looking for a battery power switch

izeman

1 GW
Joined
Jun 21, 2011
Messages
5,131
Location
Vienna, Austria
i want to disconnect the main battery from the controller when not in use, and i don't want to open the battery box and disconnect the plugs.
battery has some 80v and current will be between 30-40a max. i will use an anti spark circuit installed inside the controller so there will be no high current peaks when switching on. any suggestions what to use? all i can find and are small enough are limited to 16/25a and 250v.
http://www.conrad.at/ce/de/product/704171/Marquardt-Wippenschalter-Serie-1830-18313313-250-VAC-20-4-A-1-x-AusEin-rastendrastend/0216200&ref=list
 
The circuit in the thread below does this, and manages the capacitor inrush current when switched on. The isolation when the switch is off is very good, I couldn't measure any current at all flowing when switched off and two of my bikes have this type of switch as the only power switch, and I leave both with the battery packs connected. It also allows a small switch to be used as the main power switch, and this small switch can be located some distance from the battery pack.

http://endless-sphere.com/forums/viewtopic.php?f=3&t=40142#p586436
 
wonderful. i already thought that as well. those circuits don't draw any current when not connected, so only as soon as you engage the switch, the current flows, and you need a small switch only, as current to feed the circuit is super small.
i will just have to adopt the circuit to fb4310. i hope i can do that ;)
 
izeman said:
wonderful. i already thought that as well. those circuits don't draw any current when not connected, so only as soon as you engage the switch, the current flows, and you need a small switch only, as current to feed the circuit is super small.
i will just have to adopt the circuit to fb4310. i hope i can do that ;)

If using the much higher on resistance IRFB4310, then I would strongly advise using more of them in parallel, to reduce the total on resistance. I've found that I can just about get away with two paralleled IRFB3077 FETs (3.3 mohm on resistance each) for a 20A switch, and four IRFB4110 FETs in parallel is OK for 30A. To get the same on resistance as four paralleled IRFB4110 FETs with IRFB4310 FETs would mean using at least six IRFB4310 FETs in parallel.
 
Have you check out the kfong's smart switch thread?

It has anti-spark, capable handle the high voltage and amperes. I'm using his smart switch on Yuba Mundo cargo bike. :)

viewtopic.php?t=32135
 
Jeremy Harris said:
izeman said:
wonderful. i already thought that as well. those circuits don't draw any current when not connected, so only as soon as you engage the switch, the current flows, and you need a small switch only, as current to feed the circuit is super small.
i will just have to adopt the circuit to fb4310. i hope i can do that ;)

If using the much higher on resistance IRFB4310, then I would strongly advise using more of them in parallel, to reduce the total on resistance. I've found that I can just about get away with two paralleled IRFB3077 FETs (3.3 mohm on resistance each) for a 20A switch, and four IRFB4110 FETs in parallel is OK for 30A. To get the same on resistance as four paralleled IRFB4110 FETs with IRFB4310 FETs would mean using at least six IRFB4310 FETs in parallel.
i got some 4310 left. i just checked rds(on) and voltage for reference
3077 - 75v - 2.8mohm
4110 - 100v - 3.7mohm
4310 - 100v - 5.6mohm

let's say i want 40a to be on the safe side, and the 4310 has doubled loss compared to the 3077, then i would need 8 of them in parallel. that's a bit too much (space wise)
those will sit on an aluminium bar and this is screwed to the controller case. so some watts loss should be no big issue. what do you think?

edit: i miscalculated. i would need even more, as 40a means much more loss then 30a. but i hope it still is ok. or i'm gonna order some 4110.
 
It's best to use the data sheet maximum figures, rather than the typical, just to be safe, so 3077s are really 3.3mohms, 4110s are really 4.5mohms and 4310s are 7mohms.

If you had 4 paralleled 4310s, then the total on resistance would be around 1.75mohms max. At 40A the power loss in the FETs would be 40² x 1.75mohms = 2.8 W. This is OK, but the FETs will get a little bit warm, as each will need to dissipate 0.7 W maximum. This isn't really a problem if the FETs are on an alloy bar. I've always been pushed for space so have ended up with the FETs inside the battery box, where extra heat was always welcome.
 
izeman said:
thanks jeremy for doing the math for me :)
those fets will be installed on the inside of a 12fet infineon controller.
do you know the heat dissipation of the installed 4110s? if this is much higher then 2w i can life with it ;)

A good general rule of thumb for these controllers is that the FET dissipation is roughly double the resistive loss (because of switching losses). A 12 FET controller running at 40A will be dissipating around 3.6W per FET, around 1.8W as resistive loss, and 1.8W as switching loss.
 
izeman said:
so this is 0.7w compared to 3.6w. correct? so this should be no problem.

Yes, no problem if you are going to mount them on an alloy bar as a heat sink. My rule of thumb for TO220 FETs with no heat sink, mounted inside a sealed case (which is how mine are mounted), is no more than about 1W, but if you can bolt them to the case then that makes a big difference.
 
I am using jeremy's design and started fabing the board. So far it is first version run, so not really modular enough. I chose all surface mounted parts to keep the size down and make it look clean. I ran through the numbers and made some excel sheet on how many FETs you need for certain current setup. I ordered 3 boards and will have it back in a month because I went with batchpcb. If you are running less than 74V I would go with 3077. They are not much more expensive than 4310. I will update this thread when I get it and fully test it.
 
mvly said:
I am using jeremy's design and started fabing the board. So far it is first version run, so not really modular enough. I chose all surfaceounted parts to keep the size down. I ordered 3 boards and will have it back on a month because I went with batchpcb. If you are running less than 74V I would go with 3077. They are not much more expensive than 4310. I will update this thread when I get it and fully test it.

thanks. i got some 3077 spares. but this is gonna be 80v+. so i have to go with the 4310. no 4110 left. i will have to wait until i burn my next controller ;) those 4310 are from a burned eb309. i removed them, and tested them. 3 where burned, the rest seems ok.
i will post pictures of my circuit probably within next 24h. it's quite simple and doesn't need a board.
 
izeman said:
mvly said:
I am using jeremy's design and started fabing the board. So far it is first version run, so not really modular enough. I chose all surfaceounted parts to keep the size down. I ordered 3 boards and will have it back on a month because I went with batchpcb. If you are running less than 74V I would go with 3077. They are not much more expensive than 4310. I will update this thread when I get it and fully test it.

thanks. i got some 3077 spares. but this is gonna be 80v+. so i have to go with the 4310. no 4110 left. i will have to wait until i burn my next controller ;) those 4310 are from a burned eb309. i removed them, and tested them. 3 where burned, the rest seems ok.
i will post pictures of my circuit probably within next 24h. it's quite simple and doesn't need a board.

Personally I like things clean, simple, and neat. I am currently using a precharge circuit and it's a hassle to unplug and plug everytime. And if I forget to precharge, BAM there goes a charred connector. I would much prefer a clean and reliable solution. Sure I can solder and make myself out of electrical boards and such, but I figured, it would cost the same in terms of physical materials. The IRF FETs are the most expensive parts anyways. Moreover, I do not trust my jury rigging skills. I have seen one too many difficult, bad solder joint, loose connectors, etc. Therefore, I rather put some time to PCB the circuit. It is simple enough to not take too much time and money.

Here is the excel sheet attachments (PDF) on the number of FETs in parallel and the relative power dissipation per FETs for up to 150A with 10A increments. I also highlighted green and yellow for currents you should use for certain FETs and such base on what Jermey stated. (< 1.1W for green and < 1.5W for yellow) You will know what I mean once you see it. Hopes this helps someone. IGNORE THE FIRST SHEET.
@Jeremy, let me know if this looks right. I created base on what you recommended. I left the Typical calculation in for people who either plan to cheap out or trying to run slightly higher than the limit. : )

It turns out for for my 51.8V BMC V3 Ebike limited at 32A, I can get away with just using 3x IRF 3077 FETs. For my 74V HS3545 Ebike limited at 60A, I would need 4x IRF 4110 FETs.
 

Attachments

  • Precharge Power FET Dissipation.pdf
    114 KB · Views: 162
if you have a lifepo4 pack with a BMS you can turn off the output mosfets so the battery would not spark when connected to the controller. then turn it back on when you have the controller connected.

you can do that by putting a switch in the circuit current that drives the output mosfet or just put a switch between the gate and source of the output mosfet and turn it off by shorting the gate to source. that turns off the mosfet. that is how the thermal breaker works on the new headway BMSs.
 
izeman said:
one more questions: i only can find a 50v cap in my spares box. but i guess it should be 100v for a 82v battery. correct? how much w for the resistor? 1/2w is ok i guess.

The 1uF cap does not see more than 20V because of the voltage divider and if you calculate the resistors correctly. So you don't need that much of a margin. Keep in mind Vgs on the FETs are only +/- 20V limit so you will ruin your FETs before you ruin your Caps.

I might be wrong, but this is what I get from doing the math on the circuits.

Also for the resistors, it turns out you can get away with 0.25W or even 0.1W if you are venturous. However I would play it safe and use the 1/2W.

I will up load some for calculation later to show how much power is being dissipated by the 10K resistor and the high val resistor.

UPDATED with the resistors values and their power dissipation.
 

Attachments

  • Precharge Resistor Values - Precharge Resistor Values and VGS.pdf
    29.9 KB · Views: 106
dnmun said:
gate drive should have a zener to protect the gate. whatta lotta work to imitate the output mosfets. then place it in series.

No need for the zener at all, if the right resistor values are used. Just an extra component that adds nothing useful. The FETs need more than about 6 or 7 V on the gate to turn hard on, and less than 20 V to be safe, so there is a massive safety margin if the right resistor values are used. For example, if using a 33k and 10k resistor pair then the safe battery voltage range is from about 30 V minimum to about 86 V absolute maximum, a ludicrously wide absolute range when you consider that I very carefully stated that the usable range was 50 V to 70V for this combination, just to be extremely safe and remove the need for any more components in what is a very simple, safe and reliable switch circuit.

The thing is imitating a power switch, rather than the output FETs, and with several FETs in parallel is massively lower in on resistance than the controller. Even the small 30A switch I've been using is the equivalent of a 24 FET controller in terms of on resistance, and because there is no switching loss is about twice as efficient as even that monster controller in practice.
 
izeman,
I understand you are looking for brand of switch, capable to switch your motor current.
So staying on topic :
I can suggest NKK switches sold by Mouser or DigyKey,
I extensively used them on my conversions.
You cannot beat them for quality MadeInJapan.
http://www.alliedelec.com/search/productview.aspx?SKU=70192243
 
@miro13car: thanks for the link. a perfect switch for switching high loads.
still i soldered the mosfet switch today. i like this solution.

came out quite nicely, all components very well packed. i bent the contacts:

G - level -> to switch
D - down -> to controller 0V
S - up and flat with fet's surface -> to battery negative

the red wire, with 47k in line, goes to a conventional low amp switch, which is installed in/on the controller.

22282,mosfetBG957.jpg
 
izeman said:
@miro13car: thanks for the link. a perfect switch for switching high loads.
still i soldered the mosfet switch today. i like this solution.

came out quite nicely, all components very well packed. i bent the contacts:

G - level -> to switch
D - down -> to controller 0V
S - up and flat with fet's surface -> to battery negative

the red wire, with 47k in line, goes to a conventional low amp switch, which is installed in/on the controller.

22282,mosfetBG957.jpg


Looks pretty nice! But I would be a bit wary of the Cut insulator on the wire. Don't want those to short. They are too close together. This is why I rather save myself some headache down the line to PCB the solution. Also I can mass manufacture. Try to mass build your design. : ).
 
how can that be too close together? this is how the mosfets left the factory. pins are as far/close together as they were manufactured. and they are good for 180a! so i'm not expecting a spark to jump - are you?
and they are definately not shorted now - how should they short in the future?
if i oversaw something please let me know. i surely don't want that thing stop working somewhere in no man's land ;)

ps: i'm really looking forward to see this massproduced. i'll be your first customer! :)
 
This is so crazy. When I wanted to attach a switch between my battery and controller, I just used a thumb switch that I can flip off so I can leave my battery plugged in during charging. Granted I'm only running 36V through the switch. Also the controller has a cut off feature for when too much stress is being applied to the battery. I'm usually pretty good about cutting back on the throttle when it's about to kick in but not always. I mounted the switch down at the apex of the frame and seat tube in front so if I have to I can reach down and flip the switch off and on to bring the power back. It was kind of getting to be a pain to get off my bike on a hill to unplug and plug in the battery so I could start going again. I've also noticed since I installed the switch, the cut off feature doesn't kick in as much.
 
izeman said:
ps: i'm really looking forward to see this massproduced. i'll be your first customer! :)

And I'll be the second. :mrgreen:

I mentioned to methods that he should add these to his product line but he didn't show too much interest. I think his plate is about full these days.
 
Back
Top