germanium line amp, modify for variable gain?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

diggy fresh

Well-known member
Joined
Mar 19, 2014
Messages
107
Location
Montreal, Canada
Hi peeps, here's a cool little  germanium line amp i want to clone for historical reasons.

It is a McCurdy AT-218. from the datasheet ( thanks Byron!), it says it is a "mixer amplifier" intended to be after a 600 Ohm fader.  It has 35db of gain. also stated is "active impedance matching input",  low output impedance and may be used with input/output transformers.

It is a common emitter amplifier followed by a emitter follower. 

I am still learning and would like to know if it is possible to mod it for variable gain? and how?  without affecting the circuit too much (impedance etc.)  It says the R2 1k trimmer will " alter the gain +/- 1db without seriously altering the input impedance.
If i changed it for a bigger trimmer, or a pot, could i vary the gain more?  any adverse effect?

i'd like idealy to be able to change the sound and add distortion etc, kinda like the Chandler Germanium.

If not possible i could always use an attenuator before or after but a way to vary the gain directly would be cool, no idea how Chandler does it.


some specs:
frequency response : 30hz to 20k
input noise : -120 dbm
source impedance : 600 ohm, unbalanced
input impedance : 600 ohm
load impedance:  600 ohm, unbalanced
output impedance : 75 ohm
input level : -40 dbm
power requirement : -36v,  10 ma


Any help would be appreciated

John
 

Attachments

  • McCurdy-AT-218-schematic.jpg
    McCurdy-AT-218-schematic.jpg
    61.7 KB · Views: 215
Probably the best way of doing so is changing the feedback network resistors R7 and R1 ratio will set you gain but it will also change the input impedance. Also in this configuration the source impedance will have quite a bit of effect on the gain. Other thing, depending on the gain of the first transistor the input impedance would be pretty low and the amp is inverting the polarity, so if you add an input transformer you probably want to wire it inverting it again, so you don't get the inverted polarity.

JS
 
I think the combination of R7 and R1 only set the dc conditions at the base of the first transistor. The NFB through R7 acts to make a virtual earth at the base. The gain is therefore R7 divided by the source impedance. if this is 600R as in the spec then the gain is 39K/600 = 36dB. In practice is it a bit less than that because the open loop gain is not infinite.

As the dc operating point and the closed loop gain are both set by R7 then you cannot really vary R7 directly to alter the gai. You could reduce the gain by connecting an RC in parallel with R7 to reduce its impedance in the audio band.

If you leave it as is and use a 39K series input resistor you have a nice unity gain germanium virtual earthe mixer.

Cheers

Ian
 
Thanks alot guys!! very good info! transistors are still a mistery to me. i didn't know about the phase inversion also.

i'm just starting to learn about transistors, so this is still black magic for me, very complex to understand.

So i'm guessing the best way would be to use an attenuator (like a 600 ohm t-pad) at the input or output, or both.

btw, the transistors Hfe are :  120-150 for the first transistor (2n508a)  and around 30-60 for the second one (2n1925)
If it helps..


taken from the datasheet:


"The voltage drop across R6 provides correct base bias for Q2. the bias for Q1 is obtained
from the voltage divider R7 and R1, with R7 obtaining a negative voltage from the drop across R9
. This arrangement ensures dc stability even at higher temperatures, since the effects of
leakage current in one stage tends to cancel that in the other stage.

Current feedback is applied to Q1 emitter by the network comprising R2, R3, and R4.
Voltage feedback is applied to Q1 base, from the output load, by R7. The two forms of
feedback combine to set the input impedance close to 600 ohm, R2 provides approximately +/- 1 db
gain variation without seriously affecting the input impedance."


Also,  this is a mid 60's design, would upping the capacitance at certain spots help?

Thanks
John
 
It is a 1-transistor amplifer (plus buffer). There is only so much you can do with it.

Stock, the input impedance is like 500 Ohms.

If the source impedance is less, the forward gain is about 77, there is little NFB, the THD will be 4% near maximum output. The maximum output is around 8V peak or +17dBu.

If the input impedance is greater, such as one 10K mix-resistor, the gain approaches 39K/10K or almost 4, THD falls near 0.2%.

If the input is multiple mix-resistors, say eight 10K, the gain from each input is almost 4 but THD goes up to 2%.

I strongly suspect it is meant to be worked far below +17dBu. +3dBu will give roughly 1/4 of the above THD %s. So an 8-in mixer gives near half-percent, which is barely tolerable for old broadcast. Maybe the design level was even lower.

> load impedance:  600 ohm, unbalanced

It won't drive 600 Ohms with authority. Peak output is 2.5V into 600 Ohms, and THD will be higher.

But whatever. Put your SM58 in the front, the gain is acceptable for many uses. Pad it down after if a little too hot, pad it down in front (5K series resistor) if you are miking large drums or amps.

> no idea how Chandler does it.

They design it for variable gain. It may have more parts in it.
 
Thanks Prr! Amazing analysis! very interesting! 

