I'm designing a power amp using either 6550 or KT-88 output tubes, running
about 525V B+. I've recently changed power transformers- the new one
has no 5V winding for the vacuum rectifier, so I'm going to use 1N4007s.

Need I be concerned with cathode stripping when using INDIRECTLY HEATED
tetrodes such as these types? I know that it is CERTAINLY a concern when
using directly heated types, and especially so at higher voltages.

I realize that this subject is rife with Old Wives' Tales, Urban Legends
and all manner of B.S. and malarky, of both opinions.

I know that cathode stripping is very real, and can be a serious problem.
The question I have for the group is - Is it a problem at 525V for
indirectly heated 6550 or KT-88 types? I'm not interested in hearsay or
legends, but in your direct personal experience: Have any of you ever
built an amplifier like this, and if so, have you found cathode stripping
to be a problem?

If so, I'd rather not resort to a separate front-panel switch for B+, and
am considering a vacuum tube "delay relay". Anyone have any experience in
this area? Are such things a Good Thing, or nothing but trouble?


Marty Dippel

Marty Dippel wrote:

I'm designing a power amp using either 6550 or KT-88 output tubes, running
about 525V B+. I've recently changed power transformers- the new one
has no 5V winding for the vacuum rectifier, so I'm going to use 1N4007s.

Need I be concerned with cathode stripping when using INDIRECTLY HEATED
tetrodes such as these types? I know that it is CERTAINLY a concern when
using directly heated types, and especially so at higher voltages.

I realize that this subject is rife with Old Wives' Tales, Urban Legends
and all manner of B.S. and malarky, of both opinions.

I know that cathode stripping is very real, and can be a serious problem.
The question I have for the group is - Is it a problem at 525V for
indirectly heated 6550 or KT-88 types? I'm not interested in hearsay or
legends, but in your direct personal experience: Have any of you ever
built an amplifier like this, and if so, have you found cathode stripping
to be a problem?

If so, I'd rather not resort to a separate front-panel switch for B+, and
am considering a vacuum tube "delay relay". Anyone have any experience in
this area? Are such things a Good Thing, or nothing but trouble?

Marty Dippel

Many famous brands of amps changed from tube rectifiers to silicon types the
day after
silicons were released, and the B+ was allowed to establish even faster than
using directly
heated tube rectifiers.
Then the gradual turn on of the outputs charged the cathode bypass caps up
slowly without any pain, and then pulled the B+ down a bit.
The situation with fixed bias is better because the bias establishes almost
instantly,
and there is even less cathode trauma than with cathode bias.
But really, if you measure the peak current flowing from the PS to output
tubes during heater
warm up after turn on, it never exceeds the idle current.

Directly heated cathodes exist in many tube rectifiers and there are no
delayed turn on circuits yet such rectifiers like the 5Y3 in old radios often
last
50 years.

If you had cathode bias, and you had a separate heater transformer to get the
OPV cathodes
all hot before turning on the B+, then the sudden turn on of B+ would perhaps
strain
the be-jesus out of the cathodes with a pulse of excess current because it
takes time for the
cathode caps to charge especially if we are using "modern"values of say 1,000
uF per tube.
Repeated every time the amp is turned on, this current surge would damage
tubes,
and maybe cause the mains fuse to blow.
So a slow turn on for the HT is very desirable, or a slow turn on after a
delay.
If you insist on using silicon rectifiers for the B+ and you want to apply
the B+ when the cathodes have been heated, then use yet another transformer
to make 2.5 amps at 5v for a GZ34 acting as a series resistor after the first
cap off the rectifier,
say 100 uF, and before the choke and the second cap.
So when the B+ tranny is turned on, so too is the GZ34 heater, so the B+
slowly turns on

In your case you could add a separate 15 VA heater transformer so a GZ34 could
be used as normal,
or as a series delay diode in series with the choke in the CLC.
The GZ34 is a bit slower to warm up than the output tubes.
They seem to last OK themselves with years of turn ons.

Many amps have directly heated tube rectifiers which turn on in about 3
seconds, so the B+
soars to say 580v, and then settles down to say 525 when the current draw
causes a
voltage drop across the tube rectifier plate resistance.

The rolls royce system is where all the heaters and B+ is allowed to come on
with a resistance to limit inrush current, and then a grid bias is immediately

applied of say -80v, and then after 30 seconds this is then gradually reduced
to allow the normal
grid bias to be applied. Solid state delay circuits are easy to dream up.
This way the tubes are in cut off until 30 seconds has elapsed, and regardless
of the biasing method,
either fixed or cathode, the current surge is very slowed indeed.

Where screen supplies are used, its possible to delay its turn on, and hence
slow down the anode current turn on, and since the screen supply is lower
current
than the plate supply, a delayed turn on of a GZ32 in series will slow down
the G2 voltage
rise from zero to the final value of say 300v, which is all you need if B+ is
525v
for KT88/6550, and you are not powering a very low RLa-a,
and the load is for a lot of class A.

But if anything goes wrong, watch out, you don't want inadvertant bias
failure.

Some SE amps damn well oscillate for a second during turn on at the very
instant of current
conduction in the cathodes.
Turn on and turn off oscillation bursts are a real PITA in many amps, not just
tube types.

Various tricks sometimes have to be employed to make the amps
non self destructing, and user friendly.

Patrick Turner.

I know that cathode stripping is very real, and can be a serious problem.
The question I have for the group is - Is it a problem at 525V for
indirectly heated 6550 or KT-88 types? I'm not interested in hearsay or
legends, but in your direct personal experience: Have any of you ever
built an amplifier like this, and if so, have you found cathode stripping
to be a problem?

If so, I'd rather not resort to a separate front-panel switch for B+, and
am considering a vacuum tube "delay relay". Anyone have any experience in
this area? Are such things a Good Thing, or nothing but trouble?


Marty Dippel

Try the DLS16, thermal relay. It looks like a small tube on an octal base,
and with 6V DC glows incandescently for about 45secs before switching.
Quite a conversation piece.

Iain

Thanks, Patrick -!

I was hoping I'd hear from you in this thread. Your reply is exactly
what I was hoping for- thoughtful, based on experience, and complete.

My further comments (a very few) are interspersed below

Thanks a million!




Marty Dippel wrote:

I'm designing a power amp using either 6550 or KT-88 output tubes, running
about 525V B+. I've recently changed power transformers- the new one
has no 5V winding for the vacuum rectifier, so I'm going to use 1N4007s.

Need I be concerned with cathode stripping when using INDIRECTLY HEATED
tetrodes such as these types? I know that it is CERTAINLY a concern when
using directly heated types, and especially so at higher voltages.

I realize that this subject is rife with Old Wives' Tales, Urban Legends
and all manner of B.S. and malarky, of both opinions.

I know that cathode stripping is very real, and can be a serious problem.
The question I have for the group is - Is it a problem at 525V for
indirectly heated 6550 or KT-88 types? I'm not interested in hearsay or
legends, but in your direct personal experience: Have any of you ever
built an amplifier like this, and if so, have you found cathode stripping
to be a problem?

If so, I'd rather not resort to a separate front-panel switch for B+, and
am considering a vacuum tube "delay relay". Anyone have any experience in
this area? Are such things a Good Thing, or nothing but trouble?

Marty Dippel

Many famous brands of amps changed from tube rectifiers to silicon types the
day after
silicons were released, and the B+ was allowed to establish even faster than
using directly
heated tube rectifiers.
Then the gradual turn on of the outputs charged the cathode bypass caps up
slowly without any pain, and then pulled the B+ down a bit.
The situation with fixed bias is better because the bias establishes almost
instantly,
and there is even less cathode trauma than with cathode bias.
But really, if you measure the peak current flowing from the PS to output
tubes during heater
warm up after turn on, it never exceeds the idle current.

Directly heated cathodes exist in many tube rectifiers and there are no
delayed turn on circuits yet such rectifiers like the 5Y3 in old radios often
last
50 years.

If you had cathode bias, and you had a separate heater transformer to get the
OPV cathodes
all hot before turning on the B+, then the sudden turn on of B+ would perhaps
strain
the be-jesus out of the cathodes with a pulse of excess current because it
takes time for the
cathode caps to charge especially if we are using "modern"values of say 1,000
uF per tube.
Repeated every time the amp is turned on, this current surge would damage
tubes,
and maybe cause the mains fuse to blow.


Thanks for clarifying this. I was heretofore relating the mechanism to
the act of somehow (how?) stripping the electrons from the cathode before
the cathode was up to temperature. Your description makes much more sense.

At least I'm using fixed bias, also obtained with a silicon rectifier.
And the bias supply is available well before the output tube heaters reach
temperature.

So a slow turn on for the HT is very desirable, or a slow turn on after a
delay.
If you insist on using silicon rectifiers for the B+ and you want to apply
the B+ when the cathodes have been heated, then use yet another transformer
to make 2.5 amps at 5v for a GZ34 acting as a series resistor after the first
cap off the rectifier,
say 100 uF, and before the choke and the second cap.
So when the B+ tranny is turned on, so too is the GZ34 heater, so the B+
slowly turns on

Yeah, well... I'd rather NOT use solid-state rects, but no 5V winding on
this xfmr, and I've already got about 50 lbs of iron going here, so I'd
rather not add a separate 5V xfmr just to power the HV rectifier.
But it is a thought that I should probably (re)consider.

In your case you could add a separate 15 VA heater transformer so a GZ34 could
be used as normal,
or as a series delay diode in series with the choke in the CLC.
The GZ34 is a bit slower to warm up than the output tubes.
They seem to last OK themselves with years of turn ons.

Many amps have directly heated tube rectifiers which turn on in about 3
seconds, so the B+
soars to say 580v, and then settles down to say 525 when the current draw
causes a
voltage drop across the tube rectifier plate resistance.

I had been using a 5U4 here, and of course this effect was exactly what I
was seeing- but even the 5U4 turns on a heckuva lot slower than 1N4007s!
An indirectly heated rectifier would be better, of course- at the cost of
eating up even more forward drop...


