Prevent No-Load Failure of OPT (Outputs transformers) without speakers hooked

[GordonW]

As I previously discussed, an unloaded — i.e. no speakers connected — output transformer turns into an ignition coil, regularly emitting arcs in the several kiloVolt range. Because energy is continually pumped into the system by the audio input being amplified, voltage in the output transformer rises until current may flow, and any suppressor device must therefore CONTINUOUSLY discharge high-energy pulses of regular occurrence.

The MOV is essentially discharging an ignition coil which is continually pumped by the audio input, an application for which the MOV is not suited. Who would build such a system when all the articles and datasheets clearly specify this will not work?

The MOV cannot withstand regular transients causing it to fire. That's just a fact. The continual breakdown destroys the semiconductor structure until leakage current overheats and destroys the device. Fact. MOVs are not used in place of flyback diodes for relays, motors, etc. for very good reason: the MOV will rapidly fail in such use. Fact.

The superior solution for disconnected speakers is adding flyback diodes to the output transformers, as would be used on any coil (relay, motor, choke, etc.) to safely discharge the collapsing flux as it converts to current. Such diodes prevent the voltage from ever rising above B+, because the diode conducts at that point. DONE.

For those unwilling to add such diodes to an amplifier, a resistor or TVS diode, or both for that matter, all as above described, is the only solution suitable available at the speaker terminals. DONE.

None of this is rocket science. It is basic transient suppression and ought to be obvious from both the materials I previously set forth and those found using a google search.

Unless I am missing something important- I believe that the assumption that the plate voltage of an output tube, should never exceed B+, is not true in practice- in push-pull amplifiers.

In a PP amp, the plate idles at B+. As one tube "turns on", the other "turns off"- the plate voltage of the "on" tube drops to near zero volts (at least, at high output level), while the "off" tube plate voltage rises to nearly TWICE B+.

A flyback diode used in such a case, between B+ the tube plate lead, will prevent the unloaded half of the winding from reaching twice B+... effectively acting as a short circuit to half of the transformer primary. That would likely result in quite a catastrophe...

Regards,
Gordon.

 

[trobbins]

It's not rocket science to be able to read and understand a datasheet, or work through what actually happens in an OPT when unloaded. Shooting from the lip with underlined text is no justification for a personal view - do some testing yourself and show some results and a 1:1 description of when the plate voltage deviates from a 0V to 2xB+ window, and what current is available in each transient event - just bellowing away does not progress the discussion.

Perhaps start with a half-primary and indicate what current is passing through it, and then work out what happens to that current (and its magnitude) as it dissipates as a transient through a protective device (if it helps then use a TVS as the protective device). Remember that for repeated transients the winding current is controlled and limited by the associated valve V-I plate characteristic (which constrains the peak current to a very different level than an automotive set of points energising a coil). An amp has to use a large bottle and a high screen voltage to get anywhere near 0.5A. Then work through what happens to that transient current as the energy in the winding transfers to a voltage rise across the winding, with the shunt winding capacitance initially soaking up the transient energy as the winding current circulates and ramps down - then when the winding voltage exceeds the protective device on-voltage, the remaining winding current splits to go through the protective device and to raise the winding voltage further whilst ramping down to zero current once the energy in the transient has dissipated itself. And don't forget that this half-primary winding inductance-related energy couples to the other half-winding during the transient, so one plate exceeds 2xB+ and the other plate goes below 0V (and the use of flyback diodes from plate to 0V would therefore only act to divert current through one half-primary winding, as the other half-primary would be flying above 2xB+ with no protection local to that half-winding). Still with me?

 

[tom3]

I have a small Zenith tube stereo SE amp from a record player. The right channel has a detachable speaker and it has a 100 ohm resistor wired across the output for protection. Uses a 6AQ5A tube. No real power but still had the protection for the transformer.

 

[Retrovert]

A flyback diode used in such a case, between B+ the tube plate lead, will prevent the unloaded half of the winding from reaching twice B+... effectively acting as a short circuit to half of the transformer primary. That would likely result in quite a catastrophe...

The flyback diode is connected from plate to ground. (Anode to ground and Cathode to plate. Lenz's Law.) One diode is placed on each end of the Primary (or Plate, no difference), and since the ends are of opposite polarity only one diode may conduct at any given point. Since the center-tap is at B+ no shorting risk. The voltage on the diodes must be at least three times B+ for headroom.

Flyback diodes were widely used in the 1980s, in Ken Fischer's Trainwreck, Marshall, Peavey, Fender, MusicMan, etc. Speaker jacks were 1/4" and not tight, it was trivial for a cable to be yanked out, which would then blow the output transformer when a guitarist did a sound check. Because high-voltage diodes were not readily available I added a string of three 1N4007 (something like that, been nearly 40 years) to each plate in the fall of 1983 as this was the best readily available technology. I would not today suggest construct a diode string in that fashion.

