long digital delay build log

[~arph]

Itead studio is mucho cheapo

[cloudscapes]

Itead studio is mucho cheapo

yeah but don't ship via regular post office (last I checked). unfortunately in my case, couriers like UPS and Fedex are super complicated. limits my options somewhat.

[~arph]

DHL, UPS or registered air mail.. So they do now 

[cloudscapes]

[Marcvv]

Wow, looking really good. Well done! Sold!

[alanp]

This kinda stuff is hugely impressive -- it's not something I could aspire to do!

[cloudscapes]

This kinda stuff is hugely impressive -- it's not something I could aspire to do!

That's what I thought when I started 8 years ago

[cloudscapes]

some of the art I drew for the design

[~arph]

Nice, skilled artist too  any other special abilities you need to tell us about?

[cloudscapes]

I cook a mean batch of vindaloo, also 

[cloudscapes]

looks like it doesn't work at all. did some probing and can't figure out what yet.

microcontroller "works" at least (its executing code), at least it didn't blow up

[cloudscapes]

there seem to be at least three issues with it.

sent sine wave into the input and probed the path step by step. the buffer and gain stage appears to work as expected until it gets to the feedback mixer, where all of a sudden there is huge gain on the signal when there shouldn't be, so the sine turns into a hard square wave. resistor values for that gain stage are the correct ones (desoldered them, and put in fresh ones, no change). the diode clipping part right after protects the ADC from more than 4v, and just checked so that isn't a problem. only its getting a square wave instead of sine (4v p to p, as wanted). I really don't get what's going on here with this gain stage. it's 4 resistors and a couple caps, and they all check out individually. on the breadboard it's working, so... I don't know.

onto the next issue. ADC samples the now-square wave, and the DAC spits it back out as expected. only that the microcontroller isn't responding to control inputs at all, such as delay time and all that. and I can't even hear the square wave out the pedal's output, but I haven't yet probed where it stops and why.

this is so discouraging. I spent a full week cross-checking the breadboard with the schematic and PCB layout to make sure there weren't any errors before I had it made. but there seem to be several. one I can understand, but this many problems...... it's almost easier to redesign a brand new layout. assuming the schematic I drew even works.

fortunately I still have the working delay on breadboard. so there's that.

[~arph]

The PCB could have a production error. Like an unconnected via.

[cloudscapes]

might've been, but I checked continuity and it checks out. between the bit where the test signal is fine and as expected... to the bit where it doesn't work, there are only 4 resistors, 3 caps, one pot, one opamp, one via and a voltage reference. I replaced all the passives, and everything checks out individually. just.... not really as a circuit, at all. even though the exact same thing is on my breadboard and working. nothing's shorted either.

I'm half expecting it to be a breadboard "error" that made it work, and a pcb without the errors doesn't.

I'm done with this thing for tonight. tomorrow I'll reflash the firmware so there's no delay, only ADC in straight out to DAC, no memory/processing. then I'll probe the original breadboard at every resistor, every cap, make note of voltages, then do the same on the board to try and see a better picture.

[free electron]

Rule no1 in troubleshooting: don't try to fix everything at once, do it step by step.  Let's focus on the clipping problem first. If i got it right, the problem is where the red "!" is?

Btw. you don't need that 470k resistor in parallel to the Dry pot.
What kind of opamp is that? If a FET input like TL0XX you also don't need that 1Meg resistor between the Vref and the +input. Connect the Vref directly to the + (it's a high impedance input, it won't load the 10k/10k divider).
Then check the voltages on all opamp terminals of that stage, in+, in- and output. All of them should be close to the Vref potential.
I'd remove the 10k or 100n supplying the Wet signal (feedback pot), just to take it out when you troubleshoot the dry path.
If you are 100% about the resistor values, measure the resistance between the pins 8 & 9 directly on the chip, should be around 100k, which is your feedback resistor.
Does OSH Park E-test their boards before shipping them out?

[cloudscapes]

That green path exactly, yeah. Thanks, by the way!

You're right about the 470k resistor. It's there on the working breadboard though, and in designing the PCB I adopted a strict 100% copy mentality. it's a leftover from one of the several audio paths I had tried, I think. I'll hook up the breadboard tonight and probably end up removing it to simplify things. Good idea on the feedback path pot, as well. You never know.

will report back later

[cloudscapes]

finding a lot of mistakes, but fixing them, as I go along.

the first one, thanks to you! the 1m between vref and +input in the feedback mixer section. I shorted the 1m and that seems to have fixed the clipping problem at that stage. I don't really understand why it didnt work in the first place, though, since the 1m is still being used on the breadboard and it works fine. I put it there out of habit, because it's "safe". thanks for that one!

others I've discovered, a software issue with the microcontroller's pin assignments (was easily fixed), and really dumb mistake where I had forgotten to solder a whole row of resistors in the post DAC filter, another mistake where I actually needed a 1m between vref and +input in another part where there was none, and one I just found, a low pass rc filter where I had the resistor before the ground cap, instead of after. I'm done for tonight though. I'll continue fixing tomorrow, as well as provide more info on what I changed/fixed.

[cloudscapes]

it's working! video may take a couple weeks though, since I want to tweak the firmware first. still haven't programmed tap tempo, proper debounce on the footswitches, and a few extras like programming settings while holding down the footswitches on powerup.

[bluebunny]

That looks great, Etienne!

[~arph]

  it's perfect!

Thanks for sharing the schematic and build progress here, I enjoyed following this thread.