I have made a tri-vibe, with some changes (pot instead of mode switch and attint85 waveform generator as lfo). I found that with the mode pot in the middle position, and the lfo depth set to zero, I get a beautiful bell like tone, even on an otherwise muddy amp. What in the circuit causes this? Is it the op-amps, or the OTA, or their combination, or... ? What i would like to do is make it into a sort of preamp, without the lfo, but then i need to know what part of the circuit is doing this.
Insights much appreciated!
(https://runoffgroove.com/tri-vibe.png)
Mostly unrelated question: why did you use another waveform generator?
Hmmm. I can guess at two possibilities. One is fostered by not knowing exactly how you have the mode pot wired. Some wirings could lead to response peaks that brighten up an amp. Next, subtle distortion gets heard as a tone shift, not distortion. Also, I suppose, could be a little of both.
Quote from: fryingpan on November 08, 2024, 02:05:51 PMMostly unrelated question: why did you use another waveform generator?
Aaah...
convenience (I had already programmed it),
versatility (various waveforms, easily modifiable), and
simplicity (just 1 dip-8 IC rather than the complex circuit in the original).
Quote from: R.G. on November 08, 2024, 02:33:43 PMHmmm. I can guess at two possibilities. One is fostered by not knowing exactly how you have the mode pot wired. Some wirings could lead to response peaks that brighten up an amp. Next, subtle distortion gets heard as a tone shift, not distortion. Also, I suppose, could be a little of both.
I opened up the box to see. It is a 10k pot with the wiper as output, but looking at it, I'm not sure I didn't change other things. Looking at the schematic, I must have... I need to take a closer look.
In any case, this sound is not some known OTA feature?
Just a sec. Is the bell-like tone achieved with the rate at 0? (So with the oscillator "stopped"?) Because it might make sense. You have your all-pass filters fixed at a certain frequency, and by mixing the filtered signal with the dry signal you are going to get some peaks and troughs that might be pleasing.
EDIT: Yes, it is, because the depth is at zero. All you did is stop the filters at a certain frequency, I assume. Basically it's what Hammond organists did when they stopped the Leslie. All you need to do now is turn the rate knob into a value knob (reprogram the chip so that instead of outputting an LFO it outputs a single value), find the position where the filters sound best, ??? and profit! (To reuse an old meme).
EDIT 2: I am also guessing that you did away with the vibrato mode (no clean signal), mounted just the 47k resistor and mounted a 50k pot to go from "Swirl" to "Whirl" (more or less clean signal). A phaser and a classic-style vibrato are very similar, it's just the dry/wet mix and possibly the shape of the LFO (eg. sine for phaser, triangle for vibrato). The Tri-Vibe is very similar to Rod Elliott's take on a phaser-vibrato, with the only major difference that instead of using FETs or Vactrols as the variable element it employs OTAs (noisier and/or less linear but more "consistent" - the poor man's VCA).
Thanks for pondering this. I take it you are a man of culture, familiar with the Underpants Gnomes?
So I think I have added a depth pot, which is not in the schematic. Tomorrow I'll take the whole thing apart, such that I can take a better look at what I actually made, and I'll draw out the schematic.
Quote from: Ksander on November 08, 2024, 04:51:36 PMThanks for pondering this. I take it you are a man of culture, familiar with the Underpants Gnomes?
Heheheh.
QuoteSo I think I have added a depth pot, which is not in the schematic. Tomorrow I'll take the whole thing apart, such that I can take a better look at what I actually made, and I'll draw out the schematic.
Maybe you didn't add a depth pot, but when the rate pot goes to zero it just stops the oscillator where it is. Which is why I recommended recoding the uC to simply output a value set by the rate pot.
Quote from: R.G. on November 08, 2024, 02:33:43 PMHmmm. I can guess at two possibilities. One is fostered by not knowing exactly how you have the mode pot wired. Some wirings could lead to response peaks that brighten up an amp. Next, subtle distortion gets heard as a tone shift, not distortion. Also, I suppose, could be a little of both.
After tracing the circuit I actually made, it is as follows: I replaced the output buffer part, along with the mode switch, by the boxed in circuit in the 'panning for fun' document by yourself (thanks for this, I use it often!). Input A to the panner is the output from the buffer U1a; input B is directly after the 220nF capacitor (everything after that, i.e., the 47K resistor onwards is gone, instead there is the panning circuit).
There also is a depth pot for the LFO; the rate does not go to zero.. When the depth pot is at 0, the input to the OTA labeled LFO is Vref.
So I guess my question is: what is the role of the all-pass stages from the OTA in getting the bell-like sound?
Quote from: Ksander on November 09, 2024, 05:25:02 AMSo I guess my question is: what is the role of the all-pass stages from the OTA in getting the bell-like sound?
See The Technology of Phasers and Flangers (http://www.geofex.com/article_folders/phasers/phase.html) at geofex.com for some background.
The all-pass stages contribute phase delay, but not volume change. When you mix the original "dry" signal with the time/phase delayed signal, you get cancellation notches at frequencies where the phase shift is multiples of 180 degrees from the original signal.
