ETALON-schematics?

[JonMarsh]

They should be on the heatsink. You can verify this by simulation if you have any doubts.

Also, be careful in the types you choose; circuit behavior and bias level is sensitive to Hfe and Vce sat as well as VBE Sat. Would be a good idea to use similar process devices- for example, if you use 3281/1302 for outputs, try using the 1A Tosh parts for the control.

OTOH, you might want to try the same circuit just as a bias controller, without the attempt at error correction, and with more output devices, as Chas suggests...

~Jon

 

[fab]

...I think it sounds very good, at least it did when I built one 21 years ago. PMA mentioned here that the circuit of Upupa Epops has gained quite some fame.
Steven

Steven, even though you appreciated the sound 21 years ago you seem to not use amps with this error correction technique anymore. If that is the case, why? Do you have found some drawback?
Jonmarsh suggests to use only the NPN/PNP pairs for bias without the error correction. Do you share the same opinion?

Thanks

 

[lumanauw]

I remember somewhere someone said to skip the error correction sch. Why is that? Is it because of the sound or other consideration? For now (I havent built it yet) I think error correction has merit of possibly infinite feedback. Dont know what this sounds, but looks great.
Also he said about possible problem if the BJT output drive difficult load. Is this error correction not suitable for difficult load, like too capacitive load? Is this become less problem if we parrarel several output devices?

 

[Charles Hansen]

http://www.diyaudio.com/forums/showthread.php?postid=330982#post330982

 

[fab]

Since I have already seen the Halcro patent and the Halcro amp sells for about 40000 $ I would suspect that it must be better than average amp. So, seing the simple Etalon error correction circuit, I decided to try such a correction circuit into a modified Hafler DH-200 amp prototype. I had to adjust the resistors values to adapt with my amp.

I read the post from andy_c indicated by Charles.

Here are the electrical results:
- at 1KHz sinus full power the damping factor was raised from original 300 to more than 2000;
- It was very stable with 10 KHz square with 8 ohms and 1.2 uF load (gave normal overshoot with good damping)
- However, the THD at 1KHz stayed the same as without the error correction circuit.

I did not have the chance to "hear" the result (only a prototype).

Attached is the circuit schematic used.

 

[lumanauw]

Can error correction made by ordinary single npn vbe multiplier in ordinary amp?

 

[Steven]

Hi,

Some late replies. I was on holiday to Germany for a few days.

Steven, even though you appreciated the sound 21 years ago you seem to not use amps with this error correction technique anymore. If that is the case, why? Do you have found some drawback?
Jonmarsh suggests to use only the NPN/PNP pairs for bias without the error correction. Do you share the same opinion?

At this moment I don't use these amplifiers anymore, but I have used them for quite some years with pleasure. Then I increased the bias current to use them for a bigger part in class A which was also fine for another year. Then it looked like a good idea after all those years to build a nice cabinet around each amplifier (they were still open frame). But after some months of occasional use they were left on for some days and during a night thermal runaway occured (I guess) and one amp broke down. The combination class A and a closed cabinet was not so fortunate, after all. At that time I was not so busy anymore with audio as a hobby and I was able to buy a pair of classic Marantz mono blocks for almost nothing and started to use these. The other amps stayed in the attic, collecting dust. Later on the Marantz amps were replaced by a Marantz surround sound transceiver with DVD player (that's family life).
Recently a renewed interest in audio has started and maybe I will make myself a new set of amplifiers to this principle.

I think the Hawksford error correction circuit works fine, but it is possible to use that stage for only biasing. It would look like the Tanaka biasing circuit, resulting in a nice non-switching class AB amplifier: http://www.diyaudio.com/forums/showthread.php?postid=282339#post282339

I remember somewhere someone said to skip the error correction sch. Why is that? Is it because of the sound or other consideration? For now (I havent built it yet) I think error correction has merit of possibly infinite feedback. Dont know what this sounds, but looks great.
Also he said about possible problem if the BJT output drive difficult load. Is this error correction not suitable for difficult load, like too capacitive load? Is this become less problem if we parrarel several output devices?

I did not encounter stability problems with the Hawksford circuit. I admit that the error correction is not perfect for all frequencies, e.g. if you have adjusted the error correction stage to zero output impedance at 1kHz, it will be slightly off for 20kHz, vice versa.

Since I have already seen the Halcro patent and the Halcro amp sells for about 40000 $ I would suspect that it must be better than average amp. So, seing the simple Etalon error correction circuit, I decided to try such a correction circuit into a modified Hafler DH-200 amp prototype. I had to adjust the resistors values to adapt with my amp.

I read the post from andy_c indicated by Charles.

Here are the electrical results:
- at 1KHz sinus full power the damping factor was raised from original 300 to more than 2000;
- It was very stable with 10 KHz square with 8 ohms and 1.2 uF load (gave normal overshoot with good damping)
- However, the THD at 1KHz stayed the same as without the error correction circuit.

I did not have the chance to "hear" the result (only a prototype).

Attached is the circuit schematic used.

Maybe your 1kHz distortion was not primarily determined by the output stage?
I guess you left out the collector resistors of Qa ands Qb in your circuit diagram?
Using Sziklai pairs for the error correction stage is a nice touch.

Can error correction made by ordinary single npn vbe multiplier in ordinary amp?

I don't think so, at least I don't see a direct solution for a push-pull amplifier.

Steven

 

[fab]

QUOTE]Originally posted by Steven

Maybe your 1kHz distortion was not primarily determined by the output stage?
I guess you left out the collector resistors of Qa ands Qb in your circuit diagram?
Using Sziklai pairs for the error correction stage is a nice touch.
Steven [/QUOTE]

thanks for your feedback.

1) 1kHz distortion : yes, you are probably right. I had forgot that I changed a couple of things to experiment with (it is really a prototype with a lot of added circuits to replace every stages, etc ...). It does not look too much anymore like a DH-200!
I was experimenting in reducing significantly the open loop gain of the first stage (diff. input) to see the effect of reduced feedback on THD. Also, I have increased the tail current of the diff input stage, etc...

2) collector resistors: If I understand what you mean you are not talking about the resistors to reduce the open loop gain of the VAS stage but the one in Hawksford circuit to act a a potential divider. If you look closely at the Etalon Integra circuit in the first post you will notice that they are NOT there either!

3) "Sziklai" pairs : I did not know the name but thanks.

Also, you indicated "...I admit that the error correction is not perfect for all frequencies, e.g. if you have adjusted the error correction stage to zero output impedance at 1kHz, it will be slightly off for 20kHz, vice versa..."

Does it mean that the adjustment is only done to have the maximum damping factor regardsless of THD?

 

[lumanauw]

In FAB's modification cct (and original etalon cct) the output is connected to R10 and R11. But in Steven's explenation, it is not. Does this connection necessary in error correction, or not?

 

[fab]

In FAB's modification cct (and original etalon cct) the output is connected to R10 and R11. But in Steven's explenation, it is not. Does this connection necessary in error correction, or not?

In Steven's circuit, check intersection point of R5 resistors which is the output that directly connected to R4 then to base of the error transistors pair.

 

[Steven]

2) collector resistors: If I understand what you mean you are not talking about the resistors to reduce the open loop gain of the VAS stage but the one in Hawksford circuit to act a a potential divider. If you look closely at the Etalon Integra circuit in the first post you will notice that they are NOT there either!

These are the same resistors! In the original Hawksford circuit emitter followers were used as a buffer with a low output impedance. They were followed by a series resistor, that is R1 in my circuit diagram in post 17. This R1 is the collector resistor of the error amp and as such also the input resistor of the potential divider. The Etalon circuit does not use an emitter follower but 'collector followers', the BF469 and BF470. They are loaded by the 6k8 resistors. These 6k8 resistors reduce the gain of the amplifier, as you noted, but they act also as collector resistors for the error amp with BD139 and BD140, just like the input resistors of a potential divider.
For the error amplifier it makes no difference whether it is loaded by a voltage source with series resistor or by a current source with parallel resistor. These circuits are equivalent.
That means that the value of these resistors is important for your error correction adjustment.

