More v3205 DirtBag findings

[madbean]

After many hours of tweaking I've made some more discoveries concerning some of the noise and distortion issues with the v3205 version of the DirtBag.

R13/R14 - this divider is the main source of distortion - omitting R14 removes much of it (leaving R21 and C14 at their stock values).
R11/R12 - whether they are 10k, 24k, or 47k actually does not play a significant role in noise, but the lowest values do seem to be best for controlling distortion.
R42 - can be omitted entirely in some cases. Too low a value brings pin12 too close to ground and will kill the expandor portion of the circuit. Higher values work okay. I'm using 20k for R42 and 10k for R11/R12 right now.
C10 - stock 1uF seems to be the best value. 2u2 can cause a little "whump" with self-oscillation.
R10 - lowering this value adds some more attack, but too low and there is excessive oscillation. Trying 10k now.
Also, a cap across R10 does seem to help a bit with noise. Currently trying to find the right combination (if there is one).
C27 - (THD cap) Actually not finding much of a difference with different values
R43 - can be omitted
C9 - want to try 10uF there--have not socketed it yet

Overall: distortion can more or less be eliminated by removing R14 and adjusting GAIN1 and GAIN2 accordingly for optimal self-oscillation. The biggest issue is actually noise, not distortion. The compressor reacts very well with larger input signals--it virtually eliminates all noise. The problem is in LOWER input signals. IE, when notes are played softly, the compression threshold seems to be too high and results in "zipper" type noise. Same thing with decaying notes via delay repeats. This could be an intrinsic flaw in running the circuit at 9v....or maybe not. Don't have an answer yet.

TL;DR - eliminated distortion, working on eliminating noise i.e. faster response time or lower compression threshold, not sure if it is an actual circuit design flaw or not yet.

[jubal81]

Glad to hear you're figuring things out. Should I go ahead and make changes or wait for further testing?

[madbean]

You can remove R14 now if you like, since I'm sure that will be part of the "final" fix. You might want to hold off on making any other changes ATM.

[LaceSensor]

A sincere thanks for the work put in.
I played on mine for about an hour yesterday, its really really nice.
Still looking forward to having the final version of mods.
I find that the pedal is a bit noisy. Is this inherent in the design? and by that I just mean "hiss" when engaged. Is that anything to do with what you mention above regarding the compressor?

Other than that, I have it almost clean, and almost able to go completely crazy on the regeneration, but when you play certain notes or chord you can hear the gritty sound on the repeats. Its not heavy or bassy either. I didnt make notes but I found it around the kinda 8th -12th fret on the top 3 strings. I was playing Smiths stuff. With just the dirtbag, no other FX.

Still, I had a nice time. Again, the modulation is amazing.

[jkokura]

Lace, I could be wrong, but that sounds like many of the Analog delays I've played. They are inherently noisy, though you may have more noise than you should but I'm not there to tell you. They do feature a small amount of distortion of pushed too hard too, and that's an expected feature. It's part of the reason the digital revolution even happened in the delay world! I could be wrong, but your build may be exactly where it should be.

Jacob

[LaceSensor]

Yeah it could be.
I think im right in saying if I removed R14 Id have to put back r21 and c14 to the 1k/33n they were before.

I think I am just gonna enjoy what I have for now, then return when there is a final set of mods rather than desolder anything else. I think I will mess with some of the values I have socketed some more in the week, I have vacation now til Jan

[LaceSensor]

Interesting from Dirk Hendrix @ FSB...

"Excellent build MisticWhiskey!

Finally had the time to finish mine (ex boxing). No problems with distortion (V3205's) without altering the circuit.

But! (my error)
When making R62 7.5k that calculates to a perfect 4.5 volts according to the NE570 datasheets. IRL it biases the expanders output to approx 1 volt. Therefore replace R62 with a wire bridge or, when it's boxed already, shortcut the thing with a piece of wire on the bottom."

Interesting.

[madbean]

I'll check out R62 as well.

My take on all this is, like Jacob said, there will probably be some small noise floor that may not be able to be eliminated. And, true, this is inherent in anything but pure digital delay. However, my goal now is to address that by finding the right fix to minimize it as much as possible and maintain the integrity of the effect. I'll have it sorted out really soon. I've got just about every critical component socketed now!

Ultimately, this will be a good thing moving forward, even though it's a huge pain for everyone right now. It'll have a very good impact on making the Aquaboy Deluxe perfect out of the gate, too.

[madbean]

I picked up a Memory Boy today to do some audio probing and voltage readings. The design definitely deviates some from the "DirtBag" but I'm pretty sure I can extract the info I need from it to make an intelligent comparison.

Side note- took my first trip to Rockblock Guitars down in Nashville. What a freakin nice store! Their pedal selection was outstanding. Really nice used selection, too. They had an Echo Baby, Tremorama, Tap-A-Delay, and some other neato stuff. A few things I had never even seen before.

They also had a new Blakemore Bi-Polar Octidrive. I was really tempted to tell the owner: hey, I designed the PCB layout for that, but then I decided that would be dickish. They also had some JHS, but I don't think there was a Sweet Tea (apparently I inadvertently designed some of the PCB for that one without knowing it )

[dwstanford]

I have a deluxe memory boy.  I can take some voltages if it will help at all.

[MadeByMike]

Sorry just saw this thread.

My board currently has R21 as 100K, and R11/R12 as the originally spec'd 47Ks for the 3205s.

What would you recommend I do right now? Anything that should definitely be subbed? Should i revert R21 to 1K for example, or is it still your recommendation that I hold off until you have a final definitive list of values?

[madbean]

From the Memory Boy:

R11, R12 are 33k. R10 is 6k8 AND there is a cap in series with it, i.e. the cap a 6k8 resistor lie between pins 5 and 6 of the compander. Very interesting!

Still tracing.

[madbean]

How big of a dumass am I? I'm sitting here trying to figure out why R10 is marked 289. Looking up resistor code charts for SMD components. Can't find anything with a "9" suffix. After 5 full minutes I realized I was looking at it upside down and it's 682.

[madbean]

One thing for sure: the Memory Boy has the same "zipper" noise at long delays when you play softly...sounds just like what I'm hearing in the DirtBag. You can hear it by turning the Blend all the way up to kill the input signal and listening to the repeats.

So, basically, I'm going to trace out the portions around the compander and filters and stuff to compare.

It also has a CD4011 and an LM13700 on the breakout board which holds all the BBDs.

[mjcyates]

Side note- took my first trip to Rockblock Guitars down in Nashville. What a freakin nice store!

I was there earlier in the year. I agree, very cool shop.