The rolls royce system is where all the heaters and B+ is allowed to come on
with a resistance to limit inrush current, and then a grid bias is immediately

applied of say -80v, and then after 30 seconds this is then gradually reduced
to allow the normal
grid bias to be applied. Solid state delay circuits are easy to dream up.
This way the tubes are in cut off until 30 seconds has elapsed, and regardless
of the biasing method,
either fixed or cathode, the current surge is very slowed indeed.

Where screen supplies are used, its possible to delay its turn on, and hence
slow down the anode current turn on, and since the screen supply is lower
current
than the plate supply, a delayed turn on of a GZ32 in series will slow down
the G2 voltage
rise from zero to the final value of say 300v, which is all you need if B+ is
525v
for KT88/6550, and you are not powering a very low RLa-a,
and the load is for a lot of class A.

But if anything goes wrong, watch out, you don't want inadvertant bias
failure.

Some SE amps damn well oscillate for a second during turn on at the very
instant of current
conduction in the cathodes.
Turn on and turn off oscillation bursts are a real PITA in many amps, not just
tube types.

Yes, they are. And this particular problem can be a *real* nightmare, as
you well know, I'm sure. Guaranteeing transient behavior is sometimes
less than straightforward!

Various tricks sometimes have to be employed to make the amps
non self destructing, and user friendly.

Yes, agreed. Solid-state, especially. I've designed SS amps for 20 years
as my professional career. This tube work I'm now interested in is a
"hobby" really- and a WHOLE LOT of fun.

Solid State amps had two areas that I often ran headlong into:

1. Turn-on and turn-off oscillation
2. Turn-on and turn-off transients (sometimes a "thump" so bad as to
damage speakers!)

The few tube designs I've built so far don't seem to have these problems-
at least at the level worth worrying about. (So far...)

Patrick Turner.


Thanks again, Patrick. You've been a great help!!

On Sun, 30 Jan 2005 19:13:38 GMT, Marty Dippel
wrote:

I was heretofore relating the mechanism to
the act of somehow (how?) stripping the electrons from the cathode before
the cathode was up to temperature.

The issue is with loss of cathode material, not current.


Yeah, well... I'd rather NOT use solid-state rects, but no 5V winding on
this xfmr, and I've already got about 50 lbs of iron going here, so I'd
rather not add a separate 5V xfmr just to power the HV rectifier.
But it is a thought that I should probably (re)consider.

Another possibility is the use of a large indirectly heated diode
like a TV damping diode as a pass device, or as rectifiers. They're
rugged, cheap, plentiful, and can withstand large DC voltages
between heater and cathode. 6.3 volt versions are common.

Bi-metal thermal delays have issues of arcing just before closing,
but are fine at lower currents, or controlling a relay.

Chris Hornbeck
"Don't be foolish, like the others." _Lola Montes_, 1955

Marty Dippel wrote:


If so, I'd rather not resort to a separate front-panel switch for B+, and
am considering a vacuum tube "delay relay". Anyone have any experience in
this area? Are such things a Good Thing, or nothing but trouble?


I don't know if "cathode stripping" is worth worrying about
or not, but *if* it is a real issue, you could use a pair
of TV damper diode tubes for indirectly heated rectifiers.
TV damper tubes feature very high heater-cathode voltage specs,
and thus you could heat them with the usual 6.3V heater
winding. And they will "turn on" at about the same time
as the output tubes. But maybe the damper diodes will
suffer their own cathode stripping? Just a thought.

Another idea, if you use fixed bias, might be to create
a negative soild state rectified supply for the bias, and
use an indirectly heated vacuum tube diode or triode
cathode follower that when it warms up, pulls the bias
up high enough to allow the output tube to go into normal
operation. Haven't tried this, it might be a stupid idea :-)

On Sun, 30 Jan 2005 15:00:19 +0000, Marty Dippel wrote:

snip

If so, I'd rather not resort to a separate front-panel switch for B+, and
am considering a vacuum tube "delay relay". Anyone have any experience in
this area? Are such things a Good Thing, or nothing but trouble?


I saw a neat idea somewhere. Use a normal, diode-connected, indirectly
heated valve to drive a small DC relay (anything within the rating of the
valve). Slightly underrun the heater to give a suitable time delay on
warm-up. If the relay is powered from the bias supply then it gives an
additional interlock. I haven't tried this, but the idea looks good.

--
Mick
(no M$ software on here... :-) )
Web: http://www.nascom.info
Web: http://projectedsound.tk

Marty Dippel said:

I had been using a 5U4 here, and of course this effect was exactly what I
was seeing- but even the 5U4 turns on a heckuva lot slower than 1N4007s!
An indirectly heated rectifier would be better, of course- at the cost of
eating up even more forward drop...

Nit-picking: indirectly heated rectifiers have a lower voltage drop
than directly heated ones.

In some old guitar amp, I replaced a 5U4G with a 5AR4.
HV went up 40 volts......enough to have an electrolytic blow on me.
That was of course many years ago, when I was still a young and
handsome but less knowledgeable bloke .

Be warned!

--
Sander de Waal
" SOA of a KT88? Sufficient. "

Best tube life is obtained by powering up filaments first then slowly
bringing up B+. This can be done in a lot of ways, IMO for new build
equipment the most elegant is to use separate filament and plate
transformers. Bias can be derived from the filament transformer so it's
present as soon as A supply is powered.

Attempts to do this with rectifier tubes bring up the question: what
brings _them_ up slowly? Aren't their heaters equally worthy of
protection? Sounds like tube bigotry to me.....

wrote:
Best tube life is obtained by powering up filaments first then slowly
bringing up B+. This can be done in a lot of ways, IMO for new build
equipment the most elegant is to use separate filament and plate
transformers. Bias can be derived from the filament transformer so it's
present as soon as A supply is powered.

Attempts to do this with rectifier tubes bring up the question: what
brings _them_ up slowly? Aren't their heaters equally worthy of
protection? Sounds like tube bigotry to me.....


Yea,sorta what I thought,just dumping the problem onto another tube..
*But*, Replacing a damper diode,or rectifier every year or two has got
to be cheaper than replacing a quad of 6550's or KT88's!

On Sun, 30 Jan 2005 19:33:17 +0000, Chris Hornbeck wrote:

On Sun, 30 Jan 2005 19:13:38 GMT, Marty Dippel
wrote:

I was heretofore relating the mechanism to
the act of somehow (how?) stripping the electrons from the cathode before
the cathode was up to temperature.

The issue is with loss of cathode material, not current.

Yes. I mis-spoke, here.
You're quite correct, I realize. Thanks.


Yeah, well... I'd rather NOT use solid-state rects, but no 5V winding on
this xfmr, and I've already got about 50 lbs of iron going here, so I'd
rather not add a separate 5V xfmr just to power the HV rectifier.
But it is a thought that I should probably (re)consider.

Another possibility is the use of a large indirectly heated diode
like a TV damping diode as a pass device, or as rectifiers. They're
rugged, cheap, plentiful, and can withstand large DC voltages
between heater and cathode. 6.3 volt versions are common.


I like this idea, or the one another fellow mentioned: use a tube with a
starved filament to close a relay when it has warmend up.

If I'm understanding the issure correctly, it sounds like to really solve
the problem, the B+ needs to be brought up slowly; a simple delay isn't
really good enough. In this case, the damper diode sounds like just the
thing.

It may be that with my amp, running fixed bias, that cathode stripping
won't be much of a problem. I'll have to observe turn-on behavior to
really know.



Bi-metal thermal delays have issues of arcing just before closing,
but are fine at lower currents, or controlling a relay.

Eeeeewwww... Not exactly a Good Thing, when it comes to higher-end audio!

Chris Hornbeck
"Don't be foolish, like the others." _Lola Montes_, 1955


A great many thanks to all who have replied to this thread. I've learned
a bit here, and your answers have all been well-considered and insightful.


Marty

On Sun, 30 Jan 2005 23:18:34 +0100, Sander deWaal wrote:

Marty Dippel said:

I had been using a 5U4 here, and of course this effect was exactly what I
was seeing- but even the 5U4 turns on a heckuva lot slower than 1N4007s!
An indirectly heated rectifier would be better, of course- at the cost of
eating up even more forward drop...

Nit-picking: indirectly heated rectifiers have a lower voltage drop
than directly heated ones.

Really! -This seemed so much the other way, I had never stopped to
actually compare specs! -I'm off to do this this evening...

Thanks!!


In some old guitar amp, I replaced a 5U4G with a 5AR4.
HV went up 40 volts......enough to have an electrolytic blow on me.
That was of course many years ago, when I was still a young and
handsome but less knowledgeable bloke .

Ouch! That's more than I'd expect--Certainly something to
watch for!


Be warned!

On Mon, 31 Jan 2005 02:17:43 GMT, Marty Dippel
wrote:

I mis-spoke, here.

It wasn't you who mis-spoke, but several responders.

I like this idea, or the one another fellow mentioned: use a tube with a
starved filament to close a relay when it has warmend up.

If I'm understanding the issure correctly, it sounds like to really solve
the problem, the B+ needs to be brought up slowly; a simple delay isn't
really good enough. In this case, the damper diode sounds like just the
thing.

It may be that with my amp, running fixed bias, that cathode stripping
won't be much of a problem. I'll have to observe turn-on behavior to
really know.

The issue is with physical damage to the tubes' cathodes.

I really doubt that it will be any sort of a problem. If you're
really wanting to do something exotic, like a choke-input power
supply, or very expensive output tubes, or running driver tubes
*way* above voltage ratings, then you might want to compulse.

I've built amps doing all of the above, and included another
startup routine, resistor in the primary of the filament transformer,
shorted out after a time delay. Yeah, compulsive, so what?

Chris Hornbeck
"Don't be foolish, like the others." _Lola Montes_, 1955

Attempts to do this with rectifier tubes bring up the question: what
brings _them_ up slowly? Aren't their heaters equally worthy of
protection? Sounds like tube bigotry to me.....


Yea,sorta what I thought,just dumping the problem onto another tube..
*But*, Replacing a damper diode,or rectifier every year or two has got
to be cheaper than replacing a quad of 6550's or KT88's!


IIRC I heard a rumor that rectifier tubes don't mind this
sort of abuse anywhere near as much as other tubes. But
I'd like conformation on this....

