you are saying the same thing as me....only in a different sentence....
you have to realize that the discussion here is how to turn on a transformer used in a psu....naturally there is a rectifier and capacitor banks in the secondary and that affects the situation of the transformer at the primary side....to limit that current is the tenor of this thread.....
you have to realize that the discussion here is how to turn on a transformer used in a psu....naturally there is a rectifier and capacitor banks in the secondary and that affects the situation of the transformer at the primary side....to limit that current is the tenor of this thread.....
Hey Tony, congrats on your promotion by the way..also hope you and yours are not affected by the Manilla floods
And I think I was not clear enough - the secondary load does not contribute to the inrush current in the primary. Attached is a 100VA toroidal power transformer primary current at switch on with an open circuit ct 12V secondary.
The OP was correct in calling the effect primary saturation and his solution to power on at Vmax is also correct because this will be the point of future current zero crossing.
And I think I was not clear enough - the secondary load does not contribute to the inrush current in the primary. Attached is a 100VA toroidal power transformer primary current at switch on with an open circuit ct 12V secondary.
The OP was correct in calling the effect primary saturation and his solution to power on at Vmax is also correct because this will be the point of future current zero crossing.
Hi Johno,
is this experiment confirming that this transformer switched in this manner reaches quiescent state in just 1 cycle of the mains waveform, i.e. stabilises to quiescent current in 20ms?
Does the plot confirm ~10ms?
Can you do further experiments on other transformers and using switching at different points on the waveform and if possible using cores that have switched OFF at different points in the waveform.
This set of results could end up showing near the worst case (duration for stabilising), for switching on a mains transformer when no load is connected.
is this experiment confirming that this transformer switched in this manner reaches quiescent state in just 1 cycle of the mains waveform, i.e. stabilises to quiescent current in 20ms?
Does the plot confirm ~10ms?
Can you do further experiments on other transformers and using switching at different points on the waveform and if possible using cores that have switched OFF at different points in the waveform.
This set of results could end up showing near the worst case (duration for stabilising), for switching on a mains transformer when no load is connected.
Andrew , without any proof my instinct is to say 2.5 mS is significant if 50 Hz . I would have suspected it to be less previously . Sorry in advance if I misunderstand the intention of your question .
A tip if using relays . Some have 5 mS reaction times . Using simple peak switching might find a lower surge current . Some advantage on 60 Hz also . I noticed a product said zero voltage switching . I doubt it as it used relays . One can short them with SCR devices which could do that ( SCR off when stabilized ) . This would doubly assist each device ( heat, sparks ) . Most relays state switching times . I tested some relays and the 5 mS is surprisingly accurate ( simple pulse from DC source ) . This is in switching situations and not about inrush . I did drill holes in relays and filled them with oil to reduce sparks as books advise . Very little effect , sparks still seen . I was told it would work as sparks are ionizing air . Switching times more helpful I find . The SCR snubber a better solution . SCR's are not considered as good as relays in most applications ( reliability tests ) . The hybrid SCR / relay should be superb .
Good question Andrew .
A tip if using relays . Some have 5 mS reaction times . Using simple peak switching might find a lower surge current . Some advantage on 60 Hz also . I noticed a product said zero voltage switching . I doubt it as it used relays . One can short them with SCR devices which could do that ( SCR off when stabilized ) . This would doubly assist each device ( heat, sparks ) . Most relays state switching times . I tested some relays and the 5 mS is surprisingly accurate ( simple pulse from DC source ) . This is in switching situations and not about inrush . I did drill holes in relays and filled them with oil to reduce sparks as books advise . Very little effect , sparks still seen . I was told it would work as sparks are ionizing air . Switching times more helpful I find . The SCR snubber a better solution . SCR's are not considered as good as relays in most applications ( reliability tests ) . The hybrid SCR / relay should be superb .
Good question Andrew .
Last edited:
Andrew,
1/ the scope shot is from an analysis I did quite a few years ago. The investigation was into the reasons behind thyristor failure when switching inductive loads (eg power transformers in domestic appliances) and also why core balance or earth leakage miniature circuit breakers that we were designing at the time for domestic use, were tripping.
2/ If you look carefully at the scope trace you will see perturbation on the positive half cycle of the first 4 or so cycles - this is the magnetising flux reaching equilibrium.
