[Prune]

I've already asked something about the hybrid electrocstatic headphone amp
befo http://www.headwize.com/images5/gilmore4_1.png

I'm asking the following while operating under the assumption that
resistors can potentially make a difference, and I don't want to discuss
whether this is so or not as there are supposed experts on both sides.
Instead, let me ask the following:

I'd like to use Vishay bulk metal foil resistors in the most impactful
places. Using them all over is out of the question as they are $8 a piece.
My question is which places ARE most impactful?

On first look at the schematic, I'd say the input stage source degeneration
200 Ohm resistors, the second stage 50K resistors, and the feedback 200K
resistors are the ones in series with the signal and would make most a
difference. This far, considering that I have to use pairs for the
feedback resistors as they are rated at 300 V whereas the voltage in that
position reaches about 375 V, that makes for 16 resistors or $128.

Other positions I'm not sure about. Are the gate grounding 100 ohm
resistors of the fourth stage, or other resistors parallel to the signal
path such as the input 500 K ones as likely to make a difference? What
about say the 100 ohm ones at the emitters of Q6/7, or the 10 K ones from
the -300 V single-transistor supply biasing the cathode-driving N-channel
followers?

Thanks in advance for any help.

[Patrick Turner]

Prune wrote:

I've already asked something about the hybrid electrocstatic headphone amp
befo http://www.headwize.com/images5/gilmore4_1.png

I'm asking the following while operating under the assumption that
resistors can potentially make a difference, and I don't want to discuss
whether this is so or not as there are supposed experts on both sides.
Instead, let me ask the following:

I'd like to use Vishay bulk metal foil resistors in the most impactful
places. Using them all over is out of the question as they are $8 a piece.
My question is which places ARE most impactful?

On first look at the schematic, I'd say the input stage source degeneration
200 Ohm resistors, the second stage 50K resistors, and the feedback 200K
resistors are the ones in series with the signal and would make most a
difference. This far, considering that I have to use pairs for the
feedback resistors as they are rated at 300 V whereas the voltage in that
position reaches about 375 V, that makes for 16 resistors or $128.

Other positions I'm not sure about. Are the gate grounding 100 ohm
resistors of the fourth stage, or other resistors parallel to the signal
path such as the input 500 K ones as likely to make a difference? What
about say the 100 ohm ones at the emitters of Q6/7, or the 10 K ones from
the -300 V single-transistor supply biasing the cathode-driving N-channel
followers?

Thanks in advance for any help.

I doubt the type of resistors you might use would make any difference to the
sound.
As long as the wattage and voltage rating is high enough then the the sound
shouldn't change much
with expensive resistors. Vishay have swallowed up many small resistor makers
such as
Beyschlag and I don't here the slightest difference between Beyschlag and any
other generic metal film.
Resistors are used in such vast numbers that the reliablity and goodness factor
is about
the same for the best from many makers.

The circuit you post contains so much complex SS circuitry with token EL34
triodes that I just cannot see what difference
resistors would make because would not the SS and tubes dominate the sound
quality?
Then you have balanced global NFB loops which will further reduce any "voice"
or timbral quality
that any active or passive component may impart onto the sound.

Patrick Turner.

[Prune]

Patrick Turner wrote in
:

The circuit you post contains so much complex SS circuitry with token
EL34 triodes that I just cannot see what difference

How is it token? The tubes contribute a good part of the voltage gain
here. All solid state stages except the P-channel common source one have
mostly current gain and very little voltage gain.

Then you have balanced global NFB loops which will further reduce any
"voice" or timbral quality
that any active or passive component may impart onto the sound.

The feedback is really low, actually, the way it goes into the sources
instead of gates of the input. It only cuts gain by about two.

[Sander deWaal]

Prune said:


I'd like to use Vishay bulk metal foil resistors in the most impactful
places. Using them all over is out of the question as they are $8 a piece.
My question is which places ARE most impactful?


I'm no expert, but I use metal film in critical places (low signal,
high impedance), and carbon film in others.
Wire wound in power supplies.

$8 per resistor is a joke. Mundus vult decipi.

--
"Due knot trussed yore spell chequer two fined awl miss steaks."

[Alan Holmes]

"Prune" wrote in message
4.76...
I've already asked something about the hybrid electrocstatic headphone amp
befo http://www.headwize.com/images5/gilmore4_1.png

I'm asking the following while operating under the assumption that
resistors can potentially make a difference, and I don't want to discuss
whether this is so or not as there are supposed experts on both sides.
Instead, let me ask the following:

I'd like to use Vishay bulk metal foil resistors in the most impactful
places. Using them all over is out of the question as they are $8 a
piece.
My question is which places ARE most impactful?

