[JohnFerrier]

Quote:

Originally posted by alsq not even a puff (small capacitance in the junctions)? or a little, tiny inductance? [edit] whoops! my PC crashed. OK, back to asking inane questions. When the load on the class A finals is not resistive, there are phase issues associated with it. That should affects also the PSU, and that's the part that makes life interesting. How can that be put into the picture? When very small signals are amplified, disphasing the PSU is about neglegible. When large signals are, that is no more neglegible. How can one account for all this, or at least approximate it, using something like Dimitry test rig? Brainstorm!

Okay, L. The quick answer is that it depends upon your wiring. Compared to R, it will be small. Guess, it will be less than 1uH. However, you will need to calculate this based on the wire gauge size and length. For your test, just replicate what you have or plan to use.

C is more involved. However, I'll say (and this is only the amplifier) (<50pF || OP27 bypass caps). (And there are no bypass caps on Gilmore's schematics.)

So:
<1uH + (65-234 ohms || (<50pF || OP27 bypass caps))

What did everyone else come up with?

And I notice that no one can tell me what they are using for a volume control. Is is 10k, 20k, or 50k? This is also not in Gilmore's description. Isn't this an easy answer for someone?


JF

 

[alsq]

First try with Visio. Why shouldn't this work?

 

[JohnFerrier]

alsq,

Ha, ha, ha...

I guess I misunderstood you. I thought you asked for a Thevenin equivalent of Gilmore's (amp) board. What was I doing?

BTW: What do you expect your input voltage to the pre-regulator to be?



JF

 

[alsq]

Quote:

Originally posted by JohnFerrier I guess I misunderstood you. I thought you asked for a Thevenin equivalent of Gilmore's (amp) board.

Oh, but you were right. On that line of thought I was
thinking how to abuse the PSU trying to measure its
dynamic behavior under repeateble circumstances,
somewhat similar to real use. Except, I'm baffled by
how a can transducer would look as interpreted
through the amplifier. So you are right, after all,
I kinda asked the wrong question.
(I'm quite ignorant in this realm.)

Quote:

Originally posted by JohnFerrier What do you expect your input voltage to the pre-regulator to be?

I reckon anywhere btw. 25-32Vpp or so. The 15000uF
filter shold give about 200mV pp ripple at 400mA. Higher
than 32 may become dangerous for the LM317, less than 25 may not provide enough headroom. The
higher the supply, the smaller the filter cap needs be.
With a 30V supply, 4000uF (bit less than 1Vpp ripple
at 400mA) should be plenty already. The
value of the 1800uF is tentative. The 2.2uF caps should
be, according to the datasheets, solid tantalum
(about $0.50ea) and, according to the datasheet, should
do a world of good for high frequency transients,
like those we're trying to smother. The 47uF cap
should have the lowest possible internal resistence,
I believe Digi-Key part 4297PHCT-ND may do?

Suggestion are of course welcome! As I said, I'm
learning as I go.

 

[JohnFerrier]

Quote:

Originally posted by alsq Except, I'm baffled by how a can transducer would look as interpreted through the amplifier.

See impedance graph (versus frequency) here:
http://www.headphone.com/layout.php?...tID=0020080580



JF

 

[alsq]

Yup. Those are the good ol' HD580. I was wondering,
though, what the PSU would see when the cans are
connected to (in this case) the gilmore, and a
non-steady-state signal is fed to the amp. Also, the
graph is the magnitude of Z. The juicy part is
conveniently ignored, as usual.

 

[Voodoochile]

Yes, I was referring to Didden/Jung super regs. I think that is what Arzela was asking about? Anyway, Audio Xpress is a magazine, they also sell some circuit boards for things that have appeared in it. Several years ago the Jung reg article ran, and Jan Didden ultimately laid out a PCB for the revised circuit. Link to the boards.

Please forgive me if the history of the regs is not perfectly accurate. This is as I recall it only.

 

[tangent]

Quote:

Please forgive me if the history of the regs is not perfectly accurate. This is as I recall it only.

I've been studying this stuff recently, so here's a fuller history:

- Mike Sulzer published a linear regulator in the 2/1980 issue of The Audio Amateur, a predecessor of audioXpress. It used an NE5534 op-amp as the error amp, a regular zener as the reference, and a now-obsolete pass transistor. The unregulated supply voltage powers the error amp through a low-pass filter designed to strip the 120 Hz ripple. Even in this primitive state, I believe it's better than an LM317. The configuration is simple and straightforward. I wish more people made Sulzer regulators so you could see more info online about it, because it's good for introducing that sort of design, in the same sort of way that the CMoy amp does for other op-amp headphone amps.

- In the 1/1981 issue, Sulzer published a follow-up article. In it, he passes on a recommendation from Walt Jung to use an LM317 atop a zener as a pre-regulator. This powers the pass transistor as well as the error amp, and so the low-pass filter feeding the error amp is removed with this variation. There is also a small topology fix which improves the decoupling of the op-amp from the raw side of the power supply. This article also goes into a few other improvements, but nothing noteworthy at this late remove.

- In the 1/1983 issue, Breakall et al. did output impedance testing on on an extended Sulzer regulator, and compared it to the performance of monolithic regulators. They used the Jung LM317 pre-regulator, and replaced the zener with an LM329 precision reference.

- In the 1/1987 issue, Jan Didden chimes in with some variations on the Sulzer regulator. He uses an LM7818 as a pre-regulator and an LM7805 as a reference, and he uses a better pass transistor than the generic one Sulzer used.

