On Thu, 24 Jan 2008 08:04:04 +0000, Nick Gorham
wrote:

Don Pearce wrote:

On Wed, 23 Jan 2008 23:56:54 +0000, Nick Gorham
wrote:


Don Pearce wrote:

On Wed, 23 Jan 2008 20:31:49 +0000, tony sayer
wrote:



It is.



but these DHT filaments are very sensitive

'ere wot abaht puttin summatt around the filament to keep it in?.

This might catch on;!...



And suppose you coated it with something that was really good at
boiling off electrons. And suppose you gave it a bit of thermal mass
so you could use AC rather than DC to heat it. That would make a much
better valve - wouldn't it?

d


Unless the new emmiter couldn't supply enough electrons at high current
peaks or for class C use, then you might still prefer a thorated fill.

Having said that, it can't be that simple, a 6c33c is indirectly heated,
but it does have a huge cathode.

Anyway, more files, I emailed these links to you yesterday Don, but
maybe they got lost after the night out :-).

http://www.lurcher.org/nick/heater/2a3-sq-V.wav
http://www.lurcher.org/nick/heater/2a3-sq-I.wav
http://www.lurcher.org/nick/heater/2a3-sweep-I.wav
http://www.lurcher.org/nick/heater/2a3-sweep-V.wav


Well, apart from a tiny difference around 25Hz which I'm pretty sure
is some interaction with the leakage inductance of the output tranny,
those curves overlay each other perfectly right up to 46kHz. The
difference you were identifying aurally was at the top end, wasn't it?
They are both on here.

http://81.174.169.10/odds/dual_fr.gif

If there is a difference it isn't in the frequency response. Do you
want to try the distortion test again in case something shows that we
didn't see using the shunt C method? Incidentally did you listen to
the test amp with and without the shunt C to check if you could still
hear the difference in that mode?

d

Its still using the shunt C, I haven't had chance to hook up the reg
yet. I will, but it might not be until the weekend.

I could try and listen to it, ut of course I only threw one chan together.

I could try the shunt C on my 211 that I do listen to, but would have
been a pain to measure.

I've just redone the comparison a bit more carefully - I removed that
last bit of odd LF stuff before doing the FFT, and both are now
absolutely identical, including at 25Hz.Overlaying the two traces on
one graph, you can only see the last one you put on.

d

--
Pearce Consulting
http://www.pearce.uk.com

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

That may or may not be so. However you may find effects like audible
clicks. Note also that the effect of a bypass capacitor reduces at *LF* as
the frequencies fall below the turnover of the relevant RC time constant.

You are moving into a whole new area here. If you want to open up the
discussion to deal with issues of the filtering of mains-borne interference
that's fine. But please don't try to pretend that that is what the
discussion was about before. Of course the effect of the capacitor reduces
at LF, but then so does that of the parasitic capacitance to ground from the
PSU and it's wiring!


Again, I can't speak for DHT audio amps. But I have encountered various
items of 'audiophile' kit that show clicks, etc, for reasons like this in
SS circuitry.

Indeed.

I can't say if there will be problems or not with DHT amps
whose design I have never seen or tested. TBH I doubt you can, either,
but...

No I haven't. That doesn't mean that I see any reason why interference
should not be a problem with DHT amps as well. But as I said this isn't what
the thread was about.


Whereas the distinction I point out was that you were referring to the
internal heater resistance of the valve, and I + others had been talking
about the impedance of the PSU.


Well no. I (and the others) were talking about the output impedance of the
power supply *in relation* to the impedance of the filament. But we were
talking about the impedance across the floating supply which is in parallel
with the filament. Only you, as far as I can see, have been talking about
the impedance *to ground* of the PSU.

David.

Don Pearce wrote:
On Thu, 24 Jan 2008 08:04:04 +0000, Nick Gorham
wrote:


Don Pearce wrote:


On Wed, 23 Jan 2008 23:56:54 +0000, Nick Gorham
wrote:



Don Pearce wrote:


On Wed, 23 Jan 2008 20:31:49 +0000, tony sayer
wrote:




It is.




but these DHT filaments are very sensitive

'ere wot abaht puttin summatt around the filament to keep it in?.

This might catch on;!...



