What's your favourite voltage regs?
 

In article , David Looser
wrote:
"Jim Lesurf" wrote in message
...

Can only speculate at this point as I have never tried using these
circuits for heaters in audio amps. However...

With 'direct' heating the heater psu is - I presume - connected to
the same physical place as the cathode signal. Having a 'constant
current' PSU means the PSU looks like a high impedance connection, so
will be less likely to have a loading effect on signal drive to the
cathode if the PSU common mode isolation is poor. i.e. the problem is
that the cathode signal has to drive any loading it sees due to the
PSU, so a high impedance might be preferred.


No, it's nothing to do with loading.

So if I grounded one of the heater wires it would have no effect on the
signal? Are you sure of that? It seems implausible to me, but I haven't
done any work on DHT amps for audio. If you have, perhaps you can explain
more clearly.

Even with indirect heating with insulation between heater and cathode I'd
expect some coupling between heater and cathode due to capacitance or
inductance, so some (small) effects due to heater PSU loading could arise.
Can't presently see why this would be impossible for DHT, I'm afraid.

What is actually happening is that the resistance of the filament is
partially in series with the cathode current. So it acts as an
unbypassed cathode resistor creating a little bit of negative feedback.
Effectively shorting that out with a low impedence filament supply
reduces this negative feedback.

I think we were talking about the use of resistors *external* to the heater
element, and in which the signal current would not appear if the PSU were
genuinely floating. *Not* the resistance of the valve between its heater
terminals. The point of the resistors being to reduce any effect of the PSU
not being a perfectly floating one.

I suspect you have misunderstood the point that were made, but I will be
interested in your reply.

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.audiomisc.co.uk/index.html
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html

What's your favourite voltage regs?
 

David Looser wrote

Can only speculate at this point as I have never tried
using these circuits
for heaters in audio amps. However...

With 'direct' heating the heater psu is - I presume -
connected to the
same physical place as the cathode signal. Having a
'constant current' PSU
means the PSU looks like a high impedance connection, so
will be less
likely to have a loading effect on signal drive to the
cathode if the PSU
common mode isolation is poor. i.e. the problem is that
the cathode signal
has to drive any loading it sees due to the PSU, so a
high impedance might
be preferred.


No, it's nothing to do with loading. What is actually
happening is that the resistance of the filament is
partially in series with the cathode current. So it acts
as an unbypassed cathode resistor creating a little bit of
negative feedback. Effectively shorting that out with a
low impedence filament supply reduces this negative
feedback.

Isn't this just an alternative description of the same
phenomenon? How is cathode loading nothing to do with
feedback?

Whatever, looking at a typical filament circuit, such as the
first diagram in this quite interesting article on the
reduction of hum due to the thermal effect of AC heating:

http://members.aol.com/sbench/humbal.html

the parasitic LCR network to ground through the filament
supply source is shunted by just a few ohms, in the case of
fixed bias, or by that plus the combination of cathode
resistor and bypass capacitor, in the case of cathode bias.
I am surprised that it ends up being significant.

Also, if you replace the floating transformer winding in the
illustrated circuit with a voltage source or a current
source, also floating, it isn't immediately obvious to me
why the one should present a lower impedance to a *common
mode* signal than the other.

The big difference would be in the impedance presented to a
differential mode signal between the ends of the filament.
Assuming a DC source allows the hum pot to be dispensed
with, and that the two remaining resistors are balanced,
there should be no differential mode signal. So I don't see
why this should be significant either.

So why does it sound better?

Ian

What's your favourite voltage regs?
 

David Looser wrote:

"Jim Lesurf" wrote

Can only speculate at this point as I have never tried using these
circuits for heaters in audio amps. However...

With 'direct' heating the heater psu is - I presume - connected to the
same physical place as the cathode signal. Having a 'constant current' PSU
means the PSU looks like a high impedance connection, so will be less
likely to have a loading effect on signal drive to the cathode if the PSU
common mode isolation is poor. i.e. the problem is that the cathode signal
has to drive any loading it sees due to the PSU, so a high impedance might
be preferred.


No, it's nothing to do with loading. What is actually happening is that the
resistance of the filament is partially in series with the cathode current.
So it acts as an unbypassed cathode resistor creating a little bit of
negative feedback. Effectively shorting that out with a low impedence
filament supply reduces this negative feedback.

Finally some sense.

Graham

What's your favourite voltage regs?
 

"Ian Iveson" wrote in message
o.uk...
David Looser wrote

Can only speculate at this point as I have never tried
using these circuits
for heaters in audio amps. However...

With 'direct' heating the heater psu is - I presume -
connected to the
same physical place as the cathode signal. Having a
'constant current' PSU
means the PSU looks like a high impedance connection, so
will be less
likely to have a loading effect on signal drive to the
cathode if the PSU
common mode isolation is poor. i.e. the problem is that
the cathode signal
has to drive any loading it sees due to the PSU, so a
high impedance might
be preferred.


