Original (?) harmonic tremolo circuit troubleshooting

oxbow_lake · Post by **oxbow_lake** » 09 Oct 2022, 04:19

Hello, this is my first time posting and I'm not sure if this is the right place to do it. I tried to register with DIYstompboxes but the registration confirmation email never got through despite many attempts.

I am an amateur circuit enthusiast. I designed the attached harmonic tremolo circuit from scratch. What makes it different from other harmonic trems I have seen is that rather than always having the high and low pass filter gains oscillating 180 degrees out of phase from each other, you can have each filter oscillate independently. This leads to some kind of nice leslie-esque effects. The circuit works, however it is extremely noisy. Specifically, there is an intolerable amount of hiss/white noise regardless of knob settings. Also, in moving from breadboard to a PCB in a box, it also picked up the dreaded oscillator click bleeding through the power supply.

The basic design is a JFET input buffer with some gain, followed by active high- and low-pass filters whose input amplitude is modulated by JFETs acting as variable resistors. A buffered version of the clean signal is mixed through a simple pot to act as a blend/depth control. The LFOs are your typical bang-bang opamp LFOs because they're easy. Pots control high-pass oscillator frequency, low-pass oscillator frequency, blend/depth, and output volume. Switch toggles between traditional harmonic tremolo and independent oscillator control.

The JFETs being used are NTE458s (no longer available, but the circuit works the same using PN4393s), and despite the labeling in the schematic, the two quad opamps being used are TL074s. Power is being supplied at +9V, +5V, and 0V using an L78L05 voltage regulator.

Is there something inherently noisy about this design? And if so, could anyone offer suggestions on how to reduce the hiss? The oscillator clicking is almost certainly from insufficient power supply isolation, but suggestions on how to make that better would also be hugely appreciated.

Thanks for any input and advice, apologies for any ineptitude either in circuit design or schematic drawing, and apologies if this is not an appropriate place to post. Also if anyone knows how I can get registered with DIYstompboxes that would be awesome.

Manfred · Post by **Manfred** » 09 Oct 2022, 10:49

In my opinion, there are some questionable things in the power supply.
The capacitors are oversized.
Are the wrong ground connections only in the schematic or also in the real circuit?
The entries marked in green are the commonly recommended values.

: PSU Questions.jpg (47.35 KiB) Viewed 1817 times

oxbow_lake · Post by **oxbow_lake** » 09 Oct 2022, 15:10

Hey, thanks for your reply! I can certainly try swapping out those capacitors. As far as ground connections, it's a little confusing as written, sorry about that. To clarify:

* Power to the entire circuit is coming from a 9V DC power supply.
* In all cases in the schematic, "Vcc" is +9V, "Vee" is 0V. These are the voltages that go to the + and - power pins of the opamps. For the voltage regulator, I have +9V going to pin 3 ("VI"), 0V going to pin 2 ("GND" as internally labeled), and pin 1 is giving me +5V.
* Everywhere along the signal path between C1 and C7, "GND" is the virtual ground of 5V. So, the DC potential that the filters, buffers, and LFO opamps see at their inputs is approximately +5V.
* The chassis ground, input, and output jack sleeves are connected directly to the 0V power supply terminal. This is (I think? I mean, it should be!) DC decoupled from the 5V "ground" used in the signal path.

Hopefully that make sense? Thanks again!

mauman · Post by **mauman** » 09 Oct 2022, 15:41

Welcome to the forum! Your choice of posting locations is fine.

Using a triangle symbol for AC "ground" is unconventional and as you can see, causes confusion. I'd recommend marking it +5V, Vref, 1/2V or something similar. You could also use a voltage divider (resistor pair) instead of a regulator to provide that 4.5V bias voltage. You could use one of those 100uf caps on the 4.5V (or 5V) bus to ground.

OC26 · Post by **OC26** » 09 Oct 2022, 17:06

The clicking noises (I suppose there are two of them) comes from the quick action of the two schmidt triggers in oscillators circuitry. Try to put two small caps between pins 5&7 and 12&14 (say 10n) to slow down the switching transient and see what happens... Isolating this noise via power supply depends a lot on circuit layout IMHO.

oxbow_lake · Post by **oxbow_lake** » 10 Oct 2022, 01:06

Thanks everyone for your responses. I'll try some of these things and see how it goes, and will keep in mind the grounding convention for future schematics.

OC26 - when I had the circuit laid out on a breadboard, the clicking was absent. In that situation, I had the two TL074s at opposite ends of the breadboard. I tried to maintain that separation (of the LFOs, the filters, and the voltage regulator) physically as much as possible on the PCB, but obviously it wasn't enough. Are there other layout strategies I should take into consideration?

oxbow_lake · Post by **oxbow_lake** » 10 Oct 2022, 01:56

Tried swapping out the 100uF caps for the values suggested by Manfred, tried 10nF caps on the LFOs to reduce click as suggested by OC26, unfortunately no improvement in hiss or click. Mauman, I have a 100uF cap going from the 5V pin of the voltage regulator to the 0V pin, is that what you meant? In terms of circuit noise, would a simple voltage divider for the reference voltage be quieter?

mauman · Post by **mauman** » 10 Oct 2022, 02:21

Yes, 100 uF from +5 to ground will help prevent noise. A passive voltage divider might be a bit quieter than a regulator, but I don't think the regulator is adding the hiss you describe. Next step might be making an audio probe and tracing the noise thru the signal path to find the source.