When you add feedback, mixing some of the delayed/mixed signal to the input, you get both deeper/sharper notches, but also peaks between the notches, where the phase delay is a multiple of 360 degrees. The peak in response and sharpness of the notches depend on how much feedback you let through.
This is the thing I was referring to when I mentioned "how the pot is wired", as the mode pot looks like it selects a varying amount of feedback. Setting the LFO speed to 0 leaves the peaks/notches frozen at specific frequencies, as they no longer wobble around. The peaks/notches then act like a very complex tone control. Combine this with the propensity of an OTA to have distortion on its input for bigger signals, even with the input diode predistortion on the LM13700 series OTAs, and you can easily get soft distortion products feeding the complex tone shaping of the phase delay/feedback peaks.
That's what I was thinking. I was just too lazy to type all of that in. :)
Quote from: R.G. on November 09, 2024, 09:53:57 AMQuote from: Ksander on November 09, 2024, 05:25:02 AMSo I guess my question is: what is the role of the all-pass stages from the OTA in getting the bell-like sound?
See The Technology of Phasers and Flangers (http://www.geofex.com/article_folders/phasers/phase.html) at geofex.com for some background.
The all-pass stages contribute phase delay, but not volume change. When you mix the original "dry" signal with the time/phase delayed signal, you get cancellation notches at frequencies where the phase shift is multiples of 180 degrees from the original signal.
When you add feedback, mixing some of the delayed/mixed signal to the input, you get both deeper/sharper notches, but also peaks between the notches, where the phase delay is a multiple of 360 degrees. The peak in response and sharpness of the notches depend on how much feedback you let through.
This is the thing I was referring to when I mentioned "how the pot is wired", as the mode pot looks like it selects a varying amount of feedback. Setting the LFO speed to 0 leaves the peaks/notches frozen at specific frequencies, as they no longer wobble around. The peaks/notches then act like a very complex tone control. Combine this with the propensity of an OTA to have distortion on its input for bigger signals, even with the input diode predistortion on the LM13700 series OTAs, and you can easily get soft distortion products feeding the complex tone shaping of the phase delay/feedback peaks.
That's what I was thinking. I was just too lazy to type all of that in. :)
Woah. Thanks, I actually understand!
Quote from: Ksander on November 09, 2024, 05:25:02 AMAfter tracing the circuit I actually made, it is as follows: I replaced the output buffer part, along with the mode switch, by the boxed in circuit in the 'panning for fun' document by yourself (thanks for this, I use it often!). Input A to the panner is the output from the buffer U1a; input B is directly after the 220nF capacitor (everything after that, i.e., the 47K resistor onwards is gone, instead there is the panning circuit).
From what you are saying there you have removed the 15k + 4.7nF cap from the feedback network of U1B?
The way this circuit works is it boosts treble at the input using using the 15k+4n7 at the input. Then it cuts treble at the output using the 15k+4n7 at the output. The two EQ's exactly cancel out to produce an *overall* flat response. The purpose of doing this is to reduce noise without affecting the flatness of the response - it's call pre-emphasis and de-emphasis.
If you remove the 15k+4n7 at output the boost/cut process no longer cancels out the treble boost and you are left with the input 15k+4n7 boosting the treble. The boost starts at 550Hz and starts to flatten at 2.3kHz reaching about 12dB treble boost. (BTW, it's not unlike a bright switch on some amps.)
The rest of the circuit could be tweaking the EQ a little.
Quote from: Rob Strand on November 09, 2024, 03:50:42 PMThe rest of the circuit could be tweaking the EQ a little.
I had a quick look at it. If output mixing is set so the dry and filters are equal then it will create a deep notch filter. From what I can see when the LFO pot is set to 0 (wiper at Vref) the notch frequency is at about 600Hz. You could measure it to confirm.
Rob - try adding in a variable amount of feedback. Should be interesting.
Quote from: R.G. on November 09, 2024, 08:13:48 PMRob - try adding in a variable amount of feedback. Should be interesting.
It's only a two stage phase shifter so the feedback isn't so effective.
https://www.diystompboxes.com/smfforum/index.php?topic=132500.0
The combination of the treble boost and notch can be approximated by a twin-T notch with a resistive load. You could perhaps match the response a little better by just using an unloaded twin-T notch with a separate treble boost. If you only wanted that EQ profile these options are much simpler than an OTA.
Thanks, I'll need to do some research to figure out how to actually do that, but I'm going to try!
Well, a notch at 600Hz is close to the "middle" setting on a Fender tonestack, which is said to give "bell-like" tones. There ya go.
Quote from: Rob Strand on November 09, 2024, 08:25:39 PMQuote from: R.G. on November 09, 2024, 08:13:48 PMRob - try adding in a variable amount of feedback. Should be interesting.
It's only a two stage phase shifter so the feedback isn't so effective.
https://www.diystompboxes.com/smfforum/index.php?topic=132500.0
The combination of the treble boost and notch can be approximated by a twin-T notch with a resistive load. You could perhaps match the response a little better by just using an unloaded twin-T notch with a separate treble boost. If you only wanted that EQ profile these options are much simpler than an OTA.
