This is my second attempt to build a module with a 10 pin power header on a PCB and I think I might have this wrong again, or… I don’t know. I am confused…
The PCB in question is a PT2399 delay kind of unit by Gerbrand Sterrenburg. His website and the associated github page have the module well-documented, so I had a PCB of this fabricated and put it together. Since the last module I put together on a PCB turned out to be problematic (I have yet to resolve this), I was even more careful this time and took some measurements without the ICs and the power on… and everything came out wrong…
Take a look at the picture.
C16 received -12 on the positive side and C17 +12 on the negative when it should have been the other way round. 78L05 received -12V on the pin 3 input when it should have been +12 and the IC1 (PT2399) pin 1 (bottom right as it is oriented upside down on the PCB) received something like -4.5V. The readings on IC2 (TL074) pins 4 and 11 were reversed as well (and pin 4 was conspicuously low as well).
I checked the continuity on the PCB and indeed, that’s how the connections are supposed to be… Reversed ???
I double-checked the schematic of this module and indeed the two pins on the left of the power header (pins 9 and 10) should get -12V, as per the Eurorack/Kosmo standard. And they should connect to the negative side of C17, not the positive side of C18 as per the PCB and the continuity of the connections.
However, the picture of the built module by Gerbrand Sterrenburg shows an orientation of the header, capacitors or regulator exactly like the silkscreen on the PCB and my build.
This had me thinking that I must be missing something really obvious.
Your photo is labeled correctly. With the key notch at the bottom, the -12 V pins will be on the left and the +12 V pins on the right. Assuming your cable is made correctly, of course. Something you should check if you haven’t.
But in the KiCad Connector_IDC:IDC-Header_2x05_P2.54mm_Vertical footprint, and presumably in equivalent footprints in other EDAs, the pins on the left (the -12 V pins) are pins 1 and 2, the pins on the right (+12 V pins) are pins 9 and 10.
I use a custom footprint and symbol in KiCad but it will be similar using standard library footprints and symbols. With my symbol, these are the correct connections:
Note -12_IN connects to pins 1 and 2, +12_IN connects to pins 9 and 10.
This from the linked page is incorrect unless it’s being used with a footprint whose pin numbering is backwards from the above convention.
Which it appears it isn’t. I can only conclude that person is using backwards cables. Look at the left end of their header and the capacitor it connects to:
It seems clear the left pins (which would be -12 V with a standard Eurorack cable) connect to the + terminal of the capacitor with the - terminal connecting to ground. Which would explode if that really were the -12 V end. So their cable must be wrong.
Presumably the fix for you is to use pliers to pull the plastic shroud off the header and press it back on rotated 180° so the notch is pointing up.
Or desolder the header, put it backwards, and while I am at it change the regulator that must have been fried (there was a smell for a few seconds, so…), and have a look at the electrolytic capacitors as well?
Looks that way to me. It’d be more understandable if they were using unshrouded headers, then it kind of doesn’t matter which way around it is in the schematic as long as you plug it in the appropriate way, but shrouded headers connected backwards, that’s… remarkable.
Might be worth leaving a note on the Github Issues tab, might not.
I took a look at some of their other modules, and they’re not even consistent. The Curtis VC ADSR is also backwards:
If you want to. Non-destructively unsoldering headers can be a miserable business; pretty much impossible with an iron, maybe with hot air. Pulling the shroud off and turning it around is much easier. And yes, polarized components may need replacement. I think if the caps didn’t blow they’re probably OK though.
You are right about desoldering the header! I forgot how tricky it is to desolder multipin stuff from modern PCBs (unlike vintage PCBs and stripboard). Pulling the shroud off did the trick.
It looks like there is a warning about the backwards connection, as pins 1 and 2 are marked on the left. So if I ignored the footprint and the build picture on the website and followed the pin numbering instead (or been more careful noticing what should logically go where) I wouldn’t have had the issue.
Still, its a good warning that gerber files in github might have inconsistent silkscreen markings.
I replaced a few capacitors to be on the safe side, and had to kludge a new regulator in as it was impossible to pull the damn thing out.
I haven’t tested the module yet, but I am aready thinking that I might have actually spent less time putting together a stripboard layout from scratch.
I had a bunch of the gerbster boards fabbed a few years ago and had problems with quite a few of them. This same delay was in the pile.
I noticed the header was backwards and broke off the plastic part at which point the module started working to some extent but there were still isues and I gave up on it for some reason.
I spoke to someone else on discord who had built a lot of the gerbster stuff and it seems there’s more with issues than not, so for a newcomer it may be worth looking elsewhere. It’s a real shame though as theres quite a few nice modules on there.
This particular one is said to be a copy of a Synthrotek module. I’ve heard some disparaging remarks about the quality of Synthrotek’s designs (above and beyond the disparaging remarks about Steve Harmon) — can’t speak to that personally, I have not built or used any of them. The Synthrotek schematic doesn’t show the power connector, though, that’s on Sturrenberg.
