Active pre-charge/inrush control

[auraslip]

How about we send ourselves this circuit schematic and make a pcb that people can buy and solder the components themselves.

I'd like to learn how to print pcb's using my laserjet printer. If someone wants to create a pcb to sell I'd buy it, but I think it'd be cool to make it myself.

 

[NeilP]

Would not even consider using a PCB for a circuit like that,

Simple mount the FETs to a heat sink bar within a case of some sort and wire it all up with hard wire links/buss bar arrangement.

Once the FETs are mounted, pull the source pins up in line with each other, solder to a thick strip of copper bar/wire, one at at time, slipping heat shrink over each section between FETs, or if using insulated wire, just strip back where necessary to solder the FET leg to the wire.
Do the same with the drain pins,

Thin(ish) 'signal' wire to the gate pins and the two resistors and the cap mounted on the back of the switch body

 

[izeman]

i'd like to x-link this thread to my thread, where jeremy (and others) helped me a lot. maybe it answers some questions as well, that others may have

http://endless-sphere.com/forums/viewtopic.php?f=2&t=46543

this is the outcome:

 

[NeilP]

Almost identical to mine apart from the colour of the cap

 

[izeman]

Almost identical to mine apart from the colour of the cap

i thought that after reading your post not that i copied it. but when two got the same idea it can't be that wrong.

 

[mvly]

I finished some PCB design on this circuit. However I am still waiting to get it back from the fab. However I am a bit concerned on my design, as the traces might not be thick enough to support the current I am using, I.e. up to 60Amps continuous. I can widen the traces, but then it would increase the size of the board. I also thought of stacking boards, but then it would mean more work on my part to put something together. Finally I tried to look into using 5oz cu traces, but that can be expensive. The only good solution I can see is to not have a solder mask and so I can beef up the traces using external wires. But this will be messy. So far i estimate around $40-$50 per switch assuming 6-8 FETs. Any recommendations?

Here are my goals:
1) simple to put together
2) nice form factor
3) cheap
4) able to carry 60A continuous

Here is the website I used to calculate the width thickness of my circuit if anyone is interested.
http://circuitcalculator.com/wordpress/2006/01/31/pcb-trace-width-calculator/

 

[izeman]

just solder some solid wire on then pcb where it needs stronger traces. everyone who can solder can make the traces stronger that way.

 

[dnmun]

yes, solder some solder wire onto the trace and make contact with the mosfet legs as it goes by each one. they actually make 'bus wire' specifically to do that, but you can use regular solid wire like he said.

jeremy's design uses the irfb4110s so that it can handle up to 100V packs, but if you have only a 60V pack you can use mosfets with lower BVds which would be cheaper. i found some irfb3205s for 50 cents each on ebay which would be good to 60V. jeremy also recommends the 3077s for the 6 fet controller, they are good up to 75V as i recall.

also since this circuit is in the ground of the battery, you should use a fuse on any high voltage, such as the B+ lead that enters the space to avoid bad consequences of the inevitable short.

 

[mvly]

I think I got the best solution. I will keep the solder mask to prevent shorts, but I will include large holes so people can solder 12gauge wires from point to point. Therefore there should have enough current path to reduce the resistance and power dissipation of the PCB board.

I'll see if I can get the revised design up soon so you guys can see.

 

[fechter]

That should work. The 12 jumpers won't be very long so won't have much voltage drop.

I like layouts that have the drain and source wires coming from opposite directions so the current distribution in the FETs is more even (equal path lengths).

For me, I would also use a 12v zener across the gate as extra protection. Zeners are cheap. With the right resistors and a zener, a single design could work over a very wide voltage range.

 

[docnjoj]

Hey Jeremy. I would like to build one of those switches for my wife's trike and I have a bunch of np75np(?) fets. Could they be used for 36 volts. They are from an old BMS that failed because of the connector for the balance cables. I have about 6 new ones when I was going to repair the Ping BMS but he sent me a new one.
otherDoc

 

[Eascen]

Two quick questions:

1) Is it possible to use regen with this? For some reason my electrical knowledge and research is failing me, I venture to say yes but my lack of understanding makes me ask.
2) I have two 36v batteries. Originally, I was going to setup two sets of 6 4110's, in order to do low side switching of each pack... it's since occurred to me that's I believe that's uneccessary, I'll simply need to switch the negative of Battery1->Positive Battery2, am I correct?

 

[fechter]

Regen should work OK as when the FETs are on, they are essentially the same as a piece of wire.

I'm not sure about two 36v packs. Series or parallel? Either way, if the circuit is placed directly in line with the controller power input, you shouldn't have problems. In a series configuration you could run into problems if you tried placing the circuit between the two packs.

