Hi everyone,
A friend has built the Black Devil Amp (published in the german DIY magazine "Elrad", schematics: link).
The author recommends to select the input transistor for noise and distortion and that is what I am trying to do for him. I got a bunch of transistors, but what to measure? I don't have access to the full amp at the moment, so I want to build test circuits and to use this as an exercise to better understand distortion mechanisms.
Noise
For the noise I would build a simple common Emitter test circuit and use the QA401 to measure the noise level. However, I am wondering if transistor noise makes a difference at all with the given resistor values. --> Does that makes sense?
Distortion
My understanding is that, as the colector-emitter voltage varies, the early effect plays a major role. Hence, selection for early voltage and Beta (as a high beta mitigates the early effect) --> Does that makes sense?
Which other distortion mechanisms could be relevant here? I can imagine that the beta droop could be relevant as the collector current varies as well. But I havn‘t seen any related FOM.
The amp circuit shall not be modified, it's just about selecting the input transistor.
A friend has built the Black Devil Amp (published in the german DIY magazine "Elrad", schematics: link).
The author recommends to select the input transistor for noise and distortion and that is what I am trying to do for him. I got a bunch of transistors, but what to measure? I don't have access to the full amp at the moment, so I want to build test circuits and to use this as an exercise to better understand distortion mechanisms.
Noise
For the noise I would build a simple common Emitter test circuit and use the QA401 to measure the noise level. However, I am wondering if transistor noise makes a difference at all with the given resistor values. --> Does that makes sense?
Distortion
My understanding is that, as the colector-emitter voltage varies, the early effect plays a major role. Hence, selection for early voltage and Beta (as a high beta mitigates the early effect) --> Does that makes sense?
Which other distortion mechanisms could be relevant here? I can imagine that the beta droop could be relevant as the collector current varies as well. But I havn‘t seen any related FOM.
The amp circuit shall not be modified, it's just about selecting the input transistor.
Last edited:
Thanks, I found the thread before but it‘s about circuit modifications and does not answer my questions.
if you own such cool equipment as the QA401, why not build an emitter follower and test a bunch of transistors for noise and distortion. Maybe you will observe something or not.
Time has passed and semiconductor production most likely have improved since the author made that statement.
Here are my results. I used the following test circuit:
I used the QA401 with frequency sweep from 100Hz to 20kHz and measurement stop 80kHz. Input DUT: -4dBU, Output DUT: 8,5dBV
The originally used BC546B of which I tested 12 pieces didn't show any difference, neither a BC550C or a C1845 do:
I measured two very old devices (perhaps 70's?) and compared it with the previous results: (BC170 and BC173 are the old ones)
Even here no remarkable difference. Uce varies from 1.9V to 9.6V, Ic from 1-2.4mA (approx. - depending on device type, determined by LTSPICE)
So, what's the conclusion? Nothing to select with today's devices? I have been lucky with the two tested old ones to not grab crappy ones?
I used the QA401 with frequency sweep from 100Hz to 20kHz and measurement stop 80kHz. Input DUT: -4dBU, Output DUT: 8,5dBV
The originally used BC546B of which I tested 12 pieces didn't show any difference, neither a BC550C or a C1845 do:
I measured two very old devices (perhaps 70's?) and compared it with the previous results: (BC170 and BC173 are the old ones)
Even here no remarkable difference. Uce varies from 1.9V to 9.6V, Ic from 1-2.4mA (approx. - depending on device type, determined by LTSPICE)
So, what's the conclusion? Nothing to select with today's devices? I have been lucky with the two tested old ones to not grab crappy ones?
Last edited:
> what's the conclusion?
Your R3 R2 do not drift much from one test to the next.
They completely (70:1) overwhelm transistor variation. This IS the type of circuit where any non-defective transistor will work about the same, doing what the resistors tell the transistor to do.
I am puzzled how you have a 0.2KHz (200Hz!!) low cut. If that is the circuit you really built, it looks like 20Hz (0.020KHz).
FWIW: I glanced at the Black Devil and agree it too should work about the same with "any non-defective transistor".
Your R3 R2 do not drift much from one test to the next.
They completely (70:1) overwhelm transistor variation. This IS the type of circuit where any non-defective transistor will work about the same, doing what the resistors tell the transistor to do.
I am puzzled how you have a 0.2KHz (200Hz!!) low cut. If that is the circuit you really built, it looks like 20Hz (0.020KHz).
FWIW: I glanced at the Black Devil and agree it too should work about the same with "any non-defective transistor".
Sorry for the confusion, I used 0.47uF instead, that should explain the low cut.
We are looking at THD plots over frequency.
I am trying to understand, why this circuit is so non-sensitive to transistor variation. My suspects were initially: early effect, hfe and beta droop. However, it seems to me that early effect, hfe and beta droop - all in the end affecting beta - do not play a role unless source impedance is too high.
Taking a look on transconductance: Ic is varying from 1-2.4mA, causing the intrinsic Re to vary from 10.8 to 26Ohm, which is negligible in view of the 1kOhm emitter resistor.
Is that correct?
However, one last thing isn't clear to me: The author warns that transistor variations may cause significantly worse performance than possible (THD and noise). Firstly, I don't see that the circuit is sensitive to that and secondly, the article if of the latter 80's where semiconductor processes were already capable of producing high quality components (e.g. the first microprocessor was launched in the early 70's).
Same for noise, no matter which BJT I try, the noise floor is just a few dB above the QA401 noise floor (~-110dB).
We are looking at THD plots over frequency.
I am trying to understand, why this circuit is so non-sensitive to transistor variation. My suspects were initially: early effect, hfe and beta droop. However, it seems to me that early effect, hfe and beta droop - all in the end affecting beta - do not play a role unless source impedance is too high.
Taking a look on transconductance: Ic is varying from 1-2.4mA, causing the intrinsic Re to vary from 10.8 to 26Ohm, which is negligible in view of the 1kOhm emitter resistor.
Is that correct?
However, one last thing isn't clear to me: The author warns that transistor variations may cause significantly worse performance than possible (THD and noise). Firstly, I don't see that the circuit is sensitive to that and secondly, the article if of the latter 80's where semiconductor processes were already capable of producing high quality components (e.g. the first microprocessor was launched in the early 70's).
Same for noise, no matter which BJT I try, the noise floor is just a few dB above the QA401 noise floor (~-110dB).
Last edited:
I would not be surprised if there is some truth in it.
I have read a couple of good ELRAD amplifier related articles but the Black Devil one is kind of weird. The reasoning doesn't sound logical to me, if not even contradictory. E.g. the statement that NFB always comes too late and the conclusions taken out of that.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.