The quiescent current of a CFP output stage can be <5mA and for the whole power amp <10mA.
A few tens of ohms results in a few hundred mV of drop that when bypassed will result in a tiny current of ~100mV/1r cf 50V/5r.
Even when the quiescent current is 200mA that still results in a small current charging pulse of ~ 2000mV/1ohm
Slow charging is a completely different animal from soft starting, the requirements of the two control/limiting systems are completely different.
Last edited:
Did anyone try to use a lighting bulb in series with the primary (iso power resisters)
For instance a 50W halogen lamp or similar. Of course a relay should switch after few seconds to full 220v. Or switch manually after few seconds by using "two stage" power switch.
For instance a 50W halogen lamp or similar. Of course a relay should switch after few seconds to full 220v. Or switch manually after few seconds by using "two stage" power switch.
A light bulb is essentially a PTC, which you do not want for your application. You either want a NTC or a fixed resistance.
As I see it, those huge caps make the biggest probs at suddenly rising main voltages
when the crrent limiting device is either bypassed or saturated or whatever.
Non of the solutions I have seen so far deals with this problem and the only way out seem to be to limit the charging rate of the caps to values all circuit components can handle under all conditions. This migth make necessary to settle for much lower dc-regulation as you have hoped fore.
If you are only concerned with surge current at switch on then the limiting effect of the saturated inductor is worth trying, but unless of a sufficient coresize it wont even work for the first half cycle. It is simple, VERY reliable and clever but it migth be bigger and heavier than you thougth of.
when the crrent limiting device is either bypassed or saturated or whatever.
Non of the solutions I have seen so far deals with this problem and the only way out seem to be to limit the charging rate of the caps to values all circuit components can handle under all conditions. This migth make necessary to settle for much lower dc-regulation as you have hoped fore.
If you are only concerned with surge current at switch on then the limiting effect of the saturated inductor is worth trying, but unless of a sufficient coresize it wont even work for the first half cycle. It is simple, VERY reliable and clever but it migth be bigger and heavier than you thougth of.
I think it is unlikely that the mains voltage will change very fast. At least not nearly as fast as it does when you flip the power switch on and it jumps from zero to 120 or 240 volt in an instant. A simple NTC inrush thermister handles this case well enough. The device's resistance depends on the current going through it. So if for whatever reason there is suddenly more current going to the caps then the resistance will change and limit the current. It doesn't "saturate" except that it has a fixed maximum resistance which you have already selected to handle the worse case scenario which is the power switch.
If you are VERY concerned about the maximum current you can always install a fuse.
Mains voltage raise is going to be in the 5% range at worse. At least here in the US the electric code limits the voltage drop in branch circuits to 5% so if you have a 1,500 watt space heater that is cycling plugged into the same outlet as your amplifier you can expect the mains to drop no more then 5% as the heater cycles. Actually 2% is pretty common in larger buildings like apartment houses.
I don't think that even double the worst case is a problem with your filter caps.
ChrisA, what you thinck is unlikely is a opinion you are entitled to.
Grid power failures level maybe different in different parts of the world but they can hardly be let unnoticed as an important design criteria.
Where to draw the line? Why overvolt protection for any equippement if it is so unlikely?
Offcourse 1kV spikes will occur only once a year and will be of such short duration that they usually dont do much harm but they happen and it is a good idea to protect sensitiv equippement against them.
Just as it is a good idea to limit switch on surge currents.
At switch on of a big toroid with rectifier and big caps you need to limit the inrush current. But NTC or resistor or whatever you choose will have to be bypassed after a certain time. This bypass will usually be done with a relay that has to be fed with dc to be noiseless. Now, when the grid brakes down for whatever reason there is usually a minor timestep before the grid comes back on again. What happens during this time?
Your PSU is still connected DIRECTLY to the grid because your dc-fed relay cannot fall off imidiately when the powerinterruption occurs. There will be delay. It is possible your
big caps are discharged 30-40-50% at the time the grid comes back on again and your relay is still shortcutting the current limiting device. The surge current into only partly charged caps + magnetising current of the toroid under those unlimited conditions will most certainly be higher than the surge current under normal switch on conditions.
