John Curl Design Approach

rkernell

Member
2009-02-11 4:40 am
I have recently purchased a Parasound Halo A 21 and I am extremely impressed with the clarity, detail and definition. I am in the process of reading G. Randy Sloan's and Douglas Self's books on amplifier design and I am curious about Mr. Curl's approach. Does anyone know of any publications that he has that outlines this?
 
I am glad you like the unit. All of my Parasound designs are based on a 'formula' that I have developed over the decades. There are other approaches that work as well, and although I do disagree with Doug Self and Bob Cordell on a number of things, their writings can be helpful as well, so long as you keep an open mind about my approach, which they will often criticize.
I make amps, complementary jfet input, complementary balanced throughout, fast, and as Class A as possible. I tend NOT to use an output coil or an input cap. I servo the output for DC offset, instead. That is about all there is to it.
 

rkernell

Member
2009-02-11 4:40 am
Mr. Curl,

Thank you for the reply. I take everything that I read with a fistful of salt.

I am skeptical about your assertion "That is about all there is to it." Something of such high quality did not arise that cavalierly, it took much hard work, innovation and intellect, and might I say, love.

Thank you for your work...

Rick
 
Good to hear from you, Wavebourn.

Good to hear you too John. ;-)

The secret is, what and how to optimize. Nothing is perfect. Eliminating one type of errors you increase other errors that are more audible than what you eliminate. Each high end designer has own set of criteria of optimization, and the art is which set is better, the rest is the plain dumb math and engineering. ;-)
 

hahfran

Member
2007-12-19 8:53 pm
I make amps, complementary jfet input

This is a way for a manufacturer but not for DIY, i was semiconductor engineer and i know the reasons why it is much harder to make really complementary Jfets . One either has to have preselected devices or bet on a chance to find such pairs by buying some hundred and measuring.

A manufacturer can easily order some 1000 matched pairs but a DIY cannot.
 
This is a way for a manufacturer but not for DIY, i was semiconductor engineer and i know the reasons why it is much harder to make really complementary Jfets . One either has to have preselected devices or bet on a chance to find such pairs by buying some hundred and measuring.

A manufacturer can easily order some 1000 matched pairs but a DIY cannot.

Or you could get a matched set at the diyAudio store.
 

hahfran

Member
2007-12-19 8:53 pm
Or you could get a matched set at the diyAudio store.
I am rather skeptical about those matched sets. It is even difficult to find matching pairs of one nfet type and almost a nightmare getting matched pfet types.
The reason is the manufacturing technology.
I am not about John Curl's undisputed excellency of design, i mean it is not a good idea to clone these with off the shelf parts.
One can with good results clone bjt designs provided the pcb layout is cloned, too.
But with Curl's you better by the product , clones will be disappointing
 
Hi Guys

Topologically, how John described his circuit is dead-on and has not really changed since the 1970s. There are examples in the Audio Amateur mags from the era. For me, John's circuits fall in line with my own preference for symmetry and simplicity.

What has changed is the ease of getting jfets at all let alone complimentary ones and let alone matched ones. Both Mr. Curl and Mr. Pass buy up whatever of these stocks they can find so they can sort through and get the matched devices required.

Bob Cordell is a bit more open-minded about circuit design than Doug Self, as far as I can see from what they write. Bob's "favourite" (my opinion) circuit is that of his 1984 mosfet amp, which uses a n-jfet input and push-pull VAS stage with non-Miller compensation for high slew rate. He presents alternative input stage designs that use a single pair of n-jfets but develops symmetric drive and symmetric VA. In both of these jfet-input examples, he acknowledges the fact that n-channels are about all you can get these days, especially if you want matched ones. Further, most of his circuits do not require that these be matched too closely if at all.

On the other hand, Mr.Self is squarely in the typical UK camp of building things to the lowest cost possible, which also tends to mean using the simplest circuit possible, the ubiquitous and erroneously-named Lin amp. The actual Lin amp is an output buffer driven by a VAS/TIS with an overall inverting voltage gain. The addition of the input diff-amp was by Thompson, at least for audio. In any event, this is the circuit used in 99% of amps, so it is economically beneficial to investigate how to improve its performance and that is what Doug has spent his career doing.

Almost all of the CFA amps on this forum and many in production use complimentary symmetry. It is very rare to see a jfet input on this, and that will usually be on the asymmetric circuits.

All of these topologies have been around for decades. There are small refinements here and there. Some of these circuits are particularly sensitive to device matching, or to the circuit halves being matched in gain - the latter alleviates individual stage matching and can reduce the problem of matched complementary jfets. So, where John described his circuit plainly, he has only left out the "detail" of refinement. But... almost everyone has a simulator these days, so anyone can play with the design in the computer and find some of the refinements required.

Have fun
 
Almost all of the CFA amps on this forum and many in production use complimentary symmetry. It is very rare to see a jfet input on this,

Because of the topology and requirements on the input buffer of the CFA amplifier (low JFET transconductance, poor symmetry of NJFET and PJFET parts), it just makes sense to make it bipolar.

Note: yes complementary symmetry (not complimentary, another meaning), but not complementary differential for CFA input.
 
Last edited:

hahfran

Member
2007-12-19 8:53 pm
Because of the topology and requirements on the input buffer of the CFA amplifier (low JFET transconductance, poor symmetry of NJFET and PJFET parts), it just makes sense to make it bipolar.

Note: yes complementary symmetry (not complimentary, another meaning), but not complementary differential for CFA input.
there are dual monolithic nfets avaliable but not any dual monolithic pfet.
The semiconductor engineer has more process variables to make BJT complementary matching pairs thus a DIJ guy is far better off with bipolars in this respect.
 
Because of the topology and requirements on the input buffer of the CFA amplifier (low JFET transconductance, poor symmetry of NJFET and PJFET parts), it just makes sense to make it bipolar.

Note: yes complementary symmetry (not complimentary, another meaning), but not complementary differential for CFA input.

May I compliment you on your complement to the forum? ;)

Agreed.