Debug of Germanium Buffer

[rockhorst]

After years of absence I decided to try my hand at building effects again. Just finished all the off board connections (hate that part) and tried the effect. Something weird is going on.

Build: AMZ Germanium Buffer http://www.muzique.com/tech/ge_buff.htm
Used an alternative version: http://www.newtone-online.nl/downloads/NewtoneGEBuffer.pdf
Built it on stripboard, according to my own layout:



Just because I like the way they look, I used a big 'greenie' cap for the input. AC128 transistor. Substituted two 1M resistors for the input impedance. All other parts as specified.

Wired it up to a 3PDT switch.
Now, when testing I got the following result: Bypass signal normal, ON signal soft and muffled. Real surprise came when I unplugged the power: the effect signal was unchanged (so there actually was a soft, muffled signal even with no supply voltage). Switch problem or could it be something else?

[PRR]

Voltages?

Emitter voltage might be best first-test.

[R.G.]

I'll have to dig a bit more before I can tell you what's wrong with your implementation.

But I think that the design has a problem being a "buffer". The input impedance is low, not only because of the two input bias devices, but also because low gain germaniums don't have enough gain to prevent being an uncomfortable, treble losing load themselves.

The input impedance of an emitter follower (that's what this is) is about the current gain times the emitter resistor, roughly. In this case, that's low germanium gain, maybe 20-50, since these are devices selected as too low for fuzz faces. So the germanium base of this circuit could be as low as 200K, probably 300K to 400K. Parallel that with, say, 235K for two 470K bias resistors, and you get an input impedance of 132K. This is firmly into the "loses treble" category, even before you parallel in the 1M, which gets it down to 116K.

Maybe you could say it's a "warm sounding" or "brown sound" buffer.  

And Paul's right - what to do when it doesn't work.

[rockhorst]

Sorry, my DMM crapped out on me a while ago, but it's back in business.

Vbias: 9.36V

Q1
C =  0V (connected to ground, so that seems right)
B =  9.13V (should be half?)
E = 9.13V

Output of the effect is about 10dB below bypassed output looking at the VU meters of my DI.

EDIT: checked the voltage dividing resistors. Should be 1M, also according to the color code. Top one reads 10K, bottom one reads 260K. That's pretty consistent with the 0.3V drop from bias to base right? I'm a bit puzzled how this could've happened. Scratch that, shouldn't measure resistance with transistor connected.

[rockhorst]

Checked the transistor, seems like it wasn't making proper contact with the socket before. Now reading:

C = 0
B = 5.9 V
E = 6.2 V

Sound level is still lower than bypassed, and it still sounds rather muffled/bassy.
Might it be useful to get a fresh AC128 after the weekend?

EDIT: measured leakage current (110 uA) and collector current (950 uA), resulting in a very average gain of 93.

[PRR]

Thanks.

> B =  9.13V (should be half?)

With transistor out, yes, half or a hair lower.

With transistor in, base current will load the base bias, and with 1Meg+1Meg at base and 10Meg at emitter the loading will be "significant".

> measured leakage current (110 uA) and collector current (950 uA), resulting in a very average gain of 93.

Not making sense. The ratio of operating current to leakage current is not a normal definition of "gain", though it does give a hint about how marginal the operating point is. And the maximum current possible (transistor DEAD-short) is 9.36V/10K which is 936uA... not sure what your 950uA reading is.

R.G. stated "these are devices selected as too low for fuzz faces." If true, then indeed "9" is a plausable hFE gain.

> Vbias {battery?}: 9.36V
C = 0
B = 5.9 V
E = 6.2 V

Very reasonable voltages.

The 10K emitter resistor has 9.36V and 6.2V or 3.16V across it, 0.316mA or 316uA.

Unloaded Base voltage "should be" 9.36V/2= 4.68V behind 1Meg||1Meg or 500K. Measured base voltage is 5.9V. 4.68-5.9= 1.22V change in 500K or 2.44uA.

The apparent hFE is 316uA/2.44uA or over 100. That's unlikely.

That all said..... I dunno. 300uA/0.3mA into 10K load should "work". Assuming low hFE, it will (as R.G. says) load the guitar like a 50K-100K resistance. Bass and some mids will come through fine, the top-end of guitar will be weak. Maybe that's how it's supposed to be?

Just for fun: can you change the 10K to 100K? It might work a bit better, it might stop working at all.

[rockhorst]

Thanks for all the time PRR.

For the measurements, I followed the instructions from Small Bear (point 24): http://www.smallbearelec.com/HowTos/FuzzFaceFAQ/FFFAQ.htm

As I understand it, the circuit was designed for 'devices selected as too low for fuzz faces', as Jack Orman wanted to do something useful with a pile of left over AC128s. However, I just went out and bought a random AC128, not necessarily out of FF spec. I read that he used AC128s with leakage currents of 300 or even 500 uA. Way more than the 100uA I measured. Seems unlikely that you can have a transistor performing too well in this circuit?

I'll try the 10 -> 100 K suggestion soon (no time today). If anyone has other suggestions or experiences with this buffer? Suggestions for other simple buffers welcome as well (basically want to use it as a line driver at the end of the pedal chain).