5% mainsvoltage changes are of no concern offcourse. But 5% is not what I ment.
You can expect much more than that during and after a powergrid failure.
And one more thing, at switch on I usually have my volume down but at an unexpected powergrid failure things can be different and in a costly way fuses cant possible protect you from....
Grid power failures level maybe different in different parts of the world but they can hardly be let unnoticed as an important design criteria.
Where to draw the line? Why overvolt protection for any equippement if it is so unlikely?
Offcourse 1kV spikes will occur only once a year and will be of such short duration that they usually dont do much harm but they happen and it is a good idea to protect sensitiv equippement against them.
Just as it is a good idea to limit switch on surge currents.
At switch on of a big toroid with rectifier and big caps you need to limit the inrush current. But NTC or resistor or whatever you choose will have to be bypassed after a certain time. This bypass will usually be done with a relay that has to be fed with dc to be noiseless. Now, when the grid brakes down for whatever reason there is usually a minor timestep before the grid comes back on again. What happens during this time?
Your PSU is still connected DIRECTLY to the grid because your dc-fed relay cannot fall off imidiately when the powerinterruption occurs. There will be delay. It is possible your
big caps are discharged 30-40-50% at the time the grid comes back on again and your relay is still shortcutting the current limiting device. The surge current into only partly charged caps + magnetising current of the toroid under those unlimited conditions will most certainly be higher than the surge current under normal switch on conditions.
5% mainsvoltage changes are of no concern offcourse. But 5% is not what I ment.
You can expect much more than that during and after a powergrid failure.
And one more thing, at switch on I usually have my volume down but at an unexpected powergrid failure things can be different and in a costly way fuses cant possible protect you from....
I did not say it was unlikely.
First I think spike are easy to protect from. We use LC filters and MOVs, No need to talk more about spikes. I think what we are talking about here is longer term changes in AC mains that take place for many seconds or longer.
So, as I said above what's the worst case you want to protect from? How about we select the case where the power company doubles the voltage then leaves it there? That has to be a very unlikely case. My opinion has and is that if your in-rush thermister can handle an instant change for 0 to X volts then it can also handle an instant change from X to 2X volts. It's the exact same thing, at least in terms on in-rush current.
A good fail safe"crow bar" shut down system could also be added. Place a zenier diode across the rectifier output and choose it's voltage such that in the normal case it will not conduct. In other words if the DC is 50V use a 60V zenier Next place a fuse on the transformer output. Now if the mains raises the zenier "switches" on and shorts the transformer and then the fuse blows. This kind of setup is common and they call it for some reason a "crowbar". The best part is that it is so simple it can't fail.
If the filter cap''s voltage rating is higher than the zenier voyage then no mains over voltage can damage them. a 60V diode will keep the DC volts at or below 60V. Soon the fuse blows and DC goes to zero via the bleeder resister.
There is a problem with in-rush current blowing the fuse, so the fuse needs to be quite a bit larger than you'd like. This is why there is a thermister, to limit in-rush to well below the fuse rating. Do NOT short the therister with a relay. Place the thermister on the AC Mains side of the transformer. On a solid state amps (I assume) the mains side has lower current .
Safety device need to be simple and also designed so their failure mode is safe. You want them to be so simple they will work even if the amp is on fire. The "'crow bar" fuse/diode system is that simple
Last edited:
Maybe not everyone is blessed with a situation I am in and has not the same probs.
But I aimed at the best possible dc regulation and found out the hard way there is a price to pay. My "grid" is fed with airisolated wires to a 20kV 200kVA transformer at the corner of my house. About 15m of 3x25sqaremillimeter cabling to my fusebox. 100A fuses. Inhouse cabling 3x25A 400V only 10m to the second fusebox. Grid is hard as a rock. All in all only a few meters of cabling below 6 squaremillimeter to my wallsocket 16A 230V. I have designed the toroid with a much higher than normal coppercontent
pimary 1.8mm diam, secondary 2x3.15mm evenly spaced on a 2kVA rated core.