And suppose you coated it with something that was really good at
boiling off electrons. And suppose you gave it a bit of thermal mass
so you could use AC rather than DC to heat it. That would make a much
better valve - wouldn't it?

d


Unless the new emmiter couldn't supply enough electrons at high current
peaks or for class C use, then you might still prefer a thorated fill.

Having said that, it can't be that simple, a 6c33c is indirectly heated,
but it does have a huge cathode.

Anyway, more files, I emailed these links to you yesterday Don, but
maybe they got lost after the night out :-).

http://www.lurcher.org/nick/heater/2a3-sq-V.wav
http://www.lurcher.org/nick/heater/2a3-sq-I.wav
http://www.lurcher.org/nick/heater/2a3-sweep-I.wav
http://www.lurcher.org/nick/heater/2a3-sweep-V.wav


Well, apart from a tiny difference around 25Hz which I'm pretty sure
is some interaction with the leakage inductance of the output tranny,
those curves overlay each other perfectly right up to 46kHz. The
difference you were identifying aurally was at the top end, wasn't it?
They are both on here.

http://81.174.169.10/odds/dual_fr.gif

If there is a difference it isn't in the frequency response. Do you
want to try the distortion test again in case something shows that we
didn't see using the shunt C method? Incidentally did you listen to
the test amp with and without the shunt C to check if you could still
hear the difference in that mode?

d

Its still using the shunt C, I haven't had chance to hook up the reg
yet. I will, but it might not be until the weekend.

I could try and listen to it, ut of course I only threw one chan together.

I could try the shunt C on my 211 that I do listen to, but would have
been a pain to measure.


I've just redone the comparison a bit more carefully - I removed that
last bit of odd LF stuff before doing the FFT, and both are now
absolutely identical, including at 25Hz.Overlaying the two traces on
one graph, you can only see the last one you put on.

d


Oh, well, at least that shows that the tests are at least reproducible :-)

What I need to take great care with when I use a real voltage reg, is
that the actual voltages the fill sees are close enough.

--
Nick.

On Thu, 24 Jan 2008 09:59:36 +0000, Nick Gorham
wrote:

Oh, well, at least that shows that the tests are at least reproducible :-)

What I need to take great care with when I use a real voltage reg, is
that the actual voltages the fill sees are close enough.

Yes, a small difference in emissivity could easily mask what we are
trying to measure.

d

--
Pearce Consulting
http://www.pearce.uk.com

Don Pearce wrote:
On Thu, 24 Jan 2008 09:59:36 +0000, Nick Gorham
wrote:


Oh, well, at least that shows that the tests are at least reproducible :-)

What I need to take great care with when I use a real voltage reg, is
that the actual voltages the fill sees are close enough.


Yes, a small difference in emissivity could easily mask what we are
trying to measure.

d


Well, this is with a real voltage reg, I have set the voltage across the
fil as close as possible, within 20mv anyway. The results are rather
interesting.


http://www.lurcher.org/nick/heater/2a3-1k-I.wav
http://www.lurcher.org/nick/heater/2a3-1k-V.wav
http://www.lurcher.org/nick/heater/2a3-swp-I.wav
http://www.lurcher.org/nick/heater/2a3-swp-V.wav

--
Nick

On Thu, 24 Jan 2008 21:34:13 +0000, Nick Gorham
wrote:

Don Pearce wrote:
On Thu, 24 Jan 2008 09:59:36 +0000, Nick Gorham
wrote:


Oh, well, at least that shows that the tests are at least reproducible :-)

What I need to take great care with when I use a real voltage reg, is
that the actual voltages the fill sees are close enough.


Yes, a small difference in emissivity could easily mask what we are
trying to measure.

d


Well, this is with a real voltage reg, I have set the voltage across the
fil as close as possible, within 20mv anyway. The results are rather
interesting.


http://www.lurcher.org/nick/heater/2a3-1k-I.wav
http://www.lurcher.org/nick/heater/2a3-1k-V.wav
http://www.lurcher.org/nick/heater/2a3-swp-I.wav
http://www.lurcher.org/nick/heater/2a3-swp-V.wav

Yup - absolutely identical frequency response again between voltage
and current.

