Question from an irrelevant
On a heatssink with a thermal resistance of 0,35K/W, are mounted 4 pairs of MJL21193-21194 in full complementary symmetry. All emitter resistors are 0,22Ω non inductive type. The voltage drop accross each Re is 11mV. With a dummy load of 8Ω connected in output and with the input of amplifier shorted in GND:
After one hour of preheating the heatsink it reaches a temperature of 49,5 to 50 degrees Celsius and remains stable arround this value for the rest of time.
The Re remains relatively stable, as fluctuates from 10,5 to 11,5 mV.
In which class is biased this amplifier?
Thanks for any explanation.
Fotios
On a heatssink with a thermal resistance of 0,35K/W, are mounted 4 pairs of MJL21193-21194 in full complementary symmetry. All emitter resistors are 0,22Ω non inductive type. The voltage drop accross each Re is 11mV. With a dummy load of 8Ω connected in output and with the input of amplifier shorted in GND:
After one hour of preheating the heatsink it reaches a temperature of 49,5 to 50 degrees Celsius and remains stable arround this value for the rest of time.
The Re remains relatively stable, as fluctuates from 10,5 to 11,5 mV.
In which class is biased this amplifier?
Thanks for any explanation.
Fotios
Hello Fotios,
From the Vre and Re it appears that each pair is biased at 50mA, so 200mA total. Thus the amp can output 400mA peak in class A. This is 280mA RMS (for a sine wave). 280mA RMS in 8 ohms is 2.24VRMS, which is 0.63 Watt. Therefor the amp works in class A up to 0.63 Watts, and for higher outputs goes into class B.
Makes sense?
jd
Hey there!
Jannemans calculations are correct.
But from Fotios temperature measurement, you actually can derive pretty much.
49 dgr. C @ 0,63 Class A watts is oretty hot, but try to imagine the cooling requirements for 100 watts of class A then. And then think of 2 channels in the same enclosure.
What is also relevant is, that amplifiers with rated power output @ i.e 100 Watts pr. ch. of which 20 are biased in class A, runs a lot hotter than a traditional class A amp rated @ a fixed power output of 20 Watts.
Once you´ve realised the needs of cooling, you´ll se how rare class A operation in power amps is. Also a lot of amps claimed class A operation are not.
Jannemans calculations are correct.
But from Fotios temperature measurement, you actually can derive pretty much.
49 dgr. C @ 0,63 Class A watts is oretty hot, but try to imagine the cooling requirements for 100 watts of class A then. And then think of 2 channels in the same enclosure.
What is also relevant is, that amplifiers with rated power output @ i.e 100 Watts pr. ch. of which 20 are biased in class A, runs a lot hotter than a traditional class A amp rated @ a fixed power output of 20 Watts.
Once you´ve realised the needs of cooling, you´ll se how rare class A operation in power amps is. Also a lot of amps claimed class A operation are not.
Well, in theory, from Fotios' data on bias current and temperature and heatsink data you probably could calculate the supply voltage. It is the supply voltage * the bias current that determines the dissipation and thus the heatsink temp.
But I leave that as an exercise for the reader 😀
Yes! The 100W amp has a much higher supply voltage but the same bias current for 20W in class A, so would dissipate a LOT more!
jd
Last edited:
Hi,
I agree with most of the statements here.
Regarding the blameless optimal bias theory, obviously based on some obscure measurements and software simulations, confessedly, I have extremely limited trust in that kind of approaches. One of the main causes of bipolar transistor nonlinearity is the forward-biased base-emitter junction, giving much inconvenience in transconductance stages. Emitter resistors improve the base-emitter linearity and the poor, strongly temperature dependent Vbe/Ic relationship, but for switching devices they also prolong the switching time worsening crossover distortion. For instance, the turn-off time of the double Darlington output transistor is about three times longer than that of a single transistor.
I agree with most of the statements here.
Regarding the blameless optimal bias theory, obviously based on some obscure measurements and software simulations, confessedly, I have extremely limited trust in that kind of approaches. One of the main causes of bipolar transistor nonlinearity is the forward-biased base-emitter junction, giving much inconvenience in transconductance stages. Emitter resistors improve the base-emitter linearity and the poor, strongly temperature dependent Vbe/Ic relationship, but for switching devices they also prolong the switching time worsening crossover distortion. For instance, the turn-off time of the double Darlington output transistor is about three times longer than that of a single transistor.
