Simple and tested bipolar transistor amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi!

I'm new on this site, i just discovered it yesterday when searching information about class-D amplification and for my surprise it's full of interesting information about amplification and many projects... So i feel that i need to contribute something. I was born in Russia and 10 years ago aprox have builded an interesting amp, the schematics is not mine, but taken from a very good book (Bolotnikov and Itaev) about HIFI solid state amplification. Now you can find this circuit on internet, but always in russian and with russian semiconductors. Well, now i live in Spain and builded again this amp using the most popular and available transistors here.

I think, increasing the voltage and changing the drivers (BD139 and 140):dead: you can experiment and increase the output power. I already have working this amp on proto board (photo attached) and it's sounds better than any integrated STK-based cheap junk... :D

If you decide build this amp you will not need to make any adjust, only if you use other type of drivers and final transistors a bias should be adjusted for them changing the type of D1 diode or connecting more diodes in parallel or serial. :D

The main characteristics of this device are:

- Power source: +/- 36V 2A for every channel

- THD aprox 0,07% at 50W 1Khz

- Signal/Noise rate 89dB (my unit noise is unnapreciable)

- Standby current 80 - 100mA

- Input impedance 10Ko

- Frequency response 20 - 50000 Hz

If you find any way to improve this circuit or have any question -> mail

Godd Luck!!!
:D :D
 

Attachments

  • ampli 60w.png
    ampli 60w.png
    46.3 KB · Views: 1,231
There are several errors in the schematic you've posted. Diodes D2-D5 should be connected other way and not connected to the output node. Unless this is corrected the amp will just blow up instantly. On top of this deadly error input and feedback electrolytics are connected in a wrong polarity.

x-pro
 
The electrolytics seem allright to me. The 4.7 uF at the emiters of the LTP sees about half of the negative rail, so it is correct, the plus side to ground.
The 47 uF in the negative feed-back sees zero volt. No problem. And the 100 uF boot-strap also sees zero volt on the output rail, and about half rail in the negative side. OK for me, plus up and minus down.
Diode D1 inded is reversed. It is used to bias the output in pure class B. Connecting it that way, will blow the output in a microsecond (or less).
Diodes D2...D5 seem to protect the bases of the output stage from reversed polarty in case of saturation, and are shown the correct way.
 
Besides the diodes wrong way ( D2-D5) it is not too bad.
I mean the general idea has been seen working many times.

I would probably lower the Voltage a bit.
+- 36 Volt for a single pair of MJ15002/01 is not needed.
They wouldnt be to cope with so much.
Say when a loudspeaker have dips from 8 Ohm downto like 5 Ohm.
Not if you are unlucky
.. read: Some of your girlfriends happens to turns up the volume for FULL :D :D when not 100% sober.

For 1 pair of output devices, I would go for smoething like +-25 VDC.
Less heat, gives better performance.

Also if you lower, you may keep BC547/557, without any hazards.
But when you can consider 45 V transistors, the low-noise variants BC550C/ BC560C are much to prefer.
'C' transistors, instead of 'B' have ihgher gain. Gives less base currents, and so less output DC-offset issues. Easier to get 0.001 Volt at output terminals :D

With the current +-36, the 2 input transistors share not perfectly.
Each will not have close to 50% of the current ( 2.60 mA ) coming from the tail resistors ( 6k8 + 6k8 ).

You should try to get something like 1.25 mA in the 680 Ohm resistor at upper left.
Probably you should use 560 Ohm here, instead of 680 Ohm.

It is very important that BOTH Input transistors have exactly same currents.
48-52% is okay. But for example
1.00 mA vs. 1.60 mA is actually = 38-62% ... !!!!


Regards, Lineup, from Lineup Audio :)
 
whoandcar said:
The electrolytics seem allright to me. The 4.7 uF at the emiters of the LTP sees about half of the negative rail, so it is correct, the plus side to ground.
The 47 uF in the negative feed-back sees zero volt. No problem. And the 100 uF boot-strap also sees zero volt on the output rail, and about half rail in the negative side. OK for me, plus up and minus down.
Diode D1 inded is reversed. It is used to bias the output in pure class B. Connecting it that way, will blow the output in a microsecond (or less).
Diodes D2...D5 seem to protect the bases of the output stage from reversed polarty in case of saturation, and are shown the correct way.


Hi... sorry for D1 mistake :eek: made when drawing the diagram, y will attach a corrected schematics this evening (for Spain). Also some photos of the proto amp, that actually are working fine, but with only +/- 30V (I dont have other transformer...)

Thank for input transistor type suggestion, i will change it for BC550. I used BC547B because y don't have others now... so let's go to the store
;)
 
x-pro said:
There are several errors in the schematic you've posted. Diodes D2-D5 should be connected other way and not connected to the output node. Unless this is corrected the amp will just blow up instantly. On top of this deadly error input and feedback electrolytics are connected in a wrong polarity.

x-pro


My friend, the only really "deadly" error I made was D1 inversion, and it was a errata.

Regards.
 
