John Curl's Blowtorch preamplifier part II

Status
Not open for further replies.
Disabled Account
Joined 2012
Well, this is why I asked about what constitutes 'optimum'.... and that leads to design priority for PS. Now, as JC likes working with low level circuits (LP/phono/MC etc), I would expect a priority is noise. Use a three terminal for regulation and line isolation, then a cap mulitplier (Cx) to reduce noise to extreamly low levels. Secondarily, add more isolation by using a Cx to each stage after the regulator. However, this all fine for class A and lower current levels than for a power amp. There, the priorities might not be noise first.


TH-RNMarsh
 
Last edited:
Trudging on. '-) We have a simple schematic for an audio regulator, but no values as of yet.
Generally C1 should be between 1000uF and 10,000 uf. We found that more capacitance sounds better, I think it is because of the improved 'ground return' that the larger cap gives. Of course too much capacitance and we stress the transformer and generate noise that will come out into the line and make trouble for other components.
Cadj, and C9, can be 10uf to 100uf typically. This is consistent with the 317 app note or data sheet. Works pretty good, except for noise and transient response.
C4 should be 50uf to 220uf, depending on what will fit and what you have around, not too critical.
C8,C7 is for high frequency stability of the cap multiplier and is usually around .001 uf.
C3,C6 is up for debate. Should it be large or small? Later I will show how to make it small (0.1 uf) It takes circuit changes though to make it work right. This example is typical of the JC-80, and a couple of master recorders, and a couple of mixing boards made between 1977 and 1984.
 

Attachments

  • slightly better ps.jpg
    slightly better ps.jpg
    205.6 KB · Views: 292
Last edited:
Well, this is why I asked about what constitutes 'optimum'.... and that leads to design priority for PS. Now, as JC likes working with low level circuits (LP/phono/MC etc), I would expect a priority is noise. Use a three terminal for regulation and line isolation, then a cap mulitplier (Cx) to reduce noise to extreamly low levels. Secondarily, add more isolation by using a Cx to each stage after the regulator. However, this all fine for class A and lower current levels than for a power amp. There, the priorities might not be noise first.


TH-RNMarsh

What I have been finding important besides the low pass through and added noise that are clearly in your field of expertise is uniform damping vs frequency. Some regulator circuitry not only has decreasing regulation with increasing frequency but also has sharp resonances often at frequencies above the audio band but some within it.

As PSSR usually decreases with frequency resonance outside the audio band often appear to the ear as an increase in the noise floor.

One of the interesting bits is that switching the phasing of the AC line is said to improve the midrange. Turns out there is support for that when you measure the leakage noise from the transformer primary coupling to the chassis. When you look at the spectrum of the coupled line noise it will have a bit more effect on the midrange even with the high level of power line frequencies present.
 
It's also relatively pointless unless you're designing crap signal circuits that have lousy PSR. The only thing that matters is the output of the preamp.

No it is quite possible to design a power supply that feeds noise back into the AC line and decreases the performance of other parts of the system.

But Pavel has a point there are at least three different preamp designs and maybe as many power amplifier designs, adding actual source choices and you get at least 8! possible combinations to examine, possibly more! :)

(8! = 40,320)
 
Last edited:
Really! Do you want to post one of your designs with the measurements of the AC line noise it produces?
Virtually all power supplies inject lots of noise back into the line, not just switchers.

Yes, switching power supplies with power factor correction can be reasonably quiet.
Diode rectifier circuits with high input current pulses, even with an EMI input filter, are not.
 
See Linear Audio, vol 8. If it's throwing noise back into the line at some microscopic level (your favorite), who cares? The system is dead nuts quiet.

SY,

I have posted lots of my measurements here, lets see your actual measurement of power line noise.

Basic math will show that when you rectify an AC source into DC at the transistion from diode off to turn on, you create harmonics very much on the order of sin(x) + 1/3 sin(3x).... So it is not a microscopic level unless your power supply is only producing microwatts. At audio midrange frequencies a 20 watt supply will be putting out more than a watt into the power lines.

Have you measured how much that is attenuated into the next power supply? Now if we get into low F RFI you may be down to milliwatts, how much do you think that affects say a phono preamp?

Now what happens when you have a few hundreds of watts from a power amplifier power supply around phono level signals?

These noises get into your other gear not just through the power supply. But you really have to start there to not just filter line noise but also to not make as much.

ES
 
it always, mister ? is that any remedy until today ?

I must admit that I am in a specfic situation. I make power line filters and get rid of much of grid borne muck before it even gets into the device. Therefore, my grid supply is by default cleaner than most, and consequently, I have less to clean up later on. Or what I use has less work to do, if you like.

I do not make audio for commercial reasons, meaning I offer and sell none of it. Strictly for myself. Yet, because I know only too well what a typical grid looks like, I am perhaps a little obsessed with power line noise and possibly tend to overdo it.

On the plus side, I have been asked more than once by people who own famous brand products how come I have less noise than they do, which tells me I must be doing something right.
 
Why? My system is dead quiet. You seem to have lost sight of the goal.

What kind of cooling system do you use? Getting close enough to absolute zero to eliminate noise is very impressive! :(

Have you conducted double blind tests to determine if the system is really dead or still has some life left? :)

Maybe you have some measurements of maximum signal and noise levels so we can have an actual discussion of what passes for quiet enough?

I had one guy stick his ear into the loudspeaker horn of one of my systems to check for any noise. He could hear some power line type noise. Of course if there was maximum signal when his ears were there he would have damaged his hearing permanently.

If you want to fairly make fun of folks who proclaim "Audioisms" without any testing, please try not to do the same!

As mentioned some power supply issues can result in perceived items such as decreased midrange detail. The issue is not just noise but the reproduction of music or what some prefer to listen to and call music.
 
Last edited:
AX tech editor
Joined 2002
Paid Member
Trudging on. '-) We have a simple schematic for an audio regulator, but no values as of yet.
Generally C1 should be between 1000uF and 10,000 uf. We found that more capacitance sounds better, I think it is because of the improved 'ground return' that the larger cap gives. Of course too much capacitance and we stress the transformer and generate noise that will come out into the line and make trouble for other components.
Cadj, and C9, can be 10uf to 100uf typically. This is consistent with the 317 app note or data sheet. Works pretty good, except for noise and transient response.
C4 should be 50uf to 220uf, depending on what will fit and what you have around, not too critical.
C8,C7 is for high frequency stability of the cap multiplier and is usually around .001 uf.
C3,C6 is up for debate. Should it be large or small? Later I will show how to make it small (0.1 uf) It takes circuit changes though to make it work right. This example is typical of the JC-80, and a couple of master recorders, and a couple of mixing boards made between 1977 and 1984.

John I asked it before - what's that buffer doing? To me it looks like it just replicates all rectifier ripple at it's output - that's what a buffer does, no?
Or does it anything else?
What chip is it?

Edit - ahh - you changed the circuit - the 'buffer' is now an LM317. Sneeky :)
Looks like whoever drew that circuit is symbol challenged :)

Jan
 
Last edited:
Status
Not open for further replies.