So tonight was my first time taking my Party out to a gig. Everything worked as it should, but the noise was prohibitively outrageous. I had noticed it was noisy when putting it together, but in context, it was all but useless.

After checking some of the vids here, I think something might be wrong. It seems similar to the problems others have experienced (at least from the 'crunchy' thread.

I submit the following:



At first it's clean guitar going straight into WTPA. Then you heard the sound of recording being engaged (pretttty loud), then BOOM, the playback is like a mack truck in the ear. I do a bit of overdubbing so you can hear how fast it deteriorates into total mush.

I then do another recording with gaps in it. You can hear the processor(?) chugging away during the recording process, but on playback it's louder than the guitar. You can also hear like an aliasing attached to each note, or something funny happening.

I'm totally into the party, and have a Cocolase from ciat-lonbarde, so I'm a connoisseur of 8-bit glitchy delay, but in it's current state, it's not going to be dancing at prom much.

All the pots (bodies) are soldered onto the board, barring the 2 that are offboard (which are grounded either way). But I'll eyeball around.

What does poorly soldered RAM sound like? Or rather, do I just check continuity between the pins, or unsoldered connections?

Is the processor noise what sounds the most off about that? I can live with the blippity bloop, although it's not super awesome or anything. It's mainly the wall of static in all sample playback that makes it hard to subtly sample things.

With the selectively grounding the circuit. Are you talking about just grounding random things, or grounding ICs and all that?

I remember seeing something on here somewhere about cutting some traces to reduce noise, although that might've been just in reference to the V1.00 boards.

Man, what a mission this has been!

Ok, I grounded more and tested more (btw I'm the fellow with the mixer). I think my mixer was fucked. After poking around a while, I 'cut the cord' and went direct into the WTPA and the signal to noise ratio MASSIVELY changed, IE, my mixer is killing the signal bigtime.

WIth everything grounded nicely (touching jacks at the moment, as the enclosure is plastic) when I press record, I only hear a quiet blippity blip. And it's SUPER quiet as compared to incoming signal

I've cut the overdub traces, and now that I'm getting all 'sleeves rolled up' I think I'm going to cut the other traces and do the shielded wire thing.

Could you be more specific about the what/where of that?

It's a load off knowing the sampler part is working alright, although the mixer thing is it's own PITA (it' suuuuper hot-glued into the enclosure at the moment, so even getting it out is going to be a pain.)

Hey Rodrigo --
Good work!
To answer your questions:

It would totally be a drop in replacement to switch the TL072 to a TLV2462 or TLV2472, and would allow you to share PSU rails.
However, I'm not entirely sure that it will help other than to simplify the circuit.

Knowing what I do now, I strongly suspect that capacitive noise pickup between the two circuits is the culprit. Your mixer isn't GENERATING digital noise, obviously. This means that electrostatic noise from WTPA (the digital edges tend to radiate pretty strongly) is getting into your mixer circuit somewhere. There are a couple fixes for this:

1.) Lower the impedances in your mixer. This will be a pain in the ass, but all those 100k resistors and Hi-Z inputs are going to be prone to noise pickup especially if there isn't a ground plane or grounded metal chassis surrounding them. You rebuild the thing with 10k resistors where the 100k resistors are, buffer the pots with another quad op amp, and scale the coupling caps accordingly and you'd get 10x less noise pickup at the expense of a little more current draw.

2.) Distance. Electrostatic noise drops as a function of distance, although I can't remember if it's linear or exponential. Linear, I think.
Either way, moving your Hi-Z mixer bits away from the offending digital noise source will help a lot.
IE -- don't make the two PCBs a sandwich if you can help it.

3.) Ground plane. Put a piece of aluminum (or copper if you happen to have that lying aroung, like a radio shack board) between your offending source and the pickup location and ground it. This will help a lot with capacitively coupled noise but NOT AT ALL with EM noise, which I'm pretty sure isn't your problem.