Thanks Jan for the explanation. Yes, it seems to be right. Indeed, i have biased the amplifier output transistors at 50mA per device. Now, according to your calculations and by taking into account the reply of the dear Kurt von Kubik (thanks my lord) i have resulted in this:
If i want to have the 20Wrms (from its total 180Wrms/8Ω) of the amplifier in class A, then the Iq must be increased at 300mA per output device. Right?
The problem it is that; it is very complex (and probably very rough) the calculation of a heatsink with the appropriate thermal resistance for a proper thermal dissipation. By taking the risk to make a rough calculation, i need a heatsink 6 times bigger from this which i reffer (0,35K/W : 6 = 0,06K/W!). I don't think that this calculation it is right! I have the Fischer catalog front of me and i can see the curves of K/W vs Length of different big heatsinks.
Anyway, thanks for the lights 🙂
Fotios
Don´t leave that for the reader, since also manufacurers do not know.
You´ll find amps with partly class A specifications which imposibly would keep their heatsinks below 90-100 dgr. C, and then you even sometimes can choose a silver design version, where also the heatsinks are silver, thus loosing around 20% of their cooling capacity. 😀
Hi janneman,
The amp can output 400mA peak in class A when amplifier load is.... ?
You should specify the load amplifier when you refer to the 400mA peak, because when we say "peak value", we refer to the transient operation mode. I think you again mixed operating modes of the amplifier (stationary=unloaded running and transient=loaded running)!!!
The power of resistance R2 is dependent on its value and current. It is not correct to refer to the power dissipated by the resistance to classify the amplifier in class A, AB or B. ...
janneman, you have plot output static characteristic Ic=f(Vce) for the transistors MJL21193-21194? Only then you can determine the current field Ic=f(Vce) transistors can function in class A. And after that you can calculate resistance and power dissipation of the Re (in the stationary and transient audio amplifier operating mode), for function to the amplifier in class A. I have not checked your calculations, but I guess not apply to transistors MJL21193-21194.
So, in all this philosophy about classes audio amplifiers, matters a lot and characteristics of output transistors.
The amplifier can't operate in class A using any transistor and the same resistance Re. Matters a lot here to know the characteristic output amplfier transistor to say with certainty that.
Regards
Yes; close enough for government work 😉
Yes, it should be huge. I guess your supply is about 55VDC? If you would limit your design power to like 100W/8 ohms, you can use a lower supply voltage of say 44VDC and the dissipation problem will be less, but still substantial.
jd
No, this has nothing to do with any 'transient mode'. Peak current is just that, peak current. It can be continuous or now-and-then, but the 400mA is the max in class A because at that point the 'other' device is just cutting off.
The amp can deliver 400mA peak in any load in class A, as long as it does not clip on the supply of course. Evidently, 400mA in 1k would require a 400V supply so that is not realistic. 400mA in 16 ohms is only 6.4V so that is no problem. 400mA in 1 ohms is also not a problem. Don't make it too complex, it's much easier than you think 😉
jd
Donpetru, I have no idea what this R2 is. I have no idea why anything about the transistor has anything to do with this. You can use any transistor and any Re, for class A the important thing is how much current can be output before one of the devices cuts off. This is determined by the bias current in the output stage, and you can set it by manipulating the bias voltage and the Re value. As I said, you make it too difficult!
jd
Your statement leads me to believe that you obviously have not read &/or understood the detailed analysis that lead to this 'blameless optimal bias theory'.
OTOH, if you did, kindly make us aware of the flaws in that analysis.
jd
😀😀😀 Indeed, we have in Greece such a work. General elections in 4 October.
Yes, you have got it almost! My supply it is +/-58,5Vdc at 500VA for two channels. Ah Jan, you know me pretty much 🙂
I have two more questions, but it is time for lunch (14.25 in Greece) and the people is screaming from the dining room 😀
I will post them in the afternoon.
Bon apetite to you and to all in Europe!
Fotios
janneman, I mean Re, not R2. Sorry, a typing mistake.
I_bias amplifier is set in stationary operating mode.
