belton reverb overdriving on transients

[pinkjimiphoton]

hi fam,
built one of these:
https://guitarpcb.com/wp-content/uploads/2018/05/BD_ADD-VERB-Build-Document.pdf
lovely project, works great, sounds great...

except if i REALLY bash the input, it clips hard and sounds horrible. i'm a fairly dynamic player, and when i hit the guitar, i tend to really thwack it.

i was looking around trying to find a q&d way to limit this, without having to get crazy, and was thinking maybe  antiparallel diodes before the first stage in line witht he audio, or maybe antiparalleled led's in the feedback loop of the first stage kinda like this:

https://www.261.gr/limiters.html

figure 3.

or is there a better way to do this? suggestions? i've found this problem universally with belton brick reverbs, and 2399 delays, as well.

appreciate the advice!
peas!

[Mark Hammer]

As a device that needs to run off 5VDC, one needs to have relatively conservative expectations of the Belton Brick.  And t's not like clipping in digital devices is a pleasing and gradual thing.  Perhaps what you need to include is a compander circuit so that you get your dynamics but without the clipping.  Scott Swartz's PT-80 design uses a PT2399 as the delay element, but with a compander circuit and I find it behaves very nicely.

[pinkjimiphoton]

thanks sir mark,
was hoping you'd chime in. i was thinking a compander, but hoping something more q&d... i don't think its the brick distorting, tho it may be, i think its the input stage of the first opamp doing it... that was why i was wondering if the old dual diode trick may work to slow it down some.
i will take a look at the pt80 project, thanks for the tip.
its a really nice reverb sound, but i'm hoping to find a way around the issue. when it clips, its npt graceful at all, and the reverberation of the attack transient distortion kinda makes it impossible to actually be able go use it. it DOES have a gain stage built in, so i suppose i could attenuate the input somewhat then make up for it at the pedal, but i'd have to add a buffer bypass to it then i think, as attenuation before the circuit would be an issue when the effect is off.

you know the drill... how do i make this work the easiest cheapest way and get 200% results? lol

thanks again for the advice bro. much appreciated!

[ElectricDruid]

That initial gain stage is a bit odd. +17dB of gain and a input impedance way down in the Kohms. Plus 22n/27K gives a rolloff at 268Hz - middle C, more or less. Seems like strange choices to me.

https://guitarpcb.com/wp-content/uploads/2018/05/BD_ADD-VERB-Build-Document.pdf

My first try would be just to increase R2 a bit and trim that gain going in. Maybe 47K or 68K. Failing that, put your pair of back-to-back diodes across R3, but in series with another resistor (maybe 47 - 100K). That should soften any clipping that does occur and keep it a bit more reasonable. This is what I did on the Flangelicious input mixer to limit the signal going to the BBD, but for a flanger a slightly crunchy input sounds great whereas for a reverb maybe not so much!

[pinkjimiphoton]

thanks tom.
i will try it, i love my flangelicious!!

[R.G.]

The BTDR2 does hard-clip on transients. It's specced for max of 3V peak to peak if I remember right. I put two back to back diode pairs in a string at the input and fed it through a resistor. It still clips on peaks, but the clipping is now much softened, as it starts clipping earlier (so, feed it through a voltage divider and don't hit it so hard  ) and recover the lost gain later. Yeah, it's not perfect from a noise perspective, but you're way up at line level anyway, so you're at least out of the moving-coil mud. The clipping I got from this was much better behaved, from very little at lower levels to pleasant (by guitar standards) crunching.

At least that's how my BTDR replacment for an accutronics tanks works. I like it.

[pinkjimiphoton]

The BTDR2 does hard-clip on transients. It's specced for max of 3V peak to peak if I remember right. I put two back to back diode pairs in a string at the input and fed it through a resistor. It still clips on peaks, but the clipping is now much softened, as it starts clipping earlier (so, feed it through a voltage divider and don't hit it so hard  ) and recover the lost gain later. Yeah, it's not perfect from a noise perspective, but you're way up at line level anyway, so you're at least out of the moving-coil mud. The clipping I got from this was much better behaved, from very little at lower levels to pleasant (by guitar standards) crunching.

At least that's how my BTDR replacment for an accutronics tanks works. I like it.

elegant.

but i opted for the easiest solution. i added a pot to the input to pad it down to where it stops distorting, then used the "volume" part of the circuit to make up the lost gain on the other end. no more crackling noise or distortion, and plenty of reverb and volume. even after padding it down at the input, i have almost a half pot's throw of volume left over after reachieving unity gain!

next time, i may try your approach as well as mine... i have a feeling a touch of "soft clipping" may sound good for some applications.

[anotherjim]

Q&D might be a series resistor and a couple of clipping LED's on the way into the brick. RED would limit it to 3v-ish.
A softer clip might be a CMOS inverter running on the same 5v with equal input and feedback resistors (unity gain) followed by a trimmer volume control. The inverter soft clip will happen first and then volume control lowers the level so it can never overload the brick.

[pinkjimiphoton]

yeah, that's what RG suggested almost. great idea.

but i opted for lazy. can't get much lazier than shunting signal to ground with a pot!

[pinkjimiphoton]

when all was said and done, i ended up having to mess around a bit more than expected.

had to use a jfet to do the switching; that took care of the "pop" when engaging it. after the jfet switch, a 100k trimmer to cut a bit of the input gain, i didn't measure it yet, but about 30k resistance in series. as a pot, not a variable resistor. that helped with the brick distorting, but not enough... by the time ya trim it back enough to nullify the clipping, you lose enough input signal where the reverb really suffers.

so next try was 914's, then led's. both kinda worked, but i didn't like the clipping at all. just didn't sound good.
so i tried a pair of bat41 schotkes , with the bands facing each other, one from the ic side of the 27k input cap, and the other side to ground. BINGO. now i can turn the verb up enough to get some shimmer and high end. sounds a lot better. figure 5 in the link in the first post seemed to work best.


BUT it needed more output. so i went to the second stage of the chip, and swapped out the 10k output pot, first 50k, then 100k, then 250k, which ended up perfect, as i could turn the reverb up just a bit louder than i need, which for my purposes, was good.

i grafted the whole mess into this: https://pcbguitarmania.com/wp-content/uploads/2019/08/Tweed-Bassman-59-building-doc.pdf
which is a really nice preamp designed by bajaman.

send is from junction of r9 and r10, return is junction of r13 and r14. puts the reverb in parallel with the tone stack, basically, which works out pretty well. ideally, i should have returned it at a later stage, changed a buffer to a gain stage, but i was lazy.

so when all settles down, gotta nice little fender style preamp with reverb that sounds great.

thanks for the help, guys!