Transformer question

[AkisTzortzis]

OK, so you now have a 50 amp 12v power supply...

Let me ask you this, now that you have touched live circuits with your fingers, got shocked, and discovered they can be dangerous, are you likely to lay your fingers on one again?

Oh I have touched live bits many times, but that particular episode was 20 years ago.

There is a reason thousands of models of solid state amplifiers are made the way they are: it works well. What you are proposing is not a very efficient way to proceed.

WHile there would be about 100v between + and - supply rails in a conventional amp, in your high current scenario, failures would be hard pressed to blow fuses, and fire is a very real danger.

Yes there would need to be really thick wires and what you say makes sense. I am giving up on the 12V idea already.

But if you must, look at some of the older single supply amp designs. Instead of trying to come up with some ersatz transformer, why not a large electrolytic cap from the output point to your speaker?

Let us say you set the output of your proposed system at 6VDC out of 12v. That means the signal can swing potentially 6v either way - so the output signal is 6v peak. We are talking ideal here, in real life it would fall short of that somewhat. 6v peak is about 4.2v RMS.

[/QUOTE]

Would I need a huge capacitor? I do not know how to calculate size vs amps (and freq.). Also do capacitors not do something nasty to the output like phase shift or delay or some such? I seem to remember (20 years ago) that noone in their right mind would try to use a cap for output stages ?

Anyway what I have been playing with the last few days on the train on the way home is a class A push-pull output stage. I know how to do it with a simple potentiometer, now I need to replace it with transistor(s) driven from the previous stages (and there I am seriously stuck). But one thing is certain. If you have a power supply say plus/minus 10 Volts, and you bias the output transistors so the one on the positive rail swings between 0-10 Volts, with bias at 5, and if you setup the other on the negative rail so it swings between -10 and 0 volts biased at -5 Volts, and if you somehow guarantee a permanent 10 Volt drop between their bases at all times, then they would swing in the same direction, for example when the top one is at 5 Volts (the idle) the bottom one is at -5 Volts so you get 0 at the output. When the top one is at 6 Volts the bottom one is at -4 Volts (difference always 10 Volts) and the output now has 2 Volts. In other words for every Volt change on the output stage you get 2 Volts on the speaker. Which means I only have to amplify my signal to half the supply voltage, ie I need to achieve a plus/minus 5 Volt swing only even though my output stage is plus/minus 10 Volts - and I still get same power as a class B design only much more clean with no crossover etc

What I cannot do is establish a constant/permanent voltage drop between the bases of the two output transistors. I could perhaps do it with a series of diodes, say 650mV each, so for 10 Volts I would need like 17 diodes and if my power supply were plus/minus 35 Volts, I would need a 35 voltage drop which would mean a bucketful of diodes. Or perhaps a single transistor with a zener diode?

Am I making any sense?

[Enzo]

Output cap. No one does it anymore, but many amps used to be built with single supplies. That meant the output bus sat at 1/2 the power rail voltage, and they used a large cap to pass signal to the speaker.

Go to Schematic Heaven, under "Post 70s AMps", and in the Acounstic section find the model 320/330 file. The power amp is on page 16 of that file. It is a little blurry, but the output cap is 3900uf/75v. Large, but certainly easy to find. It is an example of a successful design. Phase issues, etc? Hell if I knoe, but the Acoustic amps of that era sounded good, clean and strong, and still sought after by many.

But if you are now looking at split supplies, the point is moot.

You seem to be trying extremely hard to reinvent the wheel here. There is no secret free power circuit. Solid state amp design is mature and there areplenty of designs that have stood the test of time. There are circuits that builders make on their own with plenty of people having already done it and able to share their insight, just as we here have built boatloads of 5E3 guitar amps. I think you will find that any sort of shortcut you come up with either won't work, or has already been done.

for example when the top one is at 5 Volts (the idle) the bottom one is at -5 Volts so you get 0 at the output.

What? SOunds like there is 10v difference. What do these connect to? Certainly not the two terminals of a speaker - it would have 10v of DC across it. Moving the two in unison is simple, but if there is a steady 10v between them, how does that help you? A speaker with a steady 10v across it will sit there with its cone extended (and will burn up in the process) and it won't matter that the whole thing is moving up and down with the signal.

