Strange regulator circuit from an EH Memory Man prototype

[cjlectronics]

The attached sketch is a reversed engineered voltage regulator circuit for an early EH 7500 Memory Man with Boost switch (economy). Supposedly, this was an early prototype before the transistor regulator circuit seen in the EH7500 memory man circuit (see dropbox link below).  I've searched data books and app notes for something similar in design but have come up with nothing.  After drawing the circuit, it looks like the feedback circuit includes the load of the SAD1024's and everything else.  I've never seen anything like it.

The pedal is very noisy and I thought it might have been this regulator circuit. However, when I pull all of the SAD1024's the voltage cleans up nicely.  I'm still playing with it and thought I'd share this unique circuit with other pedal geeks like me.

Full schematic here https://www.dropbox.com/s/xxi7e8qauxd7r8p/EH_MMecon.jpg

Prototype's voltage regulator circuit


The pedal looks like this (See the prototype image)
http://www.effectsdatabase.com/model/eh/memoryman


CJ

[theehman]

Offhand it reminds me of the voltage regulation used in the Electric Mistress.  The connections are different but the basic idea seems to be the same.

[PRR]

It is a bog-standard Negative regulator.

Re-draw it and you will see.



The negative output is grounded, but that's OK since the PT winding floats.

The funny transistor is used as a 5-cent Zener, pretty reliably 6.2V-7.0V.

The Zener is, unusually, fed from a bunch of pots returned to ground. However Zener current is not critical so this is fine.

Zener sets one input of op-amp (yes, OTA) to 7V below the positive rail.

10uFd cap bypasses some Zener hiss.

Pass-transistor (hidden in OTA) feeds the negative output trying to find a happy-point.

A 10K(?):5K network compares the negative output to the 7V Zener. The happy-point must be 21V across the output.

The OTA is not made for closed-loop operation; however for sub-sonic response rate we can glom a heavy capacitor on pin 1 so it is stable as mud.

If you find 23V raw, then 21V output is too close to going out of regulation. Also isn't 21V seems a lot for any BBD that I know? If I assume the 19K is genuine (perhaps a factory trim for the loose-spec "Zener") and the exact 5.6K value, this gives 13V-15V, which should regulate fine and is plausible BBD power(?).

[PRR]

Why do you have a "ground" (3-line triangle) symbol at pins 6, 4, and PT CT? does it go to case or any common bus? (I bet not.)

Your hand-draw lacks any Main Filtering Cap. This plan must have a big cap fron diodes common to PT CT, on the order of the 100uFd shown in the simpler version.

> very noisy

What is "noisy"?? Hiss, humm, buzz, Cuban radio stations?

Hum/buzz would relate directly to that main filter cap, also the 10uFd across the "Zener". Both are old enough to buy cigarettes without ID, so it is time to replace them without mercy.

Hiss could be that 10uFd. I doubt it, but it is quick(?) and cheap to replace.

If hiss not fixed by that, it isn't the power supply. LOTS of stuff can hiss in a BBD box.

Cuban radio (or cellphone bursts) is not power supply.

[cjlectronics]

Awesome... thanks for the input.

The noise is a 5k Hz whine.  It is everywhere, power rail, signal input, etc...  If I can find the source of this whine and correct it, the pedal sounds great.  I can hear the delays behind the whine really good.

I struggle with where to place my scope reference.  The input and output jacks aren't grounded to the PCB. Should the transformer CT connect to the jacks?

You're right, I forgot to put a 100uf cap at the rectified side of the 2 diodes.

I haven't ruled out a bad SAD1024 because the whine changes when I sweep the delay pot.  This makes me think the negative regulator is fine???

[theehman]

That sounds more like clock noise from the SAD1024A ICs.  Those early units didn't have any companding circuitry and are generally noisier than later units.

[cjlectronics]

The circuit for this memory man is not at all close to the EH7500 circuit in the first post.  The basic block diagram is the same but there is a marked difference.  I am still battling the clock noise in the circuit.  The whine changes as I turn the delay pot.  I pulled the first BBD in the cascade of BBD's and I see the whine at the input to the BBD and at the output of the op amp that drives the first BBD.  Tried swapping the op amp... no difference.

[theehman]

Is the Feedback at minimum?

[Scruffie]

The circuit for this memory man is not at all close to the EH7500 circuit in the first post.  The basic block diagram is the same but there is a marked difference.  I am still battling the clock noise in the circuit.  The whine changes as I turn the delay pot.  I pulled the first BBD in the cascade of BBD's and I see the whine at the input to the BBD and at the output of the op amp that drives the first BBD.  Tried swapping the op amp... no difference.

Can you post some gut shots? I'd like to try and trace the whole thing out.

[PRR]

> Should the transformer CT connect to the jacks?

No. This isn't a 5U4 tube-amp. The PT CT is *not* ground. It is Raw Negative.

"Ground" is surely your <8> symbol... pin 8 of '3094, pin 4 of '4558, pin 7 of CD4047, etc.

Your 10uFd cap and Zener appear to be drawn backward. Top is more positive than the end that connects to 3094 pin 2.

[cjlectronics]

Yes, the 10uf cap was in backward.  I swapped it but it made no difference as far as the clock noise goes.  In the process, I've done a few things that helped. 

1.  Added a .002uf cap across pins 1 and 2 (the feedback loop) at the op amp stage that connects to the input of the 1st BBD.  This helped a lot because I hear no clock noise when the delay pot is set for the shortest delays.  Obviously, because the clock noise is at level above our hearing. The pedal sounds great... warm delays.

2. Added a 3 prong power cord and grounded the chassis.  Now I don't hear the Mexican radio station.

The real problem is when the delay pot is set for longest delays, the clock noise is very loud. I can still hear the warm delays and repeats but the clock noise makes the pedal unusable.  At the longest delay, the clock noise is approx 5kHz. 

How do you filter 5Khz from the guitar signal path without filtering the signal that you want?

[Scruffie]

That's normal I think, I breadboarded the schematic version and the delay was useless at longer delays, there is no companding and not much clock filtering compared to designs that came later so the whine comes through quite clearly.

So not much to do about it apart from use a different design or don't turn the delay up full.

[Scruffie]

I have the very same model CJ worked on in my hands here (opamp and OTA had gone and some past their prime caps) and i've come across something very peculiar.

After fixing it I noticed one of the wires that was sneakily wrapped around the input wire and hidden had broken off in its life with no indication where it once lived but it was connected to the input jacks ground, so I added a new bit of wire and found a suitable ground point and hooked it up.

The pedal stopped working! All I got was some minor farting signal, I probed around and the input opamp reference voltages had all shot up to 11V on a 13V supply, something funny there...

Then I realised the 'boost' switch which connects to the input opamps inverting pin via resistor was also hooked up to the input jacks ground... but had no capacitor to shunt it above ground.

Looking at the output from the blend pot, the pull down resistor from the output cap (it has no input cap too, not unheard of though) was connected to the V.Ref... so I hooked up this mystery input ground wire to V.Ref and suddenly, all that whine and noise went away! Well, with a good tweak of the BBD 'cancel' trim.

Has anyone ever seen anything like this? It doesn't seem right at all to me... I know it's a virtual ground but still, having the enclosure hooked to it.