John Curl's Blowtorch preamplifier part II

Status
Not open for further replies.
Jump to Latest
dimitri

dimitri

Member
Joined 2002
I had a test jig that brought the devices under test to a constant temperature. A tech sorted through the lot after a pre-screen for minimum Idss. The parts were then binned for both Vth and Idss, and the two-parameter data was used to select matched pairs. They were then mounted in aluminum blocks with a temp sensor and heater, the notion being that they'd be operated at a regulated temperature slightly above ambient. These were to form the first stage of some discrete line-level balanced amplifiers, prefacing a switched resistor array for level matching which was then followed by buffer amplifiers.

By the time I got close to this realization, the money that was unlimited became a good deal more constrained, and I wound up dialing down the ambitious design for a less costly alternative. I have a bunch of the aluminum-mounted pairs to this day, I think at least 32 to cover an 8 channel balanced system.

But the notion was to have sufficiently low offset voltage and drift to preclude the need for d.c. blocking or servos. This was fairly easy too as the required gains were not large. Of course this supposed no offset in the sources.

The loudspeaker and power amplifier output switching was to be done with Mercury displacement relays.
Click to expand...
Thank you Brad,
We also (with diyaudio member rlenser) designed similar system many years ago. We decided to avoid any active circuitry in the signal path, we used pi attenuator with 1/2dB step for level matching, dc coupled, Takamisawa relays for attenuator, Panasonic relays for speaker. We found that pi attenuator with constant input and output resistance didn’t change audio. Our listening room was large enough and prevent placing speakers on turntable.
 
Bonsai

Bonsai

www.hifisonix.com
Joined 2003
hifisonix.com
Paid Member
Kindhornman said:
Bonsai,
I think that ostripper would disagree with you also on the use of a servo as he has done that in many of his designs. If you can do without that is great that you have worked that out but I don't see the real objection to using the servo to control dc offset.
Click to expand...

Of course there is no real objection. But, if you can get away without coupling caps and without a servo then I would say that's the best option all round.

I have the utmost respect for OS and many of the other designers here . . . So no hard feelings please.
 
B

Bob Cordell

Member
Joined 2006
cordellaudio.com
john curl said:
Darn Bcarso, you went all out! At the time, there was also the NPD5564 and the 2SK240 available, years before the 2SK389. They all worked pretty well. The NPD5564 was a monolithic matched fet pair from National, and the 2SK240 was two 2SK170's that were computer matched and factory placed inside an aluminum can.
Some people hold out for full DC operation, and I cannot disagree with them, except for the hassle. Servos are easy, BUT can be problematic if compromised in some serious way. Usually it has to do with using the servo as an audio filter, as well as a DC control, or not decoupling the servo well enough.
Click to expand...

The good old NPD5564 was indeed a great part. I think I still have a few in my basement. I can't seem to locate the old spec sheet, though. I've wondered how close the 2SK389 is to it. I think they are pretty close, but it is possible that the LSK389 was a little quieter.

BTW, I seem to remember that there were some devices around that might have been called 2N5564 that were also a two-die solution in one package.

Cheers,
Bob
 
ostripper

ostripper

Member
Joined 2008
Bonsai said:
Of course there is no real objection. But, if you can get away without coupling caps and without a servo then I would say that's the best option all round.

I have the utmost respect for OS and many of the other designers here . . . So no hard feelings please.
Click to expand...

Thank you , Bonsai. Your material is very good , as well.
Most of the 30+ year old OEM amps with servo's survive. 30 year old
capacitor based amp will go DC and blow the outputs.
We should add to our project manuals- " replace DC cap every 15 years" :D


There is a downside to the servo - settling time. Usually no worse
than a blocking cap , but you will get that short DC start up
transient with both.

Fine tuning the shunt regulators and running the servo supply together
with the input pair CCS's can minimize this "issue".

PS - I also have many designs without a servo , usually "classic" VFA
designs.
Many exceptional amps without a servo ! :cool:

OS
 
john curl

john curl

Member
Joined 2003
For Bob Cordell:
 

Attachments

  • npd5564.jpg
    npd5564.jpg
    733.9 KB · Views: 234
B

bcarso

diyAudio Member RIP
Joined 2005
Bob Cordell said:
The good old NPD5564 was indeed a great part. I think I still have a few in my basement. I can't seem to locate the old spec sheet, though. I've wondered how close the 2SK389 is to it. I think they are pretty close, but it is possible that the LSK389 was a little quieter.
...

Cheers,
Bob
Click to expand...
The 389s were appreciably larger area and about a factor two or more lower noise. Higher C of course.
 
Esperado

Esperado

Banned
Joined 2011
esperado.fr
ostripper said:
There is a downside to the servo - settling time. Usually no worse than a blocking cap , but you will get that short DC start up
transient with both.
Click to expand...
I think it is an important part of comfort (and security) to never have the slightest commutation noise from a hifi system.
My protection offers soft start, delay before to connect the speakers, and disconnect before power off.

It seems some people, here, think Servos can change the sound of an amplifier. because this IC in the feedback path. Correctly designed, they don't. The IC carry no signal in the audio range, and both the low harmonic distortion signals from the ultra low frequencies and the noise of the IC can be attenuated in such a way that they will be a lot under those of the amp itself. Non measurable. And you can set, with little caps, the FC low enough to get no phase turn at frequencies as low as 1 or 2 Hertz.
 
Last edited:
J

jcx

Member
Joined 2003
john curl said:
Sine waves do not measure DA.
Click to expand...

of course some would take that as conceding the argument - something doesn't affect sine waves then by Fourier DA therefore doesn't affect Music

which is practically the case at common feedback, XO impedance levels and 1st order DA models - just milli or even micro dB frequency response changes
 
Last edited:
jan.didden

jan.didden

AX tech editor
Joined 2002
linearaudio.net
Paid Member
Esperado said:
I think it is an important part of comfort (and security) to never have the slightest commutation noise from a hifi system.
My protection offers soft start, delay before to connect the speakers, and disconnect before power off.

It seems some people, here, think Servos can change the sound of an amplifier. because this IC in the feedback path. Correctly designed, they don't. The IC carry no signal in the audio range, and both the low harmonic distortion signals from the ultra low frequencies and the noise of the IC can be attenuated in such a way that they will be a lot under those of the amp itself. Non measurable. And you can set, with little caps, the FC low enough to get no phase turn at frequencies as low as 1 or 2 Hertz.
Click to expand...

How about this:

A servo is an integrator, and the ideal integrator needs an ideal infinite gain, infinite bandwidth gain block. Using an opamp with limited (OL) gain as a servo will cause the integrator action to break down above a certain frequency. That means that (distorted) signal components 'break through' from the servo to the point where the servo correction is in inserted. Doesn't sound (sorry for the pun) as a desireable situation.

Thoughts?

Jan
 
Status
Not open for further replies.
Top Bottom