[Ben_Tech]

Interesting discussion!

The main reason I am building (another) Gilmore is due to the fact my DC offset is 50-90mA (off to full on the volume control, respectively), causing me to worry about the life of my headphones....and there is some hum (very little but still). The amp does sound wonderful...I think all this is due to my poor part-matching skills.

Am I to understand that my servos may not be working right, and that if I raise the voltage of one rail in the power supply a smidge the DC offset may stabilize?

Pardon my noobishness.

 

[Arzela]

Ben, is this the bridged Gilmore?

 

[Ben_Tech]

No, I am now speaking of my first Gilmore, a single-ended amp.

Antness v1 PS board and a couple single-channel mini-boards, with a shallco series attenuator.

 

[jeffreyj]

Quote:

Not being snooty, i was probably just like you. 35 years ago.

So, then, you are 65-66 years old yet still resort to name-calling to add vigor, if not substance, to your arguments?

Like I already said: if there is something wrong with my reasoning, show me where I erred - preferably without resorting to ad hominem attacks - and I will gladly accept it. I don't make any claims about being the most knowledgeable person around w/r/t to audio amplifier design, but I do know a thing or two about power supplies, having made my living for many years designing them.

And, please... the user id is jeffreyj, not jefferyj.

 

[Arzela]

Ben, I thought you might be
using the balanced Gilmore,
which seems to require a servo
circuit different from that specified
by its creator.

50-90 mv is what the amp
typically, at least in my limited
experience, produces without a
servo; even less.

 

[kevin gilmore]

quote
If an amp design requires a regulated power supply to stop hum and PSRR problems, then that amp needs to be redesigned (paying better attention to trace widths, placement and routing more than anything else as it is difficult to make an amplifier so poor with modern op-amps).

Lets look at this one for a minute. What it is really saying is that
only amplifiers with huge amounts of open loop gain are good
amplifiers. This is exactly the kind of thinking that in the 1970's
generated lots of amplifiers that tested well and sounded like
crap. Who said anything about op-amps. Furthermore why
are so many people trying the n'th iteration of the multi-loop
stuff trying to get their amplifier to sound better. The open
loop gain is way out of hand. If you actually removed the
feedback and measured the distortion of such an amplifier
you would find that it is 5 to 10%. No wonder so many
opamp headphone amplifiers sound so lousy. You have
to also pay attention to slew rate induced distortion and
TIM.

Fact is PSRR has absolutely nothing to do with anything.
Some amplifiers designed to sound their best may require
fancy power supply designs. This is not a detriment. My
amplifier is a low open loop gain design. It has an open
loop distortion of less than .01%. It is fully dc coupled.
To keep the output dc at zero requires some fancy footwork.

If all amplifiers had hugh PSRR ratios, then why would
power supplies based on diden-jung and others exist.
What about single ended amplifiers whose PSRR is also
horrible.

Lets look at one extreme example. The Melos SHA-1. A lot
of people really like this amp. (i've never heard it). Very
strange design. In fact so strange that the fillament of the
power tube is in series with the minus power supply and is
tied directly to the headphones. (Its used as a current source).
What is the PSRR of this amplifier? ZERO!!!!! But evidently
this thing sounds great. Look and the design and scratch
your head. Any power supply wiggle (especially 60hz) ends
up directly in the headphones.

Some power supply designs have great static low resistance
at dc. But grado's and definitely akg's have significant inductive
components that rip right back thru the amplifer into the
power supply. So you need a power supply that is both stiff
at DC, and is stiff at all other audio frequencys too. Just
a couple of electrolytics that are by definition inductive in
nature ain't going to cut it. Same thing with simple 3 terminal
regulator supplies. You need power supplies that are active
in both directions.

Now power amplifiers are a different story. Both krell and
levinson regulate the supply to the main output transistors.
They do so for a number of reasons. One reason is that
otherwise with an unregulated power supply at low volume
levels the rails would exceed the breakdown voltage of
the transistors. So they regulate the power rails throwing
away as much as 30 volts at low power to make sure that
the amp can reach its desired power levels without blowing
up. Same thing i do in my uberamp. The best sounding output
transistors are not the ones with the highest breakdown
voltages.



quote
And, please... the user id is jeffreyj, not jefferyj.