But when it comes to distortion, the current driven version is
considerably worse - particularly the higher order harmonics. That is
sort of counter-intuitive, but it would account for the difference in
sound at the top end. I can just hear the difference between the two
by joining a one second chunk of each in sequence and letting them
loop. I can tell which bit I am in without looking at the monitor.

So yes, there is a difference now. The current feed has more of a
"valve" sound - ie more distorted. I guess that if you like that sound
you would find it preferable.

I'm totally stumped for a mechanism though. It has worked out
backwards from the way we were expecting.

d

--
Pearce Consulting
http://www.pearce.uk.com

Don Pearce wrote:
On Thu, 24 Jan 2008 21:34:13 +0000, Nick Gorham
wrote:


Don Pearce wrote:

On Thu, 24 Jan 2008 09:59:36 +0000, Nick Gorham
wrote:



Oh, well, at least that shows that the tests are at least reproducible :-)

What I need to take great care with when I use a real voltage reg, is
that the actual voltages the fill sees are close enough.


Yes, a small difference in emissivity could easily mask what we are
trying to measure.

d


Well, this is with a real voltage reg, I have set the voltage across the
fil as close as possible, within 20mv anyway. The results are rather
interesting.


http://www.lurcher.org/nick/heater/2a3-1k-I.wav
http://www.lurcher.org/nick/heater/2a3-1k-V.wav
http://www.lurcher.org/nick/heater/2a3-swp-I.wav
http://www.lurcher.org/nick/heater/2a3-swp-V.wav


Yup - absolutely identical frequency response again between voltage
and current.

But when it comes to distortion, the current driven version is
considerably worse - particularly the higher order harmonics. That is
sort of counter-intuitive, but it would account for the difference in
sound at the top end. I can just hear the difference between the two
by joining a one second chunk of each in sequence and letting them
loop. I can tell which bit I am in without looking at the monitor.

So yes, there is a difference now. The current feed has more of a
"valve" sound - ie more distorted. I guess that if you like that sound
you would find it preferable.

I'm totally stumped for a mechanism though. It has worked out
backwards from the way we were expecting.

d


Yes, I didn't expect this result. I am going to repeat tomorrow, just to
make sure I didn't do something stupid.

I would be tempted to suggest that the first stage is contributing
something, and there is some cancelation going on, but both the even and
odd order harmonics seem to be affected, so I don't believe thats the case.

I will try with a 300b tomorrow, and see if the results repeat.

I do tend to beleve the results are correct thogh, if I had done
somethiung stupid like got the fil temp differemt in the two cases, I
would have expected the fundimental to be alteres by more than it was.

--
Nick

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

That may or may not be so. However you may find effects like audible
clicks. Note also that the effect of a bypass capacitor reduces at
*LF* as the frequencies fall below the turnover of the relevant RC
time constant.

You are moving into a whole new area here. If you want to open up the
discussion to deal with issues of the filtering of mains-borne
interference that's fine. But please don't try to pretend that that is
what the discussion was about before.

Please don't confuse yourself by assuming that was what I was doing. :-)

I note you have snipped away the point you made to which the above was
a response. Plus you removed the other relevant points. This means you have
taken the above out of its context. Hence your confusion and the erronious
assumption that it was an attempt to "pretend" something.




Whereas the distinction I point out was that you were referring to the
internal heater resistance of the valve, and I + others had been
talking about the impedance of the PSU.


Well no. I (and the others) were talking about the output impedance of
the power supply *in relation* to the impedance of the filament. But we
were talking about the impedance across the floating supply which is in
parallel with the filament. Only you, as far as I can see, have been
talking about the impedance *to ground* of the PSU.

What I wrote was to clarify what Nick had said and give a better worded
explanation for a point he made. He emailed me later to confirm this.

So far as I can see, what then happened was that you dealt with another
matter to the one I was referring to, and have misinterpreted what I
was saying. I have tried to explain this to you. Simple matter of
cross purposes. Sorry if you are still unable to accept this, but - as
above - it may be because you are not keeping things in the relevant
contexts, and thus are forming inappropriate assumptions.

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

Don Pearce wrote:
On Thu, 24 Jan 2008 21:34:13 +0000, Nick Gorham
wrote:


Don Pearce wrote:

On Thu, 24 Jan 2008 09:59:36 +0000, Nick Gorham
wrote:



Oh, well, at least that shows that the tests are at least reproducible :-)

What I need to take great care with when I use a real voltage reg, is
that the actual voltages the fill sees are close enough.


