I was recently thinking about whether it would be possible to add an additional output transformer to a tube amp to provide an additional tap (never mind why). This has led me to realize that I don't really understand what is going wrong when an amp is destroyed by being run with its speaker unplugged. And to my surprise, the internet doesn't easily cough up clear answers. Maybe we can fix that here.
Some thoughts / speculation / questions as I try to get a clearer picture of the situation:
- Does a transformer with no secondary load act like inductor, as if the secondary weren't there at all? Since no current can flow in the secondary, this seems probably right to me.
- The problems are all about what happens on the primary side, right? If I'm understanding correctly, the secondary gets a voltage across its leads but nothing destructive happens there. Like when testing a power transformer to determine its open circuit secondary voltage(s).
- In fact, testing a power transformer with an open secondary is a pretty similar situation in which nothing is destroyed! 120VAC at 60Hz on the primary side, secondary side disconnected, and everything is fine. Even when you then unplug it, giving the primary nowhere to dump its current as its magnetic field collapses.
- People seem to describe two symptoms when things do go wrong: power tubes are destroyed, or high voltages in the output transformer cause arcing that destroys its insulation. Seems consistent with the idea that the problem is a flyback voltage from the primary. Meaning, the problem occurs when the primary side has no path for its current, and for some reason it's worse for an output transformer than for a power transformer. (Why?)
- But if the output tube is conducting, why would there be a flyback voltage spike? Is the problem limited to push-pull amps where the signal drives the output tubes into cutoff? I guess in a SE amp, the signal does restrict current through the tube somewhat, and anode voltage would rise to keep the current constant, which could become a problem if it exceeds ratings.
- But if all of the above is right, why don't power supply filter chokes cause the same problem? The situation is a little more complicated, but they are essentially in the anode load for the tubes, right? Are filter caps always arranged so that they provide a place for filter choke current to go?
Some thoughts / speculation / questions as I try to get a clearer picture of the situation:
- Does a transformer with no secondary load act like inductor, as if the secondary weren't there at all? Since no current can flow in the secondary, this seems probably right to me.
- The problems are all about what happens on the primary side, right? If I'm understanding correctly, the secondary gets a voltage across its leads but nothing destructive happens there. Like when testing a power transformer to determine its open circuit secondary voltage(s).
- In fact, testing a power transformer with an open secondary is a pretty similar situation in which nothing is destroyed! 120VAC at 60Hz on the primary side, secondary side disconnected, and everything is fine. Even when you then unplug it, giving the primary nowhere to dump its current as its magnetic field collapses.
- People seem to describe two symptoms when things do go wrong: power tubes are destroyed, or high voltages in the output transformer cause arcing that destroys its insulation. Seems consistent with the idea that the problem is a flyback voltage from the primary. Meaning, the problem occurs when the primary side has no path for its current, and for some reason it's worse for an output transformer than for a power transformer. (Why?)
- But if the output tube is conducting, why would there be a flyback voltage spike? Is the problem limited to push-pull amps where the signal drives the output tubes into cutoff? I guess in a SE amp, the signal does restrict current through the tube somewhat, and anode voltage would rise to keep the current constant, which could become a problem if it exceeds ratings.
- But if all of the above is right, why don't power supply filter chokes cause the same problem? The situation is a little more complicated, but they are essentially in the anode load for the tubes, right? Are filter caps always arranged so that they provide a place for filter choke current to go?
Sahazel, stay with your large iron cored inductor thought, then apply a large square wave voltage to it. What happens at the squarewave edges, knowing that during half the squarewave there is a lot of current flowing in the inductor.
PS edit: some waveforms in the linked doc
https://dalmura.com.au/static/Output%20transformer%20protection.pdf
PS edit: some waveforms in the linked doc
https://dalmura.com.au/static/Output%20transformer%20protection.pdf
Last edited:
> ...for some reason it's worse for an output transformer than for a power transformer. (Why?)
- But if the output tube is conducting, why would there be a flyback....
