Active pre-charge/inrush control

Arlo1 said:
A proper cap will still absorb this although you might see something at the fets it will be minimal compared to what you would have without.

exactly!
When the circuit is turned off under load, the voltage on controller side will jump instant to 0V (or at least very close to zero). than there is battery voltage + the voltage spike across drain and source on the FET's. a cap would help a lot for lowering the spike.
 
Big electronics dummy here. I have been reading and rereading this thread for several months and I now have more stupid questions.

Question 1:
What about this MOSFET (rated: 30 AMps @ 60 Volts) RFP30N06LE ?
https://www.amazon.com/dp/B01GNORMWA
https://www.amazon.com/dp/B00V9XF1RY
Data Sheet: https://www.sparkfun.com/datasheets/Components/General/RFP30N06LE.pdf

Question 2:
As I understand it (which is not be well) the idea is to delay/slow down the charging of the capacitors in the controler. The MOSFET is basically a variable resistor. Could the same thing be accomplished by driving the input pin of the mosfet by a MPU (Arduino) PWM output such that it takes several seconds to reach the point where the MOSFET is in the "ON" condition?
 
I'll answer backwards. Correct, the FET is just a resistor during precharge. The resistance is not a linear function of the gate voltage though. It comes on suddenly at the gate threshold, somewhere around 2v. If the Arduino PWM output went at just the right speed, it might work, but it would be hard to make it consistent as the gate threshold is temperature dependent.

The FET in the datasheet is pretty wimpy. Depends on your system voltage and expected peak current. Keep in mind that after precharging, the FET is carrying the full controller current, so needs to be rated accordingly.
 
Thank you. The bike is probably wimpy by some standards:
Golden Motors Smart Pie version 4 paired with a Lunacycle 52 Volt Shark battery.
Estimated [strike]25 Amps max[/strike] 50 Amps Max.

The real problem is the lack of an on/off switch short of unplugging/dismounting the battery (I HATE that battery).
Smart Pie 4 -elect ratings.png
 
LewTwo said:
Thank you. The bike is probably wimpy by some standards:
Golden Motors Smart Pie version 4 paired with a Lunacycle 52 Volt Shark battery.
Estimated [strike]25 Amps max[/strike] 50 Amps Max.

The real problem is the lack of an on/off switch short of unplugging/dismounting the battery (I HATE that battery).

if you use this motor at it's given specs (400w max and a 50v battery) you won't see more than 8a at the battery. if you say 50a max and 50v you're talking 2500w motor power. which of those assumptions is true?
 
izeman said:
LewTwo said:
Thank you. The bike is probably wimpy by some standards:
Golden Motors Smart Pie version 4 paired with a Lunacycle 52 Volt Shark battery.
Estimated [strike]25 Amps max[/strike] 50 Amps Max.

The real problem is the lack of an on/off switch short of unplugging/dismounting the battery (I HATE that battery).

if you use this motor at it's given specs (400w max and a 50v battery) you won't see more than 8a at the battery. if you say 50a max and 50v you're talking 2500w motor power. which of those assumptions is true?
The picture of the specs is from a website that [strike]sales[/strike] sold this version of the motor.
The 400 watts at 48 Volts is the "Rated Power" not the maximum.
With A 36 battery (charged to 42 volts) volts I never saw it draw more than 650 watts.
I believe the 50 Amp "phase" current is an instantaneous measurement but that would be the worst case scenario.
If the ratios between the rated power and max draw hold the worst I should see is about 800-900 watts which is where I came up with my original 25 Amp max estimate.
(52/36)*650= 938 watts .... I still think that a 25 Amp "switch" may more than adequate.

I have been fighting the Golden Motor software most of the night but finally got the controler reprogramed for a larger voltage range (31 to 60 volts, I think). Maybe after I get some shut eye I can try to get some actual wattage readings with the Shark battery.

EDIT: 910 watts on fully charged shark battery .... and wow what a difference those extra volts make !
 
I'd suggest a pair of IRFB4110. There are many other choices, but these are fairly inexpensive and have a good track record.
 
LewTwo said:
I believe the 50 Amp "phase" current is an instantaneous measurement but that would be the worst case scenario.
phase current has (more or less) nothing to do with battery current. phase is what the motor windings see. this is normally much much more than the battery current. a motor can have 50a phase current and still see only 20a battery current (just ball park numbers!)
this means that the data sheet is correct. :)
 
Thanks to Jeremy and Richard, my pre-charge circuit lasted for years until I lost it. Now for my present build, I need a precharge circuit for 100V, 180A. I can still use IRFB4115s but will need to change the board layout - more fets, heatsink.
What if to just use a DPDT switch wired to a contactor and controller? When switched on, contact 1 has a resistor that charges the caps in the controller for 1-2 seconds. Contact 2 has a delay timer switch(relay) or a resistor/capacitor combo that is set to power up in 4-5 seconds. So, by the time the coil is energized and switch in the contactor, the capacitors are already fully charged, then preventing a spark. The idea is to charge the caps first prior to the contactor turning on.
 
shinyballs said:
What if to just use a DPDT switch wired to a contactor and controller?
Or a circuit breaker (rated for full current) and a small pushbutton that charges the caps. Push the button for 5 seconds and then throw the breaker. If you don't hold it for long enough the breaker trips.
 