"Marty Dippel" wrote

If so, I'd rather not resort to a separate front-panel switch for
B+, and
am considering a vacuum tube "delay relay". Anyone have any
experience in
this area? Are such things a Good Thing, or nothing but trouble?

A timer costs very little to construct from logic and can be used
for a sequence of switching operations, driving suitable relays or
SS switches. You can incorporate bias current sensing quite easily,
and switch a dual-voltage heater supply to run at half power in
standby. A uP for resetting bias would be nice but I haven't seen it
done anywhere.

cheers, Ian

robert casey wrote:

Attempts to do this with rectifier tubes bring up the question: what
brings _them_ up slowly? Aren't their heaters equally worthy of
protection? Sounds like tube bigotry to me.....


Yea,sorta what I thought,just dumping the problem onto another tube..
*But*, Replacing a damper diode,or rectifier every year or two has got
to be cheaper than replacing a quad of 6550's or KT88's!


IIRC I heard a rumor that rectifier tubes don't mind this
sort of abuse anywhere near as much as other tubes. But
I'd like conformation on this....

I've heard that too,but I dunno.
Diodes have feelings (erm,I mean cathodes/filaments) too!

Well, here's two data points:

The spec sheets for REALLY big tubes (55CX500000) suggest you apply
heater voltage for many minutes before turning on plate power.


I have an old globe type-80 rectifier in this old A****er Kent, several
more in other radios. The tubes gotta be 60+ years old. None have any
warmup delay. The power transformer has a really high resistance, so
to get 300 volts out the no-load value for the first second or two is
waay up there near 500 volts, that's really pushing the PIV of the
diodes. But then again, they're fine after 60+ years, so one might
conclude that cathode or heater stripping isnt likely to happen with
the typical smaller tube.

Marty Dippel said:

Nit-picking: indirectly heated rectifiers have a lower voltage drop
than directly heated ones.

Really! -This seemed so much the other way, I had never stopped to
actually compare specs! -I'm off to do this this evening...

Thanks!!

The cathode surface is larger and more effective when a separate
cathode is used.

--
Sander de Waal
" SOA of a KT88? Sufficient. "

On Mon, 31 Jan 2005 19:56:11 +0100, Sander deWaal wrote:

Marty Dippel said:

Nit-picking: indirectly heated rectifiers have a lower voltage drop
than directly heated ones.

Really! -This seemed so much the other way, I had never stopped to
actually compare specs! -I'm off to do this this evening...

Thanks!!

The cathode surface is larger and more effective when a separate
cathode is used.

Makes sense, on both counts. Never thought of this, but having a cathode
allows one to optimize the heater to simply heat, and to optimize the
cathode to be an efficient emitter (pardon the pun). One is free to
choose both materials and geometries appropriate to each separate task.

Thanks for this! -never occurred to me..!

"robert casey" wrote in message
k.net...
Marty Dippel wrote:


If so, I'd rather not resort to a separate front-panel switch for B+, and
am considering a vacuum tube "delay relay". Anyone have any experience
in
this area? Are such things a Good Thing, or nothing but trouble?


I don't know if "cathode stripping" is worth worrying about
or not, but *if* it is a real issue, you could use a pair
of TV damper diode tubes for indirectly heated rectifiers.
TV damper tubes feature very high heater-cathode voltage specs,
and thus you could heat them with the usual 6.3V heater
winding. And they will "turn on" at about the same time
as the output tubes. But maybe the damper diodes will
suffer their own cathode stripping? Just a thought.

Another idea, if you use fixed bias, might be to create
a negative soild state rectified supply for the bias, and
use an indirectly heated vacuum tube diode or triode
cathode follower that when it warms up, pulls the bias
up high enough to allow the output tube to go into normal
operation. Haven't tried this, it might be a stupid idea :-)
That has crossed my mind as well, having a solid state bias supply that
would assure the power tubes were in full cut off until an indirectly heated
diode like a 6al5 would pull the bias to normal operating condition. Bias
supplies are low current so why not? If it's stupid, that makes two of us
;-)
Mark

On 2005-01-31 05:29:30 -0700, "Ian Iveson"
said:

"Marty Dippel" wrote

If so, I'd rather not resort to a separate front-panel switch for B+, and
am considering a vacuum tube "delay relay". Anyone have any experience in
this area? Are such things a Good Thing, or nothing but trouble?

A timer costs very little to construct from logic and can be used for a
sequence of switching operations, driving suitable relays or SS
switches. You can incorporate bias current sensing quite easily, and
switch a dual-voltage heater supply to run at half power in standby. A
uP for resetting bias would be nice but I haven't seen it done anywhere.

cheers, Ian

I've seen schematics that do B+ switch on controlled with a thermister.
According to the text that goes with them (they're in Rosinblit's
book), they take a few minutes to let the B+ come up. Whether they're
heated by their own resistance or by being in close proximity to a tube
I don't know.
--
-Jim Strickland

On Mon, 31 Jan 2005 08:45:57 +0000, gr wrote:

Well, here's two data points:

The spec sheets for REALLY big tubes (55CX500000) suggest you apply heater
voltage for many minutes before turning on plate power.

OK, I'm a bit late weighing in here.
AAMOF, one old BE who used to post here admitted to switching
on filament and B+ at the same time on powerful transmitters.
You CAN do that with certain types of tubes, as it is NOT
possible to cathode strip tubes with thoriated tungsten
filaments.

Cathode stripping on ordinary tube audio amplifiers is
a myth that's passed around to give audiophiles
the heebie-jeebies,and get them to spend lots of money
and time on fixes for an imaginary problem.
Nearly all tube audio amps have fixed bias, or a large
cathode resistor, that prevents cathode stripping,
unless the bias is lost.

Cathode stripping is not a gradual event. If you want
to see cathode stripping, look at what happens to a
rectifier when it's powered up into a shorted filter
cap. Arcing & blue light out the wazoo, then it tests
nearly dead. You can do the same thing to a power tube
or rectifier tube by flipping the power switch on & off
when it's only half warmed up (or conversely, half cooled off).
This is not something your average audiophile is going to do.

--
Ned Carlson Triode Electronics Chicago,IL USA
www.triodeelectronics.com

On Wed, 09 Feb 2005 02:31:02 -0600, Ned Carlson wrote:

On Mon, 31 Jan 2005 08:45:57 +0000, gr wrote:

Well, here's two data points:

The spec sheets for REALLY big tubes (55CX500000) suggest you apply heater
voltage for many minutes before turning on plate power.

OK, I'm a bit late weighing in here.
AAMOF, one old BE who used to post here admitted to switching
on filament and B+ at the same time on powerful transmitters.
You CAN do that with certain types of tubes, as it is NOT
possible to cathode strip tubes with thoriated tungsten
filaments.

Cathode stripping on ordinary tube audio amplifiers is
a myth that's passed around to give audiophiles
the heebie-jeebies,and get them to spend lots of money
and time on fixes for an imaginary problem.
Nearly all tube audio amps have fixed bias, or a large
cathode resistor, that prevents cathode stripping,
unless the bias is lost.

Cathode stripping is not a gradual event. If you want
to see cathode stripping, look at what happens to a
rectifier when it's powered up into a shorted filter
cap. Arcing & blue light out the wazoo, then it tests
nearly dead. You can do the same thing to a power tube
or rectifier tube by flipping the power switch on & off
when it's only half warmed up (or conversely, half cooled off).
This is not something your average audiophile is going to do.


Thanks, Ned! -I didn't realize it wasn't cumulative, but that it happened
all at once.

Judging from what you and Patrick and a few others have said, it looks
like it just won't be a problem for me, since I'm running KT88s in fixed
bias (ultralinear) with about 525 or 550V B+.

I've switched my design over from 6L6s to KT88s since I'm running higher
B+ than I'm comfortable with when using 6L6s. I'm amazed at how different
the KT88s are... Thought It'd be mostly a drop-in replacement, with a
little bias change, but Nooooooo.... It's a whole new design problem
now- but still havin' FUN!! Stopped by your shop the other day to pick up
a few JJ 12AU7s- nice tubes! -I like 'em better than the old GEs I had
been using- a little more linear, lower hum (less bleed-through from the
ac heater) and less microphony- probably due to the shorter, more rigid
structure.

Thanks for your response; I appreciate hearing from you!


Marty

Marty,
Have you ever thought of changing the KT88s to Triode Mode? What do you
think about that Thanks.
Cordially,
west

"Marty Dippel" wrote in message
news


On Wed, 09 Feb 2005 02:31:02 -0600, Ned Carlson wrote:

On Mon, 31 Jan 2005 08:45:57 +0000, gr wrote:

Well, here's two data points:

The spec sheets for REALLY big tubes (55CX500000) suggest you apply
heater
voltage for many minutes before turning on plate power.

OK, I'm a bit late weighing in here.
AAMOF, one old BE who used to post here admitted to switching
on filament and B+ at the same time on powerful transmitters.
You CAN do that with certain types of tubes, as it is NOT
possible to cathode strip tubes with thoriated tungsten
filaments.

Cathode stripping on ordinary tube audio amplifiers is
a myth that's passed around to give audiophiles
the heebie-jeebies,and get them to spend lots of money
and time on fixes for an imaginary problem.
Nearly all tube audio amps have fixed bias, or a large
cathode resistor, that prevents cathode stripping,
unless the bias is lost.

Cathode stripping is not a gradual event. If you want
to see cathode stripping, look at what happens to a
rectifier when it's powered up into a shorted filter
cap. Arcing & blue light out the wazoo, then it tests
nearly dead. You can do the same thing to a power tube
or rectifier tube by flipping the power switch on & off
when it's only half warmed up (or conversely, half cooled off).
This is not something your average audiophile is going to do.


Thanks, Ned! -I didn't realize it wasn't cumulative, but that it happened
all at once.

Judging from what you and Patrick and a few others have said, it looks
like it just won't be a problem for me, since I'm running KT88s in fixed
bias (ultralinear) with about 525 or 550V B+.

I've switched my design over from 6L6s to KT88s since I'm running higher
B+ than I'm comfortable with when using 6L6s. I'm amazed at how different
the KT88s are... Thought It'd be mostly a drop-in replacement, with a
little bias change, but Nooooooo.... It's a whole new design problem
now- but still havin' FUN!! Stopped by your shop the other day to pick up
a few JJ 12AU7s- nice tubes! -I like 'em better than the old GEs I had
been using- a little more linear, lower hum (less bleed-through from the
ac heater) and less microphony- probably due to the shorter, more rigid
structure.