- Finally we come to the Jung regulator. In the first two issues of TAA in 1995, Jung describes his circuit and gives some very detailed test results of it vs. several other linear regulator types. In issue 3, Jan Didden gives his circuit board layout and gives some general info about applying the regulator circuit. And in issue 4, Gary Galo does some subjective tests on the regulator dropped in as mods to real circuits, with side-bars written by Jung and Didden.

The Jung regulator is an extension of the Sulzer design. Major differences in the Jung regulator: better pass transistor (D44H11), different drive scheme for pass transistor, no pre-regulator or low-pass filter on the op-amp's power supply (it relies on the PSRR of the op-amp), better op-amp (AD797), precision reference (LM329), and remote sensing. In the Sulzer regulator, the pass transistor's base is driven by the op-amp. In the Jung regulator, the pass transistor's base is driven by a constant current source in parallel with the op-amp which only acts as a current sink; the op-amp corrects errors at the regulator's output by drawing current from the CCS so the pass transistor's base doesn't see it.

- In the 4/2000 issue of Audio Electronics (successor to TAA), Walt Jung published some improvements to the 1995 circuit. This version cures some start-up woes, and changes over to the AD825 for improved stability. These changes can be made as mods to the Didden circuit boards, so I think the boards distributed by audioXpress are still the original design.

There are related publications in other magazines, but this main thread is always in TAA or AE.

 

[Voodoochile]

Excellent work there. Is it on your site yet?

Maybe it ought to be.

Where does Erno Borbely mesh into the fray? Didn't he collaborate with Mike Sulzer at some point?

 

[tangent]

Quote:

Is it on your site yet?

If I base the eventual PPA PS on one of these designs, I'll have info up there. If it remains an LM317-based supply, my previous post will have to suffice.

Quote:

Where does Erno Borbely mesh into the fray? Didn't he collaborate with Mike Sulzer at some point?

You're referring to the SB-1 regulator. That page describes a PCB you can buy from them or etch yourself from art they provide, and they call the circuit a Sulzer-Borbely regulator.

Come to think of it, I wonder if it isn't a miscommunication. They might actually mean "Sulzer-Breakall". Like the 1983 article, there's an LM317 preregulator ahead of the Sulzer regulator and a precision reference instead of a zener. It doesn't use the Jungian zener-propped LM317, so I'm not certain of this guess. If Borbely did indeed have something to do with this, I don't have a publication reference for it.

 

[Voodoochile]

Odd, to say the least.
Last time I looked at the artwork for that board (SB-1), there was a dead-end trace on it somewhere. Be forewarned, anyone who plans to etch it. I'll take another look, perhaps it has been fixed by now.

 

[JohnFerrier]

Isn't there more to the story?

Year 2000 Linear Technology comes out with a family of Low Noise LDO regulators that integrates the reference voltage and error amplifier into a single part that costs less than $5. (Integrated regulators <1A have a typical noise of 20uV and 1.5A regulators with typical noise 40uV. Typical noise of SuperRegulator is about 10uV+.)

Year 2002 Texas Instruments follows with their line of "Ultralow-Noise, High PSRR, Fast, LDO" regulators. See page four below for a diagram of the integrated regulator. (Look familar?)

http://focus.ti.com/lit/ds/symlink/tps79530.pdf

Yes, the SuperRegulator may still be lower noise (for other areas it may be worse). However, my point is that this idea has been incorporated and integrated into readily available single part solutions.

alsq,

If you are still following, here is an app note on how to "extend the input voltage range".

http://focus.ti.com/lit/an/slva119/slva119.pdf



JF

 

[tangent]

Quote:

Isn't there more to the story?

It depends on what you're calling "the story". If you just mean "low noise regulators", then yes, you can get monolithic ones. Yawn.

The fact that a reference is used in these ones you speak of is also not what makes a regulator Sulzer-like. Among the regulators I mention above, four different reference schemes are used. What makes a regulator a member of the Sulzer family is:

- The error amp's gain goes to 1 at high frequency, so the error amp doesn't amplify output noise.

- For the Jung branch of the family, the pass transistor is driven by a CCS which is "governed" by the error amp. This is not a common configuration. You average monolithic linear regulator is more like the early Sulzer regulators.

- Extremely low output impedance. The TI datasheet doesn't talk about this at all, not a good sign.

- Discrete componentry, so you can improve performance incrementally as new technology becomes available. With these fancy monolithics, there are no pinout standards. Also, I'd bet that self heating is a big problem in that particular regulator you mention.

 

[alsq]

Quote:

Originally posted by JohnFerrier alsq, If you are still following

Yo, I follow, therefore I am. The TI adjustable is
1.2 to 5.5 V out and will need external parts for my
purposes. I still think Linear is waaaay more than
what I'll need already, and _truly_ simple to put
together.

 

[JohnFerrier]

Quote:

Originally posted by alsq Yo, I follow, therefore I am. The TI adjustable is 1.2 to 5.5 V out and will need external parts for my purposes. I still think Linear is waaaay more than what I'll need already, and _truly_ simple to put together.

I guess I was referring to the TI application note as an alternative way to do it. Obviously, changes need to be made to make it work for 18V. I've printed this, but haven't read it yet. I'm still thinking about my design. I think I've settled on the amplifier topology (though going back and forth about integrated versus discrete buffer) and need to work on the +/-18V PS, DC protection circuitry, and test/adjustment circuitry. One thing that is taking me some time is that I want to use two individual metal foil potentiometers for the attenuator. I want to use a "spotter" potentiometer for initial adjustment, then test circuitry to balance the metal foils. Cheaper than a stepped attenuator and less noise. I hope to make it easy to operate.

Good luck.


JF