18V Electric Mistress with reworked LFO & VCO

[rankot]

Why do you have to run the clock at 12V at all?

[DrAlx]

Why do you have to run the clock at 12V at all?

The LFO/VCO gives a different sweep range/ratio at 9V.  It was easier to take what I had done at 12V (a time-consuming piece of work BTW) and tweak one resistor value rather than rework everything I did two years ago.

[Aurae]

So is it not better to run just on 12v DC and use a 9v regulator ?

[DrAlx]

If you have a regulated 12V supply then of course you don't need to use the on-board regulator, so you could leave that part out of the board and use a jumper wire. I don't have a regulated 12v supply.

[Aurae]

I'll do the first test with the 12v supply and compare with my original mistress. I juste need a 9v reg wich is coming tomorrow. But it's really strange that the 12v reg blows up...

[diffeq]

Better still, is to use to use BBD clock itself to run a voltage multiplier (doubler@9V, x3/x4 @5V ), followed by regulator/cap multiplier  filtering. Solves the noise issue.

Sounds good, could you draw a schematic for that?

Here's a schematic.

Done like that, it'd bootstrap its own rail, which is both silly and clever.
Couple of thoughts. D1-R1-D2-R2 do the ORing 9V and the doubled voltage. 12V regulator starts to regulate when Vin is above 14V or so, and before that, it passes 9V through with some voltage drop out. The question is, will LFO-VCO even start oscillating at all at 7.2V, to start doubling? I do not know. Q1-Q2 supply higher current to doubler because clock buffers can supply only 10mA or so. C3-C5 should ideally be non-polarised (maybe several large, like 10uF, MLCCs in parallel), and that might need to be adjusted to clock frequency. It could probably be made to work, but the time you're done it makes you wonder why not try simpler ways. That doubler can develop stability issues too, mind you.

Why do you have to run the clock at 12V at all?

VCO frequency depends on power supply voltage. With the regulator you can be sure the sweep range stays the same with varying PSU voltages. One way to solve it is to change the current source in VCO, make it constant and run off 9V. Second is to use charge pump and excessive rail filtering. All of it requires time to test and make it work (it may not even after that), which is DrAlx tries to prevent by building on what time was invested in already.

[DrAlx]

I removed all the ICs in the 12v circuitry. With a measured power supply of 13.8V the regulator was warm even though there was no LFO/VCO. So then I measured voltages across all resistors in the 12v circuitry, calculated current in each one. Even if I sum all the currents (i.e. pretend all resistors are in parallell) I get a total current of about 2mA in the 12V circuitry without ICs. Such low current draw does not explain the warmth I am seeing on the 12V regulator. If I have a short then it is a strange one because the circuit still works and voltages all look ok.
Could the problem be related to the fact I have 2 regulators? I assumed that putting them in parallel was better than having the 9V reg be supplied by the output of the 12v reg.

[diffeq]

I removed all the ICs in the 12v circuitry. With a measured power supply of 13.8V the regulator was warm even though there was no LFO/VCO. So then I measured voltages across all resistors in the 12v circuitry, calculated current in each one. Even if I sum all the currents (i.e. pretend all resistors are in parallell) I get a total current of about 2mA in the 12V circuitry without ICs. Such low current draw does not explain the warmth I am seeing on the 12V regulator. If I have a short then it is a strange one because the circuit still works and voltages all look ok.
Could the problem be related to the fact I have 2 regulators? I assumed that putting them in parallel was better than having the 9V reg be supplied by the output of the 12v reg.

Could be that. Even opamps shall never get outputs connected together directly, only through some resistance. Otherwise output will 'fight' and it will seem like a short circuit to each of them. So they'll heat up a lot. Inside of 78L12 is something like an opamp so it won't be different. It can supply up to 100mA, so one is enough for this circuit.

[rankot]

Maybe you shall limit the current at LM311 output or you shall limit CD4050UB outputs? Did you check for pin shorts?

[DrAlx]

I don't think it is a short.  No short is visible on the board and all the voltages measure as expected.  The circuit works remember.  It's just that the regulator is warmer than I expected.

I removed all chips from the board and also removed the 9V regulator.  I took voltage measurements (in red below) on the resulting simplified circuit, and used those to calculate currents (in blue below).  The 12V regulator was still warming up as before with this set up.



Voltages are reasonable and total current draw is 0.47mA + 0.49mA = 0.96mA.
The measured current into the regulator (using a multimeter) is about 10.5mA.  It actually starts out at around 9.5mA and then slowly increases which I think is the result of the regulator heating up towards its final temperature.

