LoFi Tremolo - Check my work please

[SprinkleSpraycan]

I modded a Kay K-1 Tremolo and then shopped around and found a good filter to pair with it. Added in a dirty boost and stole the needed bits of circuitry to make it all work together. Im not sure how to draw it right but the lfo has its own diode filters and +/- paths to minimize ticking. hopefully leaving the breadboard will help even further with the little thats left. Before i start figuring out a stripboard layout im hoping to hear any suggestions i can get to make sure im not missing or overlooking something.

my only specific question right now is should i break the trem at the power as shown or at q4?

have at it and educate please

[ElectricDruid]

I've got a few comments. Use them or leave them as you see fit.

1) The power filtering would be more effective with a series resistance. <100R will be enough.

2) Does Q2 really work like that, with the base tied to the 9V rail? Is there a resistor missing in that bit of the schematic?

3) R23/10K doesn't do anything. That amp is wired as a unity-gain buffer.

4) R47/1M doesn't really do anything, because it's in parallel with R13/47K, so only R13 is effective. I suppose it acts to bleed C14 when the switch is open, but that's all it's doing, and I wouldn't expect much of a click from 100n.

5) I would try and re-design the filter section. You could factor the final x2 gain stage into one of the filters, which saves you one op-amp. If you can arrange it such that Q3 can drive the filter's input impedance, you can drop the buffer too, and that would save you one dual op-amp chip.

6) Pin 5 isn't really connected to pin 6 on the Hi-Cut filter U1B, is it?

There might be other stuff, but that's what I've spotted so far.

HTH

[SprinkleSpraycan]

I've got a few comments. Use them or leave them as you see fit.

1) The power filtering would be more effective with a series resistance. <100R will be enough.

Will do, can you explain how this helps?

2) Does Q2 really work like that, with the base tied to the 9V rail? Is there a resistor missing in that bit of the schematic?

i doubt it would work. Drawing error i forgot to fix.

3) R23/10K doesn't do anything. That amp is wired as a unity-gain buffer.

thank you

4) R47/1M doesn't really do anything, because it's in parallel with R13/47K, so only R13 is effective. I suppose it acts to bleed C14 when the switch is open, but that's all it's doing, and I wouldn't expect much of a click from 100n.

Yup, eliminated. thanks

5) I would try and re-design the filter section. You could factor the final x2 gain stage into one of the filters, which saves you one op-amp. If you can arrange it such that Q3 can drive the filter's input impedance, you can drop the buffer too, and that would save you one dual op-amp chip.

This is above my current understanding. but now i know what to start reading up on next.

6) Pin 5 isn't really connected to pin 6 on the Hi-Cut filter U1B, is it?

drawing error, no tie there

Thanks

[ElectricDruid]

Will do, can you explain how this helps?

The caps are there to make an RC lowpass filter, so without the "R", it's not really a filter. Of course, in reality, even a piece of wire has *some* resistance, so it's not *completely* without an R. But the filter cutoff will be much lower and much more effective if you deliberately put a small resistor in there. 47R or 68R are fairly typical. Since mains hum is the most likely noise you'll see on the power input, ideally you want the filter low enough to reduce 50/60Hz effectively.

[SprinkleSpraycan]

Fantastic. Thanks

[antonis]

@ Tom: 
3) In an ideal world..
(here it might count for input bias current..)
4) See above..
(for zero switching time..)

P.S.
You DO know that I respect you a lot..
(despite your aspect of BJT buffers..)

[ElectricDruid]

P.S.
You DO know that I respect you a lot..
(despite your aspect of BJT buffers..)

You're never going to let me forget this, are you?? hohoho

Any thoughts about (5), Antonis? Can we redesign that filter section to save poor Sprinkle two op-amps? Getting rid of one is pretty easy, but can we lose that buffer off the front too? You're much more versed in BJTs than I am (as you've already noted!!) so this is your department! We could swap the order of those two filter stages if it helps any. It's too late on a saturday night for me to think hard about that right now...

[antonis]

It's too late on a saturday night for me to think hard about that right now...

For me it's even later so we'll talk about it tomorrow, perhaps..

P.S.
I'm also waiting for duck_arse "medical report" about the health of a BJT entangled in a "Excitabat enim fluctus in simpulo" circuit of his own design, so tomorrow will be a nice day..

[PRR]

2) Does Q2 really work like that, with the base tied to the 9V rail? Is there a resistor missing in that bit of the schematic?
4) R47/1M doesn't really do anything, because it's in parallel with R13/47K, so only R13 is effective. I suppose it acts to bleed C14 when the switch is open, but that's all it's doing, and I wouldn't expect much of a click from 100n.

