Designing transformer coupled amplifiers with Ge transistors

I have been in the live music game since the 1960s. I certainly know Queen. There are plenty of things I know, and tons I don't. One thing I am not, is a fan. Some guys are amp fans. Like sports fans, they can quote specs, history, etc etc. They can tell you who designed each Fender amp model, they probably know how many grooves are on a Fender volume knob. And they likely know what year that number might have changed. They refer to amps as Red Stripe or Blue Stripe or Drip Edge, because "everyone knows" what that refers to.

Until this thread, I had never heard of a Deacy amp. Oh I like amps just fine, and I study schematics for many reasons. But what year was some amp made? Beats me.

I myself don't think of germaniums as a feature any more than carbon resistors is a feature. I have heard many germanium amps, and they no more all sound "germaniummy" than all tube amps sound alike.
 
Well, did you read my entry #13? I'm still in fond with tube guitar amplifiers, but have to find out if a Ge powered amp with both a PI and an output transformer can approach it or not. And no, I'm not that arrogant to say that I'll be able to exactly replicate the sound of a prominent rock guitarist who after all was ranked at #26 of all guitarists by the Rolling Stone magazine.

As also yet said at the very beginning, my primary intention is to get some clue about the calculation and the design of a phase inversion transformer that drives a pair of (small) power transistors. And I've encountered many really helpful hints so far.

Best regards!
 
This Forum is extremely boring to those "from outside".
I'm told pictures of women making idiotic "duck-lips" are very popular to "those outside". So are pictures of cats, especially if the cat is doing something it shouldn't be doing.

So here you go. I fixed the "extremely boring" forum. 😀

Moderators: note the guitar in the picture, making this post 100% appropriate for the "instruments and amps" forum. 😉

-Gnobuddy
 

Attachments

  • duck_lips.jpg
    duck_lips.jpg
    42.3 KB · Views: 75
  • Cat_on_Guitar_001.jpg
    Cat_on_Guitar_001.jpg
    68.6 KB · Views: 82
Of course!!!!!!!
Just search around and you will see tens of thousands (if not more) amplifier schematics, using all kinds of Technologies known to Man:
Germanium, Silicon, Tubes, Fets, MosFets, even weird IgBTs
Where have you been the last 60 years?

Of course, all trying to do (with more or less success)
the "clean" thing

maybe you missed the:
Hi Fi amps
bit.

Now, Guitar amps, intended to be overdriven (we are talking the power stage here ... of course) .... 99.9% of them are tube powered .... do you have any doubt about that?

And only very very very few SS amps are intended to overdrive, and that adding to Guitar sound, not detracting from it.

Hint:
ONE of main SS power amp problems is that they clip perfect flat top, absolutely erasing any harmonics present.
So erasing Guitar tone and character, and that´s why overdriven SS amps sound dull, lack bite and character.

Nothing subjective (you know I am anti subjective/"magic") but clearly seen and measured.
Here:

overdriven transistor amplifier:

400px-Clipping_1KHz_10V_DIV_clip_A_5ohms-1-.jpg


maxresdefault.jpg


Marshall 20 Watt (tube) well overdriven (as it should 😉 )

OutputsNoBuzzCap.jpg
 
<big snip>
ONE of main SS power amp problems is that they clip perfect flat top, absolutely erasing any harmonics present.
Every solid-state audio amplifier circuit made for the last several decades uses lots of negative feedback. One effect of negative feedback is to make the amplifier more linear - right up to the point when it runs out of headroom. When that happens, the negative feedback makes the output "slam into the wall", hard, giving you the harsh slipping and flat top (and bottom) described by Fahey.

When it's not clipping, all that negative feedback makes the amplifier have extremely low distortion - usually too low to be detected by human ears. This is great for Hi-Fi. But when you play an electric guitar through it, it is [/i]too[/i] clean, giving you a sterile, cold, thin guitar sound.

And so all modern transistor solid-state generic audio amplifiers - whether discrete, chip amp, or class D - sound bad with electric guitar. Too clean when clean, nasty and harsh and fizzy when overdriven. And nothing in between those two unpleasant sounds.

