FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 

CGS114 DUSG strange behavior
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page 1, 2  Next [all]
Author CGS114 DUSG strange behavior
Just finished building the CGS114. One side works perfectly. The other side doesn't work at all. All checked all power rail voltages to the chips and even swapped the chips between the working and non-working sides to verify all chips were OK. Working side still works and non-working side still doesn't.

I've been feeding in signals into C (sig in) and A (trig in) in order to trace them to the problem. Feeding a signal into C, it is present at pin 6 (-) of LF353 and pin 7 (output) but nothing is heard on pin 5 (it is heard at pin 5 on the working side).

Its goes out pin 7, makes it past the transitor pairs, and enters pin 2 of the 353, but doesn't exit out at 1 (the output) as it does on working side. It seems to stop at the 22n cap between pin 2 and 1. On the working side, it is present at pin 2, both ends of the cap, and outputted at pin 1. The non working side sees the signal going into one end of the cap, but not coming out the other side. Yet I can find no discontinuity there. I don't have a capacitance test mode on my DMM but both caps read the same on resistance (not sure that means anything). Should I try replacing this cap?

The obvious answer would seem to be yes, but the A -> M on this side doesn't work either. A gate sent into A is very faint at pin 13 (+ in) of the LM3900 and is not present at pin 9. With a gate going into A, the diode leading from A to pin 4 of the LM3900 (near end out) shows a very different voltage coming out of it than the one on the working side does (>10.8v on the working side vs. 0.12 or so on the non-working).

Is there some complex malfunction here that I do not understand? Is it possible for the diode to read normal on a DMM and yet not pass signal, or is there some more complex interaction going on that is jamming up the circuit?

help help Dead Banana Dead Banana

Any help from someone more conversant with this circuit than myself would be vastly appreciated and I will bestow infinite hossanahs upon thee.

Thanks wigglers.

EDIT: The non-working side is also the side where the pin 5 pad of the LM3900 lifted. Looking at the schematic, it appears as though pin 5 is left unconnected, and indeed this is precisely why the pad lifted -- because it is not connected to any traces on the PCB. So pin 5 on this LM3900 was just left floating. This is OK, I assume? There's nothing to jumper it to, right?
It's been suggested for troubleshooting this board to check if pad M is high on start up. The working side is (~5.5v) but the non-working side is not (~0.1v).

Any suggestions on how to proceed from there? I'm starting to think I'll never get this half of the board working. I am reluctant to start re-flowing all the solder joints etc because of the extreme fragility of this board and it's tendency for lifted pads.

If anybody has any suggestions on how to proceed or things to look for I would be very, very grateful!
the bad producer
I had a dead DUSG here recently which was caused by a broken (ie short circuit) zener diode, IIRC it was similar to what you're describing in the last post, but not your first post... The circuit will still work without the zener so you could lift one end to test this
Thanks for the suggestion Bad Producer. This is really frying my brain. I have inspected the traces with a magnifier and gone over them for continuity, and I can find absolutely nothing wrong with it -- except that it doesn't work. very frustrating
Sometimes it's really easy to damage a trace on single sided boards like this one. I've done it on a couple I've built over the years, but was always able to get them working. When you test continuity from part to part, make sure you are not completing the circuit by pressing a bad or broken joint into place.

Otherwise, maybe take a break for a day or two, and revisit? Sometimes troubleshooting becomes too frustrating, and we miss things we shouldn't.
ive seen bad sockets as well, where the socket pins dont go all the way through, or arent making good contact with the IC. one thing id reccomend, is disconnecting any signals, and set all pots to midrange for both halves, and then measure the DC voltages on all IC pins, transistors, and diodes. then compare these two lists. are any of them different? also, be sure to measure on the component, and not on the trace. this will help find bad component/trace connections.

one thing that seems bad from your description, is that you see signal on pin2 of the LF353. that point should be at ground and not moving. so perhaps one of those transistors is shorted out, or not the IC pin is not connected.
How are you taking your readings? I am a bit confused because you say that on the working side you can see the signal at pin 2 of the LF353, but really that pin should be at 0V all the time, unless the op amp is saturating, which it shouldn't in this circuit.

Anyway, I don't think you should be seeing the signal pass from pin 6 to pin 7 of the LF353, at least not as an unchanged signal -- it should be amplified a lot, to the point that the output clips hard, I think. For sure if the (complicated!) feedback to pin 5 is not working, that's what should happen.

So I am suspicious about the LF353 and the circuitry around there.

The whole circuit is a complicated feedback loop really so it is quite tough to find the source of the problem, but maybe that's a start.

