New Lineup IDEA - Power Follower/Output stage

[lineup]

This is an idea I have had in my head for a long long time. Now I have finally got hold of it and made some excersises in my SPICE. It turns out that my idea actually works.

As you can see from the simplified diagram below there are 2 MOSFETs arranged in push-pull output. Betweeen the gates are 4 transistors in diamond style giving feedback from the output to the input. We can say the circuit tries to keep output at same voltage level as the input.

BJTs can be used for output devices. But then you better use darlington EF pair. Otherwise the input impedance is very low.

I will post some working circuits. Just want to make them better first. But why dont you try my idea in SPICE. There is room for improvement when there are more than one head in use.

 

[Bonsai]

Nice concept.

You need to add some simple bias adjustment for the output pair.

I'm also thinking if you turned the diamond buffer types around, you may be able to figure in some error correction. Just a thought ;-)

 

[thinkhlin]

Hi, Lineup! I have a similar circuit.
links:
Çë´ó¼Ò¸øµãÒâ¼û¡¶½«·ÏÆúµÄHDAM-SA2Ä£¿éÀûÓÃÆðÀ´×ö¸ö¸ß¡°Ö¸±ê¡±µçѹ¼¶¡· - ·Å´óÆ÷DIYÂÛ̳ - HIFIDIYÂÛ̳ - Powered by Discuz!

 

[AKSA]

Lineup,

Nice idea, but matter this is not a diamond buffer since signal is passed from npn to nmos, and pnp to pmos. Reverse is usual, of course.

Your FetZilla ideas were really good, and I think you for your thought, a lot of people here built that amp and really liked it!

However, Bonsai has a great suggestions......

Cheers,

Hugh

 

[lineup]

AKSA I did not say this is a diamondbuffer, only that those four smaller transistors in the feedback is a diamond buffer.
Thanks for the nice reminder of Fetzilla. I have almost forgotten it

 

[lineup]

Hi, Lineup! I have a similar circuit.
links:
Çë´ó¼Ò¸øµãÒâ¼û¡¶½«·ÏÆúµÄHDAM-SA2Ä£¿éÀûÓÃÆðÀ´×ö¸ö¸ß¡°Ö¸±ê¡±µçѹ¼¶¡· - ·Å´óÆ÷DIYÂÛ̳ - HIFIDIYÂÛ̳ - Powered by Discuz!

Thanks thinklin, I see a diamond buffer alright. But my is not quite the same.

 

[lineup]

Nice concept.

You need to add some simple bias adjustment for the output pair.

I'm also thinking if you turned the diamond buffer types around, you may be able to figure in some error correction. Just a thought ;-)

Bonsai. First I tried the circuit with Class A current in output and instead of making bias adjust I simply throw a big output cap. 4700uF.
This solves any bias point.

As I mention in topic "Power Buffer/Output stage" this buffer can very suitable be inside the feedback loop and act as a buffer within. In fact this is how I use it. Now in Class AB.

See my diagram.
It has also got a TMC error correction. It is the first time I got TMC to work. I finetune TMC with trial and error. Changing those RC components.
Very low distortion. It is a precision discrete Opamp that controls it all.
40Vp-p THD 0.00004%
That is 25 Watt output into 8 Ohm.
See the numbers in the second image.

 

[lineup]

I repost my circuit.
The previous was not fully stable.
This one is.
With 0.00004% THD

 

[lineup]

Here is another application where the circuit is used. It acts as the output stage. OPA134 is used for the frontend and receiver of feedback.

It can be of interest to see the INPUT voltage versus the OUTPUT voltage from the buffer.
INPUT = 32.01 Vp-p
OUTPUT 32.01 Vp-p
Which tells us that what is on the input comes out. This is a very precise output buffer!

 

[Fast Eddie D]

This is a very precise output buffer!

Kudos. I like this implementation with the local feedback in the output stage.

Is this buffer linear enough to operate out of the global feedback loop? It does have local feedback. Because if you build it as a two stage amplifier, you will have less distortion fed back to the op amp. Do you think this is worth a try?

