Design Error 2 extra

[Phil Allison]

Hi to my tormenters,

---------------------

the tube amp question left most of you a tad agog.

Yet it was simply a case of noting that opening the s/b switch altered the topology of the amplifier in a not good way.

Since transient over-voltage on the plates leading to insulation failure is a common thing with tube amps generally, many makers apply a simple fix using high voltage diodes.

Two or three 1N4007s in series or a 4kV multi-chip diode is strung across the plate and ground of each tube in a pair to stop the plate voltage ever going below zero.

With them in place, the max positive plate voltage cannot exceed twice the B+ supply due to transformer action in a push pull output stage. If one side ever goes to zero, the other will be at double B+.

But this simple fix could not be safely applied to the Fender amp in my question.

Why ?

https://schematicheaven.net/fenderamps/fender_bassman50.pdf


...... Phil

 

[Winfield Hill]

Phil Allison wrote...

...
If one side ever goes to zero, the other will be at double B+.

But this simple fix could not be safely applied to the Fender
amp in my question.

You characterize the issue as the amplifier operating with
a different topology, with the output transformer and tubes
presumably operating from the 20uF cap through inductor TR2.

But at the instant of switch-off, TR2's operating current
should first make a negative-going flyback voltage, whose
magnitude may be partly attenuated by winding capacitances,
but ultimately will be limited by insulation breakdown.
A reverse diode across TR2 would eliminate any flyback.



--
Thanks,
- Win

 

[Dimitrij Klingbeil]

On 2019-10-26 08:43, Phil Allison wrote:

Two or three 1N4007s in series or a 4kV multi-chip diode is strung
across the plate and ground of each tube in a pair to stop the plate
voltage ever going below zero.

With them in place, the max positive plate voltage cannot exceed
twice the B+ supply due to transformer action in a push pull output
stage. If one side ever goes to zero, the other will be at double
B+.

But this simple fix could not be safely applied to the Fender amp in
my question.

Why ?

TR2 is a choke, no opposite winding to connect a diode network to.

The diodes from the endpoints of the TR3 primary would only protect
against TR3, but not against a flyback voltage from TR2, which would
cause a reverse breakdown of one (or both) of these diodes.

What could have offered better protection is:

- Either a 600V rated MOV between TR3 primary midpoint and GND.

- Or a a pair of at least 600V rated diodes as follows:
- one in reverse direction across the standby switch and
- the other in place of the MOV above, with anode to GND.

- Alternatively, a 600V rated TVS diode across the standby switch
(non-bidirectional, with the cathode towards the supply side)

In the first case, the MOV should be chosen to have enough thermal mass
to absorb at least 20 uF, better some 50 uF worth of stored energy.

In the last case, the TVS diode does not need to dissipate as much as
the MOV, because the flyback voltage is clamped non-dissipatively, but
it still needs to dissipate significant energy during undershoots.


Dimitrij

 

[Terry Newton]

On Sat, 26 Oct 2019 06:15:25 -0700, Winfield Hill wrote:

Phil Allison wrote...

...
If one side ever goes to zero, the other will be at double B+.

But this simple fix could not be safely applied to the Fender amp in my
question.

You characterize the issue as the amplifier operating with a different
topology, with the output transformer and tubes presumably operating
from the 20uF cap through inductor TR2.

But at the instant of switch-off, TR2's operating current should first
make a negative-going flyback voltage, whose magnitude may be partly
attenuated by winding capacitances, but ultimately will be limited by
insulation breakdown.
A reverse diode across TR2 would eliminate any flyback.

I've been working on amps using that circuit for oh what 40 years,
never had any issues. I hit the standby when at full power all the
time, no surges, no nothing. Unlike that older circuit where they
put the SB switch in series with the power transformer CT, that
one blows rectifier tubes. But the discussed circuit has always
worked fine in my experience. Some makers put a RC snubber across
the switch but never bothered, whenever I have to replace the
switch it's from mechanical damage (wore out or broken) rather
than electrical stress. Despite being ran way above its rating.

Intuitively seems like there should be significant kickback
from the choke when the SB switch is opened, but it's only
passing a few milliamps (screen and preamp supply) and the
output transformer absorbs it. With a proper load the output
transformer CT is effectively resistive (if not the frequency
response would suck!) and perfectly capable of absorbing fast
transients. But it's probably not a good idea to operate the
standby switch (or for that matter the amp itself) with no
load.. that's why the output jack has a shorting contact.