No, it's nothing to do with loading. What is actually
happening is that the resistance of the filament is
partially in series with the cathode current. So it acts
as an unbypassed cathode resistor creating a little bit of
negative feedback. Effectively shorting that out with a
low impedence filament supply reduces this negative
feedback.

Isn't this just an alternative description of the same
phenomenon? How is cathode loading nothing to do with
feedback?

Well maybe it is. In which case Jim Lesurf has a very odd way of describing
it. Is it usual to describe the function of cathode resistor bypass
capacitor in a conventional common-cathode amplifier as "loading the drive
signal to the cathode"?, I think not.


Also, if you replace the floating transformer winding in the
illustrated circuit with a voltage source or a current
source, also floating, it isn't immediately obvious to me
why the one should present a lower impedance to a *common
mode* signal than the other.

I agree 100%. The common-mode signal is the signal voltage on the cathode
(which will probably be zero anyway, unless an unbypassed cathode resistor
is used to provide local negative feedback). But the floating filament
supply cannot possibly load this signal even if it exists.

The big difference would be in the impedance presented to a
differential mode signal between the ends of the filament.
Assuming a DC source allows the hum pot to be dispensed
with, and that the two remaining resistors are balanced,
there should be no differential mode signal. So I don't see
why this should be significant either.

So why does it sound better?


Why indeed?. And why has no one mentioned the effect of reversing the
polarity of the filament supply in the DC case? This will make a far larger
difference to the operating conditions of the valve than using a current,
rather than a voltage, regulated filament supply.

Nor do I understand this sudden fashion for *single-ended* amplifiers, when
push-pull has been the preferred quality option right from the days when
DHTs were the latest technology.

David.

What's your favourite voltage regs?
 

Nor do I understand this sudden fashion for *single-ended* amplifiers, when
push-pull has been the preferred quality option right from the days when
DHTs were the latest technology.

David.- Hide quoted text -

Very good question. They sound different - push pull cleaner, SET
posessing some hard to define factor which many people like. But
PLEASE let's not go down that road in this thread - we've been there a
zillion times already. The technical discussion of how filament
supplies work is most interesting!

What's your favourite voltage regs?
 

Ian Iveson wrote:
David Looser wrote

Can only speculate at this point as I have never tried
using these circuits
for heaters in audio amps. However...

With 'direct' heating the heater psu is - I presume -
connected to the
same physical place as the cathode signal. Having a
'constant current' PSU
means the PSU looks like a high impedance connection, so
will be less
likely to have a loading effect on signal drive to the
cathode if the PSU
common mode isolation is poor. i.e. the problem is that
the cathode signal
has to drive any loading it sees due to the PSU, so a
high impedance might
be preferred.


No, it's nothing to do with loading. What is actually
happening is that the resistance of the filament is
partially in series with the cathode current. So it acts
as an unbypassed cathode resistor creating a little bit of
negative feedback. Effectively shorting that out with a
low impedence filament supply reduces this negative
feedback.


Isn't this just an alternative description of the same
phenomenon? How is cathode loading nothing to do with
feedback?

Whatever, looking at a typical filament circuit, such as the
first diagram in this quite interesting article on the
reduction of hum due to the thermal effect of AC heating:

http://members.aol.com/sbench/humbal.html

the parasitic LCR network to ground through the filament
supply source is shunted by just a few ohms, in the case of
fixed bias, or by that plus the combination of cathode
resistor and bypass capacitor, in the case of cathode bias.
I am surprised that it ends up being significant.

Also, if you replace the floating transformer winding in the
illustrated circuit with a voltage source or a current
source, also floating, it isn't immediately obvious to me
why the one should present a lower impedance to a *common
mode* signal than the other.

The big difference would be in the impedance presented to a
differential mode signal between the ends of the filament.
Assuming a DC source allows the hum pot to be dispensed
with, and that the two remaining resistors are balanced,
there should be no differential mode signal. So I don't see
why this should be significant either.

So why does it sound better?


My badly worded description was trying to suggest that the voltage
generated across the fil by the signal would be such that it would not
alter the DC current flowing, and the regulator would not be affected by
the presence of the signal across the fil. that's what I meany by common
mode.

In the case of a voltage reg, the signal would be seen as a voltage
variation and the regulator would try and track the signal to maintain a
constant voltage across the fil.

--
Nick

What's your favourite voltage regs?
 

"Nick Gorham" wrote in message
...


My badly worded description was trying to suggest that the voltage
generated across the fil by the signal would be such that it would not
alter the DC current flowing,

In fact the current flowing through the fil *will* be affected by the signal
flowing, part of the signal current will flow through the fil and add to
that from the regulated current source. This will be even more the case when
the external cathode connection is made to the +ve end of the fil as then
most of the signal current flows through the fil. When the fil is driven by
a voltage regulator then most of this signal current will flow through the
low impedence of the regulator instead.

the presence of the signal across the fil. that's what I meany by common
mode.

Though that's differential mode, not common mode.

In the case of a voltage reg, the signal would be seen as a voltage
variation and the regulator would try and track the signal to maintain a
constant voltage across the fil.

