Hi all, I have some inrush current issues with my amplifier - mainly because of large filter capacitors up to 22mF*2, and that creates a current surge from my mains (no breaker popped yet, but lights will dim). I know there are solutions such as NTC resistor, time relay delayed circuit, or a MOSFET soft start (as mentioned in https://sound-au.com/articles/soft-start.htm). But I suddenly wanted to make a universal external current limiter that you can plug between the mains and your amplifier and it solves all the inrush problems.
Having been working on current limiter for regulators, I came up with this schematic:
It's basically a MOSFET current regulator circuit, but doubled for AC. During the start, a small current through D2 and R3 (and the MOSFET body diode) will charge up C1 and C2, and Q1 and Q4 starts to conduct until they are fully on (very small Ron). The load will only see R2 in series. If the current through R2 rises above a threshold, the voltage across R2 will turn on either Q2 or Q3, and it discharges C2 or C3, thus turn off Q1 or Q4 slightly until the current is no longer above the threshold. The threshold should be Vbe/R2, so if R2=0.33ohm the current will be limited to around 0.7V/0.33ohm=2.1A. No larger inrush current anymore.
I haven't built the circuit yet because it requires working on lethal mains voltages, and I think it's better to have others to check the idea before handing on. If you find any faults or improper parts in the schematic, or you have better solutions to this problem, I need your advice! Thanks a lot.
Having been working on current limiter for regulators, I came up with this schematic:
It's basically a MOSFET current regulator circuit, but doubled for AC. During the start, a small current through D2 and R3 (and the MOSFET body diode) will charge up C1 and C2, and Q1 and Q4 starts to conduct until they are fully on (very small Ron). The load will only see R2 in series. If the current through R2 rises above a threshold, the voltage across R2 will turn on either Q2 or Q3, and it discharges C2 or C3, thus turn off Q1 or Q4 slightly until the current is no longer above the threshold. The threshold should be Vbe/R2, so if R2=0.33ohm the current will be limited to around 0.7V/0.33ohm=2.1A. No larger inrush current anymore.
I haven't built the circuit yet because it requires working on lethal mains voltages, and I think it's better to have others to check the idea before handing on. If you find any faults or improper parts in the schematic, or you have better solutions to this problem, I need your advice! Thanks a lot.
This is a common thing for induction motors, particularly those with the new modifications.
In my field of injection molding, it is common to retrofit new PLCs in place of old PLCs or contactor / relay based controls on old machines which are otherwise in good condition.
The existing motors are replaced with BLDC servo motors, with new pumps, or the cheaper and more reasonable option is a variable frequency drive, and use the existing motor - pump setup.
The modifications to use the existing motor include a soft start and current limiting system, as 3 phase squirrel cage motors on star - delta starters can draw very large current on start up, similar to your issue.
Please read up on the internet about this topic, I am sure you will find commercial products suitable for your needs.
Single phase versions, and thermal cutoff relays (with adjustable current) are also available.
Japanese electrical equipment manufacturers like Mitsubishi, Toshiba, Fuji and Hitachi may have a ready solution for your problem. And there are sources abroad as well from outside Japan.
I am not connected with them in any way.
This is a quick and ready solution, instead of designing your own circuit.
And of course, there may be ready circuits available on electrical engineering forums, with complete details, including how well they worked. That will give you a tested design.
For an amplifier, you are looking at a fractional horse power drive, about $40 at most for one.
In my field of injection molding, it is common to retrofit new PLCs in place of old PLCs or contactor / relay based controls on old machines which are otherwise in good condition.
The existing motors are replaced with BLDC servo motors, with new pumps, or the cheaper and more reasonable option is a variable frequency drive, and use the existing motor - pump setup.
The modifications to use the existing motor include a soft start and current limiting system, as 3 phase squirrel cage motors on star - delta starters can draw very large current on start up, similar to your issue.
Please read up on the internet about this topic, I am sure you will find commercial products suitable for your needs.
Single phase versions, and thermal cutoff relays (with adjustable current) are also available.
Japanese electrical equipment manufacturers like Mitsubishi, Toshiba, Fuji and Hitachi may have a ready solution for your problem. And there are sources abroad as well from outside Japan.
I am not connected with them in any way.
This is a quick and ready solution, instead of designing your own circuit.