The transformer can sit there with no or low voltage almost forever no trouble. The problem comes when you BEAT it, trying to hear something (or, because it sounds kewl).
We don't normally drive power transformers with heavy sine waves. (Acually we do, but we call them "switchers" and design accordingly.)
There is flyback when the tube STOPS conducting. (In audio driven to clipping, that happens many times a second.) Then the insulation breaks down.
We don't care about "ratio" if the secondary is open or missing.
The flyback is often about 10X the supply voltage. So the kick on a big 8 Ohm winding is hundreds of volts, the kick on a 4K ohm winding is thousands of volts. If you have both windings, the high voltage one will give trouble first (and last).
- But if the output tube is conducting, why would there be a flyback....
The transformer can sit there with no or low voltage almost forever no trouble. The problem comes when you BEAT it, trying to hear something (or, because it sounds kewl).
We don't normally drive power transformers with heavy sine waves. (Acually we do, but we call them "switchers" and design accordingly.)
There is flyback when the tube STOPS conducting. (In audio driven to clipping, that happens many times a second.) Then the insulation breaks down.
We don't care about "ratio" if the secondary is open or missing.
The flyback is often about 10X the supply voltage. So the kick on a big 8 Ohm winding is hundreds of volts, the kick on a 4K ohm winding is thousands of volts. If you have both windings, the high voltage one will give trouble first (and last).
The output transformer primary inductance will increase if secondary is not loaded. Under idling conditions, this does not cause any harm. But when a signal is applied, the magnetic field collapses suddenly and this generates very high voltage across the winding. This could be many thousands of volts. Can create sparks on terminals or puncture the winding. The tube usually survives as the winding rupture will create some shorted turns, reduces the inductance and thus the induced voltage. This is enough to prevent tube damage. Never leave the transformer unloaded. Regards.
The primary winding inductance is not related to the secondary side loading.
For a PP output transformer in an amp with no speaker loading and high signal level, then for a part of a signal cycle the inductance of a half-primary winding would reduce as winding current increases once its associated valve anode swings from cutoff to saturation (the loadline is effectively horizontal at that time) and then follows the saturation curve up for Vgk=0V and is dependant on the available screen voltage and the available inductance to support the applied voltage.
Last edited:
Primary inductance is indeed related to secondary loading. You can easily check it yourself. Connect a 100 ohms potentiometer to any transformer secondary and apply primary voltage and monitor current. As the secondary load is increased, towards zero ohms, the primary current will increase. You can also measure with any instrument that measures inductance. For technical reasons, primary inductance is mentioned without any secondary loading.
Also the maximum induced voltage in the primary is generated when the current falls to zero. The higher the change in current towards zero, the higher the induced voltage.
All transformers work the same way.
In any transformer, the iron core gets saturated with about 2-5% of full primary current. When the secondary is loaded, the current in the secondary winding produces magnetic flux in the opposite direction. This cancels some of the flux in the magnetic core and cause primary current to increase.
The primary current increases because with reduced flux in the core results in less induced voltage. The difference between applied voltage and induced voltage increase. The current stabilize when the core is again saturated to the initial level.
Regards.
Also the maximum induced voltage in the primary is generated when the current falls to zero. The higher the change in current towards zero, the higher the induced voltage.
All transformers work the same way.
In any transformer, the iron core gets saturated with about 2-5% of full primary current. When the secondary is loaded, the current in the secondary winding produces magnetic flux in the opposite direction. This cancels some of the flux in the magnetic core and cause primary current to increase.
The primary current increases because with reduced flux in the core results in less induced voltage. The difference between applied voltage and induced voltage increase. The current stabilize when the core is again saturated to the initial level.
Regards.
The primary inductance only depends on the excitation. As you increase the load current, you see an increase in primary current but it’s resistive, at right angles to the current in the inductance. You may see a decrease in that inductive primary current as the resistive part (due to loading) increases - but that’s a consequence of reduced effective primary excitation not the secondary load itself. When you add the primary and secondary side leakage reactances to the model, the part of the voltage that hits the “ideal transformer” decreases with loading. The L in parallel with that “ideal transformer” primary remains the same and only depends on the applied voltage/frequency.
