Hello Tim,
I know the thread has gone unused, but let me ask a question: why do you use series resistors along with the MOVs?
There are two different groups answering or giving advice on the matter making it a bit confused.
1. HiFi group
2. Guitar player group.
But the problem is not fully solved.
How to make sure a bad tube, being it the phase inverter or the power tubes, a short or open output, will not destroy more.
Many high tech sollution have been invented to bring distortion down in tube amps; or even raised it.
Whole libraries have been written to cancel hum, to avoid ground loops and to make you believe in the so many articles about "snake oil" (like the way the printing is done on the cabling.... for example)
But, alaas, so little is written on protecting the various parts in a tube amp by all causes mentioned before.
It would be nice to continue to think and create a good working protection scheme.
Build with common components and no sw and micro controllers as these will become obsolete further in time or the sw is lost etc...
I'm curious what the outcome will be...
1. HiFi group
2. Guitar player group.
But the problem is not fully solved.
How to make sure a bad tube, being it the phase inverter or the power tubes, a short or open output, will not destroy more.
Many high tech sollution have been invented to bring distortion down in tube amps; or even raised it.
Whole libraries have been written to cancel hum, to avoid ground loops and to make you believe in the so many articles about "snake oil" (like the way the printing is done on the cabling.... for example)
But, alaas, so little is written on protecting the various parts in a tube amp by all causes mentioned before.
It would be nice to continue to think and create a good working protection scheme.
Build with common components and no sw and micro controllers as these will become obsolete further in time or the sw is lost etc...
I'm curious what the outcome will be...
I have one of these across the primary of the OPT in my 833C monos. Just in case.
2095-600-CLF Bourns Inc. | 2095-600-CLF-ND | DigiKey
2095-600-CLF Bourns Inc. | 2095-600-CLF-ND | DigiKey
I need it for a guitar amp, where the possibilities of accidentally unplugging the speaker are bigger - still, I will use only speakon connection on the back, to limit them.
I do quite understand the use of the MOVs, but not the series resistor. I thought that if the MOVs conduct an overvoltage, then this current, distributed across the series resistor, could maintain a huge spike across the OT. Of course, the resistor could just as well blow, but this is not good anyway.
I considered it more like trying to clamp spikes using a diode + resistor - a MOV does not work exactly like a diode, but I would like to clarify this anyway.
I do quite understand the use of the MOVs, but not the series resistor. I thought that if the MOVs conduct an overvoltage, then this current, distributed across the series resistor, could maintain a huge spike across the OT. Of course, the resistor could just as well blow, but this is not good anyway.
I considered it more like trying to clamp spikes using a diode + resistor - a MOV does not work exactly like a diode, but I would like to clarify this anyway.
If the amp in question is operated at a volume level such that removing the load will not cause the output to rise above the clipping level, then an open circuit on the speaker leads will NOT blow anything. If the amp is operated at a volume level such that shorting the speaker load does not cause the tube current to rise above its peak current rating, or its dissipation rating if the short is long term, then a short will NOT blow anything. This is often the case with a HiFi amp since we tend to listen at a level well below clipping. I routinely switch loads with clip leads on my test amps at a reduced output level (maybe 1/4 the expected maximum power) during amp design. I always test my amps into a short. Never purposefully test a tube amp into an open unless you KNOW what will happen, AND are prepared for the worse. An amp with feedback, especially global feedback, may misbehave into a short or open attempting to correct the situation, causing clipping or excessive current.
This is NOT the case with a guitar amp. A well designed guitar amp should be capable of operating into extreme clipping continuously without damage.
Let's assume that Mr. Metalhead uses a pedal board with a million DB of gain and cranks everything to 11 and while performing a 15 minute trainwreck solo. The output tubes operate as switches, alternately switching each plate lead to ground at a 250 Hz or so rate. Assuming no saturation in the OPT (not a guarantee, especially with a Fender) the tubes will not see much stress since there is little average dissipation. One plate lead will be pulled toward ground by a saturated tube reaching 30 to 70 volts depending on the tube and driver circuit. The current reaches a maximum as the plate voltage goes near zero. This current is limited by the B+ and the instantaneous load IMPEDANCE reflected to the tube by the speaker and the OPT. The other plate is essentially OPEN CIRCUIT because the tube is cutoff. The half primary with the cutoff tube is essentially an unloaded secondary, and it's plate terminal will usually rise to about twice the B+ voltage if the secondary is operated into a perfect resistive load.
