Distortion Hates Symmetry

Status
Not open for further replies.
Hello all,

I have modified an elektor circuit for some distinct advantages:
1. Replaced the transistors used so that the same can be used anywhere over a wide supply voltage range. I have tested this in simulation upto +/- 70V
2. Added a DC servo circuit to avoid manual offset null procedure
3. Modified input stage to operate only from rail supply for added symmetry

Advantages of this circuit:

1. No Vbe multiplier and no dealing with thermal stability issues due to Class-B output stage!!
2. Very low distortion. Simulation results show only 0.003% !
3. Very generic and symmetric input stage. The circuit to the left of 2.2E in the supply rails can be used to drive any other topology output stage and even paralleled output stage
4. Removing R26 and R27 makes the input stage work as a current driver

Do let me know what you feel.

bimbla.
 

Attachments

Last edited:
1. No Vbe multiplier. So no dealing with thermal stability issues!!
bimbla.

I think that by not having a Vbe multiplier you invite thermal problems!
A Vbe multiplier normally is mounted on the heatsink so there is thermal feedback for thermal stability.
You have two diodes to set the bias, but as far as I can see they are not physically mounted on the heatsink so there is no thermal compensation.
Or do I miss something?

Jan
 
The article first appeared in Elektor #15. This was the July 1976 issue. It has references to the Equin amplifier published in April/May 1976 issue.
The author talks about making the crossover distortion in-audible and completely eliminate Transient Intermodulation Distortion.

The absence of Vbe multiplifier caught my attention.
I put it in simulation. There was a lot of DC offset and so I added DC servo circuit. Upon increasing the supply rails to +/- 70V, the offset went beyond the servo's control and so I have to reduce the original 10M resistors in the servo circuit to 1M and things were ok.

However, I did not like the use of BC546, BC556, BC547 and BC557 and replaced them with 2N5401 and 2N5551 and biased them straight from the rails instead of from 12V.
This brought the distortion down tremendously down to 0.003% at 1KHz. The lowest I have ever seen!!

I will be building a prototype and will only be glad to share my findings.

P.S. The circuit has also appeared in Elektor 300 Circuits. I will upload the original in a couple of minutes.
 
Last edited:
well its even more complicated

1) output stage is modified to be CFP ... In normal cfp circuits you may get away with out a VBE multiplier in the outputs but there should be one in the drivers...

2) even if you have a chance with this circuit can you tell us anything about the Ft of the transistors you used ?

3) your circuit is more or less a triple darlington bias is applied to Q1 and Q2 and then its "expected " that the output stage will follow ...you are forgetting that the output stage has a beta of 2000 and that means that it will be enough to breath over Q1 or Q2 and you will blow the bias to hell ( and that cannot be cured fast enough even with compensation above the outputs )

4) you may check all available schematics based on TIP 142-147 existing everywhere and you will notice that there is miller capacitors in between B and C leads of the output in the range of even 680pf to limit the bandwidth of the device in order to avoid that this transistor will not shelf oscillate since the beta is so high and the device will start to oscillate for no obvious reason . The above is used to darlingtons configured as EFP where the conditions are manageable...In a CFP configuration the output transistor is 10 times more unstable and 2000 times more unstable if darlington .

5) Lets suppose that you overcome all the above with some magical way and we are ready for reality here is what will happen :
-- you are not going to be able to set a desirable bias
--Bias will change over time depending on the temp in the outputs
--you may try to re adjust but this will become endless story
--If bias stabilizes for a minute if you blow your breath over Q1 or Q2 the bias will change dramatically and here you go adjusting again
--Obviously the same will happen if room temperature change by 1 degree
--Even if some how bias stabilizes no one may guarantee that both of the outputs share the exact same temp and have the exact same time response in temp variations ...In a normal transistor this is not a major issue but in a darlington with a beta of 2000 the slightest change will make the transistor go crazy and there oscillation and thermal run away will kill the amp with in minutes


I am sorry to say that but you are another victim of simulation that takes all transistors and all conditions as ideal ... this amplifier if supplied with 35+35 volt will have about 5 minutes of life before blow with no signal or load in 70+70 will be explosion without questions asked and within seconds
 
Attached is the original circuit and the text for the same. Kindly excuse me for the quality.

bimbla.
 

Attachments

  • Symmetric Power Amplifier Circuit- Original.GIF
    Symmetric Power Amplifier Circuit- Original.GIF
    92 KB · Views: 745
  • Symmetric Power Amplifier Text- Original.GIF
    Symmetric Power Amplifier Text- Original.GIF
    87.2 KB · Views: 728
statistics say that all elector amplifiers constructed from this circuit eventually failed

there is a Vbe multiplier in the circuit only the elektor people was enough stupid to leave it outside the heatsink when the pcb constructed ....

Which if placed in the heatsink i don't think the it could increase the safety margin by a lot ...
 
Looking at it again I see that the thermal stability depends only on the small signal predriver. There's a lot of local feedback from the output devices to the 0.33 ohms Re's. So maybe it is not so bad as it looks.

But I agree that simulations say nothing about this - there's no thermals in the device models, and the THD numbers are also pretty useless.