Or maybe I don't know just where the speaker connects to in all this. You might draw up a diagram of this concept and post it. REmember the whole point of an amplifier is to provide current to the speaker. Current needs a complete circuit path.

[Enzo]

Establishing a bias between two opposing output bases is done all the time, we don't usually think in terms of 10v, but the amount doesn't matter. 17 diodes would actually work, though I can't imagine doing it that way. A zener would work if you provided it with operating current. But mist circuits just use a transistor and a couple resistors - one variable - to set the current through it, and thus the voltage across it.

Go to www.ampix.org, and in the Enzo gallery open the SWR Workingmans 12-15 schematic, page 2 is the power amp. See how they used Q5 to establish the bias.

[AkisTzortzis]

I mentioned +/- 10 Volts as an example, if I ever try to build anything I think I would go to +/-30 Volts - 60 Volts end to end is quite safe and you can get some decent power from it.

The speaker is connected to 0 on one end and to the emitters of the common collector output transistors. With +/- 10 Volts, when the top is at 5 Volts then the bottom is at -5 Volts, 10 Volts between them at the base (and the emitter, let's ignore junction drops for a second). At that point there is 0 Volts at the speaker so it is "idle". If the top transistor goes to 6 Volts then the bottom will go to -4 Volts, so we are presenting 6 + -4 volts to the speaker == 2 Volts.

I have seen the Q5 transistors you mention - or sometimes a couple of diodes in series - I will also look at your plans now.

I have another idea. Looking at various power transistors specifications, it appears hfe is like 100 for Ic < 1A and then falls like a stone from 1A to 10A reaching like 10 or 15. How can you possibly design a **linear** output stage when with every volt change in the output your hfe drops and your requirements from the previous stage increase logarithmically ? I can see 3 solutions to this:

(a) do nothing let it distort
(b) use darlingtons to minimise the current needed by the driver transistors
(c) put as many output transistrors in parallel as needed so as Ic is always < 1A and then hfe is always >100 and stays there. That would be perfectly linear plus the requirements of the previous stage would be much-much less, as if I were using a darlington.

I like the idea of (c) - I have seen designs with lots of transistors in parallel at the output, but they usually do it to cope with power/heat - not to ensure linearity.

I know my ideas have probably been picked to pieces long ago ....

[Enzo]

The speaker is connected to 0 on one end and to the emitters of the common collector output transistors. With +/- 10 Volts, when the top is at 5 Volts then the bottom is at -5 Volts, 10 Volts between them at the base (and the emitter, let's ignore junction drops for a second). At that point there is 0 Volts at the speaker so it is "idle". If the top transistor goes to 6 Volts then the bottom will go to -4 Volts, so we are presenting 6 + -4 volts to the speaker == 2 Volts.

Are both positive side and negative side emitters both connected to your speaker? That means they are connected together. That means you cannot maintain 10v between them at the emitters - and yes, we can ifnore junction drops for this discussion until we get to drivers. Those meitters are each wanting to bring their own polarity of 5v to the output. COnnect them together and they will send all the current they have into each other, and it will blow up or blow fuses.

You could add resistors between them, but then those resistors would wind up in series with your speaker as far as current is concerned.

You are trying to step around the classic push pull output. That maintains no voltage difference between the emitters. The bases are separated by the two junction drops. The drivers are then separated at their bases by their own junction drops plus those of the outputs. And so on.

A properly biased amp has no crossover distortion.

GO to www.ampix.org adn find the PV 400BH schematic in the Enzo gallery. It is a very common power amp in peavey products of the 1980s. It is about as basic as it gets in commercial amps. it is what we call quasi-complimentary rather than full complimentary, but the concpet is the same. All this difference means is the negative side uses the same type transistor as the positive instead of NPN and PNP. These are all NPN. I would have used a basic complimentary circuit but there was none in my hard drive files that were simple enough for my purpose.

[Steve Conner]

I was thinking that, but Enzo explained it better

Two push-pull Class-A designs I can think of are Douglas Self's Trimodal amp and the Krell KSA50. You should probably study these. I don't know if the schematic for the Trimodal is in the public domain, but there is certainly one floating around for the KSA50, and several hobbyists have cloned it.