I choosed diodes with the lowest highcurrent voltagedrop I could find.
There is almost no wiring in the circuit. Strays are minimized by transformerdesign and using 0.5mm spaced coppersheet.
Under normal operation conditions the diodes pass nearly 100Apk at 10Adc load. The caps are a bank of 10000uF 100V each and a total of 8 in each leg. Switch-on limiting is with resistors shortcutted by 2 step operated dc-relays. Resitors are the small glasscoated thick ceramic type to assure the best heatcapacity in a small space.
I will have to built a device that cuts out instantly as soon as the grid fluctates after a gridfailure. Otherwise I have to sacrifice the excellent dc-regulation of my powersupply.
This powersupply is build to get the best possible dc-regulation so this probable is not a very common design goal but why settle for less if you have a stiff grid and use big caps?
But I aimed at the best possible dc regulation and found out the hard way there is a price to pay. My "grid" is fed with airisolated wires to a 20kV 200kVA transformer at the corner of my house. About 15m of 3x25sqaremillimeter cabling to my fusebox. 100A fuses. Inhouse cabling 3x25A 400V only 10m to the second fusebox. Grid is hard as a rock. All in all only a few meters of cabling below 6 squaremillimeter to my wallsocket 16A 230V. I have designed the toroid with a much higher than normal coppercontent
pimary 1.8mm diam, secondary 2x3.15mm evenly spaced on a 2kVA rated core.
I choosed diodes with the lowest highcurrent voltagedrop I could find.
There is almost no wiring in the circuit. Strays are minimized by transformerdesign and using 0.5mm spaced coppersheet.
Under normal operation conditions the diodes pass nearly 100Apk at 10Adc load. The caps are a bank of 10000uF 100V each and a total of 8 in each leg. Switch-on limiting is with resistors shortcutted by 2 step operated dc-relays. Resitors are the small glasscoated thick ceramic type to assure the best heatcapacity in a small space.
I will have to built a device that cuts out instantly as soon as the grid fluctates after a gridfailure. Otherwise I have to sacrifice the excellent dc-regulation of my powersupply.
This powersupply is build to get the best possible dc-regulation so this probable is not a very common design goal but why settle for less if you have a stiff grid and use big caps?
Chris, you dont understand.
Look at the following situation:
Gridfailure.
0 volt input
The powercompanies automatic switchs to a different grid.
Full voltage with loaddependend fluctations.
Time elapsed a few tens to a few hundred millisecond at most.
During that time your currentlimiting NTC will either be still hot (if not shorted) and not limit much...
Or, if it is shorted your dc-fed relay will still be in the same position than it was.
It wont fall off in time. It shorts your NTC or resistor.
So, when the grid comes automacilly back on it will feed into your trafo directly, without
NTC- or resistor- limiting the current.
And if your amplifier drained partly your caps it will feed into a circuit whose inrushcurrent will consist of current to set up the field + current to top up the caps +
loadcurrent.
Under the conditions mentioned the current will ONLY be limited by grid + inhouse wiring + transformer impedance + rectifier + caps ESR.
No inrush current limiting means huge current when the grid comes shortly back on again.
Offcourse if you have a transformer with poor regulation and schottky diodes that act like a resistor it wont mean a thing no matter what caps are involved and I understand at 115V the problem will only be half of it also....
Look at the following situation:
Gridfailure.
0 volt input
The powercompanies automatic switchs to a different grid.
Full voltage with loaddependend fluctations.
Time elapsed a few tens to a few hundred millisecond at most.
During that time your currentlimiting NTC will either be still hot (if not shorted) and not limit much...
Or, if it is shorted your dc-fed relay will still be in the same position than it was.
It wont fall off in time. It shorts your NTC or resistor.
So, when the grid comes automacilly back on it will feed into your trafo directly, without
NTC- or resistor- limiting the current.
And if your amplifier drained partly your caps it will feed into a circuit whose inrushcurrent will consist of current to set up the field + current to top up the caps +
loadcurrent.