In the interest of anyone reading this forum who wants to build this module, I should let you know that I managed to get this module to work as intended simply by removing the plastic of the header and rotating it 180 degrees as @analogoutput suggested. I am glad I was a bit cautious after the problems in my attempt on Haillant’s Simple VCA, so no ICs were harmed. I only had to replace the regulator.
Indeed, this is a Synthrotek design and not a particularly good one. There is no wet/dry blend, and the mix interacts with the feedback. Between them, there are very few usable settings, while the delay/rate pot has very limited usable travel in itself before it gets awfully noisy (not in a good way). This also means that patching almost any CV will get the rate up to the noisy setting making the sound unusable in a strictly musical sense. I could only get a decent sound when I plugged a very limited voltage in the CV input (e.g. a very slow CE-2 LFO with the depth turned all the way down) and adjusted the rate potentiometer accordingly. Still, it is not bad in these few settings. I played a few notes on a scale for a couple of minutes using a single oscillator, through a slow-modulating filter cutoff and the rate slow-modulated by another LFO, and my wife fell asleep on the couch. I knew white noise has a relaxing effect, so I guess noisy delays have that effect too? (Or it could have been the wine, I don’t know).
More importantly, it served its purpose of getting my hands wet with the PT2399, before I tried one of the nicer delay or chorus designs. I read quite a few complaints that a good fraction of the PT2399s sold in batches online at the particular auction site seem to be faulty, so I didn’t want to risk pulling my hair out trying to figure out the faults in a stripboard design. This particular design had a CV input (an attractive feature to try), and having a PCB would keep the variables that could go wrong at a minimum. Or so I thought!
As @Kelaifu mentioned, I was also impressed by the variety of designs at gerbster (like some Thomas Henry circuits that I could not find elsewhere), but I was a bit suspicious because I have not seen these repositories discussed in forums. I do appreciate it when people take the time to make their PCB designs open-source but I was just very surprised that they did not document these issues (let alone correct them). It just takes a minute to write on the website or repository a “beware… so and so” warning. I also noticed that the potentiometers are wired backwards from what one expects them intuitively. I guess not everybody is so diligent like @analogoutput and others in here!
The arrow seems to indicate counterclockwise is ground and clockwise is +5 V, as one would (I think) expect. But that assumes the symbol correlates to the footprint in the expected way, i.e. that the bottom of the symbol really is the footprint’s counterclockwise terminal. I presume this is an EasyEDA schematic; I’m not familiar with that, but I know in KiCad there are inconsistencies in the various pot footprint pin numberings.
If I had a chip of questionable provenance (not that I ever would of course, harrumph, nothing but genuine chips with a certificate of authenticity hand delivered by the CEO of Princeton Technology for me, thank you) I’d breadboard one or more application circuits from the datasheet. Then again, most of this circuit seems to be taken directly from the datasheet — they just added input and output buffers and (which I guess is the critical part) the CV section.
I checked my box of shame and found the delay and the 140 ADSR from that repo in there, the ADSR doesnt seem to have a S and R stage at all, although I wouldnt totally rule out my own error somewhere along the way.
A friend has built the Coron drum, which was one i was interested in and a few points of note are that that one of the pots is stated as 2.4M, which doesnt seem to exist in that footprint, however using a 1M seems to be a decent trade off. Also on the coron drum, there appears to be an issue on the header which connects to the 2 PCBs: The connector has a +5 and +12 voltage line and they appear to get switched on the header. However my friend who built it said the module still worked.
Indeed test circuits are the way to go, but as you pointed out the one in the datasheet is almost as complicated… the possible mistakes I could make on a breadboard increase exponentially!
Having said that, I am considering doing this for my batches of LM13700s. Unsurprisingly, someone else has already asked this question at another forum. There is a response by Matthew Skala at the very end suggesting an LM13700 LED blinker off the datasheet. I will probably try this after I finish off the other 5 (!) modules that I have on my desk that need a couple of final touches before they go on the case…
I built the triangle/square vco with the LED implemented as Matthew Skala described to test my LM13700s. Added test points so I can look at both waveforms in addition to the blinking LED. Works great.
Ouch. It’s highly unlikely that I will be able to find a non-polarized capacitor in this value. Few of the 12 or so shops that I know stock anything above 1μF and none over 2.2μF. I guess that increasing the current is not an option as the resistor will have to go down to 10k for a similar value. I guess I can try for a frequency at 25Hz and hook up an oscilloscope probe at the output instead of the LED…
Right! I did this like a year ago when I recapped that Germanium transistor radio amplifier so I must have a few 22μF leftovers… I guess I’ll be back on the breadboard tomorrow!
seem to be faulty, so I didn’t want to risk pulling my hair out trying to figure out the faults in a stripboard design. This particular design had a CV input (an attractive feature to try), and having a PCB would keep the variables that could go wrong at a minimum. Or so I thought!