Did you happen to make a frequency response graph, such that I know what to aim for?
Quote from: Ksander on November 10, 2024, 02:51:08 AMid you happen to make a frequency response graph, such that I know what to aim for?
Something like this:
(https://i.postimg.cc/FYwMRHdB/bell-tone-filters.png) (https://postimg.cc/FYwMRHdB)
It's just a set of values that work. Plenty of scope to scale impedances.
No input buffering shown.
The schematic hasn't got the details for single a supply implementation.
The assumption is I've deciphered the OTA based circuit correctly to come-up with the 600Hz frequency. It would be best to check the frequency on the OTA unit which has the sound you like.
I've shown quite a deep notch. If your pan control is off the notch might not be so deep.
Quote from: Rob Strand on November 10, 2024, 12:44:44 PMQuote from: Ksander on November 10, 2024, 02:51:08 AMid you happen to make a frequency response graph, such that I know what to aim for?
Something like this
Thank you very much. I put both circuits side by side, but find that, whereas the notch filter does something similar, it sounds a lot thinner. Is that due to the depth of the notch? Is there an easy way to control that, e.g. mixing the bell filter and dry signal with a panning circuit?
Quote from: Ksander on November 11, 2024, 02:12:34 PMThank you very much. I put both circuits side by side, but find that, whereas the notch filter does something similar, it sounds a lot thinner. Is that due to the depth of the notch? Is there an easy way to control that, e.g. mixing the bell filter and dry signal with a panning circuit?
More than likely it is related to the pan control.
If you posted your circuit and made an accurate estimate of the pan pot position it might be possible to do better.
Another possibility is my 600Hz estimate could be off. I had to estimate DC level coming out of the LFO and then the OTA response.
On useful check would be to measure the DC voltage at the output of U2B.
Another useful check would be to inject a sinewave and try to find what frequency the notch is sitting at. You do this by listening for the frequency with the lowest level output. If you take AC measurements it's possible to estimate the depth of the notch.
IF you have a software to do an impulse response or frequency response you could plot the frequency response directly.
Sure!
The pot is at 50%, and the voltage out of the LFO is 4.16v
This is the circuit (hope it is intelligible
(https://i.postimg.cc/yDMj0hzx/IMG-20241111-204854.jpg) (https://postimg.cc/yDMj0hzx)
Quote from: Ksander on November 11, 2024, 03:04:37 PMSure!
The pot is at 50%, and the voltage out of the LFO is 4.16v
This is the circuit (hope it is intelligible
OK cool thanks. I'll check it a different way.
When you built the filter circuit did you put a buffer at the input? If you drive the guitar into the filter it's going to mess up the filtering for sure.
yes, it is in the top left; S_{buf} is the output.
I saw I forgot to draw that the 'LFO' inputs to the LM13700 go in via 10k resistors
Quote from: Ksander on November 11, 2024, 03:46:30 PMyes, it is in the top left; S_{buf} is the output.
I saw I forgot to draw that the 'LFO' inputs to the LM13700 go in via 10k resistors
I checked my previous simulation and it seemed OK. However, I noticed some value differences between your new circuit and the runoff groove schematic:
- 5n6 instead of the 4n7 cap on the treble boost.
That's going to make a subtle difference to the treble boost.
- 2n2 instead of 1n5 on the second OTA.
That drops the notch frequency from 600Hz to 500Hz.
- There's also a small amount HF roll-off from the 330pF cap.
Another significant difference which will affect A/B comparisons is my filter was set to 0dB at low frequencies. However, your panning circuit has about 6.8dB boost at low frequencies. The gain difference doesn't change tone but it will make A/B comparisons difficult. There will be a "correct" gain which sounds equal volume with the dry signal. I really suspect that gain difference is making the biggest difference to your previous A/B comparison.
The notch depth on the OTA circuit is quite deep, so I'm not expecting that to be a difference. Pot settings are never accurate but it should be close with a linear pot set to the middle.
Anyway, if I incorporate the value changes then I get the following filters (~500Hz notch):
(https://i.postimg.cc/8sdY4RKp/bell-tone-filters-V20.png) (https://postimg.cc/8sdY4RKp)
The reason there's four circuits is I've let you choose if you want to use new cap values or keep some of the old values. I've also given two new circuits which have the 6dB gain to match the existing circuit. That will help with the A/B comparison.
EDIT: just to be clear the output is take at the output of the filter circuit. You don't need to pass it through the mixer. Passing through the mixer would make the filters different from the original circuit.
:-*
I just built the second to the left circuit (had those values on hand), and compared the sound of this circuit to the OTA circuit. It sounds exactly the same to me, so seems like you absolutely nailed it! Thank you!
Quote from: Ksander on November 12, 2024, 11:12:34 AMI just built the second to the left circuit (had those values on hand), and compared the sound of this circuit to the OTA circuit. It sounds exactly the same to me, so seems like you absolutely nailed it! Thank you
Very cool. Good thing there's no obscure differences left to explain.
A practical single supply circuit will need a cap between the opamp and filter to stop DC getting out.