Read the parts about "electrostatic coupling" in this article:
http://www.industrialcontroldesignline.com/212501871?printableArticle=true

OH --
And use IPA or denatured alcohol to break the bond between hot glue and other materials. Lift up a corner of the hot glue and dab a little alcohol under the lifted spot. Comes off like magic.
Thank the toy people for that trick :-)

Finally, the trace to cut and replace with a shielded wire is the one to the ADC from pin 7 of the TLV2474 to the ADC (pin 34) on the AVR. Make sure you only ground the shield at one end -- preferably the one near the op amp. Make sure you cut the trace near the AVR, too.

I haven't done this since some really early tests. I think this will help some, but I also don't think this is your major problem.

Best,
TB

OK, did the shielded cable mod. The difference is pretty remarkable. The digital cross talk is gone. At least perceivably.

Awesome.

It took a while to troubleshoot it, but I had to run a lead to the input gain pot, as nothing was making it to the ADC without that. In looking at the pictures of the board in the manual I saw that pin7 also led to the center pin of the inputgain.

If anyone is on the fence about doing the mod, I give it 3 thumbs up. Fairly easy to do once I removed the AVR (I couldn't see where the trace was coming from).

Yeah, after running the cable, I hear nothing digital when recording, or in the playback. The only noise that's there is regular static/hiss/artifact stuff. After bypass the mixer in my setup there wasn't much digital noise left, but it seems to be totally gone now.

Not sure I follow with the question there. The mixer circuit which I built/added is completely gone at the moment, so I'm just rocking a 'raw' WTPA for all of this testing/modding.

At the moment, when I hit record, I don't hear anything until I playback, which is nice too.

No problem on the feedback, like I mentioned before, I'm super jazzed about it all.


After some more researching, I think I'm going to go with a buffered mixer circuit, just have to find the right one/layout. I'm running mainly instrument level stuff into that mixer, and with the impedance as it is, and this digital clock in the same housing, I think it'll be better off being buffered.

At the moment I'm looking at this, which I believe is what you've mentioned, and it also uses 10kA pots (so I don't have to get new pots or swap out the ones I was using before).

Ok, a few questions here.

R1-R4 I think would be better off at 1M as all inputs going into it are guitar level or lower (piezos).

R5-R8 to 22k. This just makes stuff louder?

With bypass caps. The one after the diode (1n4004), it would be in series with it, or running from it to ground?
Same for the opamps. Does the bypass cap run from somewhere to ground?

Not sure on the R10/R11. R11 is already 10k. Should R10 be lowered?

I just realized now that the schematic had been changed a bit since it was posted (and is no longer working from the looks of it above).

Thanks for the input on this.

With R5-R8, is lower impedance here good for guitar/piezo level stuff, or does that not matter at this point in the circuit?

Bypass caps. How big is a 'big elec cap' to use at the start? 100uf?
How important would the smaller caps be on the opamps if I use a big one at the input? Because of how I etched the board, I'm putting the components in on the solder side, and I'm going to cover the other side of the board with copper foil, to insulate it better, so putting those caps in will have to be tricky (over top of the opamps, which will be socketted).

With the last bit there, should I change C5 to 100uf, and run a .1uf in parallel with it?

Thanks for your help with all of this. The guy who came up with the schematic isn't very helpful on the diystomboxes forum. It's very 'figure it out yourself' over there, which is great and all, but help is always welcome.


Lastly, would there be any benefit to using hi-fi opamps here? Like some burr browns, and nice caps for C1-C4? The mixer/wtpa are going to be the hub of my setup, and always in the signal path, so I don't mind going over the top if there's a benefit to it.

Happy new year to you too!

I'll try to get all the bypass caps in I can.

I ordered some parts the other day and am making all opamps socketed so I can test different opamps there(for now I have 072/074). For the caps I ordered those purrty red WIMA caps for C1-C4.


Thanks again for taking so much time to chat me up about a circuit that's not even yours.

props to Rodrigo's website miami to manchester cool trip. Can old guys say props? dunno. He played with fred frith, auuuugh... its too late, put me in the home, where are my teeth? young whipper snappers