R47 drains leakage in C14 so the switch does not pop. (I suspect there should be more switching but it is late.)

This Q2?

That's a very ordinary amp. Base is not tied to 9V. Or am I looking at the wrong plan again?

[antonis]

ny thoughts about (5), Antonis? Can we redesign that filter section to save poor Sprinkle two op-amps? Getting rid of one is pretty easy, but can we lose that buffer off the front too?

It can be done, Tom..
(but in the cost of items values rearrangement..)

After a rough estimation (not enough coffee yet), R24 can be deleted and R25 & HICut pot values should be scaled up 47 times, with C10 & C17 respective scaling down..

[ElectricDruid]

This Q2?


That's a very ordinary amp. Base is not tied to 9V. Or am I looking at the wrong plan again?

Yes, that Q2, but you've clipped off the bit I was talking about. The problem is there's no audio input to the amp since the input cap is fixed to the 9V power rail!

[antonis]

May I locate another drawing error, concerning C15 connection to D2 GND..??

[ElectricDruid]

If we include the x2 gain from U1D into the first Hi-cut stage, we can use R28=R29=10K. That's a simple gain of two for a non-inverting op-amp.

However, the gain and the Q are tied together in a Sallen-Key filter, so we have to change something else to keep the Q the same. The original filter uses a Q of 0.69, which looks to me like a butterworth design value of 0.707 within component tolerances.

Since we need R24=R25 and two equal pot sections, we have no choice but to tweak the cap values. Doing some experiments banging values into the Okawa filter tools, I found a ratio of 4 to 3 gives the required Q (comes out at 0.69, identical to the original).

http://sim.okawa-denshi.jp/en/OPseikiLowkeisan.htm

Putting 10K+100K Pot in for the resistors gives cap values of C17=820p and C15=1.1n. That's a 1n and 100p in parallel.

We could also try Antonis' suggestion of bigger pot values, so try 47K+470k pot. That gives cap values of around 180p and 240p. I don't know how you make a 240p cap.

It's worth saying that the value of both the Q and cutoff seems pretty twitchy w.r.t. the cap values, so tight tolerance caps will be required to achieve the designed performance. Otherwise just audition a few until you find a pair you like - the pot gives a wide range anyway, so it's not totally crucial.

If the x2 gain is not essential, the deisgn becomes easier and less sensitive.

[antonis]

I don't know how you make a 240p cap.

Maybe just 2 X 120pF in parallel..??

[SprinkleSpraycan]

2) Does Q2 really work like that, with the base tied to the 9V rail? Is there a resistor missing in that bit of the schematic?
4) R47/1M doesn't really do anything, because it's in parallel with R13/47K, so only R13 is effective. I suppose it acts to bleed C14 when the switch is open, but that's all it's doing, and I wouldn't expect much of a click from 100n.

R47 drains leakage in C14 so the switch does not pop. (I suspect there should be more switching but it is late.)

This Q2?


That's a very ordinary amp. Base is not tied to 9V. Or am I looking at the wrong plan again?

Q2 was a drawing error. here it is corrected along with the other suggestions implemented.

[SprinkleSpraycan]

May I locate another drawing error, concerning C15 connection to D2 GND..??

C15 should probably go to the 4.5v. going to ground is how its shown on the original schematic. ill make the change on the breadboard. i dont expect this to make an audible difference, so its more for the efficiency of the circuit?

[SprinkleSpraycan]

where the filter comes from by the by

[ElectricDruid]

I don't know how you make a 240p cap.

Maybe just 2 X 120pF in parallel..??

Doh! Yeah, E12 values would do it.

So do you think the Q3 transistor could drive any of these proposed filters directly, Antonis? Worst-case for input impedance will be with the filter's frequency up high (the pot out of the circuit, effectively). If it can, we can save two op-amps.

[antonis]

Honestly, Tom..

I DO think we've more serious design issues here..
e.g. OP claimed for C5 routed to +4.5V.. I can't see neither the reason for doing it nor the role of Q2 stage in such a case..
Also, power supply Schottky diodes should be a single one and R23 doubled after the diode..
(R23 and D4/D5 exchaged positions..)
Now, about Q6 V_CEsat set by D3 forward voltage drop..
I/'m not sure if it can work with Q4/Q5 configuration..

P.S.1
You see, the easiest thing here is to make Q3 capable to drive those filters directly..
(just a bias rearrengement for a lower R10 value..)

P.S.2
As it is, IMHO, it exhibits a very bad (risky..!!) Base bias voltage, strongly dependent on particular BJT V_BE for setting Collector current..

[SprinkleSpraycan]

So like this?

I did go with 2 diodes because it quieted the lfo ticking.