A speaker, even a mild mannered HiFi speaker is NOT a resistive load. A guitar speaker, or combination of speakers, usually has a resonance WITHIN the frequency range that a guitar operates. It's impedance can go to 30 ohms or more around resonance. (70 to 150 Hz). It can also generate considerable counter EMF when fed enough power to make the cone dance. Bass guitar speakers around resonance are worse case. A speaker also has an inductive and a capacitive component, it is not resistive.
It is not unusual to see peak plate voltages exceed THREE TIMES the B+ under normal operating conditions, and more if the (bass) player likes to slap the strings around.
The definition of inductance is the property of an electrical element that opposes a CHANGE in current flow. Any coil of wire has inductance. A lot of wire wrapped around an iron core has a lot of inductance, and so does any winding on an OPT. An OPT has at least two windings that are COUPLED magnetically. A change in CURRENT through any winding causes a change in current flow in all other windings. As long as there is a path for current to flow in at least one other winding everything is OK, the OPT will transfer the changing current flow in a primary winding to a secondary winding which becomes a changing current through the load (speaker).
If there is no load present, the conducting tube will switch one plate lead to ground causing a current to flow through the OPT half primary. This current will build up quickly to a very high level limited only by the DC resistance of the primary and the internal resistance of the tube, since there is no AC load to absorb the changing current. The property of inductance will attempt to keep this current flowing as the circuit reaches it's steady state current level, FORCING an overcurrent situation and causing the plate voltage on the conducting tube to swing NEGATIVE. Now BOTH TUBES become open circuits and there is NO path for current to flow! AT this point there is a large magnetic field in the OPT, and it MUST go somewhere. The previously conducting tube will see it's plate voltage go "flyback" to an extremely high negative voltage while the opposing plate will go to an extreme positive voltage. In an ideal (lossless) transformer the voltage will theoretically go to infinity! In reality there will be an arc somewhere. This is how the ignition coil in your car works.
I have seen voltages over 5KV on an amp with a B+ of 420 volts. I have set an OPT on FIRE by testing a cranked amp when my load resistor blew open. I since bought some 8 ohm 500 watt resistors.
This is NOT the case with a guitar amp. A well designed guitar amp should be capable of operating into extreme clipping continuously without damage.
Let's assume that Mr. Metalhead uses a pedal board with a million DB of gain and cranks everything to 11 and while performing a 15 minute trainwreck solo. The output tubes operate as switches, alternately switching each plate lead to ground at a 250 Hz or so rate. Assuming no saturation in the OPT (not a guarantee, especially with a Fender) the tubes will not see much stress since there is little average dissipation. One plate lead will be pulled toward ground by a saturated tube reaching 30 to 70 volts depending on the tube and driver circuit. The current reaches a maximum as the plate voltage goes near zero. This current is limited by the B+ and the instantaneous load IMPEDANCE reflected to the tube by the speaker and the OPT. The other plate is essentially OPEN CIRCUIT because the tube is cutoff. The half primary with the cutoff tube is essentially an unloaded secondary, and it's plate terminal will usually rise to about twice the B+ voltage if the secondary is operated into a perfect resistive load.
A speaker, even a mild mannered HiFi speaker is NOT a resistive load. A guitar speaker, or combination of speakers, usually has a resonance WITHIN the frequency range that a guitar operates. It's impedance can go to 30 ohms or more around resonance. (70 to 150 Hz). It can also generate considerable counter EMF when fed enough power to make the cone dance. Bass guitar speakers around resonance are worse case. A speaker also has an inductive and a capacitive component, it is not resistive.
It is not unusual to see peak plate voltages exceed THREE TIMES the B+ under normal operating conditions, and more if the (bass) player likes to slap the strings around.
The definition of inductance is the property of an electrical element that opposes a CHANGE in current flow. Any coil of wire has inductance. A lot of wire wrapped around an iron core has a lot of inductance, and so does any winding on an OPT. An OPT has at least two windings that are COUPLED magnetically. A change in CURRENT through any winding causes a change in current flow in all other windings. As long as there is a path for current to flow in at least one other winding everything is OK, the OPT will transfer the changing current flow in a primary winding to a secondary winding which becomes a changing current through the load (speaker).