What the devil is 'impactful'?

Alan

[Phil Allison]

"Alan Holmes"

"Prune"

My question is which places ARE most impactful?

What the devil is 'impactful'?



** Likely means that " Prune" is in urgent need of an enema.





........ Phil

[Eeyore]

Prune wrote:

I've already asked something about the hybrid electrocstatic headphone amp
befo http://www.headwize.com/images5/gilmore4_1.png

I'm asking the following while operating under the assumption that
resistors can potentially make a difference, and I don't want to discuss
whether this is so or not as there are supposed experts on both sides.
Instead, let me ask the following:

Here are some useful facts that may help.

The *excess noise* ( that's the noise generated by current flow in - voltage
across - a resistor ) is affected by construction ( e.g. bulk metal - metal
film - carbon film - carbon composition ).

The *thermal noise* is the result of thermal agitation of electrons in the
resistor material and is always there. There are no 'noise free' resistors.

So, unless you have appreciable current in ( or voltage across ) a resistor the
type of construction doesn't matter too much.

The difference between bulk metal and metal film is so minute as to be totally
unimportant for all practical reasons and tube circuitry is itself inherently so
noisy that it's rather silly to consider the use of bulk metal in it.

Since the excess noise is determined by quality of construction as well as the
material, you can get some that are poorer than others but these'll be the
cheapy 'no name' branded jobbies.

Stick to Vishay's normal metal film and you'll be totally fine.

Graham

[Bob H.]

Why not build it with ratshack carbon comps? They're dirt cheap and
not too bad, and then you can experiment with better resistors after
you give it a listen for awhile. The actual signal chain seems pretty
short (with a bunch of current controls thrown in around it). Then
you'd have a reference point without breaking the bank.

I think the feedback resistor may show the most sensitivity, as I've
seen in the past. After you're satisfied with what you end up, try
(just for ****s and giggles) putting an old carbon comp in the feedback
line. See what you hear.

Good luck
Bob H.

Prune wrote:
I've already asked something about the hybrid electrocstatic headphone amp
befo http://www.headwize.com/images5/gilmore4_1.png

I'm asking the following while operating under the assumption that
resistors can potentially make a difference, and I don't want to discuss
whether this is so or not as there are supposed experts on both sides.
Instead, let me ask the following:

I'd like to use Vishay bulk metal foil resistors in the most impactful
places. Using them all over is out of the question as they are $8 a piece.
My question is which places ARE most impactful?

On first look at the schematic, I'd say the input stage source degeneration
200 Ohm resistors, the second stage 50K resistors, and the feedback 200K
resistors are the ones in series with the signal and would make most a
difference. This far, considering that I have to use pairs for the
feedback resistors as they are rated at 300 V whereas the voltage in that
position reaches about 375 V, that makes for 16 resistors or $128.

Other positions I'm not sure about. Are the gate grounding 100 ohm
resistors of the fourth stage, or other resistors parallel to the signal
path such as the input 500 K ones as likely to make a difference? What
about say the 100 ohm ones at the emitters of Q6/7, or the 10 K ones from
the -300 V single-transistor supply biasing the cathode-driving N-channel
followers?

Thanks in advance for any help.

[Eeyore]

"Bob H." wrote:

Why not build it with ratshack carbon comps? They're dirt cheap and
not too bad, and then you can experiment with better resistors after
you give it a listen for awhile. The actual signal chain seems pretty
short (with a bunch of current controls thrown in around it). Then
you'd have a reference point without breaking the bank.

I think the feedback resistor may show the most sensitivity, as I've
seen in the past. After you're satisfied with what you end up, try
(just for ****s and giggles) putting an old carbon comp in the feedback
line. See what you hear.

That's one way to do it for sure !

Graham

[Patrick Turner]

Prune wrote:

Patrick Turner wrote in
:

The circuit you post contains so much complex SS circuitry with token
EL34 triodes that I just cannot see what difference

How is it token? The tubes contribute a good part of the voltage gain
here. All solid state stages except the P-channel common source one have
mostly current gain and very little voltage gain.

Then you have balanced global NFB loops which will further reduce any
"voice" or timbral quality
that any active or passive component may impart onto the sound.

The feedback is really low, actually, the way it goes into the sources
instead of gates of the input. It only cuts gain by about two.

But why not just all tubes?

Patrick Turner.

[Prune]

Patrick Turner wrote in
:

But why not just all tubes?