The regulator datasheet says the regulator quiescent current is about 6mA but I don't know if the test circuit they used to measure that for the datasheet means I should expect the same sort of quiescent current in my example or larger.
Assuming quiescent current is 6mA, then I am still short of couple of mA.  My scope and multimeter are not great quality so I don't know what the measurement errors are like.

The strange thing is that the regulator on my SAD1024 build doesn't get warm (even when I increase the clock rate to match the 3207 circuit).  It's the same LFO/VCO remember.  So maybe I have a bad regulator on my 3207 build. Unfortunately my soldering iron broke in the process of desoldering (handle actually snapped off and I was lucky not to burn my other hand) so my work on this is done.

[duck_arse]

The regulator datasheet says the regulator quiescent current is about 6mA but I don't know if the test circuit they used to measure that for the datasheet means I should expect the same sort of quiescent current in my example or larger.

look on that same datasheet for the standard connection circuit, whatever they call it. it will show an input capacitor and an output capacitor. your output cap is 100uF, C19. what value does the datasheet recommend?

[DrAlx]

Sheet recommends minimum of 0.01 uF.  I had already replaced C19 with a 0.1uF ceramic since I was wondering if I had damaged the electrolytic that was in there before.  The circuit works OK with 0.1 uF ceramic for C19 ... and regulator still gets warm.

[Aurae]

Hi Alex! Do you by any chance have the layout for veroboard? Thank you.

[DrAlx]

Hi Alex! Do you by any chance have the layout for veroboard? Thank you.

No.  I went straight to PCB on this one.  It was the only way to make it fit a 1590B.

[DrAlx]

Fixed the problem. I removed the regulator and 2 caps from the PCB and put them on a breadboard for testing with a dummy load. The regulator worked fine at 20V with no overheating.
So I ran jumpers from the breadboard to the PCB and that was fine too. Resoldered the parts to the PCB and it was then working fine.
So I reckon I had a short of some sort that was fixed by desoldering/resoldering.  Given that the voltages I measured before all looked correct, I am guessing the short must have been between the regulator output and ground. The circuit now draws 20mA (measured at supply into the board).

I think the circuit was probably fine when I first built it, but then I accidentally shorted the regulator to a pot casing when the pedal was boxed.  Replacing that blown regulator is what gave me the subsequent problem that took so long to resolve.

A few other things I noticed/learnt in the process...

1) C18 can be lower.  33uF is fine.  Datasheet minimium is 0.33uF.

2) C19 and C21 can (and probably should) be replaced with a MUCH lower value. It worked fine with a 100nF ceramic.

3) The circuit works fine with C20 removed and R33 shorted.
C20 and R33 are there because I had them in the SAD1024 circuit that started this thread.  I included those parts in that circuit to stop LFO ticks making it onto the audio supply.
It was an overly-cautious measure in the SAD1024 circuit (i.e. probably not needed) but in this MN3207 conversion there really is no point in having them at all, since the LFO is fed by a different regulator to the audio.

[Aurae]

Ah too bad... I only have pcb adapted to 9v so it's very complicated to put two separate power supply circuits.

Can you send me one of your PCB in France ?

Oh and why else do you use the TL062 instead of the LM339 as in the original ?

[DrAlx]

Yes I can post to France. PM me.

The original 339 based LFO/VCO is unforgiving on layout and noisy.  The whole point of this thread was to show how you could make an LFO/VCO that has the same behaviour as the original but without all the problems.
Also the 339 is too slow to give the doubled clock rates required to drive the MN3207.  I tried that as an experiment many years ago and couldn't get an acceptable sweep.  Either the top end of the sweep would disappear or the bottom end would.

[DrAlx]

3) The circuit works fine with C20 removed and R33 shorted.
C20 and R33 are there because I had them in the SAD1024 circuit that started this thread.  I included those parts in that circuit to stop LFO ticks making it onto the audio supply.
It was an overly-cautious measure in the SAD1024 circuit (i.e. probably not needed) but in this MN3207 conversion there really is no point in having them at all, since the LFO is fed by a different regulator to the audio.

In fact, R33 should ***definitely*** be shorted.  That is because Vlfo (the supply voltage to the LFO circuitry) also feeds R26 for filter matrix mode.  R33 stops the regulator from keeping Vlfo at a steady value.   In other words, the LFO sweep will cause Vlfo to vary, that means that even in filter-matrix mode there will be a variation in the CV applied to the VCO. 
So unless R33 is shorted, filter matrix mode will have a small sweep.

[DrAlx]

I corrected the schematic to show changes to the regulator section, and replaced "Vlfo" with "Vdd".



EDIT: I have updated the build document too.

[Aurae]

Ok I MP you