If there is no load present, the conducting tube will switch one plate lead to ground causing a current to flow through the OPT half primary. This current will build up quickly to a very high level limited only by the DC resistance of the primary and the internal resistance of the tube, since there is no AC load to absorb the changing current. The property of inductance will attempt to keep this current flowing as the circuit reaches it's steady state current level, FORCING an overcurrent situation and causing the plate voltage on the conducting tube to swing NEGATIVE. Now BOTH TUBES become open circuits and there is NO path for current to flow! AT this point there is a large magnetic field in the OPT, and it MUST go somewhere. The previously conducting tube will see it's plate voltage go "flyback" to an extremely high negative voltage while the opposing plate will go to an extreme positive voltage. In an ideal (lossless) transformer the voltage will theoretically go to infinity! In reality there will be an arc somewhere. This is how the ignition coil in your car works.
I have seen voltages over 5KV on an amp with a B+ of 420 volts. I have set an OPT on FIRE by testing a cranked amp when my load resistor blew open. I since bought some 8 ohm 500 watt resistors.
Tubelab, many thanks for your comment.
Be aware that I decided to provide protection for my OT, reading a comment of yours in this thread (#6). Thanks for the careful explanation though!
I will build a 50W Hiwatt clone, feeding a 150W 8 ohm Fane speaker.
I was going to use 2 600V MOVs to protect my OT, but if I understand your remarks, you suggest that this may cause undesirable clipping for my amp, if I go near its output capability of course. My B+ will be around 470-480V.
I will fuse the cathodes of the tubes.
What do you suggest as a protection method? A high wattage resistor across the speaker terminals? MOVs across the OT? MOVs with resistors? MOVs across the speaker?
My speaker is around 103 db sensitive according to specs, and even if it is not so in real life (cab construction etc) I believe that should I go near 50W output power, I could just as well blow a wall down.
But this doesn't ensure safety if the amp could be used with other cabs - I plan to let this happen if I like.
EDIT: I see that some Fender amps use R3000 diodes, each of their cathoded connected to the output tube anodes. Is this sufficient at all?
Last edited:
Adding protection in any way is a 'design' process, and so a number of aspects could be considered if you have the time and inclination.
Keep in mind that more modern OT's should withstand higher over-voltage stress than old vintage parts - but as there was/is no 'standard' applicable, it is usually the test of time that identifies those parts that were perhaps a little short on. And some amps had circuitry included that acted as a form of over-voltage protection, which may have masked a poor performer.
I like MOV's because they have a 'soft' V-I characteristic that doesn't latch, and has no delay. If the MOV 'starts' to conduct (ie. the resistance of a MOV at its 1mA spec rating is pretty damn high) at a voltage significantly higher than the worst-case 'normal' rail-to-rail voltage swing, then it will likely dampen any overvoltage transient such that the voltage doesn't get close to the likely insulation/creepage/clearance/whatever breakdown voltage, and will effectively have no noticeable affect on the amps audible behaviour.
If the thought of adding some low level of shunt capacitance is going to upset a finely tuned hi-fi OT or feedback design, then adding a series resistor to each MOV provides a zobel network that can be included in the hi-fi design (and could even be used to dampen a winding self-resonance). For guitar amps, and where you have small 7mm disks to use with suitable rating, then a series resistance of no benefit.
I put a summary together of OT protection for old PA amps in:
Output transformer protection.pdf
Keep in mind that more modern OT's should withstand higher over-voltage stress than old vintage parts - but as there was/is no 'standard' applicable, it is usually the test of time that identifies those parts that were perhaps a little short on. And some amps had circuitry included that acted as a form of over-voltage protection, which may have masked a poor performer.
I like MOV's because they have a 'soft' V-I characteristic that doesn't latch, and has no delay. If the MOV 'starts' to conduct (ie. the resistance of a MOV at its 1mA spec rating is pretty damn high) at a voltage significantly higher than the worst-case 'normal' rail-to-rail voltage swing, then it will likely dampen any overvoltage transient such that the voltage doesn't get close to the likely insulation/creepage/clearance/whatever breakdown voltage, and will effectively have no noticeable affect on the amps audible behaviour.
If the thought of adding some low level of shunt capacitance is going to upset a finely tuned hi-fi OT or feedback design, then adding a series resistor to each MOV provides a zobel network that can be included in the hi-fi design (and could even be used to dampen a winding self-resonance). For guitar amps, and where you have small 7mm disks to use with suitable rating, then a series resistance of no benefit.
I put a summary together of OT protection for old PA amps in:
Output transformer protection.pdf
Last edited:
I just read the paper before reading your post, thank you!