Indeed, thus reducing the negative feedback effect of having the signal
current flowing through the resistance of the fil.

David.

What's your favourite voltage regs?
 

In article , David Looser
wrote:
"Ian Iveson" wrote in message
o.uk...


Isn't this just an alternative description of the same phenomenon? How
is cathode loading nothing to do with feedback?

Well maybe it is. In which case Jim Lesurf has a very odd way of
describing it.

I think we are talking at cross-purposes. I was dealing with what I thought
Andy and Nick were referring to. i.e. external resistors in the heater PSU,
or their equivalent in terms of employing a 'constant current' heater PSU.
The external resistance would be in series with any the cathode *signal*
and the rest of the system via the PSU.

You are, I think, talking about the internal resistance in the valve used
for heating.

I have no idea which of these has any significant in an audio amp. :-) I
was simply commenting that if their is some loading via the PSU then
resistors would help reduce this. Seen equivalent effects in mm wave
oscillator tubes.

From the values people quote, though, the heater resistance in the valve
seems only to be of the order of 10 Ohms for the DHTs (?) If so, given the
other impedances I'd be surprised if that had a significant effect. But as
I say, I've not built or studied these types of amp...



I agree 100%. The common-mode signal is the signal voltage on the
cathode (which will probably be zero anyway, unless an unbypassed
cathode resistor is used to provide local negative feedback). But the
floating filament supply cannot possibly load this signal even if it
exists.

The problem Andy/Nick were referring to was, I think, that the PSU is *not*
perfectly isolated/floating. However if I have misunderstood, perhaps they
can say.

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.audiomisc.co.uk/index.html
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html

What's your favourite voltage regs?
 

"Jim Lesurf" wrote in message
...
In article , David Looser
wrote:

No, it's nothing to do with loading.

So if I grounded one of the heater wires it would have no effect on the
signal? Are you sure of that? It seems implausible to me, but I haven't
done any work on DHT amps for audio. If you have, perhaps you can explain
more clearly.

Generally speaking in a common-cathode amplifier the cathode *is* at ground
potential as far the signal is concerned. The only exception is when an
unbypassed cathode resistor is used to provide local negative feedback. The
filament supply is floating with respect to earth so it's not going to load
the signal at the cathode anyway.

Even with indirect heating with insulation between heater and cathode I'd
expect some coupling between heater and cathode due to capacitance or
inductance, so some (small) effects due to heater PSU loading could arise.
Can't presently see why this would be impossible for DHT, I'm afraid.

What is actually happening is that the resistance of the filament is
partially in series with the cathode current. So it acts as an
unbypassed cathode resistor creating a little bit of negative feedback.
Effectively shorting that out with a low impedence filament supply
reduces this negative feedback.

I think we were talking about the use of resistors *external* to the
heater
element, and in which the signal current would not appear if the PSU were
genuinely floating.

I'm not. I'm talking about the resistance of the filament, which is
effectively in series with the cathode current

*Not* the resistance of the valve between its heater
terminals. The point of the resistors being to reduce any effect of the
PSU
not being a perfectly floating one.

Why should the PSU *not* be a perfectly floating one?. We are talking audio
here, not HF, and the impedance to ground from the filament will be no more
than 100 ohms or so at the outside anyway.

I suspect you have misunderstood the point that were made,

Whilst it seems to me that your analysis of what is going on is totally
flawed.

but I will be
interested in your reply.

As I will be in yours

David.

What's your favourite voltage regs?
 

In article , David Looser
wrote:
"Jim Lesurf" wrote in message
...

*Not* the resistance of the valve between its heater
terminals. The point of the resistors being to reduce any effect of
the PSU not being a perfectly floating one.

Why should the PSU *not* be a perfectly floating one?.

Because in reality the PSU will exhibit parasitic capacitance, inductances,
etc. to its surroundings. For example parasitic capacitance via the
transformer windings (or electrostatic screen if present), and via the
wires. I'm afraid that in the real world you are unlikely to find a
perfectly floating PSU. Just ones which may sufficiently well isolated for
a given use. So the effects may be too small to worry about in some cases.
But that does not mean they don't exist, or can never have a significant
effect.

I had assumed you would be aware of this, and would have realised that we
were talking about different issues. If you have built and used PSUs you
will, I assume, be aware of possible problems with things like interference
getting through them, various kinds of parasitic, and how to deal with this.


I suspect you have misunderstood the point that were made,

Whilst it seems to me that your analysis of what is going on is totally
flawed.

Well, IIUR, the 'flaw' seems to be that I analysed what Andy and Nick were
talking about. Indeed I have had emails to that effect. However maybe we
are all out of step except for yourself. :-)

As I have pointed out repeatedly, though, I have never built or tested one
of the DHT audio amps, so can't comment on how significant these effects
may be in them. Can only point out that they exist. FWIW I have encountered
them in power tube PSUs of the kind I mentioned where they may affect the
control of the tube when using something like a lock loop.

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.audiomisc.co.uk/index.html
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html