And of course, there may be ready circuits available on electrical engineering forums, with complete details, including how well they worked. That will give you a tested design.
For an amplifier, you are looking at a fractional horse power drive, about $40 at most for one.
Last edited:
Checked some of the motor soft starter models and while I'm not a motor expert, I think there's a difference between a motor and an amplifier.
While both create inrush current, a motor needs it to generate a starting torque and an amplifier use it to charge the capacitor bank. With a low (but not lower than quiescent) current, the motor will have not enough torque to start, but a capacitor will be charged anyways. I saw some models of motor starters using triacs or PWM starters. Not sure if they will behave well for a capacitor bank because an amplifier mains transformer might not be rated for the modulation frequency of a motor starter (Triacs makes really nasty noises! Not a problem for a motor though). Some starter also use auxiliary windings or torque detection that are unavaliable on an amplifier. Another problem is that it's hard to find a soft starter for single phase - most of them are for 3-phase motors.
Still I saw some models with current limiting functions and will try to find a proper one. By the time being, I still think that there should be some merit from a simple current limiter that potentially anyone can build over starters intended for motors. Who knows?
While both create inrush current, a motor needs it to generate a starting torque and an amplifier use it to charge the capacitor bank. With a low (but not lower than quiescent) current, the motor will have not enough torque to start, but a capacitor will be charged anyways. I saw some models of motor starters using triacs or PWM starters. Not sure if they will behave well for a capacitor bank because an amplifier mains transformer might not be rated for the modulation frequency of a motor starter (Triacs makes really nasty noises! Not a problem for a motor though). Some starter also use auxiliary windings or torque detection that are unavaliable on an amplifier. Another problem is that it's hard to find a soft starter for single phase - most of them are for 3-phase motors.
Still I saw some models with current limiting functions and will try to find a proper one. By the time being, I still think that there should be some merit from a simple current limiter that potentially anyone can build over starters intended for motors. Who knows?
You may want to look further at issues like FET SOA as gate voltage passes through linear operation of FET, and asymmetric mains current levels during start-up and for internal fault conditions. You can test operation using a safe level of transformer secondary-side voltage that is sufficient to confirm operation.
Hi. I see problem with such limiting circuit. Problem is power dissipation in mosfets because of linear operation. You mentioned 2,1 ampere , this at shortcircuited load means all input voltage will be surged by mosfets. So at positive halfwave one mosfet must be able to dissipate 325V * 2,1A = 682W . Most mosfets now are switched mode , even they have low RDSon , they can dissipate small wattage in linear range ,lets say 50W max. You need rearrange circuit in a way , that charging will be performed by series resistor first , then when voltage after resistor is close to input voltage ( no more voltage drop on resistor) , mosfets turned on and shorted that resistor.
I guess the problem would be as mentioned here: https://sound-au.com/articles/soft-start.htm#s3
Thermistors need to be sized to particularlly working near rhe maximum current. If I were only working on one single power supply I would use it, but it isn't a plug-and-go solution if I want to use a universal circuit for every power supply. I have to find a proper size for each of them or the residual resistance will not be low enough. Personally I used NTC in a tube filament supply and the result was ... worked but hard to say. A tube requires around 0.3A and hardly any NTC will have a rating near this. Using a larger rating one? Sure, just don't expect the resistance to drop too much. (In my case a maximum 3A rated 5 ohm NTC drops to 3 ohm, while at 3A it should drop to 0.36 ohm)
Triac controller on the other hand DOES NOT limit surge current for a capacitor bank, at least not to the extent of a current regulator. As mentioned in the same article:
Basically it works by switching on the circuit at a certain phase to create the least possible surge current, and turn off in a given duty cycle to minimize the time period of that current. In a motor or transformer energizing, the current doesn't grow to the maximum because of the inductance.
For a capacitor bank, although I haven't measure my system, according to some examples like this application note, the energizing could happen in tens of microseconds. A triac would have to turn off really fast before the current hits maximum, which means lots of high frequency noises. Maybe I can construct a triac-based similar circuit, but not as satisfying as a linear regulator one.
If you still want linear current limiting , be prepared for high power rated transistors. Your schematic currently working as current limiting source , which will convert excessive voltage difference to heat, and mosfets must be able to withstand/dissipate this . At start point, your load is almost equal to short circuit.
