Active pre-charge/inrush control

[d8veh]

There must have been a short somewhere for them to blow. The FETs are protected from the 84v by the 47K resistor, so there's no way you can get enough power to blow anything. Were the battery wires connected to anything?

Try again, but with only one FET at first. Test it with nothing connected to the main wires. You can check it with your meter measuring resistance between the two main wires. You should see the resistance change from very high to zero. It'll give enough current for a gentle no-load test on your bike. When it works, add the other three.

Looking at the photo, you used two 47K resistors instead of 47K and 10K, but that only makes it switch off more slowly. What was the voltage rating of the capacitor. You need 100v.

 

[sacko]

Well spotted

Here is another board I built straight after making the first one. Its probable a bit clearer to see if there are any faults.

Red battery wire went straight to the controller, black battery wire passed through the board to the controller.

It is a 100v cap, with the negative leg on the negative track.

I have another board, and doing the tests now. Cheers

 

[sacko]

Exactly the same thing happened.

Plugged it in, switched it on (not connected to the controller) and it blew up.

 

[fechter]

The FET gate is rated for about 20v maximum. You want it to be at 10-15v to get full turn on.
If you're using two resistors of the same value, the gate will be getting half the supply voltage (poof!)

In my version, there is a zener diode that prevents the gate from going over 12v.

 

[teklektik]

This board has two significant fabrication problems:

1. Uncut trace as pointed out on the previous page. This is applying pack voltage to the FET gates - instant death.
   (Either cut the trace of jog the resistor at an angle to pick up a second trace.)
   

   resistor r1 doing nothing, no cut track

   
   View attachment 1
   
2. Bad resistor value
   
   

   Looking at the photo, you used two 47K resistors instead of 47K and 10K, but that only makes it switch off more slowly.

   No. That is not how this circuit works. This is a significant issue because this circuit attempts to be 'parts thrifty' by using that resistor to set the max gate voltage. As a result, the R1/R2 ratio is critical. Using two equal resistors raises the gate voltage to 1/2 pack voltage (40+V in your case) and bye-bye FETs...
   
   On the previous page fechter recommended using the 'improved' circuit which adds a zener specifically to address this problem.

I mentioned the non-polarized cap because I assumed you were using the later 'good circuit' (I couldn't make out the tiny linked circuit image when viewed on a tablet). I've posted it below. Apologies for any confusion there - assumption at work! - My Bad.


EDIT - oops - fechter sneaked in a post before me saying the same...

 

[alsmith]

As said- R1 is missing.
So why isn't the track cut under R1? Both ends are connected to the same track meaning it is doing nothing- it's not there.

 

[sacko]

Thanks for the help all.

Just made another one, and cut the track on this one, along with the correct resistors.

Connected it all up, and all seemed good (bike wasnt turned on), flicked the switch to power up the bike and the fets blew.

 

[fechter]

OK, you solved the wiring problem at least.

I strongly suggest using the other circuit I mentioned on page 6. With your circuit I can almost guarantee FET failures when the controller is connected. You could go through a lot of expensive FETs.

 

[teklektik]

Well - as discussed in the thread and as fechter told you a few posts back - even when built correctly - this circuit eats FETs.

You can read the thread yourself, but the gist is that your circuit drives the FETs with a non-linear current curve that slams the FETs hard immediately on power-up followed by the current tapering off. Fechter's circuit instead has a very moderate fixed current for the whole pre-charge period -- making for happy FETs.

The surge in your circuit is related to the controller capacitance so the same circuit may work on one bike but not another (translation: "Well, it worked for me!" doesn't mean it will work for you - or that if it implodes that you built it wrong or had defective parts - it's just its nature...)

 

[d8veh]

Are you sure that R2 is 10K? Measure it to make sure. My PC won't show the colours. they all look black.

 

[sacko]

Thanks all for the help on this. I saw Fechters design with the switch and may have a go in the future, but im getting tired of this 'project'!

R2 measured at 9.98k ohms.

Brown
Red
Black
Black
Red

 

[d8veh]

Measure the resistance from the positive connection point to the gate rail. It must still be shorted.

 

[alsmith]

I'm with teklektik- you've done what you can, the board is correctly built, it's time to stop wasting more time- move on to a better developed circuit
Go with the one he recommends.

 

[dawrench]

I have been using this circuit (fechter's #3 with a switch) since early June on a 48 volt system and had zero problems. Same circuit since the first of August on 18s and nary a problem. The FETs came from Mouser, I used a heat sink with thermal compound and all wiring was point-to-point.

 

[Brake]

May a well chime in.

Awesome. No issues. Works perfect.

92v 130 amps. Max.

 

[megacycle]

Would be good if it could be modded for output short circuit protection, or over current, then we wouldn't need a fuse or circuit breaker on the battery, because there'd be an electronic one.

 

[fechter]

Would be good if it could be modded for output short circuit protection, or over current, then we wouldn't need a fuse or circuit breaker on the battery, because there'd be an electronic one.

That's certainly possible, but would require a shunt that can handle the full discharge current and something like an op-amp to sense it. Most Chinese BMS circuits have such a feature.

You would still want a fuse or breaker. If the FETs shorted, you could have a big fire. If the electronic breaker functions properly, the fuse would never blow even if the output shorted.

 

[megacycle]

You would still want a fuse or breaker. If the FETs shorted, you could have a big fire. If the electronic breaker functions properly, the fuse would never blow even if the output shorted.

Yes an electrical protective device like a fuse or breaker is still needed for failsafe what was I thinking.

 

[danielrlee]

Newb question here....

If I wanted to incorporate a connection for a charger into the circuit, should the input be on the battery or controller side of the FETs?

My thought was that connected to the controller side, I could leave a charger attached without worrying about output cap drain down resistors draining the pack. Also, connecting up a charger as needed wouldn't cause the output caps to spark without plugging into the mains first.

 

[fechter]

The charger needs to go directly to the battery. If the switch is off, the controller won't drain the pack.
The FETs have an intrinsic diode in them, so you don't want to charge from the controller side since the diode will cause a slight voltage drop if the switch is off. The diode action will also not prevent sparking when you connect the charger.

 

[danielrlee]

Hmmm, it seems there's so much I still don't understand. Thanks for your quick response though. It'll probably save a whole lotta ballache when it comes to making the circuit.

 

[izeman]

i already posted a lot in this thread, and got two REALLY WELL working switches that do their job 100% reliably.
now i wanted to build another of fechter's 3A version. it seems turn on time is way below 1s and i get a (smaller than direct connection) spark. it's made of 2x 3077 resistors 1k and 1m and 1u cap with a 12v zener screwed directly to a heat sink which connects to the bike frame. max load will be 45a.
what can i do to make turn on time longer? change to cap to a higher value?

 

[fechter]

what can i do to make turn on time longer? change to cap to a higher value?

Correct. You could also use a higher value of pull-up resistor on the gates.

 

[izeman]

what can i do to make turn on time longer? change to cap to a higher value?

Correct. You could also use a higher value of pull-up resistor on the gates.

i couldn't find any spare 100v cap 2u cap so i changed the 1k resistor to 2.7k. didn't change a thing ... should i go higher?

 

[fechter]

The other resistor, the 1M one. Try 2M. This will slow it down by 2x.