Hawksford

Status
Not open for further replies.
Jump to Latest
#283
Nelson Pass said:

What I would hope to see would be Steven's comment on
the sound with and without the correction.
Click to expand...


Sorry Nelson, not very soon, I'm afraid. I made that one in 1983 and only with error correction. Not a version without. In the meantime I blew one channel after having it biased for a bigger part in class A. That worked OK for some time, but when kids came I thought it would be a good idea to make a nice cabinet around the open frame amplifier. I think in the end it became too hot and thermal runaway occured. The amps moved to the attic, since I was not so much hobbying with audio at that time and was able to buy a pair of Marantz monoblocks for almost free. They are still there.

Maybe PMA and Upupa Epops can comment on the audible differences between Upupa's design with and without correction.

Steven
 
#284
Audible differences

I am sorry, but I never had try it. Only I had try it by measuring on distortionmeter with excellent results : THD distortion fall on 10 k Hz more than five times ( BW was in this time 80 KHz ). With BW 500 k Hz was distortion less than 0.05 % at 100 k Hz, 1 dB below clipping. My connection is different than Cordell's ( I had try it too, but it was a bit unstable - maybe by my own fault :xeye: - it was maybe eighteen years ago and my experiences was not too big ). I'm using Borbelly's connection of VAS, but followed by predriver before " error correction ". This configuration have OLG cca 68 dB and BW cca 75 k Hz. High gain in VAS is not problem, 'cos I'm using in all my connections antisaturation diodes ( which I highly recomend to all circuits 😉 ). Curves of distortion was quite the same as in Cordell's papers, measured 1 dB below clipping. By lower output power they fall below noise, which was more than 123 dB, ref. to full output. 'Cos at this time I don� know nothing better, I'm still using it. And sound ? " Loud and clear " 🙂 and mainly " without tail ", 'cos with this correction you can to make very fast amp with very short settling time.
 
#285
Steven said:



Maybe PMA and Upupa Epops can comment on the audible differences between Upupa's design with and without correction.

Steven
Click to expand...

Well - I know quite well both Upupa Epops's designs with and without error correction (from the listenner's point of view). The design with error correction sounds smoother, cleaner and more natural. But I should mention that non-error correction design is based on BJT output transistors, and error correction design is based on MOSFET output stage. The other circuits of the amps are quite same.
 
#288
Hi everyone,

This is my first post and I was very interested in reading this thread and your experiments with Hawksford style error correction. I have some experience in applying this circuit in amplifiers from low power to kW powers and find it very effective at achieving low distortion performance (particularly at high frequency). It is also very effective at controlling the bias point too, if applied correctly and by thermally coupling the error correction transistors to the output stage, good thermal tracking is achieved. Very low distortion can be achieved at very low bias currents with bipolars (10mA or less!!)

Also, as an other point of interest, if you guys want to hear what this circuit sounds like all Meridian power amplifiers use the Hawksford correction system. They use different topologies (op-amp based and more recently totally discrete designs) but they all have error correction applied at the output stage.

Robin
 
#291
PMA said:
I've just started PCB and mechanical design of the amp based on an error correction as shown before:


Should be ready in 4 weeks from now.
Click to expand...

Hi PMA,

I am anxious to get your results on this circuit. This circuit (input)seems to be "usable" even with an op-amp at the input.

What is the VGSoff of the mosfet you are using at output stage? with the VB multiplier (Q5, Q6, R9,R11,R10, D1, D2) used it looks like the minimum voltage across it would be:
D1 + D2 + Q5vbe + Q6vbe which is about 2.6 depending on parts, so 1.3V per mosfet. And I did not include the voltage drop across R3 and R4 since the value is not indicated (I supposed they are very low value to minimize voltage drop since the current of the sources are probablt at least 10 ma or more?). My lateral mosfet at a VGS of 0.7v gives 270 ma bias current in a class AB amp.

Based on Hawksford (or Cordell) circuit, I thought that R1 and R2 should connect to collectors of Q5 and Q6 respectively. If not, then what is the purpose of R3 and R4?

Thanks in advance for your response.

Fab
 
#292
The output transistors are 2SK413/2SJ418 - just for the reason that I have plenty of them. I have simulated the circuit and it behaves pretty good. In my experience the simulation results correspond well to the real ones in the circuits like this. I have also simulated with IRF540/9540 - several resistor values have to be changed.
 
#296
PMA and Pass

The "NP-PMA" circuit from PMA previous post includes the Nelson Pass circuit from Threshold US patent # 3 995 228 from 1976. I wonder if Mr Pass could comment on the "audio" improvement of such a circuit compared to a normal class AB design. Was it really included as such in Threshold amps? If so, I can not have the chance to hear such an amp.

Also, the patent describes an active bias that makes an amp to behave almost like a class A while being a class AB design. It does not talk about error correction (from what I understand). Does the "NP-PMA" circuit behave like an error correction because the input drive signal is taken at emitters of Vbe multiplier transistors (Q7, Q8) instead of respective collectors of the same transistors? Does it also make the amp to behave like a class A in this arrangement?
Can you describe just a little more how it works for my and possibly other diyers benefit.