Thanks for your response; I appreciate hearing from you!


Marty

On Fri, 25 Feb 2005 12:51:06 +0000, west wrote:

Marty,
Have you ever thought of changing the KT88s to Triode Mode? What do you
think about that Thanks.
Cordially,
west


Oh, I dunno.. I haven't given it SERIOUS consideration, but I'm tempted
to stick with ultralinear. P-P U/L offers better plate efficiency than
does P-P triode mode, and also is more tolerant of a mismatched load-
which is something seldom tested for on the bench, but makes a great deal
of difference to the ear. By "more tolerant" I mean that the distortion
of the amp is flatter across a range of load impedances than is a P-P
triode configuration.

I've always been curious about P-P triodes, though- one day I will build
an amp that way just to see...

This is my first tube project in about 30 years. It started when I had
the idea of building an amp "just for fun" - and I thought to build it out
of "classic" tubes- a 6SN7 phase splitter followed by a pair of 6L6s in
push pull. I soon discovered that I had just barely enough gain to
produce clipping with 1 V RMS input- so I couldn't spare any gain for
feedback. So in goes another 6SN7 as a LTP voltage amp after the floating
paraphase splitter. So now I have almost TOO MUCH gain- 20 dB feedback is
a little high compared to most designs, but it does wonders in correcting
all manner of evils...

Hum was a bit of a problem; turned out to be "bleed-through" from the (AC)
heater to the cathode. Using a 12AU7 instead of the first 6SN7 helped by
about 10 dB. The voltage amp stays as a 6SN7 because I can use the higher
plate voltage rating to produce more undistorted output swing to feed the
output stage.

Well, it got to the point of "first listen" and I was pleasantly
surprised. Even my wife heard a distinct improvement over the solid-state
amp we had been using. -She usually can't hear squat as far as audio
nuances, but she said the improvement was obvious. And it was to me, too.
"transparency", "pace", "involvement", "extension", all that. You betcha. :-)

Anyway, she said I ought to build a nice cabinet or case or base or
something- make it look like something. So I designed one out of
hardwood- but I'm all thumbs when it comes to woodworking; even a simple
task like this presents a tremendous challenge. So I thought I'd build a
stereo amp instead of two monoblocs, to minimize the pain of woodworking.

So Hammond makes a power transformer that'll handle the current of two
amps, and it's about 25V higher on the secondary voltage. And it has no
5V winding for the 5U4- so I decide to go with 1N4007s. Which gives me
another 20 or 25 V, and by now I'm above what the 6L6s will stand. Which
is where this thread started, BTW.

So I've always been wanting to use KT88s anyway- and here's my chance.
So now it's a whole new design- the KT88 acts quite differently than does
the 6L6. As the bias current is increased, the crossover distortion
diminishes by degrees, until the plates are glowing red...
With 6L6s, the xover disappeared at some point before plate dissipation
was exceeded.

Also, I was getting about 1% THD out of my 6L6s, (no feedback) and the
KT88s produce about 3%. But some of this could be from the driver stage,
which is where I am at the present time. And 3% isn't all that bad; my
20dB feedback should reduce this to 0.3%, but still, I'd like to get the
amp working as best it can before I go applying FB.



I'll probably stick with floating paraphase for the splitter- it's good
enough, and I want to finish this design at some point in my life...

But the splitter is enormously important, and at some point I want to
revisit this aspect of the design.

Split load splitters are limited in their output swing, and they're not
really symmetrical- one output is more-or-less a cathode follower, and the
other is a plate load.

My floating paraphase has the property of almost cancelling the distortion
on one output (the mu-follower), while the other is what one might expect
from a plate-driven output.

Patrick has had good luck with his LTP configuration. It should offer
excellent symmetry, especially when fed from the bipolar current source
he's fond of. I've tried using a LTP in the "traditional" config with a
simple cathode resistor, and have had to fudge the gains of the two
sections because the cathode isn't really fed by a constant current.
I've briefly made a stab at Patrick's configuration but couldn't get it to
work in the 20 minutes I spent with it so far, so I went back to my
floating paraphase. But this is something I want to investigate further.

Really, it seems that an interstage transformer(!) may be best after all,
but such things must be handmade, by and large. They're not usually found
off-the-shelf (I'm thankful we can find OUTPUT transformers!) and they're
so specific to the design that each design requires a slightly different
one than the next.

Well, enough blathering about everything under the sun. This thread
started off as "cathode stripping" and has already digressed through
triodes Vs Pentodes, onto phase splitters.

Anyway, it's great to be working on this stuff again after all these
years; I'm amazed and pleased to find that the tiny little niche market
has grown into such a significant slice of the pie.

And it's just plain FUN to be doing this!

Marty Dippel wrote:

On Fri, 25 Feb 2005 12:51:06 +0000, west wrote:

Marty,
Have you ever thought of changing the KT88s to Triode Mode? What do you
think about that Thanks.
Cordially,
west

Oh, I dunno.. I haven't given it SERIOUS consideration, but I'm tempted
to stick with ultralinear. P-P U/L offers better plate efficiency than
does P-P triode mode, and also is more tolerant of a mismatched load-
which is something seldom tested for on the bench, but makes a great deal
of difference to the ear. By "more tolerant" I mean that the distortion
of the amp is flatter across a range of load impedances than is a P-P
triode configuration.

As long as you saty within the power ceiling of the triode amp there is
no reason to suggest UL is better in any way than triode.



I've always been curious about P-P triodes, though- one day I will build
an amp that way just to see...

Its hard to build a bad sounding PP triode amp.





This is my first tube project in about 30 years. It started when I had
the idea of building an amp "just for fun" - and I thought to build it out
of "classic" tubes- a 6SN7 phase splitter followed by a pair of 6L6s in
push pull. I soon discovered that I had just barely enough gain to
produce clipping with 1 V RMS input- so I couldn't spare any gain for
feedback. So in goes another 6SN7 as a LTP voltage amp after the floating
paraphase splitter. So now I have almost TOO MUCH gain- 20 dB feedback is
a little high compared to most designs, but it does wonders in correcting
all manner of evils...

So you have a williamson.

Not a bad design.



Hum was a bit of a problem; turned out to be "bleed-through" from the (AC)
heater to the cathode. Using a 12AU7 instead of the first 6SN7 helped by
about 10 dB. The voltage amp stays as a 6SN7 because I can use the higher
plate voltage rating to produce more undistorted output swing to feed the
output stage.

DC on the 6SN7 heaters will remove all heater-cathode leaking.


Well, it got to the point of "first listen" and I was pleasantly
surprised. Even my wife heard a distinct improvement over the solid-state
amp we had been using. -She usually can't hear squat as far as audio
nuances, but she said the improvement was obvious. And it was to me, too.
"transparency", "pace", "involvement", "extension", all that. You betcha. :-)

yep, it is obvious to many ppl...



Anyway, she said I ought to build a nice cabinet or case or base or
something- make it look like something. So I designed one out of
hardwood- but I'm all thumbs when it comes to woodworking; even a simple
task like this presents a tremendous challenge. So I thought I'd build a
stereo amp instead of two monoblocs, to minimize the pain of woodworking.

So Hammond makes a power transformer that'll handle the current of two
amps, and it's about 25V higher on the secondary voltage. And it has no
5V winding for the 5U4- so I decide to go with 1N4007s. Which gives me
another 20 or 25 V, and by now I'm above what the 6L6s will stand. Which
is where this thread started, BTW.

So I've always been wanting to use KT88s anyway- and here's my chance.
So now it's a whole new design- the KT88 acts quite differently than does
the 6L6. As the bias current is increased, the crossover distortion
diminishes by degrees, until the plates are glowing red...
With 6L6s, the xover disappeared at some point before plate dissipation
was exceeded.

There is no need to run KT88 at high Pd; 20 watts is plenty,
and if RL is just right, 30 watts of class AB1 triode is available.
The thd at normal listening levels is SFA.





Also, I was getting about 1% THD out of my 6L6s, (no feedback) and the
KT88s produce about 3%.

This seems odd to me; usually KT88 or 6550 give more power and lower thd.

But some of this could be from the driver stage,
which is where I am at the present time. And 3% isn't all that bad; my
20dB feedback should reduce this to 0.3%, but still, I'd like to get the
amp working as best it can before I go applying FB.

KT88 and KT66 and 6L6 all require a similar bias voltage and thus drive voltage.
So how can the thd from the drive amp suddenly be so much higher for
KT88 compared to 6L6?




I'll probably stick with floating paraphase for the splitter- it's good
enough, and I want to finish this design at some point in my life...

But the splitter is enormously important, and at some point I want to
revisit this aspect of the design.

Split load splitters are limited in their output swing, and they're not
really symmetrical- one output is more-or-less a cathode follower, and the
other is a plate load.

My floating paraphase has the property of almost cancelling the distortion
on one output (the mu-follower), while the other is what one might expect
from a plate-driven output.

Patrick has had good luck with his LTP configuration.

My idea follows the Mullard 520 tradition.
I think my low U triode based LTP with CCS perform much better than
the Mullard 520.

It should offer
excellent symmetry, especially when fed from the bipolar current source
he's fond of.

CCS and equal RL values gurantee equal drive voltages to each output tube.


I've tried using a LTP in the "traditional" config with a
simple cathode resistor, and have had to fudge the gains of the two
sections because the cathode isn't really fed by a constant current.

The fixed common cathode resistor of about 10k used in countless
designs and the use of disimilar anode RL and perhaps with a trimmer R
is a pretty decent way to get a low thd drive to the output tubes.
The dissimilar RLs needed with low U triodes in a diffamp
create a small level of unbalanced 2H in the outputs;
most diffamp thd is 3H.


I've briefly made a stab at Patrick's configuration but couldn't get it to
work in the 20 minutes I spent with it so far, so I went back to my
floating paraphase. But this is something I want to investigate further.

Really, it seems that an interstage transformer(!) may be best after all,
but such things must be handmade, by and large. They're not usually found
off-the-shelf (I'm thankful we can find OUTPUT transformers!) and they're
so specific to the design that each design requires a slightly different
one than the next.