If you find both of these too "op-amp-ish" for your taste, then sorry, you're on your own, as I don't know of anything simpler. The KSA50 really isn't all that complicated, though. If you ignore the nifty DDT compressor circuit, then Enzo's Peavey 400BH is about as subtle as a 10-pound sledgehammer. (and as loud as one too, those Peavey amps were great)

The output transistor linearity thing; Douglas Self covers this in his books. (Maybe you should buy his "Audio Power Amplifier Design Handbook" too.) In brief, you get special audio power transistors that maintain their beta much better at high currents, such as the MJL3281 and MJL1302. And even if you used older transistors that do droop, like the MJ15024/5, or God forbid the 2N2955/3055 then the amp's negative feedback loop will automatically supply the extra drive needed for an undistorted output. Yes, negative feedback does have a use, and people who try to design solid-state amps without it need their heads examined. Both of them.

[AkisTzortzis]

Are both positive side and negative side emitters both connected to your speaker? That means they are connected together. That means you cannot maintain 10v between them at the emitters - and yes, we can ifnore junction drops for this discussion until we get to drivers. Those meitters are each wanting to bring their own polarity of 5v to the output. COnnect them together and they will send all the current they have into each other, and it will blow up or blow fuses.

Ooops. Ooops again. And there I was thinking "why do they never try simple class A diagrams".

I will stop drawing diagrams on the way to work and back, I will stick to reading my novel (The Hound of the Baskervilles, currently).

You could add resistors between them, but then those resistors would wind up in series with your speaker as far as current is concerned.

Yes, that would end up spending energy in the wrong place.

You are trying to step around the classic push pull output. That maintains no voltage difference between the emitters. The bases are separated by the two junction drops. The drivers are then separated at their bases by their own junction drops plus those of the outputs. And so on.

I lack knowledge so I am making silly mistakes :-)

A properly biased amp has no crossover distortion.

I was thinking (just thinking) in the class B approach half your wave is amplified by one transistor while the other one is off, and the other half by the other. Which is impossible that they would produce same waves. If you were in Class A, the end result would be a blend of the two transistors' characteristics but on the full wave plus they would be operating from their middle curve outwards where the curve is much more linear.

GO to www.ampix.org

Hmm that site does not exist ?

[AkisTzortzis]

sorry, found the site now, also looked at the "krell ksa-50" diagram, it is as I remember the class-A design (with the X circuit in it), I can see the 0.5 Ohm resistors at the emitters of the output transistors and it must be that they are enough to encourage the current to flow through the intended load (the speaker) than straight into the opposite transistor... I am still not comfortable with the idea of how this works, after your comment about the falacy of my idea, however it must work, so I need to turn it around in my head a bit more.

I presume a "direct" drive such as the dual supply rails designs, means power is applied exactly where we want it, on the speaker, with some loss on the stabilising resistors (the 0.5 Ohms) at the emitters. A single sided approach as per my original post, using a transformer, would apply and expend power at the transformer and then at the speaker. Assuming a transformer of suitable design and dimensions werre found, how efficient would it be? Would we be wasting more power on the transformer - speaker coupling or on the stabilising resistors on the dual supply rail circuits?

One thing about transformers, if we think about it in terms of energy transfer: if for example my mains transformer is rated at 400 Watts and weighs, say 3Kg, why would the output transformer need to weigh more, since it would be transfering less energy and at same or higher frequencies than the wall supply?

[AkisTzortzis]

I was thinking that, but Enzo explained it better

Two push-pull Class-A designs I can think of are Douglas Self's Trimodal amp and the Krell KSA50. You should probably study these. I don't know if the schematic for the Trimodal is in the public domain, but there is certainly one floating around for the KSA50, and several hobbyists have cloned it.

If you find both of these too "op-amp-ish" for your taste, then sorry, you're on your own, as I don't know of anything simpler. The KSA50 really isn't all that complicated, though. If you ignore the nifty DDT compressor circuit, then Enzo's Peavey 400BH is about as subtle as a 10-pound sledgehammer. (and as loud as one too, those Peavey amps were great)

The output transistor linearity thing; Douglas Self covers this in his books. (Maybe you should buy his "Audio Power Amplifier Design Handbook" too.) In brief, you get special audio power transistors that maintain their beta much better at high currents, such as the MJL3281 and MJL1302. And even if you used older transistors that do droop, like the MJ15024/5, or God forbid the 2N2955/3055 then the amp's negative feedback loop will automatically supply the extra drive needed for an undistorted output. Yes, negative feedback does have a use, and people who try to design solid-state amps without it need their heads examined. Both of them.

