[Voodoochile]

I drive Senn HD580s from a KG Dynamic without any difficulty. 3mv offset on the left, 5mv on the right. It drive the Senns to an uncomforable level with ease, regardless of what the numbers might show.

Nice looking work there, Aim9x!

 

[JohnFerrier]

Quote:

Originally posted by AIM9x 118dB... the senns would practically be loudspeakers. I don't see why anybody would even bother going over 100dB... let alone nearly 4x that loud.

Well, if I listen to a performance of Edgard Varese's Ameriques, I don't intend to come up short in the "bass slam" category.


JF

 

[AIM9x]

Quote:

Originally posted by Voodoochile Nice looking work there, Aim9x!

Thanks

 

[alsq]

(1)
In another thread, there was a suggestion
to use a Linear LT1963 as a voltage regulator for
a spiffy PSU instead of the LM3x7 regulator
variety. All is fine, except the LT1963 is a positive
voltage regulator and Linear does not make a
matching negative regulator. Is there any other
optimal way to use two LT1963 regulators for a
dual PSU?

(2)
Since the Gilmore is fully class A, the PSU does not
_really_ need very fast capacitors--am I correct?
(I realize

there's a chasm btw. not using
exotic fast capacitors and using trashy ones, and a
lot of choice in between. I'm just thinking
bang-for-the-buck here.)

 

[JohnFerrier]

Quote:

Originally posted by alsq (1) In another thread, there was a suggestion to use a Linear LT1963 as a voltage regulator for a spiffy PSU instead of the LM3x7 regulator variety. All is fine, except the LT1963 is a positive voltage regulator and Linear does not make a matching negative regulator. Is there any other optimal way to use two LT1963 regulators for a dual PSU? (2) Since the Gilmore is fully class A, the PSU does not _really_ need very fast capacitors--am I correct? (I realize there's a chasm btw. not using exotic fast capacitors and using trashy ones, and a lot of choice in between. I'm just thinking bang-for-the-buck here.)

1/ Other regulator manufactures likely have similar information, I just happen to be aware of the Panasonic version. See page 8 at the following link, for how to use a positive regulator as a negative regulator. I think it's straightforward. And I am the one recommending the LT1963.

http://www.semicon.panasonic.co.jp/cat/pdf/A00036AE.pdf

2/ That the Gilmore is class A makes no difference. However, I recommend ceramic X5R capacitors--they too are fast and cheaper. I think the choice of poly film capacitors is a bit old fashion and based on old news. Ceramic capacitors have improved since the 1980 Jung/Marsh "Picking Capacitors" article.



JF

 

[alsq]

<q>That the Gilmore is class A makes no difference. </q>

Ok--I stand to learn something here. Here's the base of
my reasoning (with artistic license), I need to
understand why it is wrong.

A class A amplifier mostly wastes energy heating the
power semis, except when signal requires some of
that energy to be relayed to the load. Thus, the power
dissipation is more or less constant, ending up
more or less into heat more or less most of the time.
The power supply is operating at full load more
often than not, which is a not-very-dynamic load,
hence transients are not too significant. No need
for very fast reflexes, just steady cruising here.

A non-class-A amplifier (say AB with not much of A in
it) requires large impulses of juice from the power
supply when large impulse signals hit, hence the PSU
must be able to provide power transients with very
little delay, hence the fast capacitor.

How did I do wrong?

 

[JohnFerrier]

Quote:

Originally posted by alsq <q>That the Gilmore is class A makes no difference. </q> Ok--I stand to learn something here. Here's the base of my reasoning (with artistic license), I need to understand why it is wrong. A class A amplifier mostly wastes energy heating the power semis, except when signal requires some of that energy to be relayed to the load. Thus, the power dissipation is more or less constant, ending up more or less into heat more or less most of the time. The power supply is operating at full load more often than not, which is a not-very-dynamic load, hence transients are not too significant. No need for very fast reflexes, just steady cruising here. A non-class-A amplifier (say AB with not much of A in it) requires large impulses of juice from the power supply when large impulse signals hit, hence the PSU must be able to provide power transients with very little delay, hence the fast capacitor. How did I do wrong?

I hadn't thought about it that way.

First, with a headphone amplifier, we are not talking about large amounts of currents. The need for fast capacitors applies more for speaker amps.

More toward your inquiry, if a class A amplifier draws 200mA and jumps to 250mA and a class AB amplifier draws 25mA and jumps to 75mA, it remains a 50mA surge. To try to see it another way, I suppose that a class A has a smooth transition from one polarity to the other. That brief switch current, for class AB amps, will be slightly higher--maybe similar to what you mention. Such an effect is usually seen as typical class AB distortion. An overlooked side-effect may be what you describe--we'll call it the alsq phenomena.


JF

 

[alsq]

Off hand remark: How do you do quotes? Grrr.

