ExtremA, class-A strikes back?

[Bigun]

I have no lust for those kind of slew rates and I'm not limited to using BJTs, FETs are a perfectly good option. I like that design OS.

 

[PMA]

Enough other threads to discuss ostripper's amplifiers, isn't it.

 

[ostripper]

Enough other threads to discuss ostripper's amplifiers, isn't it.

I did not post that to discuss my OPS's .... I am about 95% "up to date"
on how this one works.

I am saying I am quite impressed by it's generic performance.
Kudo's to the designers.

This is just a stepping stone ... I want a VFET/LFET Class A (switchable to
AB) OPS. This design is too limited (but instructional).

PS - this is actually a simple version of the "YAP" HEC OPS
(syn08) , but YAP is AB only (for now).
OS

 

[ostripper]

BJT's are fine (excellent).

The "out of box " performance is stellar , as well.

Whether they build the whole amp , or tether on another
IPS .... this OPS adds nearly nothing .

Test is without NFB !! (below) ... -140db as predicted !

OS

 

[JZatopa]

The difference between distortion of the ExtremA OPS and any tube front end is several magnitudes, so you effectively cripple the OPS and nullify all the design work that has gone into creating such a low distortion OPS.

How about we stick to the ExtremA design discussed here and avoid discussion about all sorts of contraptions based on one's own preference? By all means start a new topic if you're looking to discuss alternative solutions.

ok, sounds good.

 

[Bigun]

Pete - can you post your .asc file for this OPS. I'm interested in a single ended version (i.e. I don't like this balanced bridge) but it would be handy to start with your simulation as a known good option. Instructional as you say. As requested, I'll open a new thread if it looks worthwhile to 'proudly borrow' the useful bits for something for myself.

 

[ostripper]

Pete - can you post your .asc file for this OPS. I'm interested in a single ended version (i.e. I don't like this balanced bridge) but it would be handy to start with your simulation as a known good option. Instructional as you say. As requested, I'll open a new thread if it looks worthwhile to 'proudly borrow' the useful bits for something for myself.

Actually , I am tinkering with all 3 of these HEC based stages.

One , or a few "bits" of each ... will be the final OPS for
my uses. I'll post the .asc in my thread , since it is the next
design on the table.

I will say - anyone who builds this exact amp
will be in for a low THD "treat". I simulated ALL of it.

OS

 

[Bigun]

Even better, it'll be a finished thought out design and then I can just lift it and use it

 

[ostripper]

Even better, it'll be a finished thought out design and then I can just lift it and use it

No , it be a OPS you can buy.
I'm posting it "over there" now. NO threadjack.

OS

 

[Bigun]

what you talking about 'Buy' and where the heck is 'over there' ?

 

[ostripper]

You know .. like a cheap "group buy". As cheap as
toner transfer - all the work is done. (you said it).
"over there" - you know where I post most often.
Your .asc is there , and you don't have to buy it.

I assure the threadstarter , I won't use this OPS -
educational use only.

OS

 

[PMA]

@Pavel,

1) We already discussed this, this is fixed by a simple component value rescaling and I am a bit puzzled as to why this hasn't shown up in our measurements. Bruno has measured the prototypes and final design front to back several times, I'm sure he knows what he's doing. But I can certainly ask him to see what he thinks?

2) Not seen issues of that nature before. Take notice that the DC offset trimmer needs to be set after the temperature of the OPS has somewhat stabilized. Also, all output transistors need to be on the same heatsink, do not make the mistake of using two separate heatsinks for each half of the bridged OPS as you *will* see a DC offset indeed.

Hello Sander,

as you know, I have received a request of one of the amplifier builders here to make measurements on his ExtremA. It was built on original PCBs, without visible mistakes.

During measurements, THD was low as suggested, at kHz range. Unfortunately, I am not able to measure THD 10k or THD 20k with enough precision.
But, CCIF 19+20 kHz result was not adequate, so I started to suspect low slew rate. I have measured step response, with results attached (you know them). The amplifier has balanced output, red track is this.

So I tried to change frequency compensation, by reduction of C4 and C9 to 100pF, and I have also tried another methods. The results looked promising, re step response, but distortion (at any frequency) rose more than 100x – clear indicator of local oscillations.

I had to put the original compensation back. The result (original circuit) is very low THD, not so good CCIF and low slew rate (10V/us) with not nice settling. Maybe Bruno might like to tell us more on step response settling?

Re DC – your suggestions are all fulfilled.