The one direct-drive (OTL) schematic I found is on Broskie's Tubecad
Journal; it has only two stages but it's AC coupled and simulation shows
more distortion than this one... (plus, I already have all the parts save
the resistors)

[Prune]

"Bob H." wrote in
ups.com:

Good luck

Thanks

[Prune]

Eeyore wrote in
:

So, unless you have appreciable current in ( or voltage across ) a
resistor the type of construction doesn't matter too much.

OK, but appreciable relative to what? For example, the input 500 K
parallel to the incoming signal carry much more signal current that they
shunt to ground than the current used by the JFET gates, despite their high
values. Would that mean that these resistors would be impactful? It's
hard for me to understand whether resistors that are parallel to the
signal, thus shunting it to ground or some supply voltage, are as impactful
as ones through which the signal goes in series.

The difference between bulk metal and metal film is so minute as to be
totally unimportant for all practical reasons and tube circuitry is
itself inherently so noisy that it's rather silly to consider the use
of bulk metal in it.

Noise floor of this amp is very low in the simulator. Maybe because tubes
are used only in the last stage and with grounded grid.

[Prune]

Hi Andre, thanks for the thoughtful post. I contrast this with Phil,
who's smarts get obscured by his nasty manners...

"Andre Jute" wrote in
oups.com:

amps. I buy ali-cased power resistors out of the RS catalogue; they

A what? Never heard of "ali-cased"...

years: my ears and taste are vastly superior to anything at all that
any engineer can tell me.)

Haha, I like the modesty here ;P

Boutique resistors are actually good for the *opposite* of what you

Well, I guess I never thought of Vishays as boutique resistors...

Perspective includes other things besides distance and size. The
component with the greatest sonic influence in any amp is the
transformer (check the size of transformers I use in a quarter-Watt
amp). But, given a good transformer, the next sonic signature is the
power caps

Even in amps with very high PSRR? The schematic I posted has virtually
constant current through the power supply with signal-correlated
variation in the microamps. Though one of the #diyaudio IRC channel
members insists on omitting capacitors from any place in the amps
including the power supplies (all his power supplies are resistors,
chokes, and regulator tubes)...

Every rule has an exception. There is one place where spending money
on resistors is warranted, and that is in buying a DACT attenuator: it
is build on a Swiss switch with surface mounted resistors with gold
tracks on the board: ultra-silent, a very pleasing instrument.

I went el-cheapo and got Electroswitch from digikey; it's silver instead
of gold, but I doubt I can hear the difference. I populated it with
Roederstein resistors because of the huge amount necessary.
However the design is unusual, in that it's fed directly from the DAC
chip's balanced voltage outputs (AC coupled, the only place I used a
capacitor in my signal chain, and I've got Auricaps now as they are
cheap yet I can't hear the small values in high impedance circuits),
with a series resistor on each line, and the switch resistors shunting
between the inverted and non-inverted lines. That allows the use of
only a two-decked attenuator for a balanced two channel signal. I don't
know if this shunt instead of series position on the switch resistors
changes how they may impact the sound.

If you decide to go ahead, the most important resistors in any amp are
the cathode resistors because they set the bias, followed by the load

Hmm... but the schematic I posted has the tubes as grounded grid, not
grounded cathode... so would these be the 10 K resistors from the -300 V
supply to the cathodes, or would it not apply to this configuration as
they are not really cathode resistors?

resistors, followed by the grid leak resistors which complete an
important path and play their part in the plate load so that you don't

Again, referring to the grounded grid tubes in the schematic, would
these be the 100 ohm resistors grounding the grid? I never realized
they would matter.

want them to wander, followed by resistors in constant current
devices, followed by voltage dividers in the signal path (which
includes NFB if any).

So in the schematic I have, the 200 K feedback resistors would not be as
important as the various current source resistors? I'm a bit confused
here, because this seems opposite of other replies I got, so I'm hoping
you would elaborate.

in-line fuses: the signal passes through them too.

Well, I was thinking of it as what percentage of current through the
device is signal correlated, versus the DC (or AC before rectification)
it is superimposed on.

[Eeyore]

Prune wrote:

Eeyore wrote in
:

So, unless you have appreciable current in ( or voltage across ) a
resistor the type of construction doesn't matter too much.

OK, but appreciable relative to what? For example, the input 500 K
parallel to the incoming signal carry much more signal current that they
shunt to ground than the current used by the JFET gates, despite their high
values. Would that mean that these resistors would be impactful?

In this case their effect is shunted to ground by the source, so they are
insignificant contributors. You need to do a full Norton / Thevenin equivalency
in order to determine the full contribution of any section, in this case taking
into account the unshown source.