If talking about triac circuits , they operate from point , when starting pulse is received, and then keeping on, until voltage reaches zero ,or current decreases less some threshold at that mains cycle. But as we know , capacitor bank is charged through diode bridge , and when halfwave voltage is lower than voltage already on capacitors, diodes do not conduct. So actually triac will pass high current for some small pulse period of time . Not a best way to limit current.
If talking about triac circuits , they operate from point , when starting pulse is received, and then keeping on, until voltage reaches zero ,or current decreases less some threshold at that mains cycle. But as we know , capacitor bank is charged through diode bridge , and when halfwave voltage is lower than voltage already on capacitors, diodes do not conduct. So actually triac will pass high current for some small pulse period of time . Not a best way to limit current.
Not quite right, regarding NTC usage.
The NTC is chosen to suit the average current at startup. The NTC quickly heats and the voltage from the transformer and obviously the smoothing capacitors rise. When they reach a voltage you are happy with, usually 2/3rds reqyured voltage, that pulls in a relay, the contact of which places a link across the NTC and passes all of the current allowing the NTC to cool in case the unit must be powered off and then on again.
Doesn't take much to come up with a working option from the clues I have given.
The NTC is chosen to suit the average current at startup. The NTC quickly heats and the voltage from the transformer and obviously the smoothing capacitors rise. When they reach a voltage you are happy with, usually 2/3rds reqyured voltage, that pulls in a relay, the contact of which places a link across the NTC and passes all of the current allowing the NTC to cool in case the unit must be powered off and then on again.
Doesn't take much to come up with a working option from the clues I have given.
I see. Maybe a high power resistor across the current limiter will dissipate excessive heat? I chose FCH041N65F so that it could handle a lot of power but you are right. It cannot handle the initial 680W!
Interestingly my initial goal was to take out the mechanical relay and use some electronic switches instead. But hey if a good old mechanical switch is tough enough who needs a solid state solution? I'll give it a try.
Maybe one thing I can think of is that in an external current limiter that accepts worldwide voltages, the relay has to be placed on the secondary side (I never seen a relay that works on both 110V and 230V), and the current limiter will not be external anymore. I will try this for this amplifier's power supply at this moment.
Thank you. I will give it a try.
Another thing I thought about is that maybe a current limiter operating all the time isn't a good idea. Amplifiers has fuses to protect them from fatal shorts, and during a short if the current is limited by an external circuit, the fuse will never blow (and the limiter will blow). Even if I use the limiter, I would definitely set a threshold at least 2x higher than the fuse's rated current.
Oh yes, testing on a lower voltage why didn't this come to my mind first ;-(
My choice of FCH041N65F NMOSFET has a SOA of 1.5A up to 400V DC. Not that it can contain the 2A I intended, but with another resistor in parallel to take away some of the current I think it should work, with some good heat sinks maybe.
I think this view is where you are looking for a problem that is just not there. I'd hazard a guess that 99% of amplifiers do not concern the operator about in-rush.
Is your amp (with 2x 22mF) a commercial product, or diy ?
One aspect of such large value of cap is that intrinsically it provides the energy buffer between the mains and your audio stages - the audio stages may not even draw much power from the mains, and the 22mF could be fed via some secondary side resistance with no tangible disadvantage.
Did you see that the datasheet included an SOA chart? Perhaps best to spend some time get a technical awareness of what that chart is telling you for your application.
Seems there are a lot of posts in the forum about inrush current so I guess I with them fell into the 1% concerned user group...
It's indeed a DIY project, and 22mF*2 is the least capacitance I tried that will satisfy the ripple voltage requirement.
And yes, I'm aware of the SOA chart. Unfortunately it shows pulse performance up to only 1ms so the only reliable range I can believe is the DC line. (many datasheets don't even include a DC range) As ximikas talked about the resistor earlier, using a 100R resistor first to carry most of the initial current, and then use the MOSFET to short the resistor, it will not be burdened that much, so I can try that first.
It's indeed a DIY project, and 22mF*2 is the least capacitance I tried that will satisfy the ripple voltage requirement.
And yes, I'm aware of the SOA chart. Unfortunately it shows pulse performance up to only 1ms so the only reliable range I can believe is the DC line. (many datasheets don't even include a DC range) As ximikas talked about the resistor earlier, using a 100R resistor first to carry most of the initial current, and then use the MOSFET to short the resistor, it will not be burdened that much, so I can try that first.