Hello again Mr Tiefbassuebertr
thank you very much for you help,if you don't mind,i need your help again
i follow your instruction and also your schematic,i set the bias about 20ma and let it idle about an hour,but for prevent damaging many output transistors,i only put two pairs output transistors in the circuit,and then i playing music from that amp while always check all voltage,everything fine,i let the amp near 3 hours playing music with moderate volume,it's no problem.
after that,i add two more pairs the output transisitor,so the total is four pairs,i'm check again the bias,it's go up to 30ma,i think that still fine,i play the music again while always check all voltage,they still normal.
but after 15 minutes playing music,one of output transistor is short and the emmiter resistor (R 5) 0.47R blown.
i don't know what's the problem,everything (all voltage and bias) look fine at before
would you please,give me your advice or diagnosis?
thank you very much
Now you have a oscillation problem. My simulation software dosn't respect any parasitic effects like parasitic inductivity through long connection wires and through bad layout. Please note, there are transit frequencies >20 MHz, even by the output devices. This means at the same time, you must check your amplifier until this frequency and not only until 20KHz or 100 KHz, as the most engineers try to tell me.
Actually a frequency response measuring until 100 MHz or 1 GHz by checking of an audio power amplifier is neccesary (without any low pass devices outside of the NFB loop). If there are oscillation, there is always a peak in the aera of the upper roll off, mostly between 1MHz and 10 MHz. The normal used square wave method by compex load doesn't guarantee oscillating free.
I would say, the next step is to use higher values of your emitter resistors (attention !!! without parasitic inductivity - keep in mind that you have a Super ß follower, hfe ~100.000) - 10 ohms until 20 ohms instead 0,47 ohms (to protect your output devices in case of oscillation). Check output with an oscilloskop while you play music. If there are oscillation effects, you will it see clearly.
Then you must find and remove the reason for this (hard work, even with help of simulation software resp. computer aided design).
Last edited:
0R47 are conventional values for the output devices.
But adding resistors into the drivers and outputs bases can help to get rid of oscillating problems.
There was 47 ohm in the original schematic for this purpose.
Some self across a 10 ohm resistor in the output and a zobel network too.
Last edited:
i simulated the solid state part of your amplifier, and the results
are simply astounding...
in fact this is an emmiter follower, with a first stage to balance
the high impedance output of the valve stage..
gain is about 0.9 and because of an inadapted bias, quiesent current
of the final stage is about zero, resulting in very high crossover distorsion.
output offset voltage is about 25 mV...
but the best is yet to come : if you revert the two complementary input
transistor, putting the npn in place of the pnp and vice versa, taking care
that the npn collector must be connected where the emitter of the pnp was , so
the transistors are direct biaised, the result is THE SAME..lol..
still a 25 mv output offset , still no quiescent current, and of course crossover
distorsion...it also work ...
since all you need is a current amplification, you can keep the three serial
pair, bias them correctly and drive them with you valve stage..
the first stage become then useless along with the current sources...
input impedance of a triple follower pair is high enough for you valve stage...
are simply astounding...
in fact this is an emmiter follower, with a first stage to balance
the high impedance output of the valve stage..
gain is about 0.9 and because of an inadapted bias, quiesent current
of the final stage is about zero, resulting in very high crossover distorsion.
output offset voltage is about 25 mV...
but the best is yet to come : if you revert the two complementary input
transistor, putting the npn in place of the pnp and vice versa, taking care
that the npn collector must be connected where the emitter of the pnp was , so
the transistors are direct biaised, the result is THE SAME..lol..
still a 25 mv output offset , still no quiescent current, and of course crossover
distorsion...it also work ...
since all you need is a current amplification, you can keep the three serial
pair, bias them correctly and drive them with you valve stage..
the first stage become then useless along with the current sources...
input impedance of a triple follower pair is high enough for you valve stage...
hi. Tomat,
your output stage are getting slow. your pair both are turned on in high current. you need to fasten your pair. Place 82ohm Rbe each final transistor, and double the drivers. You need to manage it from overflow.