And something from Lundahl or Plitron will work OK
but it is expensive.
The phase shifts introduced by an IST prohibit any high levels of NFB.
That's why Williamson, Walker, Leak and later McIntosh swung right away from ISTs.

They sound great in guitar amps.





Well, enough blathering about everything under the sun. This thread
started off as "cathode stripping" and has already digressed through
triodes Vs Pentodes, onto phase splitters.

Anyway, it's great to be working on this stuff again after all these
years; I'm amazed and pleased to find that the tiny little niche market
has grown into such a significant slice of the pie.

And it's just plain FUN to be doing this!

Its a satisfying experience to build a tube amp...

Patrick Turner.

On Sun, 27 Feb 2005 04:31:30 +1100, Patrick Turner wrote:



Marty Dippel wrote:

On Fri, 25 Feb 2005 12:51:06 +0000, west wrote:

Marty,
Have you ever thought of changing the KT88s to Triode Mode? What do you
think about that Thanks.
Cordially,
west

Oh, I dunno.. I haven't given it SERIOUS consideration, but I'm tempted
to stick with ultralinear. P-P U/L offers better plate efficiency than
does P-P triode mode, and also is more tolerant of a mismatched load-
which is something seldom tested for on the bench, but makes a great deal
of difference to the ear. By "more tolerant" I mean that the distortion
of the amp is flatter across a range of load impedances than is a P-P
triode configuration.

As long as you saty within the power ceiling of the triode amp there is
no reason to suggest UL is better in any way than triode.


I've always been curious about P-P triodes, though- one day I will build
an amp that way just to see...



Its hard to build a bad sounding PP triode amp.

Sounds like this would be the case; one day I'll try it and see -!






This is my first tube project in about 30 years. It started when I had
the idea of building an amp "just for fun" - and I thought to build it out
of "classic" tubes- a 6SN7 phase splitter followed by a pair of 6L6s in
push pull. I soon discovered that I had just barely enough gain to
produce clipping with 1 V RMS input- so I couldn't spare any gain for
feedback. So in goes another 6SN7 as a LTP voltage amp after the floating
paraphase splitter. So now I have almost TOO MUCH gain- 20 dB feedback is
a little high compared to most designs, but it does wonders in correcting
all manner of evils...

So you have a williamson.

Not a bad design.


Yeah; it's pretty close to the Williamson. And yes, not a bad
design.


Hum was a bit of a problem; turned out to be "bleed-through" from the (AC)
heater to the cathode. Using a 12AU7 instead of the first 6SN7 helped by
about 10 dB. The voltage amp stays as a 6SN7 because I can use the higher
plate voltage rating to produce more undistorted output swing to feed the
output stage.

DC on the 6SN7 heaters will remove all heater-cathode leaking.

If I can do it without resorting to this bother, I'm all for it -! But
yes, the idea's occurred- and it appears quite often in the literature.
This thing is already 2/3 power supply and 1/3 amplifier; I'd rather not
resort to a well-filtered DC supply for the heaters on top of everything
else...

Seems like if one is building a phono preamp, there'd be an excuse, but
this is LINE LEVEL, for goodness sake. You'd think such measures wouldn't
be required here... but I see it often enough. One of the disadvantages
of tubes, I suppose: heaters.



Well, it got to the point of "first listen" and I was pleasantly
surprised. Even my wife heard a distinct improvement over the
solid-state amp we had been using. -She usually can't hear squat as far
as audio nuances, but she said the improvement was obvious. And it was
to me, too. "transparency", "pace", "involvement", "extension", all
that. You betcha. :-)

yep, it is obvious to many ppl...

Just "more musical" to my ears (whatever that might mean...)



Anyway, she said I ought to build a nice cabinet or case or base or
something- make it look like something. So I designed one out of
hardwood- but I'm all thumbs when it comes to woodworking; even a simple
task like this presents a tremendous challenge. So I thought I'd build a
stereo amp instead of two monoblocs, to minimize the pain of woodworking.

So Hammond makes a power transformer that'll handle the current of two
amps, and it's about 25V higher on the secondary voltage. And it has no
5V winding for the 5U4- so I decide to go with 1N4007s. Which gives me
another 20 or 25 V, and by now I'm above what the 6L6s will stand. Which
is where this thread started, BTW.

So I've always been wanting to use KT88s anyway- and here's my chance.
So now it's a whole new design- the KT88 acts quite differently than does
the 6L6. As the bias current is increased, the crossover distortion
diminishes by degrees, until the plates are glowing red...
With 6L6s, the xover disappeared at some point before plate dissipation
was exceeded.

There is no need to run KT88 at high Pd; 20 watts is plenty,
and if RL is just right, 30 watts of class AB1 triode is available.
The thd at normal listening levels is SFA.


The spec sheets call for 50 mA Ib0 when running 550V B+ in U/L. So that's
about 22W, which, as you say, should be plenty...

I haven't refined my design to the point where I'm willing to argue with
the spec sheet guidelines; see below...



Also, I was getting about 1% THD out of my 6L6s, (no feedback) and the
KT88s produce about 3%.

This seems odd to me; usually KT88 or 6550 give more power and lower
thd.

Yup. So I must be doing something wrong... I'm seeing about 50W out of
a pair of KT88s at clipping; I should be approaching 100W given 575V and
U/L.

And I was seeing only 30W out of a pr of 6L6s with 450V B+.

So I've *GOT* to be doing something very wrong here... But so help me, It
escapes me what it could be. "Analysis continues", as we say...



But some of this could be from the driver stage, which is where I am at
the present time. And 3% isn't all that bad; my 20dB feedback should
reduce this to 0.3%, but still, I'd like to get the amp working as best
it can before I go applying FB.

KT88 and KT66 and 6L6 all require a similar bias voltage and thus drive
voltage. So how can the thd from the drive amp suddenly be so much
higher for KT88 compared to 6L6?


Hmmmm... Dunno about that. The 6L6 needed about -50V Vg0 for 50 mA Ib.
The KT88s need about -70V Vg0 for 50mA at 550 V.

The amount of signal DRIVE is only slightly greater for KT88s than for
6L6s- the same drive stage seems to suffice.

This is interesting; I'll have to see what the gm is of both types and
work out the drive levels req'd to produce clipping (450V for the 6L6s,
550V for the KT88)

I've been using Hammond xfmrs- the 1650P 60W 6600 ohm for the 6L6.
For the KT88, I started out with the same xfmr, (1650P) and thought to
also try the 1650N 4300ohm.

With the 6L6s, the 6600 ohm xfmr yielded about 35 Watts; the 4300 about
25.

Witht the KT88s, both xfmrs produce nearly identical output power, the
6600 ohm xfmr maybe a couple of watts more. So it looks like maybe 5k is
ideal...?


But I should be seeing about twice this, and am convinced I'm doing
something wrong. But so help me, I can't find a thing so far.

And if I have got it wrong, that could explain the THD and bias issues in
an earlier paragraph.


So I thought I'd make absolutely certain that my voltage amp (the 6SN7
LTP) has plenty of headroom. And indeed it does; It clips about 15 dB
after the output stage does. So at least the clipping isn't caused by the
driver, or something stupid like that...

What I have is entirely listenable, but it just ain't right, so I refuse
to call it done until I really understand why it behaves the way it does.



I'll probably stick with floating paraphase for the splitter- it's good
enough, and I want to finish this design at some point in my life...

But the splitter is enormously important, and at some point I want to
revisit this aspect of the design.

Split load splitters are limited in their output swing, and they're not
really symmetrical- one output is more-or-less a cathode follower, and
the other is a plate load.

My floating paraphase has the property of almost cancelling the
distortion on one output (the mu-follower), while the other is what one
might expect from a plate-driven output.

Patrick has had good luck with his LTP configuration.

My idea follows the Mullard 520 tradition. I think my low U triode based
LTP with CCS perform much better than the Mullard 520.


Indeed it should.

The "cheat" of providing a little FB around one side to lower the gain, or
of altering the plate loads to equalize the gain is offensive to me.
Plus, it should add a little 2nd harmonic distortion to the mix.

The only way to REALLY solve the problem is to use a true current sink.

But then, I still like to provide some way of trimming the gain of the the
driver stage, to compensate exactly for any gm variations in the output
tubes, even though they are matched. After getting to know my Scott 299,
I find it rather nice to have trims for DC bias, DC bias balance, and AC
drive balance. Probably not a good idea for a production amp, but for a
one-up, why not tune out as much evil as one can?


It should offer
excellent symmetry, especially when fed from the bipolar current source
he's fond of.

CCS and equal RL values gurantee equal drive voltages to each output
tube.


Yes, that's the beauty of using a CCS. And like you, I see no evil in
using a bipolar to do the job; it's just perfect.



I've tried using a LTP in the "traditional" config with a simple
cathode resistor, and have had to fudge the gains of the two sections
because the cathode isn't really fed by a constant current.

The fixed common cathode resistor of about 10k used in countless designs
and the use of disimilar anode RL and perhaps with a trimmer R is a
pretty decent way to get a low thd drive to the output tubes. The
dissimilar RLs needed with low U triodes in a diffamp create a small
level of unbalanced 2H in the outputs; most diffamp thd is 3H.


I've briefly made a stab at Patrick's configuration but couldn't get it
to work in the 20 minutes I spent with it so far, so I went back to my
floating paraphase. But this is something I want to investigate
further.

Really, it seems that an interstage transformer(!) may be best after
all, but such things must be handmade, by and large. They're not
usually found off-the-shelf (I'm thankful we can find OUTPUT
transformers!) and they're so specific to the design that each design
requires a slightly different one than the next.

And something from Lundahl or Plitron will work OK but it is expensive.
The phase shifts introduced by an IST prohibit any high levels of NFB.
That's why Williamson, Walker, Leak and later McIntosh swung right away
from ISTs.

Hmmmm... Never considered the phase shift. That would really kill the
ol' phase margin, at that. And here I thought the commercial producers
were trying to get away from the COST of the IST.



They sound great in guitar amps.
No doubt. The more distortion, phasiness, the better- to a point.
But I'd hate to offend any guiar amp folks- I really don't understand what
sounds good in such an amp. I've always focused on reproducing sound, not
on creating it.