The contactor arrangement is pretty typical in a car sized EV. A small one precharges through a resistor and a second, bigger one closes the main circuit.

Somewhere way back in this thread I did a design that uses the FETs to precharge (instead of a resistor) and once precharged, uses another FET to operate a large contactor.

See: https://endless-sphere.com/forums/viewtopic.php?f=3&t=40142&start=100
 
I may just go all fets because of the smaller density compared to contactor + dc-dc + precharge circuit combination. The Toshiba TK72E12N1,S1X fet you mentioned earlier looks like a better choice and has come down in price. It's $1.86 if you buy 10pcs in Mouser. https://www.mouser.com/ProductDetail/Toshiba/TK72E12N1S1X/?qs=sGAEpiMZZMshyDBzk1%2fWi9ZL%252bu09nRqGkiYam0gXa7k%3d
Actual power consumption I estimate is 60A continuous, 180A peak(<20secs) - based on the QSmotor version of the Kelly KLS8430H controller.
Is five TK72E12N1,S1X enough to handle the current with no heat sink?

fechter said:
The contactor arrangement is pretty typical in a car sized EV. A small one precharges through a resistor and a second, bigger one closes the main circuit.

Somewhere way back in this thread I did a design that uses the FETs to precharge (instead of a resistor) and once precharged, uses another FET to operate a large contactor.

See: https://endless-sphere.com/forums/viewtopic.php?f=3&t=40142&start=100
 
shinyballs said:
Is five TK72E12N1,S1X enough to handle the current with no heat sink?

I think so. I usually estimate 20A per FET, but the heat has to go somewhere. A TO-220 package can dissipate about 1W with no heat sink. If you shrink wrap or pot the board so the FETs have a decent thermal path to the outside, they should be fine.
 
fechter said:
shinyballs said:
Is five TK72E12N1,S1X enough to handle the current with no heat sink?

I think so. I usually estimate 20A per FET, but the heat has to go somewhere. A TO-220 package can dissipate about 1W with no heat sink. If you shrink wrap or pot the board so the FETs have a decent thermal path to the outside, they should be fine.

An aside for my education: Can you increase the load capacity of a circuit by simply adding extra FET's in parallel, or do you need to make significant other changes as well?
 
Buk___ said:
An aside for my education: Can you increase the load capacity of a circuit by simply adding extra FET's in parallel, or do you need to make significant other changes as well?

Yes, just add more parallel FETs. Since this is not switching at high frequency, things are pretty simple. I've seen some FET banks on chinese BMS boards that have 10 or more parallel FETs. This is pretty much the same thing. You can run 500A with enough FETs.
 
fechter said:
Since this is not switching at high frequency, things are pretty simple.

Ah! Okay. So FET's in parallel is simple for on/off situations, but not for PWM; which is why people spend so much time searching for higher rated FET's at reasonable prices?
 
izeman said:
no. PWM works fine for FETs as well.

Okay. If parallel FETs work fine for PWM, why do the constructors of high power controllers buy and use very expensive single FETs rather than paralleling a bunch of far cheaper lower rated ones?

(Note:I'm not arguing with you; I'm inquiring as to the reasoning.)
 
There are many thousands (probably millions) of paralleled-fet high-power controllers out there, in the ones to tens of kW and probably higher.


There are a number of reasons to use single-device-per-phase-leg instead of multiple-parallel:
space
heat removal
heat distribution
current distribution
gate drive systems
timing

I recommend reading the design/build threads for the high power controllers here on ES for the engineering details of this.

If you're only controlling a single device (or pair of them, for each driven phase) then it simplifies a number of things that get more complicated the higher the power levels you're after.


But it's more expensive to do than just paralleling a bunch of cheap small fets, so that's what many manufacturers do, who are not after the absolute peak performance they could possibly get out of the controller, and/or are willing to spend the time and money to debug and work out all the little issues caused by the mulitple paralleling.
 
hey i have been using the Active pre-charge circuit 3b for some time now and recently noticed that all this time 1 of my 2-FETs in parallel was blown.

problem is every time I solder another in parallel the one that didn't blow last time blows!! i even put 4 (fresh ones) in parallel and they all blew besides for one?! I double triple check all other connections and they were all good! by the way i though heat might be the reason why they are blowing so i added a nice heat-sink and that didn't work either.

I just spend the whole soldering and de-soldering 1 million connections!! mad am i frustrated!

Can anybody explain what is going on?
 
It seems there is more than one version of 3b floating around. If yours has a switch, then it should be OK. Without the switch, it has been determined the circuit does not work properly.

The IRF3205 is rated for 55v. This is dangerously close to a fully charged 12s pack, so may be part of the problem. A 75v or 100v rated part would be better.

It would also be good to check to make sure the zener diode isn't blown. Do this by turning on the circuit (even without the controller) and measure across the zener or across any FET outer legs. You should see 12v.
 
i am using the one with the switch.

this is all happening with the pack voltage at 46V.

.i checked the zener diode. the second i connect my probes i see 18V on the multi-meter and then it quickly shoots down to 11.5 this while the zener is still in circuit with the rest of the components.

however as the circuit turns on i see it holds the voltage of the gate at about 5v for a second or two and then jumps to 12v


quick note: after the FETs blow what happens is the GDS pins are shorted together. just in case that sheds some light on what the issue might be.

@fechter would posting a video help?
 
Back
Top