[Edit: oh, ha, I now see that guest noticed the same thing about pin 2 of the 353 and posted it while I was writing mine, which took ages! Sorry, didn't mean to repeat]
@guest -- as far as pin readings between the two halves, pad M is low on startup when it should be high. In another DUSG troubleshooting thread Zthee posted a guide and one of the first things was to check that pad M is high on startup. But he never mentioned what to do if it isn't.

I'm going to take readings as you suggested and try to isolate the problem.

LM3900 on that side is not behaving as it should according to Tim Stinchcombes' analysis of the circuit.
Swap the chips from side to side
As per guest's suggestion I went through and took some readings between the working and non-working sides of the DUSG at startup state (no signal in):

working side: LM3900 pin 5 reads 0.12, pin 4 reads ~10.9v; non-working side, LM3900 pin 5 reads ~10.9v, pin 4 reads 0.12

working side, 5v6 reads 5.56v; non-working side, 5v6 zener reads ~0.13v

working side: 4 transistors in core -- emitters read ~0.19v; non-working side read ~0.66v

working side LF353 -- all pins read 0v except power pins (which read correct); non-working side LF353 power pins read correct but pin 1 reads ~3.33v and pin 7 reads ~11v

diode to pin 4 of LM3900 -- working side reads 10.9v; non-working side reads ~0.13v


I'm sure I could keep going. But I have no idea what any of it means; the circuit is not blowing up or shorting out, but it is also operating very differently from how it should, and I can't figure out why.

I'm thinking something simple could make it come to life but I have no idea what. seriously, i just don't get it
I guess I'm lucky I have one working side. The thing is I still need another slope gen, but I don't want to roll the dice on another CGS114 board (this is the only board I've ever had these kind of problems with). Maybe I'll bananafy a befaco rampage or something.
tojpeters wrote:
Swap the chips from side to side

That's actually the first thing I did -- working side still works, broken side still broken after swapping chips. No shorts, solder bridges, broken traces that I can find, all 7 jumpers accounted for, all chips getting correct power, etc. I checked all the simple things first.
Some of the older versions of that PCB are notorious for lifted pads
I’d look there
The DUSG can be a pain to troubleshoot. I was in the same situation, took me days to figure out. I'd suggest you read this:

This is a great article on how the DUSG works, for me it was key to understanding and succesfully troubleshooting the circuit.
the bad producer
I suppose it may also be worth mentioning that at certain settings of the trimmer it won't work either, and sometimes you need to power cycle between adjusting if it locks OFF
@galingong -- yeah, I actually emailed Tim about it. Don't know how active he is now online but I'm hoping he will be able to give me some specific advice about what is going on with this thing.

@tojpeters -- W/R/T lifted pads: yeah, it's not just the older PCBs. This is the current version and I had two pads come right off -- even with a 30w iron, and even being extremely careful since I was already aware of this problem with the DUSG boards. They are not through-hole plated, and there is about a 1 nanosecond difference between thoroughly heated solder joint and melting down the elmer's glue that holds the pads to the board.

From what I understand it's basically a labor of love for Ken, so you can't really complain, but come on -- the CGS PCBs are not cheap even as DIY boards go. There's got to be a place that could fab these boards up to the general level of quality we all expect from every other PCB out there?

Seems like it's tolerated mainly because there aren't really any complete replacements for the DUSGs functionality in DIY land. I've looked into the Oakley VRG, which looks good -- but it lacks among other things the exponential input, which seems like it would really hinder the ability to use it as an oscillator/VCLFO. Same with the Slightly Nasty ASL. I was this close to ordering the boards for that, but I'm a little wary of ordering from Australia when SN hasn't posted in a year and I can't find a single build report or user review online. Right now I'm deciding between banafying a Befaco Rampage and the Oakley VRG. The Rampage is on top for sheer features, it uses TL074s for everything, and it's a modern, quality PCB, but converting it to bananas could turn out to be a nightmare. Oakley VRG is slick but it uses expensive-as-fuck parts and doesn't have the same functionality as your good ol' DUSG.
the bad producer
FWIW I must have built hundred's of modules using Ken's boards, and only had lifted pads when I started messing around replacing parts - and then only rarely.
@bad producer -- fair enough. Not knocking Ken's work. He has done the DIY community an amazing service by making his circuits and the Serge modules available to us.

I've built other CGS modules and they all worked first time, no problems.

I mean, I felt bad even complaining about it. But $35 isn't cheap, either, as PCBs go. Certainly it's not asking too much to expect the same level of quality you get from every other PCB out there.