 

[lineup]

Kudos. I like this implementation with the local feedback in the output stage.

Is this buffer linear enough to operate out of the global feedback loop? It does have local feedback. Because if you build it as a two stage amplifier, you will have less distortion fed back to the op amp. Do you think this is worth a try?

It is very precise with MOSFET output devices. Not much distortion.
My guess it is like THD 0.002%
So it can be used outside feedback loop. I will test to see the figures.

Another thing which Bonsai mentioned in post up here is the DC Offset on output. It can be an issue in real circuit. But I think we can find a way to balance output close to zero.
Another redical solution to DC Offset is to use output cap. This is something we never should be afraid of. Many old amplifiers performed very well usning big output caps.

 

[lineup]

Fast Eddie D.
Here are my testings for Power Buffer.
I put signal at buffer input. No feedback. I did test with FFT on output.
8 Ohm.
04 Vp = 01 Watt THD 0.00020%
08 Vp = 04 Watt THD 0.00042%
12 Vp = 09 Watt THD 0.00067%
16 Vp = 16 Watt THD 0.00093%

We can see the buffer performs very well. Better than 0.001% is very good. So in real circuit the distortion will be very low. My guess is better than THD 0.005%. This is better than most power buffers I know about.

 

[Fast Eddie D]

It is very precise with MOSFET output devices. Not much distortion.
My guess it is like THD 0.002%
So it can be used outside feedback loop. I will test to see the figures.

Good. You do realise the practical advantages to a two stage power amp, right? It can reduce speaker interaction with the amplifier. There are discussions of this in another thread here.

When back emf is fed back into a unity gain feedback network, it is not amplified. When it is fed all of the way back to the input voltage gain stage, it is amplified!

 

[lineup]

Yes. I know about this. The speaker has a coil and can transmit pulses into global feedback loop.
This is where a circuit like my power buffer can do a duty.
One voltage amplifier with local FB. And one buffer to follow. And no back emf to the voltage amplifier.

 

[Fast Eddie D]

Yes. I know about this. The speaker has a coil and can transmit pulses into global feedback loop.
This is where a circuit like my power buffer can do a duty.
One voltage amplifier with local FB. And one buffer to follow. And no back emf to the voltage amplifier.

That's it. My old Nakamicha TA4A reciever works that way. It is very well behaved and very powerful in an "old school" way. It uses a traditional diamond buffer with current sources. It's my favorite solid state amp. It can drive any single pair of speakers "6 ohms nominal" or higher quite well.

 

[lineup]

Nakamichi has done a lot of good things. I have own a Nachamichi taperecorder. It was the best

 

[lineup]

Here is the next part in development of this circuit.

I have added 'real' current sources.
I have added 2 compensation caps. This makes the circuit have good stability. I think it slows down the MOSFETs.

Now is only half the distortion att full output (30 Watt into 8 Ohm).
The input impedance is >11k. THis makes the buffer an easy lood for the frontend.

I now use OPA604 for voltage amp. It has max +/-25 VDC supply.
And so it can put out +/-22 Vpeak.

The diagram shows circuit at full +/-22 Vpeak output.
THD is 0.00067% at 30.25 Watt output.
Where OPA604 contributes 0.00028% and the buffer 0.00039%

These are good figures. It points to we can build a very good amplifier which is not too complicated.
THe MOSFETs run in 217mA idle.

 

[kASD]

Nice idea Lineup, one can construct a pure class-A preamp section also using your topology, as it nicely fits in some of the places of interest.

 

[lineup]

Yes.
With a discrete Class A preamp we can put out more Watt.
Chip opamps are restricted to lower voltage. (Max 30 Watt)
I have now adjusted the needed capacitance across GATE-DRAIN for MOSFETs.
One cap 1.5nF across seems to be pefect for AC values.
The input impedance is set by those two resistors to ground.

 

[lineup]

Here is the AC Analysis with those 2 caps across MOSFETs.
And the current working circuit.
1. OPA604 as voltage amplifier +/-22Vpeak.
2. The Power buffer
Total THD 0.00067% for both parts.