Project Big Marf

[Ben Lyman]

Here's a new vid with all diode combos and readings except one, oops

Here's both sides of the Ge by itself:

[pinkjimiphoton]

ben, i think he means without the ge and with the ge to see if there is a difference in the voltage.

[Ben Lyman]

cool Jimi, I think I got that option covered somewhere in those vids

[Frank_NH]

Thanks for the new videos, Ben.  What puzzles me now is why there is any difference in the DC voltage when the diode is in series with the 470 nF film cap.  The cap should block the DC, and (theoretically) the voltage should be the same (4V) on either end of the diode. 

My LTSpice simulations have shown just that, i.e. that the DC voltage to be insensitive to orientation and diode type.  However, I can simulate your video's findings if I put a 10 - 20 meg resistor in parallel with the cap (leakage?).  This is with a single 1N34A.  I can then get a 0.1V difference in the DC voltage, but only when the anode (black stripe side) points to the collector. Hmmm. 

[robthequiet]

^ @Frank, I'm thinking that Ben and PinkJimmy have magic diodes 

I've been pondering the same question, although I am not quite an engineer -- the differences in voltage are not huge, but the real test would be running a handful of diodes [edit: as AJ said back on previous page w. leakage test] through the circuit and taking the feedback loop out of circuit and building out to see at which point the difference becomes apparent. Since the stage is a gain stage, after all, by reducing gain we reduce upper harmonics. What if we are reducing gain only in the upper bands? Maybe we need to look at the combination of diodes and caps as a filter?

[Ben Lyman]

I think Jim tried to explain to us earlier that my DMM, in that instance, is the only thing applying voltage or something

[Frank_NH]

OK, so the 0.1V difference is a measurement anomaly?  I guess that's possible.  In any case, it's the noise clamping property of the germanium diode that's of interest. 

[pinkjimiphoton]

i still think it has something to do with the ge passing clipping half the wave at half the voltage. it would seem to me that may put the wave itself out of phase by the ge's wave passing thru a millisecond or so sooner.
or maybe it's the compression effect of the diode on the signal.
squashing it enough that the highs and noise get clipped off?

man. my widdle bwain hurts

[anotherjim]

So a DMM has some internal resistance between it's + and - probes. The value ought to be headline data on its front panel, but seldom is. Maybe it's on the back panel. A decent DMM might be 20M ohm or an average one, 10M. Anyway, when you measure any voltage, you are putting that internal resistance across the circuit. Who was it that said "you cannot observe a phenomenon without changing it by so doing" Heisenberger or someone?

In this bit of the BMP, those diodes are isolated at one end by a capacitor and connected at the other end to the collector. The collector is a DC voltage source, the capacitor is not - it blocks DC. Measuring from the capacitor end to 0v is putting that 20M or whatever of the DMM in circuit, and allowing current to flow - it has made a path for the DC on the collector through a diode to 0v.
Measuring the collector to ground shouldn't really alter the voltage by much - the transistor is biased on and the resistances there are very low compared to 20M.

[Ben Lyman]

Thanks Jim, I totally get it now! I looked all over my DMM and it doesn't seem to say anything about that.
I did find some specs on the manufacturer's site, for my model it says, "input impedance for DC: around 10M"

[Frank_NH]

Thanks all!  Learning more from this thread than from my thermodynamics course in college. 

[Ben Lyman]

New schematic, I'm going to box this one with a few sockets. Notable changes are:
C3 can be 470n if C13 is 10n
If C13 is 22n, then C3 needs reducing about 47n
C17 "boost" cap added with a socket, seems like 22n adds a nice sizzle, any more than 100n is a bit overwhelming
D4 Ge will be a socket for further experiments

[robthequiet]

Hey Ben, excellent job! We can save the mystery of the parallel diode for the sequel, Marf Jr.

[Ben Lyman]

Ha, ya.. or maybe "The Little Marf"

Figuring out where to drill holes is always the most nerve wracking part for me.

[anotherjim]

I've got no reason to doubt that it's the Ge leakage working the magic, but...
1. It sound good with a Ge diode.
2. Diode's a bit non-linear in the right direction, so beats adding a resistor or lowering the original 470k feedback.
3. The Ge doesn't need the Si it's shadowing, so it's no more parts.
4. Circuit noise increases with rising temperature, so does diode leakage, the Ge more so. So I'm guessing there ought to be some auto noise reduction when things warm up.
5. Very worthwhile noise reduction - makes the BMP set for full gain a bit more like a Tube Screamer. I was able to get some weird sh!t going by picking behind the bridge with the thing wide open and BMP tone a little to the bass side.

Neat, Nice one Jimi 

And well done Ben - another build on the way.

[Ben Lyman]

so far, so good... except for that one oddball B250k pot. I guess I should grind the post down a little bit.

[Frank_NH]

The build looks good, Ben!

Quick (slightly off-topic) question.  What software do you use to draw your schematics?  I've been debating what to use, and I prefer using LTSpice to generate schematics that I can actually simulate  , but they don't look all that good.  Thanks.

[Ben Lyman]

It's just a free website called schematics.com and after I draw it up I take a screen shot and save it on my desktop. It's free to sign up and all my work is saved on the site so I can open something up and change it around if I want to.

[Ben Lyman]

Big Marf Pi is complete. I used some sockets and then I forgot to put the "magic diode" in when I made the vid but the noise isn't too bad.

[bloxstompboxes]

Unless my colorblind eyes deceive me, that is the same color green I used on a 1590bb for my very first pedal, a TS808. Can't remember the name of it though. I want to say it was a Krylon color, Ben.