3) "Sziklai" pairs : I did not know the name but thanks.

Also, you indicated "...I admit that the error correction is not perfect for all frequencies, e.g. if you have adjusted the error correction stage to zero output impedance at 1kHz, it will be slightly off for 20kHz, vice versa..."

Does it mean that the adjustment is only done to have the maximum damping factor regardsless of THD?

George Sziklai was a designer at RCA and the circuit was named after him.

Minimum distortion and maximum damping factor should coincide. Zero output impedance is the easiest way to adjust the error correction.

In Steven's circuit, check intersection point of R5 resistors which is the output that directly connected to R4 then to base of the error transistors pair.

You're right. Thanks for answering.

Steven

 

[fab]

These are the same resistors! In the original Hawksford circuit emitter followers were used as a buffer with a low output impedance. They were followed by a series resistor, that is R1 in my circuit diagram in post 17. This R1 is the collector resistor of the error amp and as such also the input resistor of the potential divider. The Etalon circuit does not use an emitter follower but 'collector followers', the BF469 and BF470. They are loaded by the 6k8 resistors. These 6k8 resistors reduce the gain of the amplifier, as you noted, but they act also as collector resistors for the error amp with BD139 and BD140, just like the input resistors of a potential divider.
For the error amplifier it makes no difference whether it is loaded by a voltage source with series resistor or by a current source with parallel resistor. These circuits are equivalent.
That means that the value of these resistors is important for your error correction adjustment.
Steven

Thanks for the explanation. In my DH-200 prototype the "collector resistors" to ground are 27k so this value is high enough that it has no effect on the error correction circuit (at least no effect on the damping factor). I tried to reduce the collector resistors value to where it has an impact on the error correction, but it was too low and it increased too much the THD (lowered too much the VAS gain). So I restored the 27k. Anyway, I want to keep the VAS gain in the 200 range.

 

[Steven]

Hi fab,

You could add two emitter followers after the VAS stage. Then you can keep your 27k resistors for a high VAS gain. The emitter followers will have a lower value series resistor as the load resistor for the error correction amplifier.
See e.g. http://www.diyaudio.com/forums/showthread.php?postid=330446#post330446
The emitter followers need some biasing too, therefor a 10V zenerdiode has been added (to the circuit in the link).

Steven

 

[fab]

Hi Steven,

Yes I could use emitter followers with series resistors but this is what I want to avoid. That is why I liked the Etalon circuit. If I am able to adjust the output to zero impedance do I get what is expected from the error correction circuit?

I had seen your fabulous amp design and the basic topology is in line with the design I am working on (which is a modified DH-200 amp design).

 

[lumanauw]

Can anyone help, where can I see Halcro design or at least Halcro patent? Some said Halcro uses modified error correction, I wanted to see what it likes

 

[fab]

Hi Lumanauw,

Even the "zip" file of the patent is too big for the forum. But try this link:
http://patft.uspto.gov/netacgi/nph-...,892,398.WKU.&OS=PN/5,892,398&RS=PN/5,892,398

The error correction circuit of Halcro is similar to Hawksford or Cordell.

Enjoy

Fab

 

[lumanauw]

There is something that I dont understand. Why is this Etalon schematic doesn't fit into the Hawksford's EC equation? Steven's design, Cordell, and others fits nicely. The equation is R2/R1=R3/(R3+R4). What is the R1 in Etalon? Putting 6k8 (the one to ground) doesn't fit the equation also.
Is this Etalon dont use R1, so it can maximize efficiency with rail voltages (not so many drops)? But how it fits the equation without R1?

 

[thanh]

I re-simulated Cordell and Eltalon's circuit by orcad pspice . The result is worse than simulating in LTspice

 

[mayky]

Anybody have a original pcb layout , or a scanned picture for Etalon Integra?

Thanks!

Mayky

 

[lineup]

Interesting.
There is an op-amp used for feedback.