Thanks

Fab
 
#297
JonMarsh said:

I don't use conventional V/I limiters for short circuit protection anymore; (did use them in the past), instead I use an overload detection circuit outside the signal path measuring voltage and current, and tracking (to the extent possible) device temperature; if I get close to the SOA limit, the supply is "crowbarred" and a magnetic circuit breaker in the primary shuts down. Or the rail fuses blow. I don't really like rail fuses, but use them sometimes. None of these techniques are inexpensive, regrettably. And they require building an output stage with "excess" SOA. But this is also deirable from a signal viewpoint; i.e., if using bipolars, paralleling multipe extended beta bipolars can greatly reduce the open loop distortion and impedance.
Click to expand...

Actually, my workhorse amplifier still has rail fuses as the only protection aside from a DC detector. It runs off rail voltages of +/- 48 V. I have a heavy Pi-filter in place (15 mF each side of the common mode choke, along with some series resistors), and the output transistors (2x15 A Sanken extended beta per side) are supplied through 4 A slow blow fuses, but there are additional 470 µF caps after the fuses. These caps do not significantly slow down the fuses, but keep supply impedance low above 10 kHz.

The whole scheme seems woefully inadequate (and I wound never give this amp to anybody else), but in 8 years with lots of experimentation and abuse, I have never fried a single transistor or speaker driver. And I have also never blown a fuse due to overdrive.

JonMarsh said:

Since I only tinker with non interstage feedback amplifiers anymore (NIFA? 😉 getting the output stage characteristics as clean as possible is a primary goal. Actually, it's possible to make the output stage work so well that the main source of distortion is still in the VAS amplifier.
Click to expand...
This teaser is intrigueing. I can only second Mikeks' request for more information (in case you have not already done so on pages 6 to 29 of this thread which I have yet to browse).


Regards,

Eric
 
#298
Nelson Pass said:
Parenthetical to this discussion, the late Oscar Heil was a
firm believer that human ears are extremely sensitive to
these very small modulations, and he attributed much of
the difference in the finest stringed instruments to this.

We should also note that the distinction between phase
modulation and amplitude modulation might be more subtle
than you think. They have some interesting similarities in a
spectral plot.
Click to expand...


Let's not forget that Doppler distortion is present in practically all drivers and might easily swamp the phase modulation from the amp.
 
#299
capslock said:



This teaser is intrigueing. I can only second Mikeks' request for more information (in case you have not already done so on pages 6 to 29 of this thread which I have yet to browse).


Regards,

Eric
Click to expand...


Due to so much business activities, I don't have the time to read here or participate as much as would be fun...

But getting this low output stage distortion is not "innovative" on my part - just check out Doug Self's surprised results when evaluating extended beta transistors paralleled in duals or triples. I say "surprised", because apparently he didn't expect the improvement in both low level and high level linearity that these transistors (Toshiba, On Semi; 3281, 1302) provide.

He discusses this also in his chapters on load invariant amplifiers in his third edition of his handbook on amplifiere design.

Now, there's many folks here and elsewhere that think Self's outlook and opinions are perhaps too narrow, especially his dismissal of MOSFETs with a very cursory examination. And that's not to say there's more to learn than just what's in his book, and more ways to skin a cat, so to speak. But it's a useful resource for anyone interested in a broad education in solid state amplifier design, especially if you do want to explore what's possible with bipolar transistors.

There's other things than what Doug discusses necessary to tweak up the output stage performance; I don't like having inductors in the output path for stability, due to the attending sonic degradation, so that has to be addressed.

As Charles pointed out earlier, the use of the Hawksford style error correction is really a moot point if you've got the right transistors paralleled in sufficient quantity to give you robust SOA without conventional VI limiters. Concentrate on reducing and linearizing the open loop output impedance of the output stage.

I do have a project in the works, a rebuild of an Aragon amp, but at the rate I'm finding free time to work on it (very, very little), it may be next year before it sees the light of day...

~Jon
 
#300
andy_c said:
Apologies in advance for the verbose post that follows here 🙂. I finally took some time to read through the article www.elecdesign.com/Articles/ArticleID/7207/7207.html, and I disagree with some of the points made. His points include the following:
[...]
5) The performance can be improved for a fixed gain-bandwidth product by decreasing the open-loop gain and increasing the open-loop bandwidth. This is best done by loading the VAS with resistors to ground.

[...]

While the phase modulation concept has a valid theoretical basis, for two otherwise identical amplifiers with the same gain-bandwidth product, the one with the larger low-frequency open-loop gain and smaller open-loop bandwidth actually has an advantage with respect to theoretical performance in this area. It's a mistake to blame feedback for this.
Click to expand...


Andy,

since nobody responded to your post in the 5 following pages, I assume it has gone dead and will try to revive it.

The concept that decreasing LF gain by loading the VAS resistively will help has had me bothered, too. And I do agree that a higher unity gain bandwidth is what will ultimately improve the situation.

However, for two amps with the same unity gain bandwidth, the one with the lower LF gain and higher open loop bandwidth will be at least equal and may even have an advantage, but only IF this was achieved not through loading the VAS but by degenerating the input stage.

In this case, the gain of the input stage was spent to to increase open loop gain but to linearize the input stage. You will still see an increased voltage swing, but the gm variation is much smaller.

In the case of loading the VAS, as promoted by Jung, the gain-bandwidth plot looks the same, but the LF gain is thrown away with no linearization in return.

Greetings,

Eric
 
Status
Not open for further replies.
Top Bottom