Thanks for the comments, I had a look around, lots of schematics out there and people discussing ideas, if only one had time to read those thousands of forum pages.

[unparalleled]

The other reason is safety, having 100 volts on the PCB or across the collectors of the output (push/pull) transistors (usually mounted on the same cooler) is an accident waiting to happen. I remember once I put my finger on the transistor to check how hot it was, then I put my other finger on the other one, ouch...

It would be much better to work with smaller voltage and higher currents, but then I'd need a transformer. Back to my original question :-)

So I was looking for a safe voltage and decided 12 Volts would make sense as you're more likely to find a 12 volt power supply than some other.

Have you considered using a 1 Farad capacitor as used in car audio stiffening to couple your speaker? They are rated at 14 VDC . two of them connected in series at opposite polarity would give you .5 farad non polarised.
Such caps are now inexpensive.

[AkisTzortzis]

Are you suggesting using two big capacitors at the output transistor of a single rail stage? Is it OK to use capacitors in this case? Why do I need to care about polarity? I would simply stick the one end of the speaker to 0 and the other end at the negative of the capacitor and the positive would be the emitter?

But is the transformer not better than a capacitor? I thought driving capacitive loads was much worse?

I think I have asked this before, but surely the output transformer would be of equal or smaller size to the power supply transformer ? In which case I do not see anything wrong in having two bulky lumps of iron inside the case instead of one

[Enzo]

If we go back to that single ended idea, look at my example in post #17, the Acoustic 330. Single ended, cap for output. MAny amps were made that way back when. You don't need farad caps or anything remotely that large. The cap in the 330 was 3900uf - and it is a bass amp, plenty of low end.

But i thought we had moved on to push pull as making a lot more sense.

I was thinking (just thinking) in the class B approach half your wave is amplified by one transistor while the other one is off, and the other half by the other. Which is impossible that they would produce same waves.

What do you mean imppossible? Every one of the thousands of solid state amplifier models you see do this. The full waveform is sent to the power output stage. One transistor handles the positive stuff, and the other the negative.

I can see the 0.5 Ohm resistors at the emitters of the output transistors and it must be that they are enough to encourage the current to flow through the intended load (the speaker) than straight into the opposite transistor... I am still not comfortable with the idea of how this works

Because that is not how it works. Current from the positive transistor is not going to cram its way into the other instead of the speaker, and a .5 ohm resistor wouldn't stop it if it tried. In cases of multiple parallel transistors, those fractional ohm resistors - called ballast resustors - facilitate the transistors sharing current equally. Remember, when one side is conducting, the other side is off. The two sides do not turn on at the same time or indeed current would simply flow between them from the positive to the negative and we don't want that.

Where was it said the output transformer would be larger than the powr tranny?

[Steve Conner]

The output transformer can be just as big as the power transformer. The reason is that hi-fi amps need to reproduce frequencies lower than 50 or 60Hz, and lower frequencies need a larger core. If you half the frequency, you need twice as much iron. (Or twice as many turns of wire, which will need to be thinner to make it all fit, making the transformer less efficient.)

So if our amp was 50% efficient (plausible for a push-pull tube amp) ran off 50Hz line and had to go down to 25Hz, a reasonable starting point would be to make the OT just the same size as the PT.

For musical instrument duty, we can make it smaller: 5-string bass guitar goes down to 33Hz, 4-string to 40Hz, and guitar to 80Hz. I don't think there is any tube bass amp with an OT big enough to go down to 33Hz undistorted at full power, though. If there was, bassists would probably think it sounded too clean and thin.

For a single-ended amp, the air gap in the OT core lowers the inductance, which means that it may need to get a LOT bigger for adequate bass response. On the other hand, single-ended amps are necessarily Class-A which limits them to 16% efficiency. So again in a hi-fi design the OT ends up about the same size as the PT.

[Enzo]

Ah.

[Rick Erickson]

[I cannot drive the speaker directly because I have a DC flowing through. ]

Have you considered capacitor coupling the output to the speaker? This is a common approach to a single-ended power supply amplifier design.