It's late so, if necessary, I'll tomorrow dig out some ancient (1980s) schematics of guitar amplifier with the protection diodes.

 

[GordonW]

It's not rocket science to be able to read and understand a datasheet, or work through what actually happens in an OPT when unloaded. Shooting from the lip with underlined text is no justification for a personal view - do some testing yourself and show some results and a 1:1 description of when the plate voltage deviates from a 0V to 2xB+ window, and what current is available in each transient event - just bellowing away does not progress the discussion.

Perhaps start with a half-primary and indicate what current is passing through it, and then work out what happens to that current (and its magnitude) as it dissipates as a transient through a protective device (if it helps then use a TVS as the protective device). Remember that for repeated transients the winding current is controlled and limited by the associated valve V-I plate characteristic (which constrains the peak current to a very different level than an automotive set of points energising a coil). An amp has to use a large bottle and a high screen voltage to get anywhere near 0.5A. Then work through what happens to that transient current as the energy in the winding transfers to a voltage rise across the winding, with the shunt winding capacitance initially soaking up the transient energy as the winding current circulates and ramps down - then when the winding voltage exceeds the protective device on-voltage, the remaining winding current splits to go through the protective device and to raise the winding voltage further whilst ramping down to zero current once the energy in the transient has dissipated itself. And don't forget that this half-primary winding inductance-related energy couples to the other half-winding during the transient, so one plate exceeds 2xB+ and the other plate goes below 0V (and the use of flyback diodes from plate to 0V would therefore only act to divert current through one half-primary winding, as the other half-primary would be flying above 2xB+ with no protection local to that half-winding). Still with me?

Actually, now that I understand the flyback diode connection (between plate and ground, not plate and B+)- it appears that if one side of the winding is clamped to not go below zero volts (ground) by a diode- then, inherently, the other side CANNOT go above 2x B+.

Here's why: The center tap of the transformer is ground-coupled through the power supply caps. The power supply prevents any significant transient voltage from appearing on the power supply. In essence, the B+ lead is AC coupled to ground, for all intents and purposes.

The plate lead- the one side of the winding that is at ground- is ALSO held to ground, by the flyback diode. The plate cannot go below ground.

So, in essence, any transient is shorted through that side of the winding, limiting the voltage across that winding to between B+ and zero volts (ground). And, due to inductive coupling- the OTHER side is ALSO bound to NOT go over 2X B+ (B+ at the center tap, plus the B+ that is being limited by the other side), The voltage on the side that is seeing 2x B+, is constrained to no more voltage swing than the other side- since both are coupled together in the transfomer core (like an autotransformer, in effect).

Yes, there may be a bit of a peak, due to the fact that transformers are not perfect devices (leakage inductance and the like)- but it will be much smaller than it would be without the flyback diode.

Regards,
Gordon.

 

[Mr Boochie]

Actually, now that I understand the flyback diode connection (between plate and ground, not plate and B+)- it appears that if one side of the winding is clamped to not go below zero volts (ground) by a diode- then, inherently, the other side CANNOT go above 2x B+.

Here's why: The center tap of the transformer is ground-coupled through the power supply caps. The power supply prevents any significant transient voltage from appearing on the power supply. In essence, the B+ lead is AC coupled to ground, for all intents and purposes.

The plate lead- the one side of the winding that is at ground- is ALSO held to ground, by the flyback diode. The plate cannot go below ground.

So, in essence, any transient is shorted through that side of the winding, limiting the voltage across that winding to between B+ and zero volts (ground). And, due to inductive coupling- the OTHER side is ALSO bound to NOT go over 2X B+ (B+ at the center tap, plus the B+ that is being limited by the other side), The voltage on the side that is seeing 2x B+, is constrained to no more voltage swing than the other side- since both are coupled together in the transfomer core (like an autotransformer, in effect).

Yes, there may be a bit of a peak, due to the fact that transformers are not perfect devices (leakage inductance and the like)- but it will be much smaller than it would be without the flyback diode.

Regards,
Gordon.

Very nice explanation! Here's a link saying basically, the same thing.

https://www.electronicshub.org/flyback-diode-or-freewheeling-diode/

 

[Retrovert]

Actually, now that I understand the flyback diode connection (between plate and ground, not plate and B+)- it appears that if one side of the winding is clamped to not go below zero volts (ground) by a diode- then, inherently, the other side CANNOT go above 2x B+. [...] Yes, there may be a bit of a peak, due to the fact that transformers are not perfect devices (leakage inductance and the like)- but it will be much smaller than it would be without the flyback diode.