Well, enough blathering about everything under the sun. This thread
started off as "cathode stripping" and has already digressed through
triodes Vs Pentodes, onto phase splitters.

Anyway, it's great to be working on this stuff again after all these
years; I'm amazed and pleased to find that the tiny little niche market
has grown into such a significant slice of the pie.

And it's just plain FUN to be doing this!

Its a satisfying experience to build a tube amp...

That's what I'm hoping for. I've spent 10 years at Shure Brothers in
their circuitry development section, followed by another 10 years at
Industrial Research Products, designing sound reinforcement products.
All solid-state, of course. After being away from the field for 13 years,
I decide to build tube stuff for myself, just for fun. Just can't get
it out of my blood.

Building something for myself is MUCH more satisfying than designing
something for the marketing dep't.



And after this, the next project (taking me MANY YEARS, I assure you) is
a tubed FM multiplex tuner.

LOTS of fun with the RF stage (cascode nuvistors?), osc stability
(Vackar?) IF gain, Bandwidth, FM detector types (pulse count?) Stereo
detector (balanced modulator? Beam deflection tube like 6JH8 or 6ME8?).
front-end dynamic range and noise figure, AGC techniques, AFC (if req'd)
etc etc. And then there's the mechanical constuction of the local
oscillator and the dial mechanism (no digital readout, I still like the
old Eddystone dials- wish I could find one!) Years of endless fascination!
Or frustration... Glad there'll be no rush.



Marty Dippel

Marty Dippel wrote:

On Sun, 27 Feb 2005 04:31:30 +1100, Patrick Turner wrote:



Marty Dippel wrote:

On Fri, 25 Feb 2005 12:51:06 +0000, west wrote:

Marty,
Have you ever thought of changing the KT88s to Triode Mode? What do you
think about that Thanks.
Cordially,
west

Oh, I dunno.. I haven't given it SERIOUS consideration, but I'm tempted
to stick with ultralinear. P-P U/L offers better plate efficiency than
does P-P triode mode, and also is more tolerant of a mismatched load-
which is something seldom tested for on the bench, but makes a great deal
of difference to the ear. By "more tolerant" I mean that the distortion
of the amp is flatter across a range of load impedances than is a P-P
triode configuration.

As long as you saty within the power ceiling of the triode amp there is
no reason to suggest UL is better in any way than triode.


I've always been curious about P-P triodes, though- one day I will build
an amp that way just to see...


Its hard to build a bad sounding PP triode amp.

Sounds like this would be the case; one day I'll try it and see -!





This is my first tube project in about 30 years. It started when I had
the idea of building an amp "just for fun" - and I thought to build it out
of "classic" tubes- a 6SN7 phase splitter followed by a pair of 6L6s in
push pull. I soon discovered that I had just barely enough gain to
produce clipping with 1 V RMS input- so I couldn't spare any gain for
feedback. So in goes another 6SN7 as a LTP voltage amp after the floating
paraphase splitter. So now I have almost TOO MUCH gain- 20 dB feedback is
a little high compared to most designs, but it does wonders in correcting
all manner of evils...

So you have a williamson.

Not a bad design.


Yeah; it's pretty close to the Williamson. And yes, not a bad
design.

Hum was a bit of a problem; turned out to be "bleed-through" from the (AC)
heater to the cathode. Using a 12AU7 instead of the first 6SN7 helped by
about 10 dB. The voltage amp stays as a 6SN7 because I can use the higher
plate voltage rating to produce more undistorted output swing to feed the
output stage.

DC on the 6SN7 heaters will remove all heater-cathode leaking.

If I can do it without resorting to this bother, I'm all for it -! But
yes, the idea's occurred- and it appears quite often in the literature.
This thing is already 2/3 power supply and 1/3 amplifier; I'd rather not
resort to a well-filtered DC supply for the heaters on top of everything
else...

Seems like if one is building a phono preamp, there'd be an excuse, but
this is LINE LEVEL, for goodness sake. You'd think such measures wouldn't
be required here... but I see it often enough. One of the disadvantages
of tubes, I suppose: heaters.

Using a 100 ohm pot across the 6.3v with the wiper to 0V can also null out the
hum.




Well, it got to the point of "first listen" and I was pleasantly
surprised. Even my wife heard a distinct improvement over the
solid-state amp we had been using. -She usually can't hear squat as far
as audio nuances, but she said the improvement was obvious. And it was
to me, too. "transparency", "pace", "involvement", "extension", all
that. You betcha. :-)

yep, it is obvious to many ppl...

Just "more musical" to my ears (whatever that might mean...)

Anyway, she said I ought to build a nice cabinet or case or base or
something- make it look like something. So I designed one out of
hardwood- but I'm all thumbs when it comes to woodworking; even a simple
task like this presents a tremendous challenge. So I thought I'd build a
stereo amp instead of two monoblocs, to minimize the pain of woodworking.

So Hammond makes a power transformer that'll handle the current of two
amps, and it's about 25V higher on the secondary voltage. And it has no
5V winding for the 5U4- so I decide to go with 1N4007s. Which gives me
another 20 or 25 V, and by now I'm above what the 6L6s will stand. Which
is where this thread started, BTW.

So I've always been wanting to use KT88s anyway- and here's my chance.
So now it's a whole new design- the KT88 acts quite differently than does
the 6L6. As the bias current is increased, the crossover distortion
diminishes by degrees, until the plates are glowing red...
With 6L6s, the xover disappeared at some point before plate dissipation
was exceeded.

There is no need to run KT88 at high Pd; 20 watts is plenty,
and if RL is just right, 30 watts of class AB1 triode is available.
The thd at normal listening levels is SFA.

The spec sheets call for 50 mA Ib0 when running 550V B+ in U/L. So that's
about 22W, which, as you say, should be plenty...

RLa-a shouls be about 8 to 10k



I haven't refined my design to the point where I'm willing to argue with
the spec sheet guidelines; see below...

Also, I was getting about 1% THD out of my 6L6s, (no feedback) and the
KT88s produce about 3%.

This seems odd to me; usually KT88 or 6550 give more power and lower
thd.

Yup. So I must be doing something wrong... I'm seeing about 50W out of
a pair of KT88s at clipping; I should be approaching 100W given 575V and
U/L.

With fixed bias and 575v, about 80 watts is possible from UL with about 5k a-a
but it will be mainly class B, and you will never see 1% thd at full power
without a lot of FB.

The load determines the power max.

A pair of KT88 in UL with Ea = 500v give 34 watts into 8k, nearly all class A,
and 54 watts AB1 into 4k.




And I was seeing only 30W out of a pr of 6L6s with 450V B+.

That's about right.

If you replace the 6L6 with KT88 and have the same bias current,
and if RL is about 7ka-a, the power only marginally increases.
To reduce the thd, run a higher bias current, since it fives you more class A.
If RL was 4k a-a for the 6L6, and then you change to KT88,
expect a large power increase because the v swing can be a lot greater.
thd should also be less.




So I've *GOT* to be doing something very wrong here... But so help me, It
escapes me what it could be. "Analysis continues", as we say...

But some of this could be from the driver stage, which is where I am at
the present time. And 3% isn't all that bad; my 20dB feedback should
reduce this to 0.3%, but still, I'd like to get the amp working as best
it can before I go applying FB.

KT88 and KT66 and 6L6 all require a similar bias voltage and thus drive
voltage. So how can the thd from the drive amp suddenly be so much
higher for KT88 compared to 6L6?

Hmmmm... Dunno about that. The 6L6 needed about -50V Vg0 for 50 mA Ib.
The KT88s need about -70V Vg0 for 50mA at 550 V.

I don't recall major differences; well the KT88 with the same bias as 6L6 will
run a higher bias current.



The amount of signal DRIVE is only slightly greater for KT88s than for
6L6s- the same drive stage seems to suffice.

This is interesting; I'll have to see what the gm is of both types and
work out the drive levels req'd to produce clipping (450V for the 6L6s,
550V for the KT88)

Anyway, you driver stage should be able to make twice the required drive voltage
at 1% 3H.




I've been using Hammond xfmrs- the 1650P 60W 6600 ohm for the 6L6.
For the KT88, I started out with the same xfmr, (1650P) and thought to
also try the 1650N 4300ohm.

With the 6L6s, the 6600 ohm xfmr yielded about 35 Watts; the 4300 about
25.

Witht the KT88s, both xfmrs produce nearly identical output power, the
6600 ohm xfmr maybe a couple of watts more.

I assume you have Ea = 575v and fixed bias.
6.6k is a rather low value RL for +575v.
With 6L6 the swing on the load lines is from 575v down to about 200v,
so -375v pk = 530 vrms a-a = 42 watts and the losses in the OPT
would reduce that to 35 perhaps.

With KT88 plugged in the swing from 575v is to about 110v, and
Va-a = 557 vrms, and po = 65 watts, or around 60 watts at the secondary.

RL for fidelity needs to be around twice the 6.6k so using an 8 ohm speaker at
the 4 ohm outlet
will give you a pile more class A and lower thd because the load becomes 13.2ka-a

You only get 40 watts into 13.2k but its all nearly pure class A with KT88.


So it looks like maybe 5k is
ideal...?

What do you want?
High power and low fidelity, or low power and high fidelity?
Both high power and high fidelity are impossible without using more output tubes.





But I should be seeing about twice this, and am convinced I'm doing
something wrong. But so help me, I can't find a thing so far.

And if I have got it wrong, that could explain the THD and bias issues in
an earlier paragraph.

So I thought I'd make absolutely certain that my voltage amp (the 6SN7
LTP) has plenty of headroom. And indeed it does; It clips about 15 dB
after the output stage does.

15 dB is a lot.
It means the driver stage can produce 15 dB more max level that that needed
to make the output clip. I doubt your driver makes 400 vrms into each grid.

Without the output tube plugged in, the driver stage should be able to make about

70 vrms at each output at 1%.


So at least the clipping isn't caused by the
driver, or something stupid like that...

What I have is entirely listenable, but it just ain't right, so I refuse
to call it done until I really understand why it behaves the way it does.

I don't think you fully understand load line analysis.



