Transistor equivalent

[spec]

Nicolai I have just noticed that the output transistors are configured as complimentary emitter followers. This is bad news from the frequency stability and hifi point of view. Best change to conventional emitter flowers and put a 56 Ohm resistor directly on the base of the driver transistors.

 

[audioguru]

The bootstrap comprising C5/R10 is not to the same high standard as the rest of the circuit.

I disagree. I have used bootstrapping successfully many times.

Nicolai I have just noticed that the output transistors are configured as complimentary emitter followers. This is bad news from the frequency stability and hifi point of view. Best change to conventional emitter flowers and put a 56 Ohm resistor directly on the base of the driver transistors.

Again I disagree and so does the site where the amplifier came from (Elliot Sound Products) and from this site that shows the Complementary Feedback Pair (Szilaki Pair) produces distortion that is less than 1/20th that of ordinary emitter followers):
https://www.epanorama.net/sff/Misc/Amplifiers/Complementary_Feedback_Pair.pdf

 

[spec]

I disagree. I have used bootstrapping successfully many times.


Again I disagree and so does the site where the amplifier came from (Elliot Sound Products) and from this site that shows the Complementary Feedback Pair (Szilaki Pair) produces distortion that is less than 1/20th that of ordinary emitter followers):
https://www.epanorama.net/sff/Misc/Amplifiers/Complementary_Feedback_Pair.pdf

I'm tired of arguing about fundamental issues- you have your view I have mine.

 

[Nikolai Petrenko]

Generally no, especially with a toroid, but you must be careful with positioning the transformer in relation to sensitive electronic circuits. In extreme cases I have wrapped a shorted turn of copper or aluminum sheet around the outside of transformers to reduce hum.

This approach is not unusual in commercial amps, especially pre amps. Don't confuse an external shorted turn with an actual shorted turn, which must be avoided or the transformer will be wrecked. It is very easy to accidentally make an actual shorted turn with a torroid, so watch for this when mounting a toroid. A conductive mounting bolt thru the center of a toroid can cause an actual shorted turn. One case I came across was an amp that was growling: someone had fitted to long a bolt in the center of a toroid and generated a shorted turn. The bolt was quite hot.

No, no shorted turn because my idea like this:

In case the bad idea you talked about: the bolt will be heated by current (so it waste lots of electricity and can blow main fuse). I have solution for that bad idea: don't use metal bolt, I will use a bamboo rod then tap the head with nuts. Or use insulator between bolt-wash and chasis or bracket.
I found a website that talk about problem: **broken link removed**

 

[Nikolai Petrenko]

This amp will do 56W peak into 8 Ohms and 112W peak into 4 Ohms, unless you will be using stabilised power lines when the power values will be RMS. As you are unlikely to be using stabilised supply lines the RMS rating of that amp would be about 40W and 80W respectively, depending on the details of the PSU.

You would be strongly advised to get rid of the fuses in the supply lines unless you incorporate a circuit to blow both fuses if one fuse blows. Nothing will take out your speakers faster than a single supply line in a split supply amp. The fuses in the positions shown would be a good source of distortion too.

The quality would be greatly improved by increasing the value of C+ and C- to 2,200uF or more.

The bootstrap comprising C5/R10 is not to the same high standard as the rest of the circuit. If you like PM me and I will show you how to fix it quite simply.

You also need to put one of those 700nH home made chokes in parallel with a 10R resistor in the output to define the hf loading of the amp.

A capacitor of 220nF upwards would be good between the collector and emitter of Q9.

Thanks. I also think like you! 100uf C+ and C- are not enough and no RL output circuit to balance impedance.
The sound will be better but amp will be hotter if I adjust trim pot to increase quiescent current, right?

 

[Nikolai Petrenko]

OK, OK. I am also tired. But thank both of you any way. I may will try all opinions to fin out which is better to me .