"Peak current" characterizes a current signal (exemple: sine). This signal appears only in transient audio amplifier operating mode and the bias current is set in to the stationary audio amplifier operating mode.
I think you refer to the amount of peak current that can record, in certain circumstances, in transient audio amplifier operating mode, in class A. Isn't good to generalized, because this is not reliable classification of an amplifier work in class A, AB or B.
I_bias amplifier is set in stationary operating mode.
"Peak current" characterizes a current signal (exemple: sine). This signal appears only in transient audio amplifier operating mode and the bias current is set in to the stationary audio amplifier operating mode.
I think you refer to the amount of peak current that can record, in certain circumstances, in transient audio amplifier operating mode, in class A. Isn't good to generalized, because this is not reliable classification of an amplifier work in class A, AB or B.
@Donpetru!
As Janneman wrote: Peak current is the actual value of current delivered to the load, which could be anything. RMS current is the peak current divided by sgrt. 2, and is only a measure to compare the AC current to an equivalent DC current.
I.e the load will recieve 400 mA but only on the top of i.e. a sinusoidal wave, to compare the sinusoidal wave to DC, you have to divide the max value with the RMS factor sgrt. 2.
You see! Nothing at all about transient currents.
As Janneman wrote: Peak current is the actual value of current delivered to the load, which could be anything. RMS current is the peak current divided by sgrt. 2, and is only a measure to compare the AC current to an equivalent DC current.
I.e the load will recieve 400 mA but only on the top of i.e. a sinusoidal wave, to compare the sinusoidal wave to DC, you have to divide the max value with the RMS factor sgrt. 2.
You see! Nothing at all about transient currents.
The only way an amp can be classified is if the load is specified to.
When that is done, the bias has to be set @ 1/2*peak current, then you have a class A amp at the load specified.
If you want to force your amp out of class A it can be done in two ways, either by increasing the load or driving it into clipping.
This works for pure class A amps only, for amps with parts of their output running in class A things are different.
When that is done, the bias has to be set @ 1/2*peak current, then you have a class A amp at the load specified.
If you want to force your amp out of class A it can be done in two ways, either by increasing the load or driving it into clipping.
This works for pure class A amps only, for amps with parts of their output running in class A things are different.
Kurt von Kubik, I understand. Is very correct: "the bias has to be set @ 1/2*peak current, then you have a class A amp at the load specified".
I think the word "transient" is different interpretations for everyone (who posts here). When you quote: "Peak current is the actual value of current delivered to the load, which could be anything", it is transient audio amplifier operating mode.
Regards
I think the word "transient" is different interpretations for everyone (who posts here). When you quote: "Peak current is the actual value of current delivered to the load, which could be anything", it is transient audio amplifier operating mode.
Regards
Last edited:
No! Peak current is peak current, and that refers to the top of the sine wave @ its max. current @ 0 dgrs face shift, and nothing else.
Transient, to me at least, is sudden changes in signal magnitude, but they´ll have to be within our class A current limits, to be reproduced in class A, and they´ll have to be within our voltage limits not to be clipped.
If you have a load that varies with sudden signal level changes, you could force your amp out of class A @ lower voltages than presumed, mostly as a result of reactive loads, but then again, that´s the size of the load @ a certain time, and nothing else.
I don´t really see what the point of "transient mode" of class A is.
What I do see is that you might connect a load to your amp, that easily forces your amp out of class A as a result of reactance. Matti Otala showed this decades ago, where he proved that some 8 Ohm speakers surged more than 5-6 times the energy of a lab resistor. But that was a result of very poor speaker design by Linn, Tandberg and several others.
That did lead to better design of PSU´s and output stages, but not necessarely to better sound.
Nice Mr. Curl!
Without measuring at all, I have a few experineces about operating classes.
@ very low bias (Class B almost) I experience problems @ low level signals, which makes the decay of instruments unnatural.
@ higher bias it turns out to be somewhat more complex, as it mixes up with signals @ higher levels.
Class A solves these specific problems completely, but a funny thingis, even class A wins if biased further up. So i.e. class A bias @ 8 Ohms gets even better if bias is set to class A @ 6 Ohms etc.
And from there it goes on just until the firedepartment knoks on your dor. 😀
Last edited:
- Status
- Not open for further replies.