But remember, this characteristic comes from the fact that they all use an enormous amount of negative feedback.

Here is the interesting thing: very old transistor audio amps from the germanium-transistor era didn't have much voltage gain to spare, and designers still thought the same way they did when they designed valve ("toob") circuits. Heavy negative feedback requires excess voltage gain to throw away, and simple valve radio amplifiers didn't use much negative (voltage) feedback, either.

The result is that simple early solid-state transistor radio amplifiers used little or no negative feedback.

The schematic of the "Deacy" amp contains an error (I notated it in pink in the attached image), and also shows that there is a little negative feedback via C14 and the extra secondary winding on the output transformer. You can bet your boots that it was only a very small amount of negative feedback - a few decibels - in order to (a) not throw away too much precious gain, and (b) maintain stability with two cheap transformers inside the feedback loop.

So:

1) Thing One is that we can be sure that the cheap-n-dirty transistor radio amp (that became the Deacy) will have relatively high distortion (not as squeaky-clean as today's SS amps), and, when you start to overdrive it, it will distort more progressively, and clip more softly, than today's high-NFB audio circuits.

These are bad things in a radio amplifier, really. But they are not bad things when you plug an electric guitar in, rather than connect to an AM tuner. The higher distortion will give you a more interesting "clean tone". The more progressive distortion and softer clipping is much better for overdriven guitar, giving you a more bluesy sort of sound, rather than the "wasp in a tin can" buzz you get from an overdriven high-NFB SS amp.

2) Thing Two is that the cheap little transformers will have very poor bass response. This is not a bad thing when you only have a watt or less of audio power, and a tiny 3" or 4" radio speaker to put it through. And if you listen to Brian May's lead guitar tone, it's midrangey, with very little bass or high treble. Deacy fans may like to believe in magic transformers, but the engineering reality is that you can get the same frequency response with a graphic EQ set properly.

3) Thing Three is that the cheap little speaker itself probably had very little treble response. I haven't delved deeply into the details of the original bookshelf speaker used in the build of the original Deacy amp, but most likely there either wasn't a tweeter, or John Deacon made sure not to connect it; he was studying electronics, and had a musicians ears, and intended to play bass guitar through the amp. He would have known that a guitar would probably blow the tweeter if used, and also this was an era when bass guitar amps didn't use tweeters.

Again, this can be replicated with a graphic EQ pedal, this time set after the device producing the distortion.

4) Thing Four is that Brian May didn't plug his guitar straight into the Deacy; he used a Dallas Rangemaster "Treble Booster" in between. You can see the schematic of the Rangemaster here: https://fuzzcentral.ssguitar.com/rangemaster.php

Notice that it uses no negative feedback at all - the emitter resistor of the OC44 transistor is fully bypassed with a big 47uF electrolytic cap.

Without negative feedback, transistors are much less linear than triode valves (tubes), and the OC44 would have distorted far more than, say, the input stage of a (valve) Fender Princeton. It is quite likely that much of the supposed sound of the "Deacy" amp actually came mostly from the Rangemaster in front of it!

5) Thing Five is that we don't know what sort of signal processing was done in the recording studio, after the sound of the Deacy had been captured by a microphone. Was delay added? Was reverb added? Was EQ added? The answers are most likely yes, yes, and yes, but we don't know (at least, I don't know).

6) Thing 6 is that an eye-wateringly expensive recent re-creation of the original Deacy amplifier does not sound very inspiring at all:

Truthfully, I wouldn't even pay five bucks for that uninspiring little buzz. It's not an attractive sound. In fact, if I had a guitar pedal that sounded like that, I would chuck it in the electronics recycling bin!

I'm not much of a Queen fan, so I'm unfamiliar with the original version of that track. Did Brian May's playing sound equally uninspiring and buzzy? Dunno, can't be bothered to find out, but I doubt it; he was an excellent musician and guitarist. He wouldn't put such a drab little sound on a record, would he?