OOPS. My middle name is jeffery...

 

[Ben_Tech]

Thanks again, Kevin, for the wonderful design (I'm sure you have heard that many times!).

Arzela, are you saying that 50-90mA is OK, and will not lay waste to headphones over time?

Thirdly, does anyone have a really compact Gilmore Dynamic Power Supply PCB made up or in .pcb format? It seems only the amplifier is readily available in pre-made PCB form. I am aware of the Justin Wilson board (I have one)...it's a little big.

 

[Arzela]

Ben, sorry if I was ambiguous;
no, 50 mv is not ok. But my
gilmore puts this much out
if I disconnect the servo.
After it the sevo gets to it, DC is 2-4 mv.

BoyElroy has some nice power
supply layouts. They aren't hard
to etch. I think, with the transfomer,
the're around 4 X 8". I made up
a layout that omits the 5 uf caps
that's 4 X 5.

 

[jasonhanjk]

The servo acts as a real time feedback to adjust headphone output to near 0. Very good concept indeed...

I belive it will be 0 once you match the LED and tune it a bit?

 

[Ben_Tech]

My mini-boards have the servo on them. I did match the LED's...*shrugs*

Thanks again for the input. I think I'll just make another, rather than do a lot of unsoldering.

By the way, Arzela, do you prefer the BoyElroy or Original configuration on your board? I am guessing the former. I will need to set the jumper when I build that amp.

I plan to have both a balanced and non-balanced headphone output. The + and - outputs from one channel of the amp will go to a separate jack, and the + and - outputs from the other channel will go to another separate jack. Both of these together will comprise the balanced output. Just the + from each channel will go to a single jack for the non-balanced. I think that is correct, anyway.

 

[Arzela]

Ben, before soldering, check which
servo circuit works first for the balanced lines

(BoyElroy's circuit is what I always had to use for my
balanced amps).

As for single ended:

Your wiring scheme is correct; however,
a single ended output from the balanced amp
(that is, O+ left and O+ right) needs a servo circuit different
from that needed for balanced outputs
(in which the left O+ and O- are at the same, possibly large,
potential. Same story for the right O+ and O-).

I think I was lucky in that the original
servo worked for me when I ran single ended
outputs from the first balanced amp I
built. The single ended DC
from another amp I built is not nulled by either circuit.

However, perhaps switching
the -inputs on the opa to ground, and
the +input on the opa's to O+ will
null the DC on the O+ inputs. If so, we
have a useable single ended output.
I will test this theory out after my next (and last) amp.

A 4pdt switch could then be used to change
between nulling DC for bridged output,
or single ended output. My board can accomadate this.

 

[Ben_Tech]

Neat. I will look forward to your observations on that.

Back to power supplies....

On the Justin Wilson Gilmore PS board (below) there are pads for a pair of bridge rectifiers...is it possible to substitute Schottky diodes for these? I am fairly clueless as to how that might work. Would I use eight diodes, one per hole, with the ends twisted together into four pairs?

I've never been able to get an answer to that one. *chuckle*

Thanks again!

 

[Arzela]

You'd need four diodes per rectifier.
With a bit of creativity, it could be done

 

[Ben_Tech]

Quote:

Originally posted by Arzela You'd need four diodes per rectifier. With a bit of creativity, it could be done

This I know, but things like polarity and how they would need to be attached to each other have me stumped.

 

[Arzela]

It's a commom circuit, I think. A "full-wave bridge rectifier". Someone correct me please,
if I'm wrong.

Like in the power supply here:
http://headwize2.powerpill.org/proje...lerano_prj.htm

From a google search:

http://members.aol.com/_ht_a/RAdelkopf/rect.htm

Now that I've looked at this,
it can't be done easily

You'd have to stick two diodes in
separate holes (from the tranny), tie the other ends of them together, and
stick this in a third hole (dc out).

The other two diodes would have
to share the trannie-end ties with the first two diodes. Thier ends would be tied, and placed in the
other DC out pad. What a mess...