Joachim, your message box is full, please empty it, so I can send messages, or send me an email with your email address (preferred).
(Sorry for being of topic, John)
Regards,
Sigurd
(Sorry for being of topic, John)
Regards,
Sigurd
The material for mains transformer cores is magnetically soft, but that only means that the coercive field strength needed is low to keep the losses in each cycle low.
But nevertheless the core stays at a certain remanence level after switching off the mains voltage. And it depends on the point on the cycle where this level would be.
The mains voltage waveform droves the core through the complete hystheresis curve and the absolute worst case is if the mains voltage during the switch on tries to further magnetise the core in the same direction where it did remain during the switch off.
The core will quickly saturate in this case.
Combine that with a transformer with very low losses (i.e. a toroid transformer) and you´ll get the highest possible inrush current if switched on in the zero crossing moment.
@ janneman,
there is a nice pdf from the german "kupferinstitut" which described the situation in the case for mains voltage dc levels; unfortunately written in german, but even so the graphs are very informative:
http://www.kupferinstitut.de/front_frame/pdf/s193.pdf
The interesting part starts at page 11/12 in the pdf.
If the transformer core saturates partly during the mains voltage cycle the transformer normally not only starts to hum but the stray fields become very disturbing too.
Wishes
But nevertheless the core stays at a certain remanence level after switching off the mains voltage. And it depends on the point on the cycle where this level would be.
The mains voltage waveform droves the core through the complete hystheresis curve and the absolute worst case is if the mains voltage during the switch on tries to further magnetise the core in the same direction where it did remain during the switch off.
The core will quickly saturate in this case.
Combine that with a transformer with very low losses (i.e. a toroid transformer) and you´ll get the highest possible inrush current if switched on in the zero crossing moment.
@ janneman,
there is a nice pdf from the german "kupferinstitut" which described the situation in the case for mains voltage dc levels; unfortunately written in german, but even so the graphs are very informative:
http://www.kupferinstitut.de/front_frame/pdf/s193.pdf
The interesting part starts at page 11/12 in the pdf.
If the transformer core saturates partly during the mains voltage cycle the transformer normally not only starts to hum but the stray fields become very disturbing too.
Wishes
Last edited:
German is my 2nd language (after English), thanks for the link.
Another question, if I may:
Power transformers for amplifiers only deliver power to a load at or near the top of the mains voltage; that's when the rectifier diodes start to conduct, and they switch off again after a small part of the cycle (40-60 degrees max).
So, doesn't that mean that at the mains zero crossing, by definition the load on the transformer is infinite and there's no secondary (and thus no primary) current?
jd
I remembered that from the earlier exchange.
Do you mean that question in relation to the "dc problem" or the "inrush current case" or more for transformer in use with conventional rectifying and filter circuits in general?
Wishes
The context is remote switch-on of power amps. I also used to do that at mains zero-crossing, because I know that the transformer secondary current is zero at that point. Not sure what the primary current is at that point.
jd
I see, that relates directly to the worst case i did mention before.
If the core was left for example halfways magnetised (the remanence) in the same direction in which it will be further magnetised during the first part of the mains cycle that follows the switch on in the zero crossing, then the core will saturate and left is approximately an air coil with the dc resistance of the primary winding.
Not a problem with small transfomers but if it gets to big toroidals with just a dc resistance of the primary of ~2-3 Ohms.....
Even switching on at the peak of the mains voltage does not fully solve the problem but betters the peak of the inrush current.
If you have not to take into account short drop outs of the mains, then a ntc in combination with a relay still works (despite the problem that it needs some time to cool down, so you have to wait between sequential switching).
Wishes
OK, but how do I know what the remanence and direction is at switch-off? If I also swicht off always at zero crossing, when the load current is zero, does that not solve the problem?
jd
No,
The BH loop does not drop to zero when you remove the drive. It actually requires the drive to drop a bit in the other direction before it erases the remnant field. However it is lower than if you switched off at maximum drive.
OK I understand that, there's some hysteresis on the BH curve. So switching off at say 1 or 1.5 mS after zero crossing, that would get me close to the point of zero remanence.
Then next switch on at zero crossing. Hmm. Easy, just to change a few constants in my Flowcode program
.Need to do some measurements though....
jd
This is how I do it now.
The third curve down is the zero-crossing: To is the start, and at around 0.5mS it ends. (Don't ask, it's analog).
So the actual zero crossing is at around 0.25mS.
The top is the SPI command to the switching unit, while the 2nd down is the strobe that latches the command and executes it.
So the actual switch-on is about 50uS late.
Sorry to bore you guys with this digital stuff, but its really fun, a nice break now and then from audio (duck and runs).
jd
The third curve down is the zero-crossing: To is the start, and at around 0.5mS it ends. (Don't ask, it's analog
).So the actual zero crossing is at around 0.25mS.
The top is the SPI command to the switching unit, while the 2nd down is the strobe that latches the command and executes it.
