Yet another Machinedrum repair thread

Elektron Gear Machinedrum
#1

Hey all, my MD mk1 is borked. I’ve post a picture, but it’s identical to the first post from this old thread, which, unfortunately was not followed up on by OP: Help! MD starts no more!. Doesn’t boot at all, can’t enter the early start menu. Critically, all LEDs are lit, which seems to differentiate this issue from the typical malfunctions. Screen is mostly black with a few garbled bits that sometimes change.

So, here’s what I’ve done:

  1. Cleaned/deoxit-ed all connectors.
  2. Swapped in a fresh battery cell.
  3. Contact Elektron support. The tech said the closest they’ve come to diagnosing this involved problems with the UI controller being corrupted. It’s a PIC chip, which I totally have the tools to flash, however, they apparently don’t have the firmware for that chip any more.
  4. Just for the heck of it I swapped in a new crystal oscillator for the PIC chip because I had a spare.

What I have left:

  1. Getting a new power supply (though the tech specifically said that is unlikely to be the problem).
  2. Asking if anyone happens to have the UI controller firmware. I ran across an old thread that I cannot find for the life of me where Elektron sent someone a firmware file for a programmable chip. Maybe this mystery person will see this… and still have that file… I dunno this seems impossibly unlikely, but worth a shot.
  3. Reflowing chips on boards? Can’t hurt.

Any and all further suggestions are welcome. My guess is that this is a lost cause, but I’m willing to get my hands dirty. Maybe some incredible person out there is willing and able to try dumping that firmware? Thank all you fine folks for your time and attention!

6 Likes
#3

Diagnosing this sounds not too dissimilar to diagnosing a boot failure in a new design, except in your case you have no source code or schematics !

First think through the basic architecture. From the way you’ve described things, the display sits off, and is controlled by the PIC. Make sure this is true. Hence the stray and occasionally blinking display would concur with what Elektron support says that this sounds like a problem with the PIC. It’s been many years since i designed with a PIC, but i recall that it has nonvolatile bits that are set to tell it details about things like the oscillator etc. If those bits got blasted somehow, the PIC will not boot properly.

You ought to read about the specific PIC involved, and the boot sequence. Here’s a general version on the configuration bits for a 16 bit PIC:

https://microchipdeveloper.com/16bit:configuration-bits

If you’ve got good test equipment, like a logic analyzer you can watch the early boot sequence of the PIC and see that it vectors correctly, and that the data makes general sense, but again you’re a little in the dark. EDIT: Actually i don’t recall how much of this activity you can see.

Don’t eliminate the chance that the problem is somewhere else (there’s a cable connector to communicate with the main processor for instance, right?) and what you are seeing is cascading from that — the PIC side of things could be entirely fine. Hopefully this is a single source problem, but don’t count on that either.

You might be well served to find someone that more regularly services the Machinedrum — it sounds like Elektron is out of that business now. What — “We don’t have the source code for that anymore” ??

This sounds like the PIC, if that is the processor controlling the display is basically working but is off in the weedz.

ADDED: Thinking some more about this, if the PIC is controlling the display it likely knows very little of the higher level graphics, and more than likely only knows how to draw lines, and dots, and takes commands from the main processor. In which case a failure in communications could be more involved.

#4

All great suggestions, thank you! I’ll do some more thorough continuity testing and check all the cables. Unfortunately I don’t have a logic analyzer because I’m a scrub who is a passable technician but otherwise knows little about what he’s doing BUT my dad… has a job that’s impossible to describe… forensic technologist? Basically he professionally reverse engineers things to act as an expert witness in patent disputes for big chip companies. In other words, he has test equipment out the ass and I reckon I can bug him for some help.

The big thing is the always on LEDs. That sure seems like… I dunno, some screwed up firmware or communication somewhere. I recall similar behavior in modules I’ve built. Anyway yeah, I’ll start work on checking things end to end.

Another strange thing I’ve noticed is that the main board has some jumper wires on it. For all I know that’s typical of the vanilla mk1s but there’s not exhaustive photo documentation out there so it’s hard to say. I am not the original buyer of the instrument, but the person I bought it from was, and they implied it was relatively early in the production runs, so maybe bodge wires were de rigueur in the early days. I’ll get some photos while I’m testing.

#5

Oh, and yeah, not having the firmware any more seems… Borderline preposterous. I mean I get it, they do not formally support the hardware any more, they don’t owe anybody moving heaven and earth to find it, but yeah… slightly strains credulity. My entitlement brain says “open source the old stuff so we can repair the machines and mod in +Drive surrogates since there’s no reason for that to be a state secret” but I know that’s super unfair.

#6

Seems like it’s getting a solid 5 volts. So continues the continuity test-athon.

#7

I’ve had this off and on with my uw mkii+ and usually after a few reseats of all the connections it finally ends up working again. Bummer that it’s being such a pain for ya.