So, having a B+ of 480V, I intend to use 575-660Vac MOV's. That would allow a level of overdrive above the nominal specs, so that at full volumes I will not clip my waveform.
Can you tell me what calculation you used to derive the corner frequency of the conjuctive filter?
Selecting a suitable MOV is a juggle - it's definitely worth trying to chase down the Vdc at 1mA spec level and tolerance range - as that needs to be a good margin above the worst-case B+ (at high mains and low idle).
At the other end of the design is the likely nebulous withstand voltage of the OT primary. With a PT, the insulation system is often designed to cope with a continuous 1 minute test at some high VAC (eg. 1k, 1.5k, 2k, 2.5k) depending on the winding and standard being complied with. It would be nice to think that an OT primary for a low-middle wattage part could cope with 1kVrms withstand. I certainly megger test old OTs to 1kV for insulation resistance check.
The filter is just the simple series RC corner frequency - where the R could pedantically be set at the effective half-primary impedance - but that may be too high a series resistance and significantly soften the clamping effect. The C of the filter will depend on the MOV spec - for 7mm discs that corner frequency gets pretty damn high, even for hi-fiers I'd expect.
At the other end of the design is the likely nebulous withstand voltage of the OT primary. With a PT, the insulation system is often designed to cope with a continuous 1 minute test at some high VAC (eg. 1k, 1.5k, 2k, 2.5k) depending on the winding and standard being complied with. It would be nice to think that an OT primary for a low-middle wattage part could cope with 1kVrms withstand. I certainly megger test old OTs to 1kV for insulation resistance check.
The filter is just the simple series RC corner frequency - where the R could pedantically be set at the effective half-primary impedance - but that may be too high a series resistance and significantly soften the clamping effect. The C of the filter will depend on the MOV spec - for 7mm discs that corner frequency gets pretty damn high, even for hi-fiers I'd expect.
I just don't understand why it is ok to omit the resistor for guitar amps, but not for hi-fi amps. Could I just use a 10R resistor? The frequency then would become very high, according to your calculation of the corner frequency.
Sorry for confusion.
The MOV is inert to the circuit operation up until a voltage transient causes the MOV resistance to drop. So for normal operation, the MOV just shunts the OT half-primary winding with less than 100pF. It's likely the self-capacitance of that winding far exceeds the added MOV capacitance, and so adding the MOV has negligible effect on audio operation. So there is no advantage to adding a resistor in series with the MOV, because the MOV capacitance is effectively negligible.
The winding self-capacitance can be deduced by testing the self-resonant frequency, and measuring the low-frequency inductance of the winding. The SRF is likely to be well above 20kHz, and so adding 100pF has little impact. But for a hi-fier, any change to impedances and SRFs etc may be an issue up to and exceeding a 100kHz region of interest.
The MOV is inert to the circuit operation up until a voltage transient causes the MOV resistance to drop. So for normal operation, the MOV just shunts the OT half-primary winding with less than 100pF. It's likely the self-capacitance of that winding far exceeds the added MOV capacitance, and so adding the MOV has negligible effect on audio operation. So there is no advantage to adding a resistor in series with the MOV, because the MOV capacitance is effectively negligible.
The winding self-capacitance can be deduced by testing the self-resonant frequency, and measuring the low-frequency inductance of the winding. The SRF is likely to be well above 20kHz, and so adding 100pF has little impact. But for a hi-fier, any change to impedances and SRFs etc may be an issue up to and exceeding a 100kHz region of interest.
Hi,
Resurrecting this thread since I think its so important....
I've been investigating these instead of MOV's - have anyone else used them?
I'm looking at trying out littlefuse CG21000L for example.
Ian
> protecting tubes and the output transformers
Use more/bigger tubes at lower voltage.
This is basic engineering. Your bridge falls down. The fix is a sturdier bridge, with more wood/steel in it. Structural designers aim for 3X to 10X safety factors because people die. Electronic engineers push right to the edge of melt-down, and stuff fails.
I had a 60W known for eating 7027. After a G2-K short smoked the PT, I rebuilt it at much lower B+. Also cathode bias and large G2 resistor. Got a third of the power, but about the worst could happen is smoke a resistor. Went from LOUD and fragile to mellow and forgiving.
A pair of 6550 at 20W and 250V may not notice any kind of abuse. If you truly need 100W, use 5 pair. While standard insulation should take well over a 500V kick, since since this 100W @ 250V OT will be a custom wind, you can ask for two layers of Mylar over the HV wraps.