Best regards,
Pavel

 

[SSassen]

Hello Sander,

as you know, I have received a request of one of the amplifier builders here to make measurements on his ExtremA. It was built on original PCBs, without visible mistakes. ... <snip>

@Pavel,

It helps to illustrate what you're doing here exactly, from what you've told me the input signal is a squarewave with Trise/Tfall = 100ns. I've discussed this with Bruno and forwarded you his reply, I'll post it here as well so we may end this off-topic discussion. Feel free to contact me privately if you want to discuss this further, I'd be happy to hear from you.

Higher order loops tend to do that when you drive them beyond their slew rate limit. One designs higher order loops for their ability to reproduce low-frequency signals with exquisite precision, not to follow impossibly fast signals reasonably well. It's an audio amp. Play music through it.

 

[PMA]

Sander,

I agree this has been somewhat off topic discussion here, however, you have provided a schematics here in the thread with respect to no output coil. Very low slew rate may be the real reason why the circuit does not need output inductor.

Re Bruno's answer, I find it inadequate and I cannot admit such behavior in time domain. With such kind of reasoning, that Bruno has shown, we are back in the seventies in times of THD wars.

I promise you that I will not continue in this discussion here, however the circuit is out for public and they should now how it behaves in time domain, if the author has not provided such measurement himself.

Regards,
Pavel

 

[bogdan_borko]

@Bob,

My apologies, I forgot to mention that the OPS is a class-A self-bias design, hence it always runs in class-A (100W/8R as shown in the schematic). Only when pushed hard (>rated class-A output power) it enters class-B territory by means of a different mechanism that modulates the OPS input transistors. The THD hits a proverbial brick wall at that point.

The distortion figures of that Allison type buffer stage are great, BUT what about real loudspeakers with impedance dips at some frequency? The amps goes out of class A and distortion goes up to the roof. I think this type of buffer is very good for line level amp buffers, but not for amps, except you know what kind of load is driven and at even then you should have some headroom for transients...

I also don`t understand why this kind of input stage Bruno used, btw it`s overcomplicated for no reason IMO.

 

[SSassen]

The distortion figures of that Allison type buffer stage are great, BUT what about real loudspeakers with impedance dips at some frequency? The amps goes out of class A and distortion goes up to the roof. I think this type of buffer is very good for line level amp buffers, but not for amps, except you know what kind of load is driven and at even then you should have some headroom for transients...

It has a 100W/8R power rating for a reason, for typical listening volumes that provides ample reserve.

I also don`t understand why this kind of input stage Bruno used, btw it`s overcomplicated for no reason IMO.

It isn't overcomplicated when you consider that most of the transistors are used for current sources etc. anyway. Furthermore the performance speaks for itself, I would welcome you to try and match that with a less 'complicated' design.

 

[PMA]

In case this debate continues - I made a comparison of the original circuit vs. ExtremA output stage driven by OPA627. The version with OPA627 has both lower distortion and much better step response with higher slew rate. The step response is RC like, following exponential formula

Vout(t) = Vo * (1 - exp(-t/Tau))

The issue that Bogdan has addressed is class A to AB transition. As you know, it creates sudden distortion in the output stage, as had already been shown in the ExtremA thread. I admire excellent low distortion when it stays in class A, however the circuit has issues. This transition point depends on idle current and one has to stop somewhere to be able to cool down the whole thing.

 

[bogdan_borko]

It has a 100W/8R power rating for a reason, for typical listening volumes that provides ample reserve.



It isn't overcomplicated when you consider that most of the transistors are used for current sources etc. anyway. Furthermore the performance speaks for itself, I would welcome you to try and match that with a less 'complicated' design.

Yes on the site you mentioned that it Could be scaled to 100W/8Ohm max but as you know it is briged amp so and you have only one output pairs for the bridge, and I don`t think so it would be safe to push it to 100W because of the output par SOA and real loudspeaker loads... I guess it could be good for high sens. speakers >95dB.
I said overcomplicated IMO Pavel, said all about about slew rate issues... so I think it is not good solution for an input stage again IMO
Cheers Borko.

Maybe moderators should move these posts to ExtremA thread... sorry for the offtopic.

 

[scott wurcer]

It has a 100W/8R power rating for a reason, for typical listening volumes that provides ample reserve.

So the "power cube" is unimportant? I'm honestly curious a lot of folks make a big deal about it.

AP High Performance Audio Analyzer & Audio Test Instruments : Service & Support

 

[scott wurcer]

the ExtremA thread?

Where is that? Still trying to find out if dynamic headroom/power cube issues are just bogus or what.