I'll probably stick with floating paraphase for the splitter- it's good
enough, and I want to finish this design at some point in my life...

But the splitter is enormously important, and at some point I want to
revisit this aspect of the design.

Split load splitters are limited in their output swing, and they're not
really symmetrical- one output is more-or-less a cathode follower, and
the other is a plate load.

My floating paraphase has the property of almost cancelling the
distortion on one output (the mu-follower), while the other is what one
might expect from a plate-driven output.

Patrick has had good luck with his LTP configuration.

My idea follows the Mullard 520 tradition. I think my low U triode based
LTP with CCS perform much better than the Mullard 520.


Indeed it should.

The "cheat" of providing a little FB around one side to lower the gain, or
of altering the plate loads to equalize the gain is offensive to me.
Plus, it should add a little 2nd harmonic distortion to the mix.

The only way to REALLY solve the problem is to use a true current sink.

The CCS can be taken to a -ve supply and one grid of the LTP grounded.
This makes the effective Ea for tha LTP triodes effectively maximal,
so the V swing is maximised, and thd minimised.

Try the circuit at
http://www.turneraudio.com.au/htmlwe...0ulabinteg.htm



But then, I still like to provide some way of trimming the gain of the the
driver stage, to compensate exactly for any gm variations in the output
tubes, even though they are matched.

If they are gm matched, there is no need for ac drive adjutments.
At low levels trimming drive makes very little difference even with
slightly unmatched tubes because each tube is class A.

After getting to know my Scott 299,
I find it rather nice to have trims for DC bias, DC bias balance, and AC
drive balance. Probably not a good idea for a production amp, but for a
one-up, why not tune out as much evil as one can?

Because the evil in a pair of KT88 is negligible.

Pots in plate circuits don't belong.



It should offer
excellent symmetry, especially when fed from the bipolar current source
he's fond of.

CCS and equal RL values gurantee equal drive voltages to each output
tube.

Yes, that's the beauty of using a CCS. And like you, I see no evil in
using a bipolar to do the job; it's just perfect.

I've tried using a LTP in the "traditional" config with a simple
cathode resistor, and have had to fudge the gains of the two sections
because the cathode isn't really fed by a constant current.

The fixed common cathode resistor of about 10k used in countless designs
and the use of disimilar anode RL and perhaps with a trimmer R is a
pretty decent way to get a low thd drive to the output tubes. The
dissimilar RLs needed with low U triodes in a diffamp create a small
level of unbalanced 2H in the outputs; most diffamp thd is 3H.


I've briefly made a stab at Patrick's configuration but couldn't get it
to work in the 20 minutes I spent with it so far, so I went back to my
floating paraphase. But this is something I want to investigate
further.

Really, it seems that an interstage transformer(!) may be best after
all, but such things must be handmade, by and large. They're not
usually found off-the-shelf (I'm thankful we can find OUTPUT
transformers!) and they're so specific to the design that each design
requires a slightly different one than the next.

And something from Lundahl or Plitron will work OK but it is expensive.
The phase shifts introduced by an IST prohibit any high levels of NFB.
That's why Williamson, Walker, Leak and later McIntosh swung right away
from ISTs.

Hmmmm... Never considered the phase shift. That would really kill the
ol' phase margin, at that. And here I thought the commercial producers
were trying to get away from the COST of the IST.

ISTs have phase shift at the extreme ends of the bw caused by
shunt inductance and shunt capacitance, and both introduce yet more poles,
and larger phase shift at LF and HF so stabilizing with 20 dB FB becomes
difficult
if not impossible.



They sound great in guitar amps.
No doubt. The more distortion, phasiness, the better- to a point.
But I'd hate to offend any guiar amp folks- I really don't understand what
sounds good in such an amp. I've always focused on reproducing sound, not
on creating it.

Well forget the techniques used for guitar amps.



Well, enough blathering about everything under the sun. This thread
started off as "cathode stripping" and has already digressed through
triodes Vs Pentodes, onto phase splitters.

Anyway, it's great to be working on this stuff again after all these
years; I'm amazed and pleased to find that the tiny little niche market
has grown into such a significant slice of the pie.

And it's just plain FUN to be doing this!

Its a satisfying experience to build a tube amp...

That's what I'm hoping for. I've spent 10 years at Shure Brothers in
their circuitry development section, followed by another 10 years at
Industrial Research Products, designing sound reinforcement products.
All solid-state, of course. After being away from the field for 13 years,
I decide to build tube stuff for myself, just for fun. Just can't get
it out of my blood.

Building something for myself is MUCH more satisfying than designing
something for the marketing dep't.

And after this, the next project (taking me MANY YEARS, I assure you) is
a tubed FM multiplex tuner.

It took me a month of R&D at a local university library to find the old circuits,
work
out what they did, work out that an LTP with 76 kHz would do the
stereo decoding better than all the rest with transfromers, and I got a nice
tubed
decoder.

It shouldn't take years.
Focus and DO it.



LOTS of fun with the RF stage (cascode nuvistors?), osc stability
(Vackar?)

cascode nuvistors are not needed.

Just about any RF and IF section will work fine.
The linearity is unimportant since the IF signal needs to have no amplitude
variations.
The IF output signal is the result of 6BA6 and 6AU6 operating in grossly
overloaded conditions.

IF gain, Bandwidth, FM detector types (pulse count?)

Foster Seely or Ratio detector is OK. Also a quadrature detector with a humble
6DT6.


Stereo
detector (balanced modulator? Beam deflection tube like 6JH8 or 6ME8?).

Too high falootin. 6BN6 was another "difficult to get right" gated beam tube.

Later they used the 6DT6 quadrature detector since it was simple, linear, and
high output.
Thousands of TV sound stages used this tube.

The stereo decoding is far more critical than the mono detector
for the L+R signal.

See http://www.turneraudio.com.au/htmlwe...mpxdecoder.htm




front-end dynamic range and noise figure, AGC techniques, AFC (if req'd)
etc etc. And then there's the mechanical constuction of the local
oscillator and the dial mechanism (no digital readout, I still like the
old Eddystone dials- wish I could find one!) Years of endless fascination!
Or frustration... Glad there'll be no rush.

No need to spend more than 2 months on a tubed FM set.

Then you get years to listen to it.

Patrick Turner.



Marty Dippel

On Sun, 27 Feb 2005 14:28:39 +1100, Patrick Turner
wrote:

Marty Dippel wrote:
So I've *GOT* to be doing something very wrong here... But so help me, It
escapes me what it could be. "Analysis continues", as we say...

But some of this could be from the driver stage, which is where I am at
the present time. And 3% isn't all that bad; my 20dB feedback should
reduce this to 0.3%, but still, I'd like to get the amp working as best
it can before I go applying FB.

KT88 and KT66 and 6L6 all require a similar bias voltage and thus drive
voltage. So how can the thd from the drive amp suddenly be so much
higher for KT88 compared to 6L6?

Hmmmm... Dunno about that. The 6L6 needed about -50V Vg0 for 50 mA Ib.
The KT88s need about -70V Vg0 for 50mA at 550 V.

I don't recall major differences; well the KT88 with the same bias as 6L6 will
run a higher bias current.

The 6550/KT88's higher perveance implifies higher drive voltage
requirements for full output. And higher bias voltages for similar
idle currents. They are *not* equivalent to 6L6 family tubes.

IF gain, Bandwidth, FM detector types (pulse count?)

Foster Seely or Ratio detector is OK. Also a quadrature detector with a humble
6DT6.


Stereo
detector (balanced modulator? Beam deflection tube like 6JH8 or 6ME8?).

Too high falootin. 6BN6 was another "difficult to get right" gated beam tube.
Later they used the 6DT6 quadrature detector since it was simple, linear, and
high output.
Thousands of TV sound stages used this tube.

The stereo decoding is far more critical than the mono detector
for the L+R signal.

Beam deflection tubes are a different critter than quadrature
detectors. They have a single cathode, signal grid, and accelerating
screen and a pair of deflecting electrodes to a pair of plates.
They were used in TV's, then the 7360 brought the concept to
multiplication. Still the most elegant solution to many problems, if
you can find the devices. Collins chose 7360's and sealed their fate.

Please don't choose these for new designs: classic ham gear needs 'em
and there are no sub's. Modern designs, including FM stereo demod
can be done, and even better, with diode rings.

Same-same for front ends. Nuvistors are NOT wonderful magic. Just say
no to kitsch.

Chris Hornbeck

Chris Hornbeck wrote:

On Sun, 27 Feb 2005 14:28:39 +1100, Patrick Turner
wrote:

Marty Dippel wrote:
So I've *GOT* to be doing something very wrong here... But so help me, It
escapes me what it could be. "Analysis continues", as we say...

But some of this could be from the driver stage, which is where I am at
the present time. And 3% isn't all that bad; my 20dB feedback should
reduce this to 0.3%, but still, I'd like to get the amp working as best
it can before I go applying FB.

KT88 and KT66 and 6L6 all require a similar bias voltage and thus drive
voltage. So how can the thd from the drive amp suddenly be so much
higher for KT88 compared to 6L6?

Hmmmm... Dunno about that. The 6L6 needed about -50V Vg0 for 50 mA Ib.
The KT88s need about -70V Vg0 for 50mA at 550 V.

I don't recall major differences; well the KT88 with the same bias as 6L6 will
run a higher bias current.

The 6550/KT88's higher perveance implifies higher drive voltage
requirements for full output. And higher bias voltages for similar
idle currents. They are *not* equivalent to 6L6 family tubes.

IF gain, Bandwidth, FM detector types (pulse count?)

Foster Seely or Ratio detector is OK. Also a quadrature detector with a humble
6DT6.


Stereo
detector (balanced modulator? Beam deflection tube like 6JH8 or 6ME8?).

Too high falootin. 6BN6 was another "difficult to get right" gated beam tube.
Later they used the 6DT6 quadrature detector since it was simple, linear, and
high output.
Thousands of TV sound stages used this tube.

The stereo decoding is far more critical than the mono detector
for the L+R signal.

Beam deflection tubes are a different critter than quadrature
detectors.