 

[spec]

No, no shorted turn because my idea like this: View attachment 96973
In case the bad idea you talked about: the bolt will be heated by current (so it waste lots of electricity and can blow main fuse). I have solution for that bad idea: don't use metal bolt, I will use a bamboo rod then tap the head with nuts. Or use insulator between bolt-wash and chasis or bracket.
I found a website that talk about problem: **broken link removed**

You have a handle on the position. A conductive mounting bolt can be and is used provided it does not form a shorted turn.

 

[spec]

The sound will be better but amp will be hotter if I adjust trim pot to increase quiescent current, right?

No, the amp will be exactly the same as the quiescent current will be adjusted to be the same with the pot. The quiescent current would be about 40mA. If you go for conventional Darlington output stage it will probably be necessary to adjust a resistor value around the pot. This is no big deal. I will take a look. The quiescent current in the Voltage Amplification Stage (VAS) is the same 6mA for both bootstrap and constant current configurations. With a conventional Darlington output stage you have a defining dominant pole in the VAS so the open loop frequency response will be nice. The output transistors will also turn on and off fast. You could even make the drivers class A if you wanted.

 

[KeepItSimpleStupid]

No, no shorted turn because my idea like this: View attachment 96973
In case the bad idea you talked about: the bolt will be heated by current (so it waste lots of electricity and can blow main fuse). I have solution for that bad idea: don't use metal bolt, I will use a bamboo rod then tap the head with nuts. Or use insulator between bolt-wash and chasis or bracket.
I found a website that talk about problem: **broken link removed**

Very close:

The typical Toroid mounting is a Bolt, large metal washer. large rubber washer, torroid, large rubber washer. When mean large I mean HUGE washers.

See: https://toroid.com/Transformer-Products/Mounting-Hardware Mounting washers with rubber pads. That's what I used.

 

[KeepItSimpleStupid]

The sound will be better but amp will be hotter if I adjust trim pot to increase quiescent current, right?

What? No frying Anchovies, please!

When you increase quiescent current, the amp goes into thermal run-away.

The idea is to TRACK the changes the output transistors are doing and bring the difference in base voltages in line.
1) with the difference between the bases = 0, you have crossover distortion.
2) With too much difference, that amp heats up a lot.
3) With just the right amount of bias and the right feedback, there will be no difference in quiescent current as the transistors heat up.
That's the trick.

The regulator is often called a Vbe multiplier. It's guessing the Vbe of the output transistors based on their temperature and it wants to keep the NPN and PNP bases very close to n*2*Vbe+(some slop) as things warm up. I used n*2*Vbe because it depends on the driver stage.

 

[Nikolai Petrenko]

OK, I will set Iq around 70mA.
About transformer mounting, my construction: Bolt, shakeproof washer, washer, large ferro plate, cardboard plate, transformer, cardboard plate, chassis, washer, nut.
Why don't use free cardboards, steel sheets around instead of buy from store. Of course the cardboard must be thicker than 2mm.
I will wrap all sides of toroid core with thick cardboard, use masking tape to make it unmovable. Wind primary coil. Wrap cardboard again. Wind secondary coils. Wrap cardboard one more time. Wrap a copper sheet around to prevent magnetism leakage. Wire addition coil to feed preamp, LEDs, fans. Finally cover all transformer with some layers of tape. Done!

 

[spec]

When you increase quiescent current, the amp goes into thermal run-away.

Perhaps "may go into thermal run away if not properly designed and laid out"

The VBE of a BJT typically drops at 2mV deg C. The VBE multiplier does just that and produces a voltage of either 2x VBE for complementary output stages or 4 x VBE for Darlington output stages. It is important to have decent heat sinking for the output trans and drivers and to ensure good thermal tracking and thus reasonably stable quiescent current. The VBE multiplier transistor body should be in thermal contact with one of the output transistor cases. The low value resistors in the output path also help the thermal stabilisation, but on a well designed amp are a pain as far as audio performance is concerned. In fact those resistors are a bit on the high side at 0R33. OR1 would probably be better.

I have often run AB amps with 200mA quiescent currents and more, without any thermal run away. With the original germanium amps thermal run-away was always a worry and there you may see output resistors as high as 2R2.