Inductance of a iron core coil is proportional to the number of turns and magnetic property of that iron. If the magnetic property of iron is varied, then the inductance of the coil will vary. Loading the secondary changes the magnetic property of the core.
If the inductance of primary does not vary as you say, the impedance offered at a frequency is fixed, and so the primary current will be same at all loads, which is wrong. Regards
If the inductance of primary does not vary as you say, the impedance offered at a frequency is fixed, and so the primary current will be same at all loads, which is wrong. Regards
Inductance or impedance?
The "technical reason" for not loading the secondary to measure primary inductance is to separate resistance from inductance.
In case of output transformer.. When the transformer secondary is loaded with a constant load (impedance), it exhibits constant load on the primary (near constant impedance) over varying frequency.
Here we specify as impedance matching because of the turns ratio. The inductance of the transformer keeps changing to the frequency to maintain constant impedance.
Reactance is a factor of inductance and frequency. If reactance (here impedance) is maintained constant over a frequency range, the inductance will change to adjust.
In case of a mains transformer, the frequency is fixed. When the secondary load changes and the reflected load on the primary also change. The reactance lowers as the load increase, resulting in lower induced voltage. The primary current is depended on the supply voltage minus induced voltage. So primary current increases.
For the output transformer, If the secondary is disconnected, the primary becomes a single coil. The impedance now becomes frequency dependent, but has the inductance is constant. So the term impedance is not valid in this context unless a fixed frequency is mentioned.
I suggest anyone do the following simple test if you have access to any LCR meter.
Connect the primary of a small transformer say 10 VA to the LCR meter and read the inductance in the following cases.
1. With secondary wires open. 2. With secondary loaded with a low ohms resistor. 3. With secondary shorted.
Regards.
Here we specify as impedance matching because of the turns ratio. The inductance of the transformer keeps changing to the frequency to maintain constant impedance.
Reactance is a factor of inductance and frequency. If reactance (here impedance) is maintained constant over a frequency range, the inductance will change to adjust.
In case of a mains transformer, the frequency is fixed. When the secondary load changes and the reflected load on the primary also change. The reactance lowers as the load increase, resulting in lower induced voltage. The primary current is depended on the supply voltage minus induced voltage. So primary current increases.
For the output transformer, If the secondary is disconnected, the primary becomes a single coil. The impedance now becomes frequency dependent, but has the inductance is constant. So the term impedance is not valid in this context unless a fixed frequency is mentioned.
I suggest anyone do the following simple test if you have access to any LCR meter.
Connect the primary of a small transformer say 10 VA to the LCR meter and read the inductance in the following cases.
1. With secondary wires open. 2. With secondary loaded with a low ohms resistor. 3. With secondary shorted.
Regards.
Last edited:
Perhaps best to represent an output transformer as an inductance in parallel with a resistive load. The resistive load is the reflected secondary load seen by the primary winding terminals at mid-band frequencies. The inductance is typically very large and so its shunting impedance is negligible at mid-band frequencies. It is only at low bass frequencies where the inductance shunt element starts to become a dominant player. But the inductance doesn't vary due to secondary side loading. The inductance parameter value is determined by the level of primary winding excitation voltage (in a sinewave sense - with inductance rising with signal level but then plateauing and dropping as core saturation is encroached on by the B-H operating loci), and any DC bias, and there can be a minor influence from core lamination permeability change with frequency.
The measured inductance depends on the test frequency and whether the test frequency is influenced by the primary winding resonance (winding inductance and shunt capacitance). That is why primary inductance is typically measured around 50Hz, as the primary winding resonance for all windings open for a moderately sized PT or OPT is circa 1kHz. For measuring leakage inductance of the primary, with all other windings shorted, the test frequency used is typically 1kHz, as that is often well below any primary resonance between leakage inductance and shunt capacitance, but well above low frequency core induced affects where the phase is falling back to being a resistive character.
