Reducing noise in voltage regulators

[HarryHaller]

The AD 823 is supposed to be a quite good sounding op amp and the specs are still good at +5 volts. Its a dual though and needs a buffer follower before the pass transistor, which I believe you have in your design. It would be on the top of my list. You could solder a SMT part on a component header to translate dual SMT to single DIP for testing. I bet you wish I had brought all this up before you started! I will email you on where to send my consulting fees. I am getting samples for everything I ask for in about 3 or 4 days including some $30 voltage references. I had to buy some LT parts but they were not too bad. If you only knew the stuff I am holding back........ Sure hope I am not helping any of your competitors for the ultimate regulator board design.........

H.H.

[ALW]

> Its a dual though and needs a buffer follower before the pass transistor, which I believe you have in your design.

Yep, the board can be used without it for some small price reduction, but I like the isolation it provides to the op-amp o/p stage and it allows some serious high current versions to be built.

An important point to note is the inefficiency of this circuit at low o/p currents - the op-amp / buffer transistor has to sink all the unwanted drive current from the base of the pass device, some care in operating criteria needs to be observed. I have some cooked looking 2N2907's on some early boards where I'd got a bit excited with the source current!

>I bet you wish I had brought all this up before you started!

One can spend so long designing that you never get to an end result though, I'm so pleased with the sonic results I'm getting, I'm not overtly bothered. I am wondering though whether to make the next PCB, originally intended to be a negative version, dual purpose - some careful thought and one board may be able to fulfil both roles with some different stuffing and a couple of simple links (damn - there's another good idea for the competition!)

>I will email you on where to send my consulting fees.

Likewise

>I am getting samples for everything I ask for in about 3 or 4 days including some $30 voltage references.

Hmm, location or identity I wonder - will have to try again, different ID...

>If you only knew the stuff I am holding back........

I took the decision that this was a fairly easy option to a big improvement, at sensible cost, for all areas of my system. My next design would almost definitely be discrete, since op-amps are just too limiting (CF possibly being an exception). That design has somehow morphed into a headphone amp now though, whilst I learn more about discrete amp design

>Sure hope I am not helping any of your competitors for the ultimate regulator board design.........

I don't really consider myself to be in competition, I'm too much of a philanthropist for that, but do intend to make a small sum to cover investment and future releases I wouldn't post here if I had any business sense.

If you want to make money from this, you have to sell complete products - I'm too busy for that at present (damn again - more good advice for others to capitalise on). See what I mean, all the business acumen of goldfish...

A.

[HarryHaller]

I like to think it is my personel charm.........

"since op-amps are just too limiting"

Half the circuit is discrete now. It takes some pretty serious design to get the PSRR, Gain bandwidth product, and distortion with a discreet design. You had better be a top notch PCB designer too, since the board parasitics will eat you alive for high speed analog design. Plus there is the issue of building and testing. The guys designing op amps are more talented than the majority of high end designers out there. Maybe an op amp is and pretty good choice for the error amp after all. With a buffer
it is not swinging much current and as an error amp it isn't swinging much voltage. A pretty sweet job for any amplifier.

H.H.

[ALW]

Seems mine work too, just a little slower than yours...

....I got a parcel today )))

More after the weekend

A.

[NickC]

if i power the whole circuit with 12 volts from a battery. is it possible that i utilize ad825 for the 5 volts configuration without dropping the voltage via the 22.1 ohm resistor. ad848 is hard to source out

[artnyos]

NO

A 5 volt version has been just discussed in this thread.

Which schematic are you refering to?

Go to http://products.analog.com/products/...?product=AD825 and read the data sheet.

Everyone , please read the whole thread and provide enough information in your question to answer it in one shot. People are less likely to answer if they have go back and forth for 5 or 6 post to figure out what you are asking.

Art

[jcarr]

ALW Andy: Looking at the schematic that you posted on the first page of this thread, how about doing a version of your regulator with a push-pull output? Should be interesting, and I wouldn't be surprised if it sounds even better.

regards, jonathan carr

[artnyos]

Since the load is almost pulling (in every application I can think of) , I don't see the point. Tthe addition of emmiter resistors for bias stability wound degrade the output impedance. Preloading or a more Class A bias with a resistor load would make more sense for improvements in Ft, Hfe, and Zout.

Art

[jcarr]

Well, I've done a rather similar vreg circuit in the past, and when I changed to a P-P output, I felt that the sound was more cohesive and retained a given level of resolution better when reproducing densely orchestrated music. Your experience, of course, may be different.

regards, jonathan carr

[artnyos]

There is more than one path up the mountain. Emitter resistors can add high frequency stability. The possible increased bias for a push pull verses non preloaded single ended follower can lower output Z. They are many variables. The circuit is complicated enough as it is now; and more parts complicate stability, lengthen signal paths (and complicate ground current return paths which are critical) and make PCB layout more difficult. One would have to get down to particular circuit details to really compare theory and results. Why not post a schematic if interest is high.

Art