 

[mvly]

Two quick questions:

1) Is it possible to use regen with this? For some reason my electrical knowledge and research is failing me, I venture to say yes but my lack of understanding makes me ask.
2) I have two 36v batteries. Originally, I was going to setup two sets of 6 4110's, in order to do low side switching of each pack... it's since occurred to me that's I believe that's uneccessary, I'll simply need to switch the negative of Battery1->Positive Battery2, am I correct?

For 2, you need to draw us a diagram. For 1) I would say fetcher is right. The reason it works is the FETs are essentially switchs. However you should be careful as it is also a diode if current flow the other direction. In our case nothing changes, as current can flow either direction without any problem.

 

[izeman]

i wrote in the 2nd thread about my problem, but maybe it's better suited here:
i built the mosfet switch as posted above. connected it, and everything was great. disconnected it, connected it again, and the spark are there again. 0ohm between all pins of the mosfets. desoldered all cables. one mosfet was damaged.
i replaced the broken mosfet. connected it again. all working great. disconnected it (removed the positive feed), disconnected the positive battery line, reconnected it - BAM. spark again.
what's going wrong here? any ideas?

 

[fechter]

Hard to say exactly. Are you using a zener diode from gate to source?

 

[izeman]

No zener in the circuit.

 

[fechter]

After thinking about it a bit more, most likely the FETs are failing during use and go shorted. When the thing is on, you don't notice them shorting since it's supposed to be close to a short anyway. You only notice it the next time you hook up.

How many and what type FETs are you using?

What is the maximum amp draw of the controller?

 

[izeman]

there is no load attached but the ca-v3. so only mA.
i'm still in the testing phase. maybe the resistors are wrong and gate voltage is too high. it's strange.

 

[fechter]

Personally, I'd recommend using a zener no matter what. They're like $.05ea and can't hurt anything.
A transient spike to the gate could cause failure, but the resistors should prevent that. A 12v zener should make the gate input bullet proof.

You could try measuring the voltage on the gate to source when it's powered up (even with FETs removed) to verify it's in the healthy range. Ideally, around 10-15v. Over 20v blows it.

It could also be failing due to excessive current or dissipation during precharge or operation. This might be harder to test.

 

[izeman]

ok. i'll add the zener. regarding power distribution: does it make a difference it the power wires are connected in line, or symetrically?

in this picture, power(ground) lines enter from one side, and leave on the same side. would it be better to feed the 10/12awg wires right between the 4 mosfets? would that make any difference. from my minimal electronic engineer's knowledge it makes no difference. but who knows.

i also found a circuit with a 47uF cap instead of the 1uF one. would that make any difference? i think this is just for leveraging the voltage, correct?

i calculated the voltage divider. i use a 47k and 10k resistor. with 90v max, Vgs is 16v.

when i connect positive/controller and negative/circuit to the battery the first time, there is absolutely NO spark. then i connect the circuit's 47k line to positive battery, and the connected CA turns on. i don't think that the CA booting is a too high load (especially compared the the charging of the controller's caps)

 

[fechter]

In theory it would be better if the negative wires came in from opposite sides so the current is shared more evenly between the FETs. As is, the one on the left in the picutre would take more current since the path is shorter. If the wires are heavy enough, it won't matter much.

I'm not sure about the cap. A larger cap will slow down the turn on, and reduce the peak current when the controller caps are charging. With no capacitor, it would charge very quickly and possilby cause the current to go over the rating enough to cook something. I think a larger gate cap would help.

What FETs are you using?

 

[izeman]

In theory it would be better if the negative wires came in from opposite sides so the current is shared more evenly between the FETs. As is, the one on the left in the picutre would take more current since the path is shorter. If the wires are heavy enough, it won't matter much.

I'm not sure about the cap. A larger cap will slow down the turn on, and reduce the peak current when the controller caps are charging. With no capacitor, it would charge very quickly and possilby cause the current to go over the rating enough to cook something. I think a larger gate cap would help.

What FETs are you using?

those are 4310. capable of 140a@100v. so "stronger" than the 4110, but with a high RDSon 5.6mOhm compared to 3.7mOhm.

 

[izeman]

In theory it would be better if the negative wires came in from opposite sides so the current is shared more evenly between the FETs.

btw: it was the third from the left that burned through. and there was NO other load connected than the CA. so no motor connected.

 

[izeman]

that's strange. there was one dead fet (no1) out of four. swapped that. worked once, then sparked again.
no1 was dead again. tested no2. no problem. tested several times, on/off. no problem.
seems the fets only seem to be ok. the come from a broken controller. maybe they were damaged somehow. they did pass the simple diode test, but maybe not 100% ok.
measured Vgs. it was a stable 14v.