Under the conditions mentioned the current will ONLY be limited by grid + inhouse wiring + transformer impedance + rectifier + caps ESR.
No inrush current limiting means huge current when the grid comes shortly back on again.
Offcourse if you have a transformer with poor regulation and schottky diodes that act like a resistor it wont mean a thing no matter what caps are involved and I understand at 115V the problem will only be half of it also....
It depends on the design surge current, Transformer impedance is around 5%. I design for a peak surge of 4 or 5 times rated current so shorting the resistor out at 80% fulfils that design requirement (the 5% transformer impedance limits the current after the resistor is shorted)
I use 4 times rated current as a starting surge because it is a minor surge compared to a motor start and it ensures that a power supply will start under full load properly. This value has been tested in both single and 3 phase designs and no trips of a standard C curve circuit breaker have been observed.
With +/- 35V rails charging 0.4F takes 245 Joules A 20 ohm 25W resistor would do with a 230V supply with the capacitors 80% charged in around 140 ms. Peak current draw from the supply is only 9A.
If the load application can be delayed until after the start sequence then it is possible to use higher value starting resistor, with 2 times rated load the relay should close at 90% with rated current it should close at 95%
@gorgon53
What is you concern actually? Does the sound quality suffer audibly when your mains fluctuate? And does it happen so often that you need to do something about it? Or is it just an obsession with a theoretical concept?
That is true, and often overlooked in DIY when it comes to PSUs. So what needs to be done?
The capacitor's share of the inrush current on the primary side is negligible compared to the transformer's own current. No need for action there. With that monster transformer of yours it is probably of no concern on the secondary side either. For normal transformers the secondary fuses should take care of it.
What remains are the rectifiers and secondary fuses which makes the solution simple. Instead of limiting the current do the opposite. Choose components that can handle the charging current. Read the rectifier's datasheet and find the fuse rating, usually given in A²s. You will find that figure in fuse datasheets as well. Make sure that the fuse has a lower number than the rectifier. If that fuse blows from the inrush current (without inrush limiter!), then you need a bigger rectifier and to adjust the fuse accordingly. Voila!
Has it occured to you to limit the capacitor current by using a smaller transformer with higher inner resistance? Or by using less capacitance?
Just joking, it is obvious from your posts that you would not consider either approach.
What is you concern actually? Does the sound quality suffer audibly when your mains fluctuate? And does it happen so often that you need to do something about it? Or is it just an obsession with a theoretical concept?
That is true, and often overlooked in DIY when it comes to PSUs. So what needs to be done?
The capacitor's share of the inrush current on the primary side is negligible compared to the transformer's own current. No need for action there. With that monster transformer of yours it is probably of no concern on the secondary side either. For normal transformers the secondary fuses should take care of it.
What remains are the rectifiers and secondary fuses which makes the solution simple. Instead of limiting the current do the opposite. Choose components that can handle the charging current. Read the rectifier's datasheet and find the fuse rating, usually given in A²s. You will find that figure in fuse datasheets as well. Make sure that the fuse has a lower number than the rectifier. If that fuse blows from the inrush current (without inrush limiter!), then you need a bigger rectifier and to adjust the fuse accordingly. Voila!
Has it occured to you to limit the capacitor current by using a smaller transformer with higher inner resistance? Or by using less capacitance?
Just joking, it is obvious from your posts that you would not consider either approach.
If the primary of the transformer is the limiting factor and unaffected by capacitor charging load, how can the secondary be an issue? Can the secondary current increase without having the primary increase?
The primary is not the limiting factor. Transformer specs quote the secondary current or secondary power rating. Very few manufacturers tell you the actual primary current.
In industrial applications, where inrush current limiting is rarely used for transformers below several kVA, the actual overload protection is done on the secondary side.
Of course not. If you do the math you will find that the transformer's own inrush current is so much bigger than the capacitor's inrush current after being transformed from the secondaries to the primary, it makes no significant difference for most transformers.