Yes, a small difference in emissivity could easily mask what we are
trying to measure.

d


Well, this is with a real voltage reg, I have set the voltage across the
fil as close as possible, within 20mv anyway. The results are rather
interesting.


http://www.lurcher.org/nick/heater/2a3-1k-I.wav
http://www.lurcher.org/nick/heater/2a3-1k-V.wav
http://www.lurcher.org/nick/heater/2a3-swp-I.wav
http://www.lurcher.org/nick/heater/2a3-swp-V.wav


Yup - absolutely identical frequency response again between voltage
and current.

But when it comes to distortion, the current driven version is
considerably worse - particularly the higher order harmonics. That is
sort of counter-intuitive, but it would account for the difference in
sound at the top end. I can just hear the difference between the two
by joining a one second chunk of each in sequence and letting them
loop. I can tell which bit I am in without looking at the monitor.

So yes, there is a difference now. The current feed has more of a
"valve" sound - ie more distorted. I guess that if you like that sound
you would find it preferable.

I'm totally stumped for a mechanism though. It has worked out
backwards from the way we were expecting.

d


I tried to repeat the process, using a 300b this time, no sign of the
difference that seemed to exist with a 2a3.

I will repeat with the 2a3 again later.

But I just thought of a possible reason, see wht you think. unlike a
300b, the 2a3 I used has dual anodes, its effectivly two 45 in parallel
in the one envelope. I wonder if regulating the fil supply with a
current source instead of a voltage source is allowing different fil
temperatures in each half, and that inbalance is whats causing the
higher distortion.

Just a thought.

--
Nick

On Fri, 25 Jan 2008 22:12:26 +0000, Nick Gorham
wrote:

Don Pearce wrote:
On Thu, 24 Jan 2008 21:34:13 +0000, Nick Gorham
wrote:


Don Pearce wrote:

On Thu, 24 Jan 2008 09:59:36 +0000, Nick Gorham
wrote:



Oh, well, at least that shows that the tests are at least reproducible :-)

What I need to take great care with when I use a real voltage reg, is
that the actual voltages the fill sees are close enough.


Yes, a small difference in emissivity could easily mask what we are
trying to measure.

d


Well, this is with a real voltage reg, I have set the voltage across the
fil as close as possible, within 20mv anyway. The results are rather
interesting.


http://www.lurcher.org/nick/heater/2a3-1k-I.wav
http://www.lurcher.org/nick/heater/2a3-1k-V.wav
http://www.lurcher.org/nick/heater/2a3-swp-I.wav
http://www.lurcher.org/nick/heater/2a3-swp-V.wav


Yup - absolutely identical frequency response again between voltage
and current.

But when it comes to distortion, the current driven version is
considerably worse - particularly the higher order harmonics. That is
sort of counter-intuitive, but it would account for the difference in
sound at the top end. I can just hear the difference between the two
by joining a one second chunk of each in sequence and letting them
loop. I can tell which bit I am in without looking at the monitor.

So yes, there is a difference now. The current feed has more of a
"valve" sound - ie more distorted. I guess that if you like that sound
you would find it preferable.

I'm totally stumped for a mechanism though. It has worked out
backwards from the way we were expecting.

d


I tried to repeat the process, using a 300b this time, no sign of the
difference that seemed to exist with a 2a3.

I will repeat with the 2a3 again later.

But I just thought of a possible reason, see wht you think. unlike a
300b, the 2a3 I used has dual anodes, its effectivly two 45 in parallel
in the one envelope. I wonder if regulating the fil supply with a
current source instead of a voltage source is allowing different fil
temperatures in each half, and that inbalance is whats causing the
higher distortion.

Just a thought.

Provided they are wired in parallel they are pretty much
self-aligning. If a fil is at a lower temperature its resistance will
be lower, so it will take more current, hence more power. That will
apply whether they are voltage or current driven.. Of course if you
run them in series, whichever finds itself a little hotter will tend
to run away at the expense of the other.

d
--
Pearce Consulting
http://www.pearce.uk.com

--
Posted via a free Usenet account from http://www.teranews.com