3/ However, the magnetising current has done most of its job when the current starts to fall. With a series resistor (or higher impedance source) I recall that the spike was shorter but over a longer duration and always in the one direction, as expected.
4/ If the secondary was loaded one would see an over-damped sinusoidal decay over several cycles superimposed on that basic waveform.
Unfortunately I no longer have the means to run those type of measurements.
1/ the scope shot is from an analysis I did quite a few years ago. The investigation was into the reasons behind thyristor failure when switching inductive loads (eg power transformers in domestic appliances) and also why core balance or earth leakage miniature circuit breakers that we were designing at the time for domestic use, were tripping.
2/ If you look carefully at the scope trace you will see perturbation on the positive half cycle of the first 4 or so cycles - this is the magnetising flux reaching equilibrium.
3/ However, the magnetising current has done most of its job when the current starts to fall. With a series resistor (or higher impedance source) I recall that the spike was shorter but over a longer duration and always in the one direction, as expected.
4/ If the secondary was loaded one would see an over-damped sinusoidal decay over several cycles superimposed on that basic waveform.
Unfortunately I no longer have the means to run those type of measurements.
Measurements are always a good thing, and here they simply confirm something we already know: transformers are designed for steady state conditions, and at switch on, they can create overcurrents.
The magnitude of the overcurrent depends on a number of well-known factors: mains voltage at the time of the experiment, design margins regarding the volt.second product/saturation limit of the core, construction of the transformer and the instant of switching wrt. to the voltage waveform. That's about all.
In a worst case condition (tightly designed toroid at the max input voltage, switched on at the zero crossing), only the winding resistance will limit the current.
Thanks Andrew and Johno , an eye opener . Keep this going as it is far from obvious what is being said here . <10 mS was my wildest guess .
the very short stabilising duration when the transformer is on no load, confirms my own anecdotal evidence.
I always (well nearly always) use a bulb tester when powering up another transformer, and going back to a project that I had running the previous day.
A lone transformer rarely blinks the bulb, whereas a cap bank always makes the blink visible.
That tallies with what Johno is describing.
The bulb does not have time to blink in that <10ms current pulse. This is the common transformer only test.
But the extended damped oscillating current does give the bulb filament time to heat up and then decay down to near cold over that few tens/hundreds of milli-seconds. This is a small cap bank, or large cap bank, or enormous cap bank effect. With familiarity all these different set-ups show different bulb responses and they are recognised in the second of switch on and I am ready to switch off immediately I see something that is not expected.
The PTC in the bulb obviously behaves quite differently from an NTC, or a plain resistor, but the visible effect is useful for de-bugging.
I always (well nearly always) use a bulb tester when powering up another transformer, and going back to a project that I had running the previous day.
A lone transformer rarely blinks the bulb, whereas a cap bank always makes the blink visible.
That tallies with what Johno is describing.
The bulb does not have time to blink in that <10ms current pulse. This is the common transformer only test.
But the extended damped oscillating current does give the bulb filament time to heat up and then decay down to near cold over that few tens/hundreds of milli-seconds. This is a small cap bank, or large cap bank, or enormous cap bank effect. With familiarity all these different set-ups show different bulb responses and they are recognised in the second of switch on and I am ready to switch off immediately I see something that is not expected.
The PTC in the bulb obviously behaves quite differently from an NTC, or a plain resistor, but the visible effect is useful for de-bugging.
The bulb is a PTC thermistor . Just measured the one I use . 100 W 230 V = 40 R cold . It would be about 530 R at the intended load . If no other means available ( variac ) I use a light bulb to start up ancient equipment . A French electrician I worked with always used a light bulb . He said he trusted it far more than a meter . He added others would ridicule him for that . Small light bulbs make excellent oscillator stabilizers considering how simple they are . Thorens TD125 used one .
OK, I think I at last understand what happens when a transformer is switched on.
At the point of switch on the transformer core is fully involved, not effectively missing as some of you have claimed. The effective inductance will be somewhere between normal primary inductance and leakage inductance, depending on how the secondary is loaded. The current will rise. Unless switched on at a voltage peak, the current will have a DC component in addition to the AC. The two together are very likely to push the core into saturation, as most power transformer have very little overhead as this would be too expensive.