Another issue is with going to +/-70 V supply. Please google 'safe operating area' and you will realise that this asks for disasters.
You best bet is to build this amp to work correctly as-is which is harder than you think. THEN you can go try mods if you like.

BTW Even the Elektor article says that THD is pretty mediocre - they spec 0.05% at 1kHz iirc.

Jan
 
Well, it seems from the first couple of replies that my intention to trigger a discussion is served.

Gentleman, this is not my original work. I just modified it and I think made it a little better.
The absence of Vbe multipliier was what caught my attention in the first place.
Remember it is a Class-B design with none of its side effects, PER the ORIGINAL ARTICLE.
Out of curiosity, I put it in simulation and was shocked at the results. It has triggered interest in me to make a proto. Don't you want to get rid of the bias current settings, somehow
Let's wait for my proto results.

b.t.w its quite generic too. I used DsX output stage (elsewhere where in the Solid State forum), to the right of the 1N4007 diodes. It not only works but improves the distortion figure from 0.006% to 0.003%
 
Last edited:
statistics say that all elector amplifiers constructed from this circuit eventually failed

I think you must be referring to the Current Dumping Amplifier by G. Schmidt. Yes, that is not for DIY. Unless you have precision matching in the bridge, the results will be worse.

That depends on how you read it. They say the crossover is not audible - that's a nice way of saying its pretty bad but don't worry you won't hear it 😉.
0.05% at 1kHz IS pretty mediocre!
Jan

Well, its subjective. If its stable and sounds good, then why worry? Needs to sound good first.
Sometimes I feel that distortion figures are like megapixel war in Cameras. Cramping megapixels does not increase the light gathering ability but only reduces the signal to noise ratio. But yes, it can be a guideline. What I mean is that we pay some much more for Class-A that does not have impressive distortion figures. But it sounds GOOD! But yes, as a guide line, no problem.

Thanks!
 
I think you must be referring to the Current Dumping Amplifier by G. Schmidt. Yes, that is not for DIY. Unless you have precision matching in the bridge, the results will be worse.



Well, its subjective. If its stable and sounds good, then why worry? Needs to sound good first.
Sometimes I feel that distortion figures are like megapixel war in Cameras. Cramping megapixels does not increase the light gathering ability but only reduces the signal to noise ratio. But yes, it can be a guideline. What I mean is that we pay some much more for Class-A that does not have impressive distortion figures. But it sounds GOOD! But yes, as a guide line, no problem.

Thanks!

Yes you have a good point there!
 
...I have modified an elektor circuit for some distinct advantages:
1. Replaced the transistors used so that the same can be used anywhere over a wide supply voltage range. I have tested this in simulation upto +/- 70V.....
Advantages of this circuit:
1. No Vbe multiplier. So no dealing with thermal stability issues!!...
You forgot to replace the OPTs. TIP142/147 will not survive at UCE=140V.
The VBE multiplier is not the cause of thermal stability issues but a possible cure.
 
You forgot to replace the OPTs. TIP142/147 will not survive at UCE=140V.

You are correct. Circuit to the left of the 2.2E resistors in the supply rails can be used as is for higher supplies with an appropriate output stage.

The VBE multiplier is not the cause of thermal stability issues but a possible cure.

You are correct. My text to be read as: No Vbe multiplier. No dealing with thermal issues since the output is Classs-B.

Thanks!!
 
Short of ...off topic question Bimbla how old are you ?

Youngsters most of the time ( that will include me and many others in the forum that already been there ) have the passion to support an idea in this style ( for a change though and since we have all been there sometime in the past ...It is better to listen to it the hard way than having an amplifier blown in your face)

remember that forum except diyers has a lot of real experts with a very valid opinion ..farther more in questions like that most of the times its enough to see what consumer manufacturers did ( since they have complete teams with mechanics behind them to debug or investigate a circuit )

""No vbe multiplier "" doesn't apply to any class AB amplifier , Today's standards have left behind ...very behind Class B amplifiers

And yes if you skip the vbe thing you get rid of thermal problems is a very wrong approach

Kind regards
Sakis
 
Vbe multiplier & thermal issues

Hello Sakis,

"No Vbe multiplier and no dealing with thermal stability issues due to Class-B output stage!!"

The above is my sentence. If you are interpreting it as: No Vbe multiplifier SO no thermal issues then I cannot be faulted.

Anyway, I understand that all you are doing is guiding me to the right direction and asking me to be aware of what I am doing and I am very thankful for that.

Thanks!
 
Again in a way this is wrong ...Class B operation doesn't mean that you have no issues ...you have actually less issues but still its there

not to forget that class B amps where supposed to be small things of 15-20 W and sometimes managed to keep stable hardly with 50 w ... So supplying an amplifier like that with 70+70 rails expected to produce a bit over 150W is just a dream ( regardless the ability of an output stage with tip 142-147 )

Common practice of many manufactures is to make some class B amplifier but with a much smaller transformer than maths suggested often this transformer had small wattage but enough voltage to produce nice dynamic peaks at low listening levels while if the 50% of "rated " power was the duty transformer voltage started to "dive" distortion hit the sky but output stages was safe enough to survive ...

Bad practice ..this how it was done though
 
Status
Not open for further replies.