#9

Yeah, it’s a pain. When it happens to you are all the LEDs lit at all times? That particular quirk seems way less common, but if that happens on yours that definitely gives me hope.

#10

Just as a general question, since Elektron abandoned repair of Mk 1s has anyone taken up their repair and service? They cost enough to be worth repair. Often when a company drops repair they pass the business on to another business, or at least others step in.

#11

That’s a good question. I’ve done a liiiiitle bit of research but I wanna give it an honest try before really tracking folks down. I’ve heard tell some current/former techs from the US office may be reachable and might be willing to work on them? I was half tempted to call the US office in case their techs have more to say than the ones I talked to over email. Though if anyone glancing at this thread has any references, I’m sure we’d all love any info anyone has.

I feel like the 90s/early 00s are gonna be a real tough period for repair going forward in general, lots of microcontrollers with proprietary firmware produced amidst some notable capacitor blights. Good times.

#12

I see a shop in Rotterdam has several YT videos with older Elektron gear repairs. They’re called Human Error Synth Repair – human-error.nl

#13

Not sure. it’s been a while since I’ve had any issues. I’ll check next time I see it though

#14

http://elektron-users.com/index.php?option=com_fireboard&Itemid=2&func=view&catid=9&id=168844

check out this old thread from elektron users. i guess they used to have a repair “center” in georgia, which according to a reddit user is “just some guy they hooked up with to do repairs”. could be worth trying to find/get in touch with him, i’m sure the phone number is outdated but it’s worth a shot. Don Hassler

https://www.bbb.org/us/ga/lilburn/profile/musical-instrument-repair/don-hassler-0443-28097946

i guess he also does repairs for Teenage Engineering. this bbb link has a review from a disgruntled customer who sent him their OP1, but it looks like it got sorted out. this is from 2019

#15

Thank you! I had seen that name come up but didn’t have any solid leads. I’ll definitely try getting ahold of him!

#16

A kind user thought they might have a spare control board from a busted machine. Unfortunately they didn’t, however, in the extra exceedingly unlikely case any of ya’ll have an old busted MD control PCB please get in touch. Worth a try harvesting the chip off of it.

#17

Alright, I have some updates:

I’ve talked to the former US Elektron tech. His guess was issues with the DSPs or the CPU and gave a little debugging advice.

I’ve been able to test out the PIC on the UI board, looks like it’s fine. Which means it’s probably an issue that’s happening on the CPU board. I’m going to do some more digging to see what the CPU and ROM are doing because nothing is being transmitted to the UI board. I’m waiting on a logic analyzer to get that moving.

In the process of all this, I’m taking some pictures of the boards. As far as I can tell there’s not super exhaustive archives of, like, model or board revision images. If I’m feeling really ambitious, I’ll see if I can dump the firmware from the PIC but I’ll cross that bridge when I get there.

4 Likes
#18

Excellent work tumes !!!

I’ve always wondered what communication protocols Elektron is using between the boards ? Whether its I2C (times two), or SPI, something else, or some sort of homebrewed protocol. I am suspecting I2C but you never know. You may not need to dig to far into the protocol to solve your problem, you still might get through this without digging into all the bits. Many logic analyzers have protocol analyzers built in, but again you probably don’t need to get that deep.

I am sure you have already beeped out the cable (or cables) between the two boards. I’d be concerned too with an intermittent problem (break or short) with that cable which you might miss with a continuity test.

You’re doing the right things, divide and conquer. Eventually you will be able to grind this down.

If you only had two Machinedrums (one working!), you could swap out main assemblies and isolate. Maybe someday when you’re in the Machinedrum repair business. :smile:

#20

As far as I can tell, nothing is even coming from the CPU. Though to be 100% honest, I am doing a real hack job testing things while I wait on some proper probes, so I’ll update you once I can say for sure whether or not it might be simple break in the connection somewhere.

Unfortunately, it’s a 4 layer board so figuring out traces is… just a real boatload of fun lemme tell ya.

2 Likes
#21

Thanks! Yeah I have a couple things left to try on my list. I found a random website selling debugger interfaces for the CPU at splurge prices so, uh, things are getting serious. As far as I can tell, the +Drive connector adheres to the BDM interface proposed in the CPU datasheet so fingers crossed it’ll be plug and play (I mean, assuming I can get the 15 year old debugging software running ;p).

I just wanna be embarrassingly crystal clear, I have almost no idea what I’m doing and I’m learning all this stuff as I go. So, uh, any and all advice and braindumps are super appreciated because while I am a programmer, I am not a hardware person so this is all mostly new to me.

1 Like
#22

Oh, I forgot to mention, I’ll probably be trying to replace the cables regardless because this board was from the good old days where it’s a connector on one end and the other end is just through hole soldered. One might say that should have been the first thing I did!

#23

Ouch, that makes cable replacement messy. What, did someone think this would save the cost of a connector and plug ? Mechanical parts are always the first place to look.

Hope this is fun, more or less, for you.