 

[KeepItSimpleStupid]

Iq will ultimately be a number that with the amplifier idling, the amp should get barely warm. The voltage across the emitter resistor (Iq) is what you generally moniitor.

In post #732 the voltage between the base of Q5 and the base of Q6 is important. The voltage across R13 and R14 are other important numbers. then there is the DC presented to the speaker terminals.

You don't want cardboard around the core. Equal spaced windings. You can save time for the primary if you wind it in the Bifilar mode. Hand winding of large toroids can be done with a flat piece of wood with notches. Wind the amount of wire on the stick and then keep passing it through the toroid. Then if you need 240 V, wind as two concentric 120 V windings. i.e.two wires together.

The secondary is done the same way. So for 35 CT, you wind two 35 V windings. Mark the phase.

**broken link removed**

tape may prove to be very useful especially between the primary and secondary.

The flux of a toroid is for the most part confined, but an electrostatic shield may or may not be required.

 

[KeepItSimpleStupid]

#352. Can't really argue, but in the Leach amp it's about 3.6 V which is neither or your options. Hence my use of n*2 Vbe. That output stage is far from normal.

The only stressed point is that the bias regulator has to track. This is where thermal and mechanical design are important.

 

[spec]

The optimum quiescent current for an AB hifi amp is based on only one thing: minimizing x-over distortion. x-over distortion is gross compared to other distortions and is particularly nasty sounding. For this reason it often sets the characteristic sound of an amp. The optimum quiescent current for a typical output transistor pair is 20mA, but below this figure xover distortion incresses rapidly. Above this figure x over distortion increases gradually, so many amps chose 30mA to 40mA as the best compromise. Without dedicated Iq stabilization circuits, the Ig will vary, depending on junction temperatures.

 

[KeepItSimpleStupid]

Your right. I never had the luxury of a distortion analyzer to set Q current. Ultimately it's the best way.

==

I remember once I had nothing to go on, so I adjusted based on a bit over Vbe multiplied. Then monitored the AC current, Q current and heat sink temperature every 10 minutes or so. Got something that was stable and let it idle for 8 hours or so. Then I said done. Probably overkill. Amp never came back for repair.

 

[spec]

#352. ... the Leach amp it's about 3.6 V which is neither or your options...

I think you mean #752. I don't see why all amps in the world should comply with two options that I mention for two output configurations of a particular amp.That is illogical.

The complementary output stage has 2VEBs = 1.2V in the signal path, the Darlington ha 4 VBEs in the siganal path= 2.4V, and the Leach has 6 VBEs = 3.6V in the signal path ( see post 542)
Note that the Leach has a tripple Darlington output stage.

Hence my use of n*2 Vbe.

Although you have not defined what N is in your equation, if it means the number of VBEs between the VAS and the output, that is exactly what I am saying, and have been saying all along. It is so fundamental to this type of amp design that we do not to discuss it, but the information may be of intrest to Nikolai and others reading this long long thread.

That output stage is far from normal.

I have described two output stages which one are you saying is not normal. In any case they are both normal, with the Darligton being the most widely used. I do not know what you are trying to prove here.

 

[KeepItSimpleStupid]

You just stated it a LOT better than I did. That's nothing unusual. You understood if just fine which I expected. I apologize for not defining n.

My version of "Normal" output stage. One complimentary pair. I haven't been to sleep yet. Just lying in bed. Class D is probably the new Normal anyway.

 

[spec]

then there is the DC presented to the speaker terminals.

The DC presented to the speaker terminal has nothing whatsoever to do with the VBE multiplier or the quiescent current or its setting. The DC offset is controlled by the completely independent overall DC servo loop between the input and output. DC offset is simply not an issue with modern amp designs, some even without the undesirable DC bocking capacitor in the feedback loop of this amp.

 

[spec]

You just stated it a LOT better than I did. That's nothing unusual. You understood if just fine which I expected. I apologize for not defining n.

My version of "Normal" output stage. One complimentary pair. I haven't been to sleep yet. Just lying in bed. Class D is probably the new Normal anyway.

Now I feel bad. My excuse- I didn't sleep last night with this $%#!! flu.