Now look at the transformer description in post #48 and guess how much less difference it will make to that whopper.
pacificblue, I thinck I mentioned that I choosed diodes so I would get the best possible
dc regulation. Offcourse if I choose something I do it by checking the datasheets for option avaiable. Sadly datasheets arent perfect so the data for the high formfactor I use just is not there.
Anyway, I selected the BYW93/200 diodes in order to get the best dc-regulation.
Higher rated diodes are either much slower or have higher forwardvoltage drop or both.
Please observe that I did not want the diodes to be the weakest link in the chain.
The typical forwardvoltage drop of a BYW93 is below 0,8V at my peakcurrent of nearly 100A. They can handle a nonrepetive halfsinewave for 10mS at 800A for and handle 3200A2s.
During a 24h fullpower dummyloadtest it happend that the powergrid broke down for a splitsecond (I noticed it from a flickering ligthbulb and some nasty noise). At the same time the fuse 16A mainfuse burned. For obvious reasons I dont need any other fuses.
That event showed me clearly that inrushcurrent at switch on is not the only thing to be concerned about.
And yes you are rigth, I want it to be perfect ;o)
dc regulation. Offcourse if I choose something I do it by checking the datasheets for option avaiable. Sadly datasheets arent perfect so the data for the high formfactor I use just is not there.
Anyway, I selected the BYW93/200 diodes in order to get the best dc-regulation.
Higher rated diodes are either much slower or have higher forwardvoltage drop or both.
Please observe that I did not want the diodes to be the weakest link in the chain.
The typical forwardvoltage drop of a BYW93 is below 0,8V at my peakcurrent of nearly 100A. They can handle a nonrepetive halfsinewave for 10mS at 800A for and handle 3200A2s.
During a 24h fullpower dummyloadtest it happend that the powergrid broke down for a splitsecond (I noticed it from a flickering ligthbulb and some nasty noise). At the same time the fuse 16A mainfuse burned. For obvious reasons I dont need any other fuses.
That event showed me clearly that inrushcurrent at switch on is not the only thing to be concerned about.
And yes you are rigth, I want it to be perfect ;o)
ChrisA, I know you mean it well, so thank you for your advices alltough , mind me saying so, they are no good.
1.
A crowbar must contain a high power switching device, a SCR or relay or whatever, a simple zener wont do even if it where a highpower device.
Anyway, no crowbar needed.
2.
Except for the special case of a class A amplifier it is not a good idea to leave the NTC in the circuit without shorting it out after a suitable time. Metalsculptor described a proper approach for resistors, goes for NTCs too.
1.
A crowbar must contain a high power switching device, a SCR or relay or whatever, a simple zener wont do even if it where a highpower device.
Anyway, no crowbar needed.
2.
Except for the special case of a class A amplifier it is not a good idea to leave the NTC in the circuit without shorting it out after a suitable time. Metalsculptor described a proper approach for resistors, goes for NTCs too.
about point 2.
not even in class A its a good idea. to leave the thermistor in circuit.
then the power down need to last 60-90sec. before its back down to ambient temp. if you bypass it, it will be ready to power back up right away.
and about grid spikes and all that. don't loose sleep over it. in that short of a time, the cap bank won't discharge mutch. and if it does. and the relè is still in a short, the main fuse will pop long before any damages can be done.
it's a reason they are type fastblow. however. if the cap bank has discharged so mutch, then the DC for the relè has done the same. and the relè is not shorted when the power is back on.
AudioSan, offcourse it is not a good idea even in class A, but you can get away with it because as the NTC is hot so is the transformer. Anyway, with class A your transformer hardly really needs any inrushcurrent limiting at all.
WHAT????? thats way out------->....there.
class A hasthe biggest x.formers of all. and THEY don't need inrushcurret limiting? you need a softstart of any kind for all transformers over 500VA. and ad a monster cap bank you realy need it. i had to modify a hypex softstart for my aleph2 classA so it did not blow main fuse and resistors.
1000VA 408.000uF capbank.
Last edited:
- Status
- Not open for further replies.