Utterly different

They have a single cathode, signal grid, and accelerating
screen and a pair of deflecting electrodes to a pair of plates.
They were used in TV's, then the 7360 brought the concept to
multiplication. Still the most elegant solution to many problems, if
you can find the devices. Collins chose 7360's and sealed their fate.

I think I have a pair of 7360 around somewhere.
They are spares for a radio for SSB.

Please don't choose these for new designs: classic ham gear needs 'em
and there are no sub's.

The dude only wants to build himself one MPX decoder.

But I have never come across a schematic using 7360 for MPX.


Modern designs, including FM stereo demod
can be done, and even better, with diode rings.

The hi-fi cognescenti would never agree that diode rings are the best way.

Most chip based decoders use arrays of differential pairs and no diodes for
the MPX action. This could be done with tubes, but one would need
a few to duplicate what happens in a chip.
One could expect the thd to thus be very low because the dynamic range of
triodes and their linearity is better than bjts.


Same-same for front ends. Nuvistors are NOT wonderful magic. Just say
no to kitsch.

I cannot see any great benefits with nuvistors except size.

I would think a fet based front end would be more effective as a
small package if one wanted smallness, but with diy that
isn't a concern.
Nevertheless, front ends and IF strips have to be constructed very carefully
with small path lengths and careful set out of components since
they are prone to parastics and faults if one doesn't get it right.
Its a brave DIYer who starts out with a couple of twin triodes and a tuning gang
and some IFTs and tries to design and build his own FM tuner.
Its much harder to get it right than building an AM radio.
But the coils are easy; just a few self supporting turns of thick wire....

Patrick Turner.









Chris Hornbeck

On Mon, 28 Feb 2005 11:01:53 +1100, Patrick Turner wrote:



Chris Hornbeck wrote:

On Sun, 27 Feb 2005 14:28:39 +1100, Patrick Turner
wrote:

Marty Dippel wrote:
So I've *GOT* to be doing something very wrong here... But so help me, It
escapes me what it could be. "Analysis continues", as we say...

But some of this could be from the driver stage, which is where I am at
the present time. And 3% isn't all that bad; my 20dB feedback should
reduce this to 0.3%, but still, I'd like to get the amp working as best
it can before I go applying FB.

KT88 and KT66 and 6L6 all require a similar bias voltage and thus drive
voltage. So how can the thd from the drive amp suddenly be so much
higher for KT88 compared to 6L6?

Hmmmm... Dunno about that. The 6L6 needed about -50V Vg0 for 50 mA Ib.
The KT88s need about -70V Vg0 for 50mA at 550 V.

I don't recall major differences; well the KT88 with the same bias as 6L6 will
run a higher bias current.

Yes; that's certainly true. The KT88s need something on the order of -70V
Eg to produce 50 mA Ib, whereas the 6L6 needs something around -45 to do
the same thing.



The 6550/KT88's higher perveance implies higher drive voltage
requirements for full output. And higher bias voltages for similar
idle currents. They are *not* equivalent to 6L6 family tubes.

Right. The KT66 comes closer to being more similar to the 6L6. But there
are probably differences there, too.



IF gain, Bandwidth, FM detector types (pulse count?)

Foster Seeley or Ratio detector is OK. Also a quadrature detector
with a humble 6DT6.


Stereo
detector (balanced modulator? Beam deflection tube like 6JH8 or
6ME8?).

Too high falootin. 6BN6 was another "difficult to get right" gated
beam tube. Later they used the 6DT6 quadrature detector since it was
simple, linear, and high output.
Thousands of TV sound stages used this tube.

The stereo decoding is far more critical than the mono detector for
the L+R signal.

Yes, it certainly is. It's hard to get this done REALLY right.


Beam deflection tubes are a different critter than quadrature
detectors.

Utterly different

Yes, that's true. Nothing in common at all. Just thought I'd like to
fool around with one once and see if it's a reasonable way to do MPX
detection, or if I'd be better off with more traditional methods.



They have a single cathode, signal grid, and accelerating screen and a
pair of deflecting electrodes to a pair of plates. They were used in
TV's, then the 7360 brought the concept to multiplication. Still the
most elegant solution to many problems, if you can find the devices.
Collins chose 7360's and sealed their fate.

I think I have a pair of 7360 around somewhere. They are spares for a
radio for SSB.

Please don't choose these for new designs: classic ham gear needs 'em
and there are no sub's.

The dude only wants to build himself one MPX decoder.

But I have never come across a schematic using 7360 for MPX.

Me neither. So either I have a fresh idea here, or the ancients found out
early in the game that it's just not the thing to do.

I thought I'd try it myself and see, having found nothing in the
literature.



Modern designs, including FM stereo demod can be done, and even better,
with diode rings.

The hi-fi cognescenti would never agree that diode rings are the best
way.

Most chip based decoders use arrays of differential pairs and no diodes
for the MPX action. This could be done with tubes, but one would need a
few to duplicate what happens in a chip. One could expect the thd to
thus be very low because the dynamic range of triodes and their
linearity is better than bjts.




Same-same for front ends. Nuvistors are NOT wonderful magic. Just say
no to kitsch.

I cannot see any great benefits with nuvistors except size.

I thought nuvistors were supposed to be a little quieter than most of the
miniature types. They're a newer design than most; I'd have thought the
designers would have learned a little more by the time they designed the
nuvistors. Or maybe they were just after miniaturization; in that case,
I'd be just as well off with a 6BA6 or something. Haven't chosen a tube
type for the front end yet.

I need to really take a look at the 6CW4 before I commit to it, of course.
There has to be a sound REASON. Just thought it'd be worth a look.



I would think a fet based front end would be more effective as a small
package if one wanted smallness, but with diy that isn't a concern.

I think that seriously, one will be able to do a much better job in
general *ON AN FM STEREO TUNER* with solid state than with tubes. quieter
front ends, oscillators that don't generate so much heat, and then there's
the availability of PLLs and Double Balanced Modulators, all nicely
integrated.

But I thought I'd see just how well I could do using tubes only. Call me
crazy. (Ooops! -This is probably NOT the place to tempt people with THAT
invitation...)



Nevertheless, front ends and IF strips have to be constructed very
carefully with small path lengths and careful set out of components
since they are prone to parasitics and faults if one doesn't get it
right. Its a brave DIYer who starts out with a couple of twin triodes
and a tuning gang and some IFTs and tries to design and build his own FM
tuner. Its much harder to get it right than building an AM radio. But
the coils are easy; just a few self supporting turns of thick wire....

Well, I figure if I could build my own 10 meter- to- 2 meter transverter
and a host of other 144 MHz projects, I can probably handle FM BCB lead
dress problems, too. But of course this is not certain... :-)

I know I'm in for it, wanting to build a Local Osc, mixer, and IF strip
from scratch. I wonder if off-the-shelf (Murata, et al) crystal or
ceramic filters would work in a tubed design... I've never looked into
this idea.

Still, I'm up against tremendous challenges in every stage. That's why
I'm building it.


Marty Dippel

I cannot see any great benefits with nuvistors except size.

I thought nuvistors were supposed to be a little quieter than most of the
miniature types. They're a newer design than most; I'd have thought the
designers would have learned a little more by the time they designed the
nuvistors. Or maybe they were just after miniaturization; in that case,
I'd be just as well off with a 6BA6 or something. Haven't chosen a tube
type for the front end yet.

A pair of 12AT7 or 6AQ8 are all you need for a front end
grounded grid RF amp, oscillator, and mixer and AFC.



I need to really take a look at the 6CW4 before I commit to it, of course.
There has to be a sound REASON. Just thought it'd be worth a look.

Worth a look. Have your look. Place back in the box and forgetabout.



I would think a fet based front end would be more effective as a small
package if one wanted smallness, but with diy that isn't a concern.

I think that seriously, one will be able to do a much better job in
general *ON AN FM STEREO TUNER* with solid state than with tubes. quieter
front ends, oscillators that don't generate so much heat, and then there's
the availability of PLLs and Double Balanced Modulators, all nicely
integrated.

The pair of twin triodes are excellent, easily understandable.



But I thought I'd see just how well I could do using tubes only. Call me
crazy. (Ooops! -This is probably NOT the place to tempt people with THAT
invitation...)

Nevertheless, front ends and IF strips have to be constructed very
carefully with small path lengths and careful set out of components
since they are prone to parasitics and faults if one doesn't get it
right. Its a brave DIYer who starts out with a couple of twin triodes
and a tuning gang and some IFTs and tries to design and build his own FM
tuner. Its much harder to get it right than building an AM radio. But
the coils are easy; just a few self supporting turns of thick wire....

Well, I figure if I could build my own 10 meter- to- 2 meter transverter
and a host of other 144 MHz projects, I can probably handle FM BCB lead
dress problems, too. But of course this is not certain... :-)

Broadcast FM should then be easy.

Better hurry, they are all going to be digital soon....



I know I'm in for it, wanting to build a Local Osc, mixer, and IF strip
from scratch. I wonder if off-the-shelf (Murata, et al) crystal or
ceramic filters would work in a tubed design... I've never looked into
this idea.

No need, use IFTs; 6AU6 and 6BA6 are ok with 10.7 MHz.


Still, I'm up against tremendous challenges in every stage. That's why
I'm building it.

Marty Dippel

I though a radio was a little challenge with each stage, but a large challenge to get
them
all to work properly together.

A transistor has bug 3 legs, and to teach it to walk is easy.
To get 2 million of the leetle buggers to all march in step in a PC
took the boffins rather a long time, and then they still manage
to have properties like impudence.

Patrick Turner.

On Mon, 28 Feb 2005 11:01:53 +1100, Patrick Turner
wrote:

The dude only wants to build himself one MPX decoder.

But I have never come across a schematic using 7360 for MPX.

Yeah, you're right. I shouldn't be so twichy about it.
And, come to think of it, I've never seen a 7360 used for
FM subcarrier detection either. Might be fun.


Modern designs, including FM stereo demod
can be done, and even better, with diode rings.

The hi-fi cognescenti would never agree that diode rings are the best way.

Diode rings can be better if used with a trickle DC bias and
brute-force carrier levels. Radio guys do stuff like this, and
the OP seems to already be pretty familiar with that world.

Not as elegant as a 7360, though...

Thanks,

Chris Hornbeck
"Anarchy only works among those who can control themselves." -ha