DOD FX90: max delay and volume adjustment?

[Rob Strand]

Woah, I didn't know they made trimpot caps. That's a great idea. Time to pick up a few.

I'll see if I can dig up a 3M3 for experimentation. With all this variation, I'm thinking the trimpot might simply be there to allow for different part values. I remember reading somewhere that at least one company's explanation for differing part numbers was "it was the most economical batch we could get", and the 1M pot is a nice big range to account for variations.

There's no real point using a variable cap the variable resistor already does the same job.

For your specific unit there is a resistor value which will provide the required maximum delay.   The only thing you have to do is work out the value.   If you run out of trimpot range then simply increase/decrease the 2M4/2M7/3M3 resistor as required.   Ideally you want the trimpot centered when you have the desired delay time since that gives you the largest range of adjustment up or down.  If you want to experiment with longer delays then perhaps target 300ms with the pot shorted then the 1M pot will allow you to increase the delay from that point.   It's a very easy experiment.

If you doing production at DOD *then* what you want is the correct delay with the trimpot centered.   And that will allow the largest range of adjustments to compensate for part variations.   If the pot value is too small in comparison to the series resistor the range of adjustment will be overly restricted.   I think that's the root problem with the DOD FX90 so DOD have to keep tweaking the other resistor as well for coarse changes in the parts they get from different batches.   A 2M2 trim pot and a smaller series would solve the issue.  If they wanted to keep the 1M trimpot then you need to make the series resistor smaller and make the cap larger.    If you look at 2M7 + 1M trimpot the adjustment range is 2M7 to 3M7.  However 2M2 + 2M2 trimpot has an adjustment range of 2M2 to 4M4.

Just noticed your 2.4M post, so I might try a few different values, but I also don't want to fry the board with all this soldering. Might be time to leave it alone lo

It's only an estimate to show things are in the ball park and working roughly as expected - no weird stuff.   There's no reason to try it since it won't be the correct value.    Your unit is what it is. There is a single correct resistor value which will give you 300ms delay (or whatever)  you just need to find that value by experiment.

injected a 1kHz signal using my phone and found the top and bottom voltage levels where the signal bottoms and tops out, and then used my scope as best I could to find the midpoint, or at least the point where the top and bottom aren't clipping too much. Man, I need a signal generator...

Maybe back off the level a bit and re-adjust.   To me the level is too high (crappy -ve swing) and that makes it a little difficult to see where the best bias point is.   Typically you adjust the bias for lowest distortion, then raise the drive a little to see if it is symmetrical, and perhaps repeat until you are convinced the bias is centered.

[tootsMcgee]

Ahhh, good call. That's much better. It might be a little offset still, but it's close enough.

On my Aion FX Blueshift build instructions (still in progress), it mentions that a properly biased BBD will always be a little asymmetrical, and the example given looks almost perfectly like this.



I soldered a 10Meg next to the 2.4Meg to get ~2Meg. Now I can push the delay time up to 380ms, which is actually about where I was aiming, and I can reduce it to precisely 300 now with the trimpot. I can't actually hear any more distortion than the normal amount of noise, so I'm happy.

Thanks for all your advice and hints. I love learning more about how things work. Now I can clean this up and eventually will add it to my Toots Project thread with some longer sound demos.

For informational purposes for future people encountering this thread:

• Added a 10pF cap in parallel with the 25pF one to push the delay time a bit; probably not necessary, instead do this:
• Added a large (10Meg) resistor in parallel with the 2.7Meg resistor to yield around 2Meg. Lower resistor here will slow down the M3N101 clock frequency. The trimpot will allow further tweaking for the max delay time.
• Rebiased the output of the MN3005 BBD (because I definitely didn't touch it or anything). I used a 1kHz signal from my phone and set the trim pot to roughly 50%, and put the delay as short as possible. I probed pin 3 or 4 (they are tied together) of the MN3005 and moved the trimpot to either extreme to measure where the min and max voltage are. Finally, I adjusted it until the sine wave was centered between those two extremes. If the sine wave is very asymmetrical, reduce the signal level and try again. A proper bias will have a very slight amount of asymmetry.

I had mentioned that the bias pot also makes the square wave off the MN3101 look slightly different at either extreme. I'm pretty sure it's just a coincidence.

[Rob Strand]

On my Aion FX Blueshift build instructions (still in progress), it mentions that a properly biased BBD will always be a little asymmetrical, and the example given looks almost perfectly like this.

Looks good.  Agreed, you can't expect 100% perfect symmetry.  Inside the chip you will find some asymmetry.

I soldered a 10Meg next to the 2.4Meg to get ~2Meg. Now I can push the delay time up to 380ms, which is actually about where I was aiming, and I can reduce it to precisely 300 now with the trimpot. I can't actually hear any more distortion than the normal amount of noise, so I'm happy.

Thanks for all your advice and hints. I love learning more about how things work. Now I can clean this up and eventually will add it to my Toots Project thread with some longer sound demos.

All good.

For informational purposes for future people encountering this thread:

Summarizing the thread gets a big thumbs up from me.   All too often these threads do your head in when you read them a year later!  even though on the day it was all perfectly clear.

[Rob Strand]

FYI:
For the hell of it I had a look at the range of clock frequencies on my ltspice simulation when the delay pot was set to min and max.   The DOD FX90 manual specs 60ms to 300ms, a 5:1 range.   What I found is the 2N5089 did not give 5:1 but a 2N3904 did.  It could be a spice model issue.  I used a Fairchild 2N5089 model and a Mr Cordell's 2N3904. (Cause is BR and TR parameters in the spice model for the 2N5089.   I'm not saying these are incorrect.  The parameters have *very* significant effect on the upper frequency of the oscillator.  Even though the transistor is a current source it seems to saturate at start of the narrow oscillator pulse.)