Inruesh current handling?

[Mosaic]

Hi all:

I am making a linear pwr supply with a 16VAC input....a 35Amp FW bridge and a 15000uf cap...with .033 ESR!. Xformer secondary resistance is about .04 ohm.


Kinda worried about Inrush current killing the bridge rectifier.

Any suggestions on handling this...thx!! I don't want to lose ripple handling!

 

[crutschow]

They make negative coefficient thermistors just for that purpose such as these.

 

[Mosaic]

Well. it's kinda high current perhaps up to 20A... so Thermistors add losses.

I was thinking perhaps a very low on resistance N fet that would switch on based on the +ve charge across the cap. The cap will charge to around 23VDC, a IRLB3036 hexfet (195Amps continuous, 60 Vds) with 2 millihohm resistance needs 10 Vgs or so., but gets going from logic levels.
Perhaps a pair of 10k resistors to split the voltage across the cap to about 11 VDC would slow charge the Gate over a few milliseconds and turn on the N fet after the cap is fully charged for inrush current control.
Edit: A parallel Drain-source 1 to 3 ohm resistor would start the charging of the cap

 

[KeepItSimpleStupid]

I know the feeling really well. I have like 40,000 uf of capacitance on 2 x 35 CT windings for an audio amp I built. I use an triac power switch too.
I made a nice turn on circuit 31 years ago that still works. I put a metal oxide resistor in the 120 VAC line and keep the audio off and speakers off untill the caps reach about 2/3 of 50 V and then short out the resistor. It will/and has protected against accidently reversing the NPN and PNP outputs with no damage other than the resistor. A fuse failure on any supply will pop the resistor too. To make it better it needs a timeout: If it cannot achieve turn-on in x seconds then disconnect the line. As an aside, the amp has a 500 W AC line voltage regulator on it that I picked up for $100. It retailed for about $1000 maybe 20 years ago.

 

[Mosaic]

I guess a low wattage resistor is best as a failsafe....if the power supply has a high current drain or short circuit upon turn on the resistor will drop a lot of voltage which could prevent the nfet from full turn on, causing the resistor to pop.

Or perhaps one of these e-fuses
https://www.newark.com/jsp/search/productdetail.jsp?SKU=58K1837

So it won't 'pop' but just disconnect on excessive start up load.

Then there is perhaps a .1 uf ceramic instead of the e-fuse, which would only charge pump the 15000uf cap if there is no significant load . Any excess load would never allow the Nfet to turn on.

 

[JimB]

One way of limiting inrush current is to use a resistor in series with the primary of the transformer.
After a short delay the resistor is shorted by the contacts of a relay.
The coil of the relay is energised from the rectified DC with a resistor in series with the relay coil and a capacitor in parallel with the coil.

By putting the resistor in the primary of the transformer, the resistor can be a higher value and the relay contacts can be rated for a lower current.
The down side is that the resistor and the relay contacts are at mains voltage, but this should not be a problem if the thing is built correctly using correctly rated components.

JimB

 

[KeepItSimpleStupid]

JimB:

Which was exactly what I did, except I made the delay mostly dependent on k*R*C, where R is MOSTLY the one I inserted and C is the power supply cap.

 

[crutschow]

Well. it's kinda high current perhaps up to 20A... so Thermistors add losses.

If you look at my link, they make ones that are useful up to 50A.

Yes they add a small power loss, but it's normally not significant compared to the power being drawn. For example the 20A unit has a resistance at rated current of 0.035Ω or less which is 0.7W at 20A .

 

[Mosaic]

I don't see how a .035 ohm unit is gonna help with high inrush current. 23V peak across that will be 635 Amps!

 

[crutschow]

I don't see how a .035 ohm unit is gonna help with high inrush current. 23V peak across that will be 635 Amps!

If you read the data you would see how it's gonna help. It's a thermistor which changes resistance with temperature. Its cold resistance is around 1 to 5 ohms depending upon the model and changes to the low resistance as it heats up from the current.

 

[Mosaic]

Ok...I see...

Overall I guess the MOSFET/TRIAC approach adds a measure of protection from a power up short circuit.

 

[Mosaic]

Just another question. Won't a 500uH power inductor between the bridge and the cap. block fast inrush currents & transients as well?

 

[crutschow]

Just another question. Won't a 500uH power inductor between the bridge and the cap. block fast inrush currents & transients as well?

Fairly well. I did a SPICE simulation applying the voltage during the peak of the 16V sine-wave (which can occur with random switch closure) into 500µH in series with 15,000µF and 0.073Ω, and the current surge was about 69A. That's well within the 35A diode's surge rating.

 

[Mosaic]

Hi, is it LTspice u did it in?

Can u share the sim?

I'd like to test it with a 220uH Inductor ...also has a .1ohm resistance

..Your'e a great help!Thanks!!

 

[crutschow]

I originally did the simulation with Electronic Workbench (Multisim) but here's the simulation with LTspice. I used a different diode so the peak current is slightly different.


View attachment Surge Current.asc

 

[Mosaic]

Semiconductor version

Thanks for that:
here is my 1st attempt at LTSpice!

A Pnp/PFet pair to limit the current based on a 1 ohm current sense induced drop across the PNP base & Zener.

Max inrush is now 61 amps. With less losses than an inductor system at high current. No thermistor thermal lag.

Blue= current.
Red= AC in.
Green= Capacitor voltage.

 

[crutschow]

What is the purpose of the forward biased zener in the base of Q1? Zeners generate their voltage in the reverse direction.

You need a resistor in the base of Q1, otherwise high current will flow through the base emitter junction during start-up and likely blow the transistor (the blowing of components is not directly simulated in LTspice).

If you post your .asc file I'd like to simulate the circuit.

The thermistor thermal lag is quite short. Is that lag a problem? It would seem that one thermistor is much simpler than adding two transistors, a diode, and several resistors (doesn't follow the kiss philosophy)

 

[Mosaic]

yes, you're right.....btw how can i calc avg power in ltspice?

 

[crutschow]

yes, you're right.....btw how can i calc avg power in ltspice?

Go to the help file and look up "waveform arithmetic". It will show you how to multiply the current and voltage together in a plot to get the RMS power.

 

[Mosaic]

New sim

I did another design, this one appears to not need inrush handling.