Terry

 

[Phil Allison]

Dimitrij Klingbeil wrote:

-------------------------

Phil Allison wrote:

Two or three 1N4007s in series or a 4kV multi-chip diode is strung
across the plate and ground of each tube in a pair to stop the plate
voltage ever going below zero.

With them in place, the max positive plate voltage cannot exceed
twice the B+ supply due to transformer action in a push pull output
stage. If one side ever goes to zero, the other will be at double
B+.

But this simple fix could not be safely applied to the Fender amp in
my question.

Why ?

TR2 is a choke, no opposite winding to connect a diode network to.

The diodes from the endpoints of the TR3 primary would only protect
against TR3, but not against a flyback voltage from TR2, which would
cause a reverse breakdown of one (or both) of these diodes.

** Spot on. We have ourselves a design genius.

That exact event happened on my bench when I fitted HV, fast diodes to the 6L6 tube sockets.

First shot, under drive and load.

What could have offered better protection is:

- Either a 600V rated MOV between TR3 primary midpoint and GND.

** That could work.

- Or a a pair of at least 600V rated diodes as follows:
- one in reverse direction across the standby switch and
- the other in place of the MOV above, with anode to GND.

** Not good practice, bridging a switch with a big safety aspect to it with a diode. Safety for some poor repair tech that is.

- Alternatively, a 600V rated TVS diode across the standby switch
(non-bidirectional, with the cathode towards the supply side)

In the first case, the MOV should be chosen to have enough thermal mass
to absorb at least 20 uF, better some 50 uF worth of stored energy.

** Hmmm - about 5 Joules. Doable.

In the last case, the TVS diode does not need to dissipate as much as
the MOV, because the flyback voltage is clamped non-dissipatively, but
it still needs to dissipate significant energy during undershoots.

** YOU are not just a pretty face...


...... Phil

 

[Phil Allison]

Winfield Hill wrote:

---------------------

If one side ever goes to zero, the other will be at double B+.

But this simple fix could not be safely applied to the Fender
amp in my question.

You characterize the issue as the amplifier operating with
a different topology, with the output transformer and tubes
presumably operating from the 20uF cap through inductor TR2.

** Eee - yep.

But at the instant of switch-off, TR2's operating current
should first make a negative-going flyback voltage

** Eee - nope.

Not with a 20uf cap at one end and a 100mA of load on the other.

whose
magnitude may be partly attenuated by winding capacitances,
but ultimately will be limited by insulation breakdown.

** Fanciful with loads at both ends.

A reverse diode across TR2 would eliminate any flyback.

** There is about 50 volts p-p of 100/120 Hz ripple appearing across the choke at full power. 35uF at 250mA load, I = C.dv/dt

But only about 10 volts of DC drop.

See the issue?


FYI to all:

Win walked into that one like a revolving door ...

How did the man ever get to be a circuit designer ?

This stuff is soooo basic.


.... Phil

 

[Phil Allison]

Terry Newton wrote:

----------------------

I've been working on amps using that circuit for oh what 40 years,
never had any issues. I hit the standby when at full power all the
time, no surges, no nothing.

** That is not possible.

Unlike that older circuit where they
put the SB switch in series with the power transformer CT, that
one blows rectifier tubes.

** From the inrush current surge at switch on.

Some recent all tube amps have the same flaw - like the new Vox AC30s made in China.

But the discussed circuit has always
worked fine in my experience. Some makers put a RC snubber across
the switch but never bothered,

** They would not do that if there were NO issue.

In fact, if you *slowly* operate the s/b switch under drive and load - you can hear an arc form across the contacts. Open one and you can see all the cabonisation after years of such abuse.

whenever I have to replace the
switch it's from mechanical damage (wore out or broken) rather
than electrical stress. Despite being ran way above its rating.

** Ones I've seen all have badly burnt contacts.

Intuitively seems like there should be significant kickback
from the choke when the SB switch is opened,

** Not kickback, the 6L6s are driving the choke like IT is the load.

but it's only
passing a few milliamps (screen and preamp supply)

** At full power, screen current is not small any more.

When clipping, the plate voltage drops to about 70V while the screens stay at full supply - so pass heaps of electrons. This is the major reason for so many EL34 burn outs in Marshalls - the thin screen wires glow and melt.

But it's probably not a good idea to operate the
standby switch (or for that matter the amp itself) with no
load.. that's why the output jack has a shorting contact.

** Shame that jack is rendered useless if a lead is plugged in and not terminated.


..... Phil