I wasn't planning on using it as a mic amp, I am cloning it because i found a Mccurdy AT-220 mic preamp with 30db of gain ( which is supposed to be followed by a 600 ohm attenuator and a booster amp, the At-218 (this one).

Indeed it is supposed to come from a Canadian broadcast console mixer from the mid sixties.


How will this circuit interface with modern line level gear though ??  beeing driven by a 50 ohm output ( from a soundcard)  and driving a modern line input?  or should i add transformers? 1:1 ?


I'm toying with the idea of building two of these with a 600 ohm T-pad between them, to have some kind of line level germanium distortion box.  (maybe with other pads at the input/output. transformers etc.


 
put a variable  heater  in there,

gain will drop with heat,

like the old Arbiter Fuzz Face circuit that quits working when it gets hot,



 
diggy fresh said:
How will this circuit interface with modern line level gear though ??  beeing driven by a 50 ohm output ( from a soundcard)  and driving a modern line input?  or should i add transformers? 1:1 ?

You will need to add a series resistor at the input to set the gain.  39K will give unity gain, 10K will give about 12dB gain. A 25K pot in series with 12K resistor will give you just over 10dB range of gain adjustment. If you want a balanced input you could precede this with a 10K:10K transformer.
I'm toying with the idea of building two of these with a 600 ohm T-pad between them, to have some kind of line level germanium distortion box.  (maybe with other pads at the input/output. transformers etc.

As PRR mentioned, it will not drive 600 ohms. If you want to cascade two of these then you could use the same series resistor technique to vary the gain of the second stage.

Cheers

Ian
 
Thanks guys!


But what about a modern load ?  Like a soundcard input?

I was confused because the datasheet says " the amplifier operates into a 600 ohm unbalanced load"  .  Then in the maintenance section it says : connect a generator set for 600ohm unbalanced, -40dbm (7.7mv)".  then it says: " the reading across the 600 ohm load should be
-5dbm (0.44v) , a gain of 35db.".

So i guess -5dbm (0.44v) is not realy "enough".  But i'm interested to know how would it drive a modern line input.

@ian thanks for the input tips, shows my ignorance about transistors circuits, i learn!

 
A 'modern' line input will look like 10K at least. Your germanium preamp will have no problem driving that. Its just low impedance load like 600 it will have a problem with.

Cheers

ian
 
Any news? did you end up building it?
I just got a whole lot of germanium transistor, including some OC44, 72s with heat sinks etc and I thought this amp could be a nice passive EQ makeup amp? What do you think?
 
Germanium circuits are very fickle and IMO only useful with certain circuits matched to a particular transistor.

But first, you have to really look carefully at the transistors you have. I have probably about 200+ Germanium transistors. Some of them I paid maybe $60 USD for two and they were completely useless. The biggest problem is leakage. There are two major groups of Germanium transistors. There are the original "grown" ones and then there are the Germanium alloy types.

The grown Germaniums almost always have horribly high leakage. At least anything that you can actually buy does. These transistors have been picked over so many times, that everything for sale is horribly and dreadfully high leakage making them basically useless. For a Fuzz for example, you need a leakage of ~40 uA. Note that the Icb values quoted in datasheets and in Ebay sales refer to the base leakage. To get an idea of what the leakage is through the transistor you have to multiply that by the hfe. So if they *claim* the hfe is 80 and the leakage is 1uA, that means the leakage through the emitter is theoretically 80 uA which won't really bias well in a Fuzz [1]. At 80 uA, it will have a chance of working if it's relatively cool. But it has been my experience that the values posted for leakage in Ebay listings are completely false anyway. They probably took measurements with the windows open in the middle of winter. When you take hfe measurements, you have to be careful not to handle them too much or it will take 10-15 minutes for them to cool back down to the point where you'll get a meaningful measurement.

The Germanium alloy types are completely different. The leakage is too LOW. And the bandwidth is too good. At least for a Fuzz. A Fuzz sounds best when the first transistor is in the 40uA range and the particular transistor has limited bandwidth at high gain. Also, it could be that there is some gain droop at high current. These highly non-linear characteristics can only be had by the grown type of Germaniums. The alloy Germaniums are too "good" for a Fuzz. But for something like a Chandler style circuit, they are almost certainly using alloy junction Germaniums. That's the only thing they could use that would give them any kind of consistency. And the alloy types do of course have a lower diode drop which does give them an inherently different behavior from Silicon transistors.

So the class-A type circuit cited might sound good with an alloy junction type. And if you happen to have a genuinely low leakage grown type, it might sound good as well but IMO it would be a waste of that transistor. The circuit cited would be suitable with a medium leakage transistor but a really low leakage grown type with the right gain and limited bandwidth MUST be used to build a Fuzz or Tonebender or something of that nature. Anything else would just be irresponsible because you would be putting a $200 transistor into a circuit that would work better with a $2 transistor.