Correct, and the winding's insulation ought to safely withstand least twice B+ so failures will not occur.

This is why the flyback diode is the superior solution to adding a parallel resistor. Modern high-voltage diodes are available at 3,000 V @ 1/8 to 1/4 Amps for under $0.50 apiece. No need to use a failure-prone string of the 1N4007.

Additionally adding a TVS diode rated at slightly over B+ is beneficial, as it can withstand the repeated striking of the ignition coil. That way the diode first strikes, then the TVS addresses any unclamped portion. An MOV simply cannot endure the repeated strikes without damage. The MOV is a consumable and sacrificial device, each time slightly ruined by clamping. All of the papers and design guides stress this issue, as I above set forth. No secret here. The Gas Discharge Tube (GDT) is slower than a TVS diode and is also a sacrificial device, being degraded after clamping either (a) a small number of large transients, or (b) a larger number of small transients. The only virtue of the GDT is being inexpensive at very high voltages.

Better and inexpensive technology now exists. Modern components are simply amazing. Solid-state truly is the ideal solution for clamping, as it may conduct away the inductive kick, including that stimulated by the back-EMF of a driver, every single audio cycle without damage. Pump (amplifier) and dump (diode) all day long.

I would not have thought the modification of a flyback diode to be either controversial or arcane.

 

[mroboto]

Keep the FB circuit in mind. I'm sure Sansui engineered the FB to accomodate the protection load, but if you just add a resistor in parallel to the primary FB resistor you will reduce the FB, too. What if you have a 5K primary FB resistor and you put a 500R in parallel with it hanging off the 16R tap, or the 8R tap? It will lower the FB signal. You would need to adjust the FB resistor lower to get the correct FB signal back going to the driver. If you do that then you may also be reducing the total resistance back from the driver cathode and changing the biasing voltage of the driver, creating a need for a different cathode resistor to get that voltage back to the design value. But hey, .... it probably won't hurt anything and probably won't hear any difference, ... but can you just say do it to every amp?

Some of my amps have a feedback, and I removed it on some.

Can't tell what protection resistors change on sound.

 

[OBMG74]

Slightly off topic but was discussed:

I am totally with Retrovert on the use of MOV's in power strips.
They cause more harm than good. You won't find any in my house.

I use a Square D whole-house surge protector at my panel.
Not sure if it really protects against surges (certainly not against lightning as the packaging on some power-strips suggests).
But I am sure that it won't burn the house down.

Back on topic, I am going to use the 470ohm resistor on my tube amps.
Has been on my to-do list for quite some time.

 

[peterh]

A FB system will do more to protect an amp from spikes than anything else. It works in real time and will attenuate the spike generated from whatever source since the spike is not part of the input signal it will kill it as fast as it rises. It's just distortion to the FB. So FB will rise and counter the plate voltage spike.

FB will not work when amp is clipping. And it's at clipping time that no-load is most dangerous.

 

[gadget73]

Extra resistor across the secondary won't affect feedback or driver bias in any meaningful way. The DC path is through the transformer, which is usually not more than an ohm. AC signal would be much more affected by the speaker load, another 500 ohms in parallel with even a 16 ohm speaker is insignificant.

can't believe we're at 3 pages of hand wringing over this. Put the resistor in and call it good, or just be careful to not wind up the loud knob with the speakers unhooked. Its not that complicated.

 

[Retrovert]

Some of my amps have a feedback, and I removed it on some. Can't tell what protection resistors change on sound.

The feedback claim is utter nonsense, as has been pointed out. Feedback cannot possibly stop an arc, and certainly not when the amplifier is clipping. I would cite terms like "slew rate" but that would go over the heads of those promulgating such nonsense. Brandolini's Law applies.

I've explained the effects of the resistor, that it must be placed i parallel with driver on the same tap, that flyback diodes and a TVS diode are the superior solution, that MOV is a self-sacrificing and degrading component, etc. etc. etc.

I don't know why my time is being wasted defending standard electrical engineering practice. Again, Brandolini's Law..

 

[AmadeusMozart]

I find the whole reasoning for using flyback diodes a tad silly. It is like reasoning that one should not use fuses either because once they blow they stop working.

I wonder who in their right mind would repeatedly disconnect the speakers while having the amplifier going full tilt. Or is running an amplifier continiously into clipping, the mind boggles.

If running an amp within normal specifications then protection is hardly necessary and a MOV is cheap insurance.

 

[peterh]

I find the whole reasoning for using flyback diodes a tad silly. It is like reasoning that one should not use fuses either because once they blow they stop working.