Hi onto aban
thanks for your advice
the drivers using 2N6488 - 2N6491,i think they can handle enough current,is it still need to double those drivers?
Hello again
thanks a lot for your help
, ..sound bad for me,my scope can't reach above 20mhz.is it placing paralel L and R on speaker output will help to prevent oscillation?last night i replace burned transistor and emmiter resistor,and i'm lowering the bias till 10 to 15ma with load condition,i play the music again with moderate volume for about 1.5 to 2hours,and everything is fine,but the first five minutes the amp turn on,the sound is bad,maybe i set the bias too low,but after five minutes,the sound going normal
Hi Wahab,
you are right,this amp is really astounding
and also really make me headache
thank you for good advice
hello tomat, nowadays, the concept of your amplifier is original,
driving solid state using a valve as voltage amplification,
all without any negative feedback...
i will send you a schematic of final stage using the power devices of your amp
i added an active offset cancelling using an op amp, the final stage is
class AB ,i think this is the better solution for you as you ve got about
+- 50 V rail...
driving solid state using a valve as voltage amplification,
all without any negative feedback...
i will send you a schematic of final stage using the power devices of your amp
i added an active offset cancelling using an op amp, the final stage is
class AB ,i think this is the better solution for you as you ve got about
+- 50 V rail...
Hello bobodioulasso
yes,that's right,the tube stage is srpp,with output from the anode,it's simply voltage amp
here the schematic...quiescent current is tweaked by adjusting
R19/R20....do it gradually, i think that you know how to procede..
the input capacitance replace the one at the output of your valve stage,
make sure it can stand the valve high voltage ....if you have no such
22 uF, a lower value can be used as low a 2.2 uf, but the value of
0.22 uF AT THE VALVE OUTPUT is too low...
R19/R20....do it gradually, i think that you know how to procede..
the input capacitance replace the one at the output of your valve stage,
make sure it can stand the valve high voltage ....if you have no such
22 uF, a lower value can be used as low a 2.2 uf, but the value of
0.22 uF AT THE VALVE OUTPUT is too low...
yes, it s enough, it will make 1.2 mA collector current for the relevant
bjts...don t forget that it s an emitter follower stage, not a vas...
all three transistors in a rail work in class AB...
the only thing that seems to be forgot is to replace R16 by two
resistor in serial mode and put the common point to ground
through a capacitance of a few uF....
This is a big change to the original schematic.
Triple EF commonly use 330 to 680ohm at this place such as in the original.
Do you want to lower the gain of the whole stage? or What is the idea?
Depending on the lay out, resistors into the outputs base's devices can be necessary to prevent oscillations problems.
Anyway the original Tomat's shematic is rather elegant and has proven to work well in the past.
Why to throw it into the bin?
It's exiting to search for what has turned wrong.
His problem might be a poor temperature regulation...
Triple EF commonly use 330 to 680ohm at this place such as in the original.
Do you want to lower the gain of the whole stage? or What is the idea?
Depending on the lay out, resistors into the outputs base's devices can be necessary to prevent oscillations problems.
Anyway the original Tomat's shematic is rather elegant and has proven to work well in the past.
Why to throw it into the bin?
It's exiting to search for what has turned wrong.
His problem might be a poor temperature regulation...
the only thing i really did was to add an offset cancelation..
for the rest, you can see that i replaced his vbe multipliers
by a strip of diodes.does it make a difference for you?
...i admit that the 10K resistors are somewhat high,
but it can be adjusted to 1K or even 680r...
So, it works with 2 pairs only!
It looks like one of the power transistors you add in a second time had a problem.
...When several output devices are paralleled, it is better to match them in Hfe, for current sharing reasons.
One can measure the voltages across each 0R47 (one side) to chek the sharing of the current.
This is ok.
Though, bootstraping is not an improvement compared to current sources.
Neither diodes are an improvement compared to a Vbe mult.
Ofset cancellation is OK if there have been offset issues before. It can sometimes affect the sounding performance.
To improve the things i would rather see a little bit of local feedback in order to lower distortion figure and increase the damping factor.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.