7) Thing seven: here's the same overpriced replica Deacy amp, sounding quite good this time: but, on this track, the sound is very much like David Gilmour's guitar playing with Pink Floyd, and it is quite obvious that the Deacy mic signal has been heavily post-processed with lots of delay, and most likely also some reverb and EQ:

Summing up: personally, if I wanted to try and replicate Brian May's guitar sound, I wouldn't start with an attempt to recreate a Deacy amp. Instead, I'd start with replicating a Dallas Rangemaster circuit, follow that with a graphic EQ pedal to shape the frequency response (reduce bass to match the frequency response of the original tiny audio transformers), follow the EQ with any generic smooth overdrive pedal (a Boss OD3, for instance), follow that with a second graphic EQ pedal to remove treble the same way the tiny original speaker would have, follow that with a delay pedal, then finally a reverb pedal.

I bet you could come quite close to the better Brian May guitar sounds with this chain. If you already play electric guitar, chances are you already have most of the pedals needed to try this already. (Except that most amateur electric guitarists tend not to have discovered the amazing usefulness of graphic EQ pedals, for some reason I cannot fathom.)

On the other hand, if the purpose of the project is really just to tinker with PNP transistors, positive-ground circuits, and a hefty dose of vintage nostalgia, then by all means build your own Deacy amp!

This thread also takes me back to my youth; I was very young when I started building my first electronic circuits, and since I had no money and was much too young to catch a bus to go to the electronics store by myself anyway, at first all my components were salvaged from a dead pocket radio - meaning they were these same metal-can germanium PNP transistors. I remember building an entire AM radio whose audio section was extremely similar to this Deacy amp schematic.

I have very little nostalgia for that sort of circuit, however. I was very happy when fussy, leaky, low-gain Ge transistors were replaced by better-behaved, higher-gain silicon planar epitaxial transistors like the metal can BC107 / 108, and those in turn were quickly replaced by cheaper plastic-encapsulated versions like the BC 147/148/149.

-Gnobuddy
 

Attachments

  • Deacy_Amp_Schematic_001.gif
    Deacy_Amp_Schematic_001.gif
    285 KB · Views: 149
Yes, the little or no feedback detail is a most important one.

Also, any Germanium or transformer iron nonlinearities (remember the driver was run Class A so iron was biased into a non symmetrical area ) were NOT compensated.

No Magic really,Engineering or Physics can explain every bit of it, if you start digging.

As of the Dallas Rangemaster, it was a powerful tool on its own, even if unintended and often ignored by reviewers.
 
Biggest difference between Si and Ge in this type of circuit is the fact that Ge doesn’t sound anywhere near as BUZZY (ie high order harmonics) if they are underbiased. With Ge you HAD to underbias it or it ran away. With Si you can fix that with one miserable diode attached to the heat sink - you could actually maintain an optimal bias. You would get something “better” if properly biased, but far worse and nasty sounding if you don’t. Getting the crossover region exact is almost impossible. Getting the rest of the characteristics similar is quite possible, but requires some non-obvious choices of transistor type. No, you do not want to use 2N3055’s - it sounds closer to the classic circuit with sustained-beta types with a parasitic pole added to emulate the low fT.
 
Every solid-state audio amplifier circuit made for the last several decades uses lots of negative feedback. One effect of negative feedback is to make the amplifier more linear - right up to the point when it runs out of headroom. When that happens, the negative feedback makes the output "slam into the wall", hard, giving you the harsh slipping and flat top (and bottom) described by Fahey.

-Gnobuddy
That is not the effect of the negative feedback, it's the effect of the overall gain while the falling of the feedback gain in hard clipping happens.
Feedback cannot prevent clipping alone.
Anyone can cranck the volume up enough to hit the supply rails or just load a standard 8 ohms output with a 2 ohm speaker ...
Excessively loading a stage gives you the same behavior...Any amplifier with or without negative feedback has the same behavior when clipping.The feedback may affect the signal harmonic spectrum, but that happens just below full saturation while saturation itself is a progressive effect that happens within a small window up until it hits that wall where feedback colapses entirely.
You have some quasi saturation effects depending on the type of transistors the collector currents and the voltage used in any transistor where nothing is perfectly black on white and feedback gain has its variable window to act depending on topology .
The feedback causing output distortions in clipping is a myth...It is the lack of feedback that's causing it with circuits having feedback.
Open loop circuits can prevent bard clipping due to their design to not hit the rail supplies but cannot prevent same clipping effect when overloaded.The VCE is the one that causes hard clipping.
Put a 2 ohm 1000watts speaker on a 20watts 8ohm germanium speaker then tell me how it sounds with or without feedback ! Solid state bipolar devices can't avoid their own gm if enough base current is available.
 