I wonder who in their right mind would repeatedly disconnect the speakers while having the amplifier going full tilt. Or is running an amplifier continiously into clipping, the mind boggles.

If running an amp within normal specifications then protection is hardly necessary and a MOV is cheap insurance.

The discussion so far deals with involuntary disconnected speakers possibly combined with a large input signal.
As an example one could think of a user that turns the amp on, but hears no music and turns the volume up until the sparks occur. The speaker connectors was disconnected by the dog while chasing the cat. Or whatever scenario one can think of.

What have been discussed is ways of protecting the amp in those cases. Not an evil guest that deliberately disconnects the speakers and turns volume fully up.

The disagreements is how to protect the amp, not why.

 

[OBMG74]

^^^^^^ Yes, to protect me from myself

 

[Retrovert]

I find the whole reasoning for using flyback diodes a tad silly. It is like reasoning that one should not use fuses either because once they blow they stop working.

Flyback diodes never degrade or blow.

The point about the MOV being unsuitable is that in the event clamping is needed the MOV rapidly self-destructs and protection ends, causing destruction of the output transformer, output tubes, and whatever else in in the path of the arc.

I wonder who in their right mind would repeatedly disconnect the speakers while having the amplifier going full tilt. Or is running an amplifier continiously into clipping, the mind boggles.

Ummm, I wonder who in their right mind would ever use a cleaning service which might hook a vacuum over a speaker cord and disconnect it. Or who in their right mind would ever be poking around in an audio rack and accidentally disconnect a speaker wire or banana plug.

**** happens in life. This is why insurance exists. The flyback diodes are inexpensive insurance.

If running an amp within normal specifications then protection is hardly necessary and a MOV is cheap insurance.

Nuh-uh. The MOV is false insurance, hardly a bargain at the low price. The MOV will rapidly fail because the audio cycle continuously pumps energy in, over and over, like an ignition coil. An MOV is only for rare occurrences, as it degrades after each strike. The typically use is inexpensive protection from lightning surges and that will not long work.

 

[AmadeusMozart]

The discussion so far deals with involuntary disconnected speakers possibly combined with a large input signal.
As an example one could think of a user that turns the amp on, but hears no music and turns the volume up until the sparks occur. The speaker connectors was disconnected by the dog while chasing the cat. Or whatever scenario one can think of.

What have been discussed is ways of protecting the amp in those cases. Not an evil guest that deliberately disconnects the speakers and turns volume fully up.

The disagreements is how to protect the amp, not why.

If the dog or the cat disconnect speakers then there is something sorely lacking in how the equipment is installed.

What happened to locking speaker conenctions or using the screw down? Banana plugs for permanent connections are just plain dumb.

And if you do not hear anything when you switch the amplifier on then you should under no circumstances turn volume all the way up.

But hey, there is no protection against stupidity.

 

[peterh]

If the dog or the cat disconnect speakers then there is something sorely lacking in how the equipment is installed.

What happened to locking speaker conenctions or using the screw down? Banana plugs for permanent connections are just plain dumb.

And if you do not hear anything when you switch the amplifier on then you should under no circumstances turn volume all the way up.

But hey, there is no protection against stupidity.

True. **** happens. That's why one should try to make equip accident-safe. It's a very low cost to do it, just do it right !

 

[AmadeusMozart]

Flyback diodes never degrade or blow.

The point about the MOV being unsuitable is that in the event clamping is needed the MOV rapidly self-destructs and protection ends, causing destruction of the output transformer, output tubes, and whatever else in in the path of the arc.



Ummm, I wonder who in their right mind would ever use a cleaning service which might hook a vacuum over a speaker cord and disconnect it. Or who in their right mind would ever be poking around in an audio rack and accidentally disconnect a speaker wire or banana plug.

**** happens in life. This is why insurance exists. The flyback diodes are inexpensive insurance.



Nuh-uh. The MOV is false insurance, hardly a bargain at the low price. The MOV will rapidly fail because the audio cycle continuously pumps energy in, over and over, like an ignition coil. An MOV is only for rare occurrences, as it degrades after each strike. The typically use is inexpensive protection from lightning surges and that will not long work.

It has pointed out that your percepotion of rapid self destruction is flawed. The protection you seem to propose appear to be tailored for those who continously abuse an amplifier.

As I mentioned just before there is no protection against stupidity in either operating (turning volume up when no sound comes out) or installing badly (using banana connectors): if cables are screwed down then they will not be disconnected.

 

[mroboto]

Just note I have flimsy wires in a corner where we vacuum.

The amplifier binding posts holes don't even retain the cable as the holes are large and the wire is thin, I have to tighten all the way for the cable to be tied.

I was anxious before the addition of resistors.