like I did, leave the paper backing on!

  1. What is your opinion on other fixes that users claims to be working. Have you tried them?
    1. Cleaning the inside of the potis fromi slicone lube.
    2. Insulation (baking paper) in between the boards, to prevent them from touching each other.
    3. Tightening screws, to stop flexing and therefore preventing boards from touching.
    4. Reposition the ribbon cable (claims that this is actually what happens during other fixes)

    1. I tried variations all of these methods except cleaning the pots, and I did one better on that account and actually replaced them… none of them worked.

      As for the shield, anything conductive would work including aluminum, and something both conductive and ferrous to directly block the magnetic fields (steel) would be ideal. I don’t recall what ground point I used but you can find one by checking continuity between something you know is definitely ground (the ground lug of one of the output jacks for example) and places you think might be ground until you find one that’s conveniently located for your objective.

  2. Hi Alison,

    I wanted to know if your ESX is still working with the capacitors or if the knob jittering has come back because I read on a French forum a guy who used the method you proposed here but the knob jittering came back to his EMX.

  3. Hi Alison, Glad to hear you won’t give up on this issue. What is strange with the knob jittering is that it stops everytime you open up your ESX or EMX, but why ? Fresh air seems to make it work again for a few days or months.

Leave a Reply

Your email address will not be published. Required fields are marked *