Last edited:
We are discussing what happens when secondary is open.
The output transformer changes from a constant impedance to constant inductance when the secondary is open.
The induced voltage is also frequency dependent. Higher frequency, very short time current drops and produces large voltage. Signal will contain various frequencies. Frequencies that are higher or large current swings is enough to damage the transformer. Touching the phono plug with fingers is enough.
When the current drops fast to near zero, the resistance of the winding is immaterial. The high voltage induced will destroy the transformer. This behavior is absent with mains transformers as the frequency is fixed in design itself.
You can test with any LCR meter. Most of them as I know use constant test frequency.
The test is to suggest one to observe transformer behavior.
Regards.
The output transformer changes from a constant impedance to constant inductance when the secondary is open.
The induced voltage is also frequency dependent. Higher frequency, very short time current drops and produces large voltage. Signal will contain various frequencies. Frequencies that are higher or large current swings is enough to damage the transformer. Touching the phono plug with fingers is enough.
When the current drops fast to near zero, the resistance of the winding is immaterial. The high voltage induced will destroy the transformer. This behavior is absent with mains transformers as the frequency is fixed in design itself.
You can test with any LCR meter. Most of them as I know use constant test frequency.
The test is to suggest one to observe transformer behavior.
Regards.
Mandu, in a PT operating with mains frequency and sinewave voltage and no loading on other windings, the current level is normally low at the magnetising current level and is distorted but still basically sinusoidal. However for a particular PT if either the voltage gets too high, or the frequency too low, then the core BH loci will enter saturation region and the inductance will drop for that part of the excitation waveform and the current through the winding will peak up. The current waveform is seen to have a very peaky portion during each half-cycle. Similar situation for an inrush event, but the winding current is very peaky only during half the cycle. Perhaps if you google for some example primary current waveforms. That all happens for loaded or unloaded secondary, except that for a loaded secondary the primary current carries both the magnetising current plus the transformed current requirement of the secondary (which normally dominates, but not if conditions are forcing core saturation).
In an OPT in a valve amp that has excessive signal applied to the output stage, the voltage applied to the OPT changes from sine to a square wave. The squarewave maintains the full B+ across the primary initially - like a large applied DC level - forcing the primary current to ramp up according to the available primary inductance, but the incremental inductance starts to fall if there is sufficient time for the core BH to enter saturation section, and so the primary current rises up to a significant level (as in an inductor or a DC relay coil) but is pegged by the available resistance of the conducting valve (and its Vgk=0V anode curve), the DCR of the primary winding, and the sag of the power supply B+ and screen rails. Then the output stage valve gets turned off rapidly, and the current in the primary wants to stay at the level but can't, and so the inductive energy in the winding is transferred in to a flying voltage across the winding shunt capacitance as the valve has moved rapidly in to cutoff and so is not providing a current path.
In an OPT in a valve amp that has excessive signal applied to the output stage, the voltage applied to the OPT changes from sine to a square wave. The squarewave maintains the full B+ across the primary initially - like a large applied DC level - forcing the primary current to ramp up according to the available primary inductance, but the incremental inductance starts to fall if there is sufficient time for the core BH to enter saturation section, and so the primary current rises up to a significant level (as in an inductor or a DC relay coil) but is pegged by the available resistance of the conducting valve (and its Vgk=0V anode curve), the DCR of the primary winding, and the sag of the power supply B+ and screen rails. Then the output stage valve gets turned off rapidly, and the current in the primary wants to stay at the level but can't, and so the inductive energy in the winding is transferred in to a flying voltage across the winding shunt capacitance as the valve has moved rapidly in to cutoff and so is not providing a current path.
Last edited:
The power transformers are designed for 50 or 60 Hz +/- 2% and supply voltage +/- 5%.