Backward diodes on the ends of the OT winding nominally "catch" kickback. These are known to fail, because young designers (the old guys never did this) don't understand the voltage and current stresses. Again, triple-up your wood/steel (or Silicon).
Yes, you can "add protection". On that bridge, you could rig strain-cells to close the bridge when it sags. Early transistor amps "had" to do this. Unlike tubes, transistors eagerly pass excessive current. Voltage kills transistors, does not kill tubes (but note remarks on OTs). Transistors were too costly to use an ample amount. These protection circuits not only cut off part of "normal" waves, they failed to protect in some abnormal ways. The real cure for sand-world was to drive the cost per device down so low they could use dozens of devices in big amps. Protection could then be conservative yet still pass huge output safely.
_______________________
> I just don't understand .... guitar amps, but .... hi-fi amps.
A relaxed hi-fi guy, turned-up a bit and getting no sound, will turn-up cautiously then investigate.
A keyed-up musician, normally playing at 7 on the volume knob, not hearing anything, may well turn-up to 11 and strum harder. Aside from performance stress, the drummer and the venue staff make a lot of pre-show noise.
Use more/bigger tubes at lower voltage.
This is basic engineering. Your bridge falls down. The fix is a sturdier bridge, with more wood/steel in it. Structural designers aim for 3X to 10X safety factors because people die. Electronic engineers push right to the edge of melt-down, and stuff fails.
I had a 60W known for eating 7027. After a G2-K short smoked the PT, I rebuilt it at much lower B+. Also cathode bias and large G2 resistor. Got a third of the power, but about the worst could happen is smoke a resistor. Went from LOUD and fragile to mellow and forgiving.
A pair of 6550 at 20W and 250V may not notice any kind of abuse. If you truly need 100W, use 5 pair. While standard insulation should take well over a 500V kick, since since this 100W @ 250V OT will be a custom wind, you can ask for two layers of Mylar over the HV wraps.
Backward diodes on the ends of the OT winding nominally "catch" kickback. These are known to fail, because young designers (the old guys never did this) don't understand the voltage and current stresses. Again, triple-up your wood/steel (or Silicon).
Yes, you can "add protection". On that bridge, you could rig strain-cells to close the bridge when it sags. Early transistor amps "had" to do this. Unlike tubes, transistors eagerly pass excessive current. Voltage kills transistors, does not kill tubes (but note remarks on OTs). Transistors were too costly to use an ample amount. These protection circuits not only cut off part of "normal" waves, they failed to protect in some abnormal ways. The real cure for sand-world was to drive the cost per device down so low they could use dozens of devices in big amps. Protection could then be conservative yet still pass huge output safely.
_______________________
> I just don't understand .... guitar amps, but .... hi-fi amps.
A relaxed hi-fi guy, turned-up a bit and getting no sound, will turn-up cautiously then investigate.
A keyed-up musician, normally playing at 7 on the volume knob, not hearing anything, may well turn-up to 11 and strum harder. Aside from performance stress, the drummer and the venue staff make a lot of pre-show noise.
I stick a 400 ohm 12 W resistor across the speaker jack. That away it still has a load. I started doing that when I added DI out on my guitar amps so the outputs are still loaded.
That is the best solution.
with a slow blow fuse to the cathode to ground.
You can always insert properly rated neon discharge from transformer to ground which strike if the voltage reach 1000V or something.
Use diodes before the tube rectifier and fuse each diode.
I'm not worried about the power transformer.
But a glass discharge tube won't work if placed on the primary side of the output transformer so it will trip the mains fuse?
I have never had the problem of an unhooked speaker and my current OPT can handle B+ that would happen if all output tubes fail, so I am not terribly worried. That was part of my 'design process' as trobbins suggests.
But if I wanted to build something with lesser quality parts, then wouldn't this be worth considering? Glass discharge tubes can handle transients nicely it seems... and have far lower capacitance than any MOV. But I don't know if they are fast enough to protect a really cheap OPT...
Maybe I should simply get a few and test them.
Last edited:
Yeah... Maybe I am just a purist. Will a 400 Ohm resistor on the speaker terminals change my sound?
I guess this is so simple that it is worth trying out too.
btw - the fuse in the cathode can work very well, but finding one rated for say 65mA at a reasonable price is a bit challenging.
Last edited: