Active pre-charge/inrush control

[Arlo1]

Ok rebuilt form scratch. With a 1uf cap and 12v zener and a cleaner build with less rats nest and it works! Today I got the horn and signals wired. The bike works now its time to finish cleaning the wiring and do some riding! Thank you so much fetcher! I tried it a bunch of times and so far so good!

 

[fechter]

Awesome! thanks for testing.

It would be interesting to measure the precharge time. I guess it will be almost exactly how long between throwing the switch and the contactor kicking in. Knowing this, the voltage and the size of the main caps in the controller, the current can be calculated. If you measured the actual current during precharge, it should work out pretty close.

 

[Arlo1]

Awesome! thanks for testing.

It would be interesting to measure the precharge time. I guess it will be almost exactly how long between throwing the switch and the contactor kicking in. Knowing this, the voltage and the size of the main caps in the controller, the current can be calculated. If you measured the actual current during precharge, it should work out pretty close.

I will see what I can do. Its about .8-1 second when dead and almost nothing when its only been off for a few seconds and the caps are partially charged still.

 

[fechter]

What is your pack voltage?

 

[Arlo1]

What is your pack voltage?

84v (it was 80 when testing) I have some big caps in that controller. My next controller is basically done... 150v components which I plan to run up to 28 or 32s and ~350 amps After I teach my first day tomorrow I will try to get back to the shop and clean up some wiring and do my first test ride as a finished bike

 

[fechter]

If the controller caps were 10,000uF, the precharge current would be about 1A at that rate. I think that's a good range. This keeps the FET in the safe operation area and still provides a fast precharge. If the controller caps are smaller, the current will be less.

 

[Arlo1]

If the controller caps were 10,000uF, the precharge current would be about 1A at that rate. I think that's a good range. This keeps the FET in the safe operation area and still provides a fast precharge. If the controller caps are smaller, the current will be less.

Yeh If I remember right I have ~5000 uf of caps on this one.

 

[Arlo1]

Its a lot less then a second. I tried it again tonight. I will try to scope it tomorrow. I hear the click in the switch then the contactor about .2 seconds later. But the click in the switch is likely a little after the contact moved into position to conduct power.

 

[fechter]

You might have to put a resistor across the controller input to make sure the caps are completely discharged first. If there is any charge on the caps, it will shorten the time.

Still, even at 0.2 seconds into 5,000uF, that's going to be a little over 2A, which should be within the safe area (though that's getting pretty close for a IRFB4115). Increasing the capacitor value until it takes about 1 second, it would make sure.

 

[Arlo1]

You might have to put a resistor across the controller input to make sure the caps are completely discharged first. If there is any charge on the caps, it will shorten the time.

Still, even at 0.2 seconds into 5,000uF, that's going to be a little over 2A, which should be within the safe area (though that's getting pretty close for a IRFB4115). Increasing the capacitor value until it takes about 1 second, it would make sure.

IRFB4115 is rated for 104amps.....
http://www.irf.com/product-info/datasheets/data/irfb4115gpbf.pdf
74 at 100deg and fig8 max safe operating area says at 80-90v its about 15amps at 100uS

 

[fechter]

That safe operating area graph is the one I'm not so sure about. According to the graph, it looks like 1A at 80V would be outside the safe area. It's going to take more than 10msec to get the voltage down to inside the DC line.
Obviously your FET didn't blow up, so maybe it's OK for a non-repetitive operation in that area. At faster charge times, the time it spends in the danger area is less, but at a slow enough rate it should be inside the safe area. The graph doesn't show what happens under 1A, but you can extapolate. Interestingly, the 4110 has a wider safe area and the graph shows you can run continuous below 0.1A

I guess to be safe, you might want to increase the charging time to decrease the current. 1-2 seconds isn't too terribly long to wait.

 

[Arlo1]

Maybe I will just run 2 1uf caps in parallel.

 

[madin88]

Hi, i want to build a precharge switch for about 150A and 90V Battery. The caps are about 4000µF all together. I dont want to use a heatspreader.

is somebody able to tell me if i can use this FET's for the switch and if yes, how many do i need.

I want to use the Richard Fechter V3 sheme.

Should i better take other rated parts for R1, R2, C and Zener or is it ok to use the standard parts for my project? What about charging time with 4000µF caps to not blow the FETs?

thx in advance

 

[fechter]

That is a very robust FET. Also a bit expensive but a good choice for running high power.

At that voltage, I would recommend increasing R2 from 1K to 2.2K or 4.7K.
A 1W rated zener diode is cheap insurance. A 1/2W would probably be OK but the price difference is minimal. R1 and C1 should be OK.

The number of FETs required is determined by how much heat they can dissipate and how hot you are willing to let them get. Datasheet says 40C/W to ambient air. A single running at about 22A would dissipate 1W. In most applications, you won't be running at the limit for more than a few seconds, but you might run 70% or so for longer periods.

My wild guess is use 4 of those in parallel. The copper wire will carry away a significant amount of heat, so make sure the connecting wires are big and heavy.

With enough heat sink, a single one of those would handle the power.

 

[teklektik]

A quick calc for 4,5,6,7 FETs is shown below (no heatsink). Ambient temp is allowed to run to 100degF.

The table cells are colored according to allowable power dissipation per FET to hit 85degC (green), 125degC (red), or in between (yellow) - this is 1.18W@85degC and 2.18W@125degC for the IRFP4468 device.

It looks like 6 FETs are a workable choice for the 150A requirement.


Spreadsheet calculator is available in this post.
Edit lower left table to replace an existing device with a new one.
Ambient temp or 85degC/125degC operating temp fields can be revised as needed.

 

[fechter]

Awesome! A calculator beats the snot out of a wild guess.

In my wild guess, I assumed 150A would be for short peaks and the average current would be lower.

Any kind of moving air or conduction to an outside surface will enhance the cooling a lot. Bolting the circuit to the frame or metal box/sheet metal will make a big difference.

 

[madin88]

THX for the quick Infos.

At that voltage, I would recommend increasing R2 from 1K to 2.2K or 4.7K.
A 1W rated zener diode is cheap insurance. A 1/2W would probably be OK but the price difference is minimal. R1 and C1 should be OK.

how long would it take to fully turn on the circuit with R2 2.2k and standard 1µF C1?
Wouldn't it be safer to take 100k for R1 instead of 1M ohm?

Yes, 150A is more peak than continuos. It's the battery current limit of my controller and i just want to be on the safe side. Currently i only set it to 60A.
I think i will only take 3 FET's and mount them to a little heatsink as you suggestet. They are really not cheap but indeed very good (18 of them are in my Adaptto Controller )

 

[teklektik]

In my wild guess, I assumed 150A would be for short peaks and the average current would be lower.

Any kind of moving air or conduction to an outside surface will enhance the cooling a lot. Bolting the circuit to the frame or metal box/sheet metal will make a big difference.

Agree 100%. As you can see in the spreadsheet, your 4 FET estimate will safety handle 115W continuous.

Unfortunately, anything but 'no heatsink in still air' is essentially impossible to calculate. You can fake it a bit in the spreadsheet by reducing the R(theta-ja) to values under 30, but that's changing the reliable calculation into a SWAG (Scientific Wild Ass Guess)...

Yes, 150A is more peak than continuos. It's the battery current limit of my controller and i just want to be on the safe side. Currently i only set it to 60A.
I think i will only take 3 FET's and mount them to a little heatsink as you suggestet.

A heatsink really makes sense, particularly with more expensive FETs. As fechter points out, if you can get a little moving air, the power capacity will go way up.

Here's a run showing 3 FETs, no heatsink, and still air. As you can see - it will handle 60A with no issue and will go up to 80-85A just a bit hotter. With a heatsink, you will have a substantial safety margin.

 

[fechter]

how long would it take to fully turn on the circuit with R2 2.2k and standard 1µF C1?
Wouldn't it be safer to take 100k for R1 instead of 1M ohm?

Yes, 150A is more peak than continuos. It's the battery current limit of my controller and i just want to be on the safe side. Currently i only set it to 60A.
I think i will only take 3 FET's and mount them to a little heatsink as you suggestet. They are really not cheap but indeed very good (18 of them are in my Adaptto Controller )

The 2.2K won't change the charging time much. It should still be right around 1 second.
Don't change R1 to 100K. That would speed up things x10. 1M works good.
Don't underestimate how much heat those fat wires can carry away from the FETs. They will have a fairly large effective surface area. Even better if you can get them in the wind a short distance from the board.

 

[madin88]

@ teklektik

THX for your information and the calculation pics. This is very helpful.
I will take 3 or 4 FET's with heatsink to be on the safe side.

@ fechter

ok so i will use 1M for R1
The wires have 8mm² but they are in a tube. Its anyway a good idea to place them in the wind right after the casing.

THX for your help. Now i can order these parts and a nice case in which i also will place a charging plug and some extra low esr caps.

 

[madin88]

can i use this smd 1µF cap?

http://www.conrad.at/ce/de/product/...-V-10-Murata-GRM31CR72A105KA01L-1-St?ref=list

in this shop i cannot find 100V MLCC with wires

 

[fechter]

can i use this smd 1µF cap?

http://www.conrad.at/ce/de/product/...-V-10-Murata-GRM31CR72A105KA01L-1-St?ref=list

in this shop i cannot find 100V MLCC with wires

That should work fine.

You could also use a polyester cap or a pair of 2.2uF, 100V aluminum electrolytic caps placed back-to-back (+ to +) in series.

 

[madin88]

THX fechter!

i ordered 13V 1W zener because 12V was not in stock. Can i use them or is 1V more a problem?

I still have some 12V 0,5W at home. Would one 0,5W sure be enough for my 4 x TO-247 FET's or sould i use 2 parallel?

 

[fechter]

13v zener is fine. I think the 1/2w ones are fine too. That's what I used. When the switch is on, there is less than a mA flowing through the zener, so it won't even get close to heating. The issue is when you turn off, the zener has to take a pretty good pulse from the capacitor depending on how fast the load drops. Still, its going to be fast, so not much chance of heating. The number of FETs won't really affect the zener.

 

[wannesd]

First of all, I'd like to thank Fechter for the schematics.
I too built the circuit you designed.
I used every part you specified, and used genuiune IRFB4110's.

I am only looking to switch 20A, but used 3 FETS to be sure, and to have some room for growth.

Ramp-up time was quite long (4 seconds), but I could live with that, and made sure never to apply load until the voltage was fully on.

I have so far only tested it with teh wheels of the ground.

Yesterday, I turned the keyswitch, and the voltage immediately went to full pack voltage. I knew then that something was wrong, and indeed, it wouldn't turn of anymore.

I think I'mm go with the simpeler version that was posted on page 1 instead, there seem to be too many people with the same problem (fets shorting out)

I'm running 22S lipo. Unless I needed to change a component out for the 22S setup?