Hi,

most of us might know the »Deacy« amplifier that John Deacon of Queen had built from an old radio and which is responsible for Brian May's signature guitar sound. It's main features are the germanium transistors used in any stage and it's transformer coupled output section. Even phase inversion was done with a transformer.

As there are many Ge transistors in my stashes, I have the idea to design an amplifier with these features. I think the math can easily be done for the output transformer, but I'm completely lost with the PI tranny. How do I calculate the power transistors' drive power demands and the winding ratio?

Best regards!
Chapter 14 of the 'Mullard reference manual of transistor circuits', 1960, may be of interest to you. Here you can read it, on the World Radio History site, wich is an awsome collection by the way.
 
.Except that most amateur electric guitarists tend not to have discovered the amazing usefulness of graphic EQ pedals, for some reason I cannot fathom.
Because most amateur electric guitarists don't know about gain scheduling and clip them. As they're built from high-feedback op amp circuits they sound (to quote Slipperman) "like c*ck and balls".

One day I'll put a 9V to +/-15V convertor + input limiter into one and create an (almost) foolproof GEq
 
That is not the effect of the negative feedback, it's the effect of the overall gain while the falling of the feedback gain in hard clipping happens.
I'm not sure I understand you. But there is no doubt at all that negative feedback makes clipping harder and more abrupt. It can be shown mathematically, and it can be shown experimentally as well.

If you're talking about "slow" op-amps allegedly clipping more softly in guitar FX pedals, that has to do mostly with the slew-rate limit of the op-amp (nothing to do with negative feedback). If the output cannot slew very fast, it also cannot "turn corners" very fast, so the clipped waveform has softer corners, and the clipping sounds less harsh.

There is a small effect due to open-loop gain falling as frequency rises, but it is not a major one. If you take an op-amp with an open-loop voltage gain of 100,000, and then apply feedback to reduce that to 100, the amount of negative feedback is (100000/100), or 1000, usually expressed as 60 dB of negative feedback.

However, the op-amp usually only has a gain of 100,000 at very low frequencies, often only from DC up to, say, 10 Hz (if it's a "slow" op-amp, with a gain bandwidth product of one million in this case.)

In this example, above 10 Hz, the gain falls at 6 dB/octave (this is true of all internally-compensated op-amps, which is almost all of them.) By 1 kHz, the open-loop gain is only 1000, instead of 100,000.

So, at 1 kHz, since we are trying to get a gain of 100 times with feedback, but the open-loop gain is itself only 1000, that means the amount of negative feedback is only (1000/100) or 10x - usually described as 20 dB. By modern Hi-Fi standards, 20 dB of NFB is not very much. But, it is still enough to cause harsh clipping. So, for guitar purposes, it is far too much NFB.

An electric guitar doesn't have any fundamental frequencies higher than 1.32 kHz (and that's only if you have a 24-fret neck, rare these days). Our example op-amp still has an open-loop gain of 758 at 1.32 kHz. If we apply enough negative feedback to reduce that to 100, as before, that means we've just applied 18 dB of negative feedback. This is more than enough NFB to cause hard clipping.

In short, even an old, slow op-amp still has enough NFB at all fundamental guitar frequencies to cause hard clipping, right up to the highest note on the thinnest string of the guitar.
Feedback cannot prevent clipping alone.
I don't understand you. Feedback does not prevent clipping in any way whatsoever. Clipping is caused when the output device (or devices) run out of voltage or current headroom, and go into saturation or cutoff during parts of the waveform. This has nothing at all to do with negative feedback.

-Gnobuddy
 
No error. Point 74 is fed from point 1 via a jack. It WILL be 9V unless you poke a plug (not normal operation).

Yes the drawing suggests an error but it is just not drawn as clearly as we would wish.
Okay. That's not visible in the only reasonable image I found of the circuit, the one I posted. But that image has been cropped quite a bit (not by me).

Do you have a more complete version of the schematic that shows this connection between point 74 and point 1?

I did find a second alleged schematic for the Deacy amp - that one was riddled with mistakes, so I discarded it after a cursory glance.

-Gnobuddy
 
In the absence of negative feedback, due to the tight exponential current/voltage characteristic of a transistor, distortion is very high even for quite small signal voltages.

For very small input voltages, a rough rule of thumb is that you will have 1% THD for every millivolt of input signal; i.e., 1% THD for a sinewave input with a peak value of 1 mV, and so on. Remember that a 1 mV signal is really tiny.

For bigger input signal voltages, the rule of thumb becomes increasingly inaccurate as the signal voltage grows. I've worked out the mathematics that predicts the THD at any input signal voltage (not just very small voltages), but that bit of math is too complex to post here.

The good news is that, while the THD is complicated to calculate, it is very easy to plot the predicted output waveform, and let you see for yourself just how much distortion is created. The output waveform is very far from sinusoidal even for quite small input voltages.

As an example, the attached plot shows the predicted output waveform for a sinewave input signal with a peak voltage of 26 mV. As you can see, the signal is quite heavily distorted. This, remember, is for an ideal transistor with no negative feedback at all, in common-emitter mode, at room temperature.

Modern silicon transistors are close to ideal. Old Ge transistors were less so. For a leaky old Ge transistor, the actual output voltage wouldn't be exactly as shown in the attached figure - but it would be pretty similar. A signal generator, an OC44, and a 'scope would tell the truth.

To put it in perspective, even low-output single coil guitar pickups, on a guitar played with a light touch, will easily put out 26 mV peak voltage. So we can expect the OC44 inside the Dallas Rangemaster used by Brian May, to have been producing as much, or more, distortion than shown in the attached plot.

There is a reason why I chose the rather odd value of 26 mV, by the way. This number equals the constant (kT/q) at room temperature, where k is the Boltzmann constant, T is the temperature of the transistor in Kelvin, and q is the magnitude of charge of an electron (the elementary charge). The term (kT/q) plays a direct role in the transistor equation connecting collector current to base-emitter voltage. It is sometimes called Vt in electronics textbooks.

One thing we didn't discuss earlier in the thread, is that the Dallas Rangemaster has quite a low input impedance. These days it would be considered quite unacceptably low for a guitar input. But the low input impedance will have the side effect of drastically reducing treble response from the guitar, which probably contributes a lot to the smooth distorted sound - without harsh distortion - that Brian May became famous for. Transistor-based versions of the Big Muff Pi (guitar overdrive / distortion pedal) had the same low input impedance, and originally became famous for what was then perceived as smooth, creamy distortion.

-Gnobuddy
 

Attachments

  • CE_Transistor_Predicted_Output_Voltage_26mV_Vp_Input_Sinewave.gif
    CE_Transistor_Predicted_Output_Voltage_26mV_Vp_Input_Sinewave.gif
    23.3 KB · Views: 68
...One day I'll put a 9V to +/-15V convertor + input limiter into one and create an (almost) foolproof GEq
I found Traco Power's family of tiny 1W and 2W DC to DC converters at Digikey just a few days ago. Here is one that will spit out +/- 15 V DC at up to 66 mA when powered by a single +5V DC power supply (USB charger,say): https://www.digikey.ca/en/products/detail/traco-power/TMH-0515D/9383703

Switching frequency is variable but always above 70 kHz, so with a little care in shielding and filtering, one of these looks eminently suitable for powering a few op-amps inside a guitar pedal.

This one costs a few bucks more, but can be run on 9V DC input: https://www.digikey.ca/en/products/detail/traco-power/TMR-1-1223/9382814

-Gnobuddy