Lani · Post by **Lani** » 10 Oct 2022, 04:44

Welcome!!
May I also suggest switching the TL074 used for the LFO, to a lower current device such as the TL064.
https://www.ti.com/lit/gpn/tl064
https://www.ti.com/lit/gpn/lm324
But also, do as mauman suggests and build yourself an audio probe so you can narrow it down a bit.
Also, shouldn't C5's negative terminal go to VEE (GND) instead of GND (+5v)?

stolen · Post by **stolen** » 13 Oct 2022, 17:13

Hello friend!

We noticed one tiny thing that might help improve performance! Unfortunately it comes down to your power supply setup once again, but from a different angle: Regulators almost always are unipolar and make for very bad fake ground. Imagine the regulator like a resistor between VCC and its output that adjusts itself to produce 5V at the output; this works well if there's a load pulling it down, however it is absolutely powerless against being pulled up! In your circuit, only R12/14/15/16 pull it down with a light load of 55k, while your input buffer is a 5k-ish-ish pullup! Did you ever measure if the output of the regulator actually is at 5V? We wouldn't be surprised if it was a bunch higher.

We'd advise to use a voltage divider + bypass cap buffered by an OPA instead. You'll get a much stiffer reference that way. The oscillators should get their own reference to avoid input clamping effects or similar parasitics - in your case another buffered divider would work well. This isn't the most efficient way to execute the circuit idea, but it should do the trick.

R19 does nothing but increase noise. We'd omit it.

Lastly, using less than half of the supply for your input buffer reduces its headroom dramatically, so we suggest to tie R2/R3/C3 to VEE instead of GND.

Hope this helps!

oxbow_lake · Post by **oxbow_lake** » 21 Oct 2022, 17:02

Hey, thanks for the response! I'll give it a shot. When originally modeling the circuit in software using a voltage divider for the reference gave me some weird output from the FET variable resistors, though I never understood why, though in retrospect it was probably because I didn't buffer it with an opamp. I will try it in real life and see how it goes.

The headroom on the input buffer was intentionally low so that I could overdrive it easily if I wanted to. I don't think that should increase the noise floor?

Thanks again!

Manfred · Post by **Manfred** » 21 Oct 2022, 21:02

using a voltage divider for the reference gave me some weird output from the FET variable resistors

What values did you take for the voltage divider resistors?

oxbow_lake · Post by **oxbow_lake** » 22 Oct 2022, 00:15

I forget now, but I believe some something like 33k? Trying to balance between excessive current draw while still holding voltage as a reference. A buffer is obviously a better way to do it. This time I was going to try a two 10k resistor voltage divider feeding the input of an voltage follower, with that input also shunted to ground (VEE) by a 100nF cap, as shown in the diagram. Any reason not to use higher value resistors? Actually I see in other schematics on this site people using 1Meg resistors. I should probably spend more time looking at other people's circuits.

I would use the other half of the tl072 to make a duplicate reference voltage, one for each tl074 (oscillators and signal path) as suggested by stolen.

Lani · Post by **Lani** » 29 Oct 2022, 02:12

I'd put at least a 10uF on the divider... Other than that, it looks good.

oxbow_lake · Post by **oxbow_lake** » 08 Apr 2023, 13:47

Just an update here, this version of the circuit works nicely with no clicking and little hiss. Happy with the result! If anyone cares I can post some audio clips.

Thanks again for everyone's help!

Nocentelli · Post by **Nocentelli** » 09 Apr 2023, 09:14

oxbow_lake wrote: ↑08 Apr 2023, 13:47 Just an update here, this version of the circuit works nicely with no clicking and little hiss. Happy with the result! If anyone cares I can post some audio clips.

I would love to hear audio and I'm sure many others would too, if it's not too much hassle for you to set up, record and post clips.

I'm planning a stereo panner pedal, I was going to do some sort of tremulus lune thing with an inverting opamp stage but this design popped up.
I assume that in the second mode switch position, the hi- and lo-LFO control signals are sync'd to the left-hand LFO only, but are they out of phase as well? i.e. might this work as a panner if the two outputs go to separate L+R?

oxbow_lake · Post by **oxbow_lake** » 14 Apr 2023, 01:40

Cool, I'm out of town for a couple of weeks but will try to post something when I'm back.

Yes, in one switch position only the left-hand LFO has any effect, and the two control signals are 180 degrees out of phase. There's no real advantage to using the BJT to phase split instead of using an inverting op-amp, but it would have put an odd number of opamps in the circuit and that bugged me.

oxbow_lake · Post by **oxbow_lake** » 23 Apr 2023, 11:15

Hmm, so having had more time to play with this pedal I've noticed another glitch. When the high-pass FET is ramping up or down, there is a white noise "scratch" sound when ramping in either direction. When the FET is fully conductive there is no noise, and when fully non-conductive there is no noise. This ONLY happens when the FET gate is fed by the LFO signal coming from the BJT collector - not the 2nd LFO, and not the emitter side of the BJT. I can make the low-pass FET make the same noise if I feed it the collector-side LFO signal. The LFO waveforms are identical after the coupling capacitors regardless of the signal source. Possibly the output impedance at the collector side is high enough that there is some oscillation or other interference being created when the FET is neither fully on or off?

Your thoughts and suggestions are more than welcome. Perhaps I could put a p-channel FET for the low-pass channel, eliminate the phase splitter altogether and just feed both FETs the same control signal? Otherwise, I could try to work around it by adding an emitter follower to feed the high-pass channel, or I could implement op-amps to do it instead of the phase splitter, but that would either require more components or an odd number of op-amps.

Sorry for posting the previous schematic prematurely!