This is 25 USD:

that's a beautiful fucking PCB. Surely the DUSG board could be brought up to the same standard, given it's unbeatable utility-to-space-to-dollar ratio. Maybe of one of the layout wizards around Muffs should do ken the solid and make a real, double-sided, through-hole plated DUSG pcb. I'd buy several.

EDIT: It should be noted this is (sort of) unrelated to my problems with this particular build, however. I did have some lifted pads but they were caught and repaired straight away. I'm not convinced the problem I'm having with this board is because of the quality of the PCB. I mean, I can't rule it out -- because of the well-known issues with this pcb, I can't even rule out that there aren't some lifted pads or broken traces somewhere that are invisible even to a magnifier or undetectable without an electron microscope. That seems to be the way this board is. But I don't *THINK* that's what's wrong with it. It's probably my fault -- I just can't figure out how. I am content for now to just have 1/2 of a DUSG, and I don't want to start reworking it trying to find the problem and end up with 0/2 of a DUSG. I'm just looking for an alternative PCB that isn't quite so perilous at this point.
Tim Stinchcombe was helpful when I was trying to figure something out on a DUSG a few months ago (wonky waveforms at high frequency I think) so you may well hear back from him.

If I were debugging this I would next look at the voltages at LM3900 pins 9 and 10, to determine which way the DUSG thinks it should be moving, at all the LF353 pins.

Did your post imply that they are all zero other than 1 and 7? I wasn't sure... but if so, if pin 5 is at 0V and pin 7 is around 3V, something is wrong between pin 7 and pin 5, so I would want to know what is going on at the 2N3906/220k resistor / diode in that path.

It is likely to be a very slow process going over these readings on a forum but I bet we can find the problem if we try!
artilect99 wrote:
But $35 isn't cheap, either, as PCBs go. Certainly it's not asking too much to expect the same level of quality you get from every other PCB out there.

Gosh, I didn't realise they cost that much from Modular Addict and Synthcube. The cost from Elby is about half that, but then you have to pay the shipping of course.

(FWIW I should have a build report on the Slightly Nasty ASL in a few days. I keep forgetting to order the final parts.)
the bad producer
I think with the CGS Serge boards the cost is higher as there is a license fee for Serge too, and if someone else made a licensed PCB it would also carry this fee.

Maybe some pictures of the build would be useful, I know it is easy to miss something that someone else might spot instantly!

And yeah, +1 on the Oakley boards!
@cygmu -- yeah, when I power up the board with no inputs, all the pins on the LF353 (on the bad side) are at ~0v except the power pins. (EDIT: I'll check that again just to be sure.)

I did read that if the LM3900 won't flip-flop, it prevents any signal from getting through, because the diode clamps down -- and it is starting low, when it should be high. also LM3900 pin 4 and 5 voltages are flipped WRT the working side. I'm going to try what Tim suggested to another poster and try shorting the power rail to the wiper of the trimpot and see if I can get it to flip over.

(What's discouraging tho is that in that other post at E-M, Tim got really in depth with the guy, and he never got it sorted out either -- the thread just stops, presumably out of total loss of the will to live. cry)

Oh, and the trimmers are both centered. I've tried adjusting/resetting the bad side trimmer but it seems to do nothing at all insofar as the unit won't cycle or pass any signal. That's what is weird -- even if it won't cycle, it should still be able to pass a signal from C to DCO out, as I understand. But it doesn't. So it has *two* bad components? And they both ended up randomly on that side of the board? Or there is one component that is preventing the whole thing from working? cry

Every time I think about it my head hurts.
@cygmu -- regarding the 2n3906/220k/68k coming off pin 9 of LM3900 -- the base of the transistor is sitting just below positive rail, about 11.25v. Same for the resistors.

EDIT: Woops -- I see you meant the 220k and diode going to negative rail. Will take those readings when I get back from work. Hate it when real life interferes with my obsessions!
ill have a closer look in a bit, but it sounds like your zener might be in backwards, or be busted.
guest -- Bad Producer suggested that very thing. Here's hoping that's all it is! That did strike me as odd, but I think I assumed from looking at the schematic that the zener would be sitting low if the LM3900 is in it's low state (which it boots up to for some reason). Fingers crossed!

EDIT: I can tell you it's not backwards -- or at least the stripe is aligned with the silkscreen as it normally should be. (Unless this is one of those components you are supposed to install backward on this PCB.) but broken/malfunctioning IDK.
follow cygmus suggestion and check out that diode/220k resistor hanging off pin 5 of the LF353. do those voltages match the working one?

then have a look at the bases of the 4 transistor core. do those match?
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page 1, 2  Next [all]
Page 1 of 2
Powered by phpBB © phpBB Group