So the actual switch-on is about 50uS late.
Sorry to bore you guys with this digital stuff, but its really fun, a nice break now and then from audio (duck and runs).
jd
I am a latecomer here and have not read back more than a few pages so if I am repeating something, my apologies.
http://relays.tycoelectronics.com/appnotes/app_pdfs/13c3206.pdf
http://relays.tycoelectronics.com/appnotes/app_pdfs/13c3206.pdf
I would like to point out something that is akin to the topic just discussed. This is essentially the problem of large peak currents that charge the filter caps, and how to minimize or control them.
First, it should be pointed out why we have extra large charge currents in many hi end amps and even preamps.
In the early days, especially with vacuum tube designs, another kind of power supply was standard. This was the so called pi network or cap-inductor-cap, or sometimes cap-resistor-cap power supply. The reason for this was two-fold. First, the vacuum tube rectifiers could not tolerate very high peak charge currents, and the first cap might be limited to 20-40uf, in order to keep the rectifier tube from failing.
Secondly, high capacitance was VERY difficult to obtain in the early days and it was actually cheaper to get your extra filtering with a series inductor. The impedances of each component in the pi network MULTIPLY, rather than ADD, as would be typical when paralleling caps.
As years passed, inductors got more expensive, diodes changed to solid state, and caps became smaller and cheaper. Therefore the transition to solid state with the need for lower voltage and higher current, tended to remove the series inductor.
At first, it was relatively successful. For example, a pair of 5,000 uF caps might do in an early solid state design. This caused some extra peak current pulses at the wall outlet, but not too much.
However, people found that the caps actually did DOUBLE DUTY in the power supply. They both filtered the power supply ripple, AND they provided the low frequency ground return for the audio signal. If the caps were only 5,000 uF or so, the bass appeared to suffer. I find this true today, with my cheaper power amps. Serious listeners generally complain about the deep bass, first, because the caps are only 10,000 uF or so.
What to do? Well, adding more capacitance fixes the ground return problem, and gives more peak current capability, but look at what it does at the AC outlet! (To be continued.)
First, it should be pointed out why we have extra large charge currents in many hi end amps and even preamps.
In the early days, especially with vacuum tube designs, another kind of power supply was standard. This was the so called pi network or cap-inductor-cap, or sometimes cap-resistor-cap power supply. The reason for this was two-fold. First, the vacuum tube rectifiers could not tolerate very high peak charge currents, and the first cap might be limited to 20-40uf, in order to keep the rectifier tube from failing.
Secondly, high capacitance was VERY difficult to obtain in the early days and it was actually cheaper to get your extra filtering with a series inductor. The impedances of each component in the pi network MULTIPLY, rather than ADD, as would be typical when paralleling caps.
As years passed, inductors got more expensive, diodes changed to solid state, and caps became smaller and cheaper. Therefore the transition to solid state with the need for lower voltage and higher current, tended to remove the series inductor.
At first, it was relatively successful. For example, a pair of 5,000 uF caps might do in an early solid state design. This caused some extra peak current pulses at the wall outlet, but not too much.
However, people found that the caps actually did DOUBLE DUTY in the power supply. They both filtered the power supply ripple, AND they provided the low frequency ground return for the audio signal. If the caps were only 5,000 uF or so, the bass appeared to suffer. I find this true today, with my cheaper power amps. Serious listeners generally complain about the deep bass, first, because the caps are only 10,000 uF or so.
What to do? Well, adding more capacitance fixes the ground return problem, and gives more peak current capability, but look at what it does at the AC outlet! (To be continued.)
Last edited:
John,
Real tube amp designers never use pi-filters (C-L-C). They use the classic old stye choke input filter (L-C) which completely handles the peak current issue, and add addition filtering/regulatoin after that.
This also works VERY well in class A solid state poweramps - althogh adding a bit of mass and $$, it's well worth it sonically in my experience.
Regards, Allen (Vacuum State)
Real tube amp designers never use pi-filters (C-L-C). They use the classic old stye choke input filter (L-C) which completely handles the peak current issue, and add addition filtering/regulatoin after that.
This also works VERY well in class A solid state poweramps - althogh adding a bit of mass and $$, it's well worth it sonically in my experience.
Regards, Allen (Vacuum State)
Sorry I just have to interject...
I recently rebuilt a matched pair of Langevin 101C amplifiers built in August 1946. Following the rectifier they used an oil and paper capacitor (10/630) to a choke to an electrolytic. The oil and paper cap was still good which was very nice as replacing it would be very tough. All the electrolytic capacitors got replaced.
In this series of amplifier the rectifiers were paralleled 5Z3s. Later versions used the 5U4. Also noteworthy was the amplifier used only one capacitor in the signal path for the push side (as I call it) and two in parallel for the pull side. Phase split was done after the push side!
When rebuilt at first it did not sound right until we realized my tech was playing his MP3s!
So how about "all modern real tube designers..."
- Status
- Not open for further replies.