Now I'm really curious, might build a Gilmore
just to find this out. Tqakes a bit to explain "this."

According to Linear, the transient response of
the LT1963A-3.3 is -50mV on the on ramp, +100mV
on the off ramp for ~ 1 uS each for a ~.5A
load demand surge (square wave), or for 20uS/50us
for a ~1.5A surge. Each channel of a Gilmore should
usually need about 200mA; the surges will be way
smaller that .5A, but the supply voltage will be
higher, say 18V (assume 300 Ohm load). So, by way
of speculation, the transient may last, oh, <5uS with a
voltage swing in the range of 300 mV or so.
Now 5uS is already way supersonic in frequency, so it
should be inaudible. Yet, the issue is about
common-mode rejection perhaps, since Gilmore says his
amp is not too good at that. Rather than tracking,
maybe relying on fast regulators may be another
approach to the problem at a good value point
($5.25 per LT1963) without too many compromises.

My wicked

thought is the
following: How about building a sensible supply with
LT1963, and testing how the whole thing does with
a 50uS +ve pulse to kill this beast? Not to test the amp,
but to test the power supply and the ability to
avoid tickling the Gilmore on the bad side of
common mode. Feedback, anyone?

 

[JohnFerrier]

Quote:

Originally posted by alsq Off hand remark: How do you do quotes? Grrr. Now I'm really curious, might build a Gilmore just to find this out. Tqakes a bit to explain "this." My wicked thought is the following: How about building a sensible supply with LT1963, and testing how the whole thing does with a 50uS +ve pulse to kill this beast? Not to test the amp, but to test the power supply and the ability to avoid tickling the Gilmore on the bad side of common mode. Feedback, anyone?

Less than 1/2 inch above the "Post Reply" button there is a "Quote" button. You can delete text between the brakets [][]. Works for me. If that is not what you mean, try again.

I don't know what to say. I think that the adjustable LT1963s will work just as good as Gilmore PS or even the Jung SuperRegulator. I like Gilmore's idea a lot. It's basically a differential pair of DC amplifiers driven by a voltage reference. Jung's twist to this sort of idea is to bootstrap the reference power from the output of the "super/ultra" regulator. The biggest knock on the LT's is that they may not have as low of output impedance as the comparable alternative. I don't know, but for the Gilmore amp I don't think that matters. Remember Gilmore is a creative individual, so he will come up with new ideas. I hope he admits that there is more than one way to power an amp.



JF

 

[Glassman]

I recomand you the Jung Super Regulator - I'll be building it for my Gilmore.. I think it is simply the best you can do and relatively cheap..

Jung Super Regulator PDF

 

[JohnFerrier]

Quote:

Originally posted by Glassman I recomand you the Jung Super Regulator - I'll be building it for my Gilmore.. I think it is simply the best you can do and relatively cheap.. Jung Super Regulator PDF

Well, even if you didn't think that the low noise, low drop out LT1963s were good enough in themselves, it may be an improvement to the SuperRegulator to use the LT's as pre-regulators. The LT1963 (with an integrated reference) has a typical output noise of 40uV and the reference for the JSR (the LM329) has a typical noise of 7uV. Add the noise from the rest of the circuit and it is a close call. I haven't read the article you attached. Thanks. I think a few people have been looking for this type of information.


JF

 

[alsq]

glassman:

what are you getting for Q1, and where would
you buy just a couple of AD825?

 

[JohnFerrier]

Quote:

Originally posted by alsq glassman: what are you getting for Q1, and where would you buy just a couple of AD825?

http://www.newark.com/NewarkWebComme...jsp?id=06F4790

http://www.newark.com/NewarkWebComme...jsp?id=83F3403


JF

 

[alsq]

The Sultzer-Borbley supply is a LM3x7 followed by a more
sophisticated error correction stage. Would it make sense
to use a LM3x7 followed by a LT1963 in the same manner?
That would be _very_ simple and fairly inexpensive to do;
also, the LT1963 does not like power-in higher than 20V;
to get 18V out you'd need to bite the edge of the specs
on the unregulated side (just too close for comfort), so
the LM3x7 would be also cheap insurance to prevent
toasting the LT1963.

 

[JohnFerrier]

Quote:

Originally posted by alsq The Sultzer-Borbley supply is a LM3x7 followed by a more sophisticated error correction stage. Would it make sense to use a LM3x7 followed by a LT1963 in the same manner? That would be _very_ simple and fairly inexpensive to do; also, the LT1963 does not like power-in higher than 20V; to get 18V out you'd need to bite the edge of the specs on the unregulated side (just too close for comfort), so the LM3x7 would be also cheap insurance to prevent toasting the LT1963.

Besides low noise, it is also low drop out (meaning it doesn't need much over 18V in to work well). For 18Vdc out, you'll be find. That is exactly what I'm going to do. I'll have to watch that maximum input though...


JF