Hi,

You have pretty much a bog standard amp here, built successfully millions of times by people around the world. One note though: using only one diode (D1) to bias your output stage will mean that your transistors are completely turned off at idle, giving rise to quite some crossover distortion. Your feedback loop will deal with this to a certain degree, but some idle current in the output stage will eliminate the problem at the source. As your amp is nearly identical to Rod Elliott's project 3A, I'd suggest implementing the Vbe multiplier in his design and make the bias adjustable. Take a look at it here:

project 3A

[edit] Make sure the transistor used in the Vbe multiplier is thermally coupled to the output transistors, i.e. on the same heatsink.

Regards,
Jurgen
 
The input filter (10uF) will allow a significant AC voltage to develop across the NFB filter (47uF).
I recommend that the 47uF be changed to 330uF or even 470uF 16V. Add a pair of diodes across the NFB to limit the reverse voltage to 1.4Vdc. Consider adding a parallel cap of about 1uF to the NFB & input filters.

I agree with Xpro.
The bases of the LTP will sit at a small negative voltage. Both these filter caps should be reversed.
Make sure the transistor used in the Vbe multiplier is thermally coupled to the output transistors, i.e. on the same heatsink.
no. The Vbe multiplier must be coupled to the drivers in the Shiklai (CFP) output stage.
 
timpert said:
Hi,

You have pretty much a bog standard amp here, built successfully millions of times by people around the world. One note though: using only one diode (D1) to bias your output stage will mean that your transistors are completely turned off at idle, giving rise to quite some crossover distortion. Your feedback loop will deal with this to a certain degree, but some idle current in the output stage will eliminate the problem at the source. As your amp is nearly identical to Rod Elliott's project 3A, I'd suggest implementing the Vbe multiplier in his design and make the bias adjustable. Take a look at it here:

project 3A

[edit] Make sure the transistor used in the Vbe multiplier is thermally coupled to the output transistors, i.e. on the same heatsink.

Regards,
Jurgen



The D1 diode in the original design was 2 diodes... but I deleted one during adjusting the bias experimentally. No distortion is apreciated at low volume levels.

Thank you all for the feedback about this amp, I will take into account some of them (like transistor type change and 680R replace to 560R in order to balance the input stage, so when I return home today I will check the operation of it...

:D
 
timpert said:
Hi,

Two diodes is indeed what I would expect for class B operation. It seems pretty odd to me that you just use one diode and still get an idle current at all. How and where did you measure it?

Regards,
Jurgen


I deleted one diode because the output stage was getting much more heat and the standby current was 2,5 times bigger (150mA aprox.) So this way I reduced the standby current (more near to pure class B) without any visible distortion hearing classical music at any level (I don't have any measuring equipment more than a fluke multimeter and old frequence meter... sorry :bawling: )

regards,

Markos.
 
sekanix said:
I deleted one diode because the output stage was getting much more heat and the standby current was 2,5 times bigger (150mA aprox.) So this way I reduced the standby current (more near to pure class B)
150mA of output bias with 2diodes and 60mA of output bias with 1diode. definitely not ClassB.
Something does not add up.
Can you post the schematic with operating voltages shown around it.

I wonder if the amp is oscillating. AM radio test?
 
Hi,

The Vbe multiplier must be coupled to the drivers in the Shiklai (CFP) output stage.

Indeed, I just assumed that all four output stage transistors would be on the same heatsink, but of course they don't have to be.

Your Fluke meter should get you on your way quite nicely. But, as said, your output stage should'nt draw any idle current with just one diode, and the total current drawn by the amp idling is only the current through the input differential amplifier, about 3 mA, plus the current through the class A gain stage, about 4.5 mA.

like Andrew says, some actual measured voltages annotated on the schematic would be of great help to debug this thing. Also post a picture of the actual physical build, I also suspect some kind of oscillation going on.

Jurgen
 
Yes, i know that is AB class, but in my opinion is not necessary to have a big bias for this type of amplifier, just the necessary to avoid distortion. I'm actually experimenting with this project, so is possible that something is going wrong, but the amp is not oscillating for sure...:whazzat: and sounds very clear with a minimum heat...
 
timpert said:
Hi,



Indeed, I just assumed that all four output stage transistors would be on the same heatsink, but of course they don't have to be.

Your Fluke meter should get you on your way quite nicely. But, as said, your output stage should'nt draw any idle current with just one diode, and the total current drawn by the amp idling is only the current through the input differential amplifier, about 3 mA, plus the current through the class A gain stage, about 4.5 mA.

like Andrew says, some actual measured voltages annotated on the schematic would be of great help to debug this thing. Also post a picture of the actual physical build, I also suspect some kind of oscillation going on.

Jurgen

Only Mj15001/2 are on the heatsink, driver are "at air" and no heating is observe at normal operation. I will inlcude a photo of the stereo (2 amps on the same proto board, but separated wiring). Both of them have similar voltages and currents, but I will measure it newly this evening when return home and inlcude and updated diagram with the result

:D

Thanks for your help
 
Frequency meters need some amplitude before they start counting. So a low-level oscillation is not ruled out. I am not saying that your amp does oscillate, it just might.

A photo of your setup will be helpful too, because a boo-boo in the wiring layout is perfectly capable of causing oscillation, amongst other grief.

The Sziklai pair is less sensitive to thermal runaway than the complementary Darlington, but with thermal coupling the bias will be more stable.

Good luck!
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.