It's
hard for me to understand whether resistors that are parallel to the
signal, thus shunting it to ground or some supply voltage, are as impactful
as ones through which the signal goes in series.

It is quite a game ! Hence the need to do a full circuit analysis if you feel
the need. I wouldn't bother myself though since I know from experience that
ordinary metal film is 'overkill' anyway.


The difference between bulk metal and metal film is so minute as to be
totally unimportant for all practical reasons and tube circuitry is
itself inherently so noisy that it's rather silly to consider the use
of bulk metal in it.

Noise floor of this amp is very low in the simulator. Maybe because tubes
are used only in the last stage and with grounded grid.

I can't see a figure for the input noise voltage on the data sheet. It doesn't
look spectacularly low to me but I'm used to *very* quiet stuff.

Graham

[Eeyore]

Prune wrote:

Hi Andre, thanks for the thoughtful post. I contrast this with Phil,
who's smarts get obscured by his nasty manners...

"Andre Jute" wrote in
oups.com:

amps. I buy ali-cased power resistors out of the RS catalogue; they

A what? Never heard of "ali-cased"...

http://cgi.ebay.co.uk/100R-10W-Wirew...QQcmdZViewItem

Graham

[Prune]

Eeyore wrote in
:

In this case their effect is shunted to ground by the source, so they
are insignificant contributors. You need to do a full Norton /
Thevenin equivalency in order to determine the full contribution of
any section, in this case taking into account the unshown source.

It is quite a game ! Hence the need to do a full circuit analysis if
you feel the need. I wouldn't bother myself though since I know from
experience that ordinary metal film is 'overkill' anyway.

I wonder if I can use the simulator or another software to help with
this...

I can't see a figure for the input noise voltage on the data sheet. It
doesn't look spectacularly low to me but I'm used to *very* quiet
stuff.

I'm not sure what you mean by input noise voltage...

[Eeyore]

Prune wrote:

Eeyore wrote in
:

In this case their effect is shunted to ground by the source, so they
are insignificant contributors. You need to do a full Norton /
Thevenin equivalency in order to determine the full contribution of
any section, in this case taking into account the unshown source.

It is quite a game ! Hence the need to do a full circuit analysis if
you feel the need. I wouldn't bother myself though since I know from
experience that ordinary metal film is 'overkill' anyway.

I wonder if I can use the simulator or another software to help with
this...

Simply add a resistor across the input terminals representing the source
impedance.


I can't see a figure for the input noise voltage on the data sheet. It
doesn't look spectacularly low to me but I'm used to *very* quiet
stuff.

I'm not sure what you mean by input noise voltage...

It's specified in nV/sqrt Hz normally. Sometimes as however many uV ( but you
have to know the measuring bandwidth ).

Graham

[Prune]

Eeyore wrote in
:

Simply add a resistor across the input terminals representing the
source impedance.

Oh I see. In simulation I have a voltage source and I've set its impedance
to that of my source (~600 Ohms).

It's specified in nV/sqrt Hz normally. Sometimes as however many uV (
but you have to know the measuring bandwidth ).

You mean of the source? It's a voltage-output DAC driving the attenuator
and one-stage JFET buffer. So there is a single active device after the
DAC, and DAC specs are 120 dynamic range 102 THD+noise. Measuring noise I
only have the soundcard to do with and that's below its noise floor, so I
can't tell how close I'm to the datashee's optimal implementation.

[Eeyore]

Prune wrote:

Eeyore wrote in
:

Simply add a resistor across the input terminals representing the
source impedance.

Oh I see. In simulation I have a voltage source and I've set its impedance
to that of my source (~600 Ohms).

It's specified in nV/sqrt Hz normally. Sometimes as however many uV (
but you have to know the measuring bandwidth ).

You mean of the source? It's a voltage-output DAC driving the attenuator
and one-stage JFET buffer. So there is a single active device after the
DAC, and DAC specs are 120 dynamic range 102 THD+noise. Measuring noise I
only have the soundcard to do with and that's below its noise floor, so I
can't tell how close I'm to the datashee's optimal implementation.

Actually, I meant the noise of the 2SK389 JFET but it's unlikely to contribute
much noise to a signal that's already passed through an op-amp stage.

Graham

[Prune]

Eeyore wrote in
:

Actually, I meant the noise of the 2SK389 JFET but it's unlikely to
contribute much noise to a signal that's already passed through an
op-amp stage.

The DAC chip doesn't have an internal op amp; the switched capacitor output
stage produces voltage output directly.