At this point the core effectively disappears, so then the current rises rapidly - limited by the DC resistance. So the first half-cycle current can be huge. The second half-cycle has the DC and AC current opposed so no core saturation. By the time the third half-cycle comes along the DC current may have decayed sufficiently that the transformer is no longer pushed into saturation.
So for the first few cycles, after the initial transient, the DC current decays according to the L/R time constant. The initial transient could perhaps be worse for an unloaded transformer, as there is no secondary current to oppose the primary current so flux is maximised.
What confused me was the insistence that the core disappears from the beginning. It does not, it disappears shortly after the beginning. You were right to say that resistance dominates, but your explanation was not quite right which is why I queried it. For a simple explanation see here.
Adding resistance both reduces the size of the initial transient current and speeds up the decay of the DC.
At the point of switch on the transformer core is fully involved, not effectively missing as some of you have claimed. The effective inductance will be somewhere between normal primary inductance and leakage inductance, depending on how the secondary is loaded. The current will rise. Unless switched on at a voltage peak, the current will have a DC component in addition to the AC. The two together are very likely to push the core into saturation, as most power transformer have very little overhead as this would be too expensive.
At this point the core effectively disappears, so then the current rises rapidly - limited by the DC resistance. So the first half-cycle current can be huge. The second half-cycle has the DC and AC current opposed so no core saturation. By the time the third half-cycle comes along the DC current may have decayed sufficiently that the transformer is no longer pushed into saturation.
So for the first few cycles, after the initial transient, the DC current decays according to the L/R time constant. The initial transient could perhaps be worse for an unloaded transformer, as there is no secondary current to oppose the primary current so flux is maximised.
What confused me was the insistence that the core disappears from the beginning. It does not, it disappears shortly after the beginning. You were right to say that resistance dominates, but your explanation was not quite right which is why I queried it. For a simple explanation see here.
Adding resistance both reduces the size of the initial transient current and speeds up the decay of the DC.
I don't think that the added primary circuit resistance speeds up the decay of the DC component.
I think it delays/slows down the move towards the quiescent state.
But Taking your revised explanation, I will go back to one of my very early analogies.
At start up the core has no flux. Therefore the core can have no effect on the inductance of the primary winding.
At this instant in time the winding behaves as if it was an air-cored inductor.
This air-cored inductor will have a tiny inductance in comparison to the final fully fluxed inductance of the transformer.
The instantaneous start up current is resisted by the total of the primary circuit resistances plus the impedance of that "analogy of an air-cored inductor". The instantaneous current can be assumed to be slightly less than the current that would be defined by the primary circuit resistances alone, since the inductive impedance is essentially near zero.
I recognise that this analogy is technically wrong, not just inaccurate, but it allows a first guess estimation of the maximum start up instantaneous current.
I recall a year or so back another Member did some current peak tests or simulations and found the instantaneous peak was >50% of the primary resistances estimate. I can't remember the name, nor the method, nor the numbers.
Last edited:
No, that is the fallacy I suspected some of you might be making. I am glad you have made it explicit. If you are right, then every cored inductor would have a major flux glitch at the current zero-crossing. This does not happen.AndrewT said:
At zero current there may be no magnetic flux, but there is still the same magnetic coupling from the windings to the core so the effective permeability of the core still magnifies the inductance. It may be the case that some irons have lower permeability at low flux, but that is a separate issue and the relative permeability certainly does not fall anywhere near 1.
No, it is core saturation which causes the initial current spike.
Regarding time constants, it is CR for cap-resistor and L/R for inductor-resistor first-order circuits. Bigger R slows down CR circuits and speeds up LR circuits. This is just elementary circuit theory.
It won't measure things that's for sure . Dam good cheap substitute .
Wien bridge oscillator - Wikipedia, the free encyclopedia
Self inductance exists independently of the flux. It is determined by the geometric and material properties of the core/windings.
This does not mean that the flux (or the induction to be more accurate) cannot influence the inductance, but that is a separate issue.
The core gives the primary winding its magnetizing inductance, even at 0 flux or induction, and this inductance is ~= µr times the primary inductance when the core is completely saturated.
The explanations of DF96 are entirely correct.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.