As for making circuits like this variable gain, you have two options. One is two use a multi-pole switch so that the bias can be adjusted along with the gain because biasing a circuit like this and making it stable over a reasonable temperature range is difficult. But the best way is to adjust the series resistance with the bypass capacitor. For example, if you made part of R5 a potentiometer and connected the bypass capacitor to the wiper, that would give you good gain control. And it would work very nicely in that at lower gain settings the emitter resistance increases feedback and thus linearity. So at low gain you have clean tones and at high gain you might get bandwidth limiting and smoother tones.

In general, regarding adding "dirt" to a signal using circuits like this, there is a very narrow threshold where the circuit will add noticeable character and not sound awful. Tubes are actually more linear than transistors (by a lot - consider transfer func exponent is 1.5 vs 2.0 which is a BIG difference). So you might think tubes are adding dirt but in reality they're actually already clean enough that they don't even need feedback to sound good. So take care adding dirt to things. You should consider that you might be just making a really bad circuit.

[1] I'm sure there are many people that will dispute this but if you have every heard a Fuzz that really sounds good you will understand how everything else is just fuzz with a lower-case f.
 
Wow! Thanks for your feedback (no pun intented!)
I think I'll stick to fuzzes, and yes I have heard loads of okay to crappy fuzzes, and very few great units. But man, when they're good...
Anyway! I just thought I could gave a shot at this circuit since I have ge transistors with heatsinks.. But you're right, I had forgotten about leakage.
Maybe another day!
 
Hi, Just saw the thread was ressurected.

What i ended up doing was just use 600 ohm T-pads from Capi before and after and left the amp with fixed gain, worked like a charm!

I also played with unknown Hammond transformers i had lying around for the In/out ( not sure of the ratios, i tried a few i had and flipped them around until i had a good level/sound) but i think i used a step-down on input and a litte step up on output.

I had the Og specified transistor for this circuit which is a 2n508a, Hfe around 90-200 and no leakage measured on my peak atlas thing.  Surprizingly very low noise,  i made two and used wirewound resistors on one, almost no hiss hearable.


It ended up beeing very cool, when overdriven it has a very cool distortion/compression which is dialable from "clean" to splattery over compressed.

I'd say try it, the circuit is very simple. even if it ended beeing too lo-fi it you might still like it.

 
for using it as summing amp it's possible to use a bootstrap to increse the open loop gain, actually the open loop gain is a bit small for this application.
 
ppa said:
for using it as summing amp it's possible to use a bootstrap to increse the open loop gain, actually the open loop gain is a bit small for this application.
This module is described as a "mixer amplifier", which does not necessarily mean it's a summing amplifier. Rather it seems to be a more or less general purpose amp intended to be used in the construction of mixers.
At the time, summing amplifiers as we know them today (VE summing) were just a twinkle in Jeep Harned's eyes.
Mixing was usually done in the manner that we now call "passive", although it requires an active make-up gain element.
With vacuum tubes, it was commonly done with a general-purpose amplifier optimized for low source impedance via a step-up xfmr. Designers of early SS electronics knew that they were intrinsically suited for low-to-medium impedance, which allowed them to dispense with an input xfmr.
When/if this module was used as a summing amplifier, it would need a line amplifier to be able to drive the customary 600r impedance of the era.
 
by incresing the open loop gain you can have a lower input impedance than 600 ohm, that're the nominal impedance of this amp.
Obvius, it can run with the actual gain the same (and infact it runs), but who would want to have a "more active summing amp" from this stage can use the bootstrap, just a my modest idea  ;)
 
ppa said:
by incresing the open loop gain you can have a lower input impedance than 600 ohm, that're the nominal impedance of this amp.
Obvius, it can run with the actual gain the same (and infact it runs), but who would want to have a "more active summing amp" from this stage can use the bootstrap, just a my modest idea  ;)
I agree 100%, but in the context of the era, I think the idea of putting to market an amplifier with ca. 65dB gain and an input Z of 60r was somewhat too advanced. Many mixers were built according to a cookbook; terminating a mixing network into such a low impedance was too exotic. And so much gain was certainly bound to be noisy, eh?
 
Adding to what squarewave said, I've had and repaired a pretty good sample of 1st generation American germanium broadcast amps, and for the most part the reasons they quickly moved away from those types is valid today.  If an original seems noisy, in virtually no case have I ever found a replacement germanium, vintage or modern, that improved upon it.  Circuits with alloy type 2N404's on the other hand are quiet, and also don't behave much differently than comparable silicon circuits. 
 
abbey road d enfer said:
ppa said:
by incresing the open loop gain you can have a lower input impedance than 600 ohm, that're the nominal impedance of this amp.
Obvius, it can run with the actual gain the same (and infact it runs), but who would want to have a "more active summing amp" from this stage can use the bootstrap, just a my modest idea  ;)
I agree 100%, but in the context of the era, I think the idea of putting to market an amplifier with ca. 65dB gain and an input Z of 60r was somewhat too advanced. Many mixers were built according to a cookbook; terminating a mixing network into such a low impedance was too exotic. And so much gain was certainly bound to be noisy, eh?

but in this case using 22k- 47k res on summing net the closed gain is only a bit more than without bootstrap so the noise is not increased pratically.   
 
Back
Top