Only during the power up (inrush) time, the first few cycles of current are higher because of no/low magnetic field in the core. Once the magnetic field is established, the collapsing field during every cycle produces back EMF aka induced voltage. This opposes the primary voltage and keeps the magnetizing current constant afterwards. When the secondary is loaded, the current in the secondary produces a magnetic field in opposite direction to that of the primary. This effectively reduces the available magnetic field in the core and reduces the induced voltage in the primary. Since the opposing induced voltage is lower now, more current flows in primary until the magnetic field reach the saturation level. (The reactance changes as the load changes) This is basically how every power transformer work. At no load up, the transformer exhibits the highest impedance. The inductance of the transformer reduces as the secondary is loaded. Though the impedance of the transformer primary is a function of primary winding resistance and inductance, the change in the inductance is what matters when loading a transformer secondary which causes the change in primary current. The function of the resistance will be only to calculate copper losses, and some secondary voltage drop, which I ignore for this topic.
The power transformer is designed to operated with no load i.e secondary open i.e maximum inductance.
We are discussing the case with the OPT secondary open.
Under normal operation, with loaded secondary, the transformer functions as a constant impedance device for the frequency range it is designed. With secondary open, functions as a constant inductance device. The ratio between the resistance and reactance can be more than 100 with open secondary. And this ratio goes up higher with frequency. So the DC resistance is not considered in the discussion.
Regards
Only during the power up (inrush) time, the first few cycles of current are higher because of no/low magnetic field in the core. Once the magnetic field is established, the collapsing field during every cycle produces back EMF aka induced voltage. This opposes the primary voltage and keeps the magnetizing current constant afterwards. When the secondary is loaded, the current in the secondary produces a magnetic field in opposite direction to that of the primary. This effectively reduces the available magnetic field in the core and reduces the induced voltage in the primary. Since the opposing induced voltage is lower now, more current flows in primary until the magnetic field reach the saturation level. (The reactance changes as the load changes) This is basically how every power transformer work. At no load up, the transformer exhibits the highest impedance. The inductance of the transformer reduces as the secondary is loaded. Though the impedance of the transformer primary is a function of primary winding resistance and inductance, the change in the inductance is what matters when loading a transformer secondary which causes the change in primary current. The function of the resistance will be only to calculate copper losses, and some secondary voltage drop, which I ignore for this topic.
The power transformer is designed to operated with no load i.e secondary open i.e maximum inductance.
We are discussing the case with the OPT secondary open.
Under normal operation, with loaded secondary, the transformer functions as a constant impedance device for the frequency range it is designed. With secondary open, functions as a constant inductance device. The ratio between the resistance and reactance can be more than 100 with open secondary. And this ratio goes up higher with frequency. So the DC resistance is not considered in the discussion.
Regards
Power wants to flow from the output transformer to the speaker. When the transformer is impedance matched to the load maximum power is transferred, but when there is a mismatch of impedance less power is transferred. With the speaker unplugged you have an extreme impedance mismatch. But the tubes are still generating power so where does it go?
Power is reflected back into the amplifier, and where there is a build up of power there is heat. This is similar to reflections and build up of standing waves on a transmission line. At least that’s the way I think about it.
Power is reflected back into the amplifier, and where there is a build up of power there is heat. This is similar to reflections and build up of standing waves on a transmission line. At least that’s the way I think about it.
Last edited by a moderator:
Such a reflection (in phase at audio frequency, from the open circuit) won’t result in excess power dissipation within the amp (heat), but will result in excess voltage. It allows anything built up in the primary inductance to go unchecked. The energy is limited to what is stored in the primary inductance (1/2 LI squared - but with nothing to damp it even a few joules goes a long way. An out of phase reflection results from a *short circuit*. That’s not destructive for the transformer, but the tubes might get a little red in the face if you let it run that way long enough. At audio frequency, other than zero or 180 degree reflection phase is hard to generate without an explicit pure inductor or capacitor load - the wavelength is too long to generate rotation along a “transmission line”.