eBike Master Switch Design

Update on using the Hammond 1551K enclosure. The board easily fits. I have not been able to locally find the #2 self taping screws to mount the board in the box (they are not supplied with the enclosure). One issue for the box will probably concern the leads to the battery and controller. I was rather generous in the length of the 2 sets of 4 leads to make it easier to solder them to the board. Each set of 4 leads were crimped in a copper collar to a 10 awg lead. Just the volume of the 5 wires crimped into the copper collar is probably too much to fit into the Hammond box. So that means all 8 of the 12 awg leads will need to pass through the lid of the box. It may be possible to get around this with shorter leads/different methods of connecting the leads. If this issue isn't clear, it should become clear when I can get to posting a photo.

In view of the unexpected behavior of my build of the board upon shut-off, is anyone else also building the board with a Supertex TP2104N3-G for Q5? It seems like that would be useful to see if this was just something peculiar to my setup and/or build. I know that izeman has had difficulty in obtaining that part in Europe. I have an extra of that FET and could possibly mail it to someone. Let me know if you are interested.

Rich
 
Thanks for the updates. I have not been able to analyze in detail the results you have seen. Which main FETs did you use?

I would expect the wires to be brought out of the box, there is not much room to combine them internally. I would combine them into a connector like a PowerPole 75 so there are no extra junctions.
 
I used 4 x IRFB4110 main FETs. On my bike I rarely exceed 2 kw, so I am not anticipating any heating problems for the switch.

Rich
 
For all components, I tried to use what was specified by the BOM. It is of course possible that I didn't understand something. So I have listed the BOM below followed by the parts I used. If anything isn't clear, I can include the Mouser part number.

R1,R2 1M - 1 Mohm metal film resistor, 1%
R3 1K - 1 Kohm metal film resistor, 1%
R4-7 1K - 1 Kohm metal film resistor, 1%
C1-3 1uF (NonPolar) - One 1 uF MLCC leaded capacitor at the C2 position
D1 12V zener - 12 volt 1.3 watt 2% vishay zener diode
D2,3 1N4148 - Fairchild diodes, IN4148
Q1-4 Main FETs, nominally for 90V: IRFP4110PBF - 4 x IRFP4110
Q5 TP2104N3-G - TP2104N3-G

Rich
 
Look like good values. I'll review the symptoms and the circuit soon and see what I find, though it may take a little time.

Offhand, I would consider taking R3 to a higher value and see if that changes things, make it 4700 or 10K perhaps, if you wish to try something. It is not critical, just several factors higher than the 1K it is now. You might measure it and make sure it is 1K, that should be easy with a cap on one side. Do not subject yourself to the high voltage, disconnect before testing.

Also make sure there isn't anything wonky going on with the control line. No capacitors on that line, just connects to B+ or open, nothing else in parallel with it that might backfeed some voltage into it, etc. This is a high Z circuit, it won't take much to mess with it.

Thanks for your effort.
 
nice to see a finished board. what cross section do these wires have? it looks a bit overkill :)

what about putting the board in a bigger box with waterproof wire outlets?
 
12 gauge wires, four in parallel for input and same for output.

Wires are part of the heatsink and keeping resistance low.

A bit of overkill needed to handle big currents, this setup probably good for 60A continuous and 120A peak, possibly more. :)

Trying to keep it small and light, this box was recommended by a customer and it didn't require a lot more board area. Each increase in board area drives price up. Could always put the same board in a larger box. Lots of possibilities for repackaging.
 
i tested the board today. as i couldn't find the suggested part on page #7, i tried 2N3906 with resistor values as stated on page #7 and attached it to a 12s lipo battery - 47v.
ramp up was very nice. around 1s as soon as i connected pin2 to battery voltage.
removing pin2 from battery+ let the voltage drop slowly. no immediate voltage drop to zero as expected.
second strange thing that did happen: connecting the circuit to battery GND only made is switch to ON. positive wire was not connected to anything. and battery GND was not attached to anything but the voltmeter. i have no idea how this circuit can switch to on without supply voltage??!?
 
Thanks for making the test.

Testing with only a meter doesn't tell you much. The impedance of the meter can be so high that tiny leakage currents can still present full pack voltage. Need to put a load on to make it work properly. Even a 10K resistor across the meter.

Too bad the PNP didn't work. It looked like it could, but it depends on a low leakage PNP with enough gain. The FET has a lot more gain.

What values did you use for all the parts? Just to understand the test correctly and record it here.
 
Alan B said:
Could always put the same board in a larger box. Lots of possibilities for repackaging.
Click to expand...
These boxes are available in a few different depths. I don't know which one is pictured, but maybe the wires would have enough room in the next deeper box but still fit the board the same way.
 
Alan B said:
What values did you use for all the parts? Just to understand the test correctly and record it here.
Click to expand...

Code: 
R1,R2 1M nominal 1/2w
R3    1K
R4+7  1K
C1-3  1uF (NonPolar) 100v - one C installed on the upper side of the board @C1
D1    12V zener
D2,3  1N4148
Q1-4  Main FETs, nominally for 90V: IRFP4110PBF - installed @Q1&Q4
Q5    2N3906

i behaves "not bad" , just like every other fet-power-switch with similar design i built over the last 2 years. i just don't can't see any fast voltage drop, and i see an ON situation w/o applying any power. i will test this with controller as well, and see.
 
OK. Testing with a many million ohm load (meter on high voltage range) is just not a good test. A controller turned off generally is a 10K ohm load. Solid state switches aren't quite like open contacts.

Also, we discussed in the thread that for the PNP test some values should be changed. If you can try those changes the PNP may work. R1/R2 should be lower, try 100K. This should be tried before we give up on the PNP.
 
Alan B said:
R1/R2 should be lower, try 100K. This should be tried before we give up on the PNP.
Click to expand...
do you mean the relation between R1 and R2 should be lower? or that both' value should be lower, so both are 100k? just to be sure ;)
i will then connect the controller and see how it behaves.
one more question: pin2 only connects to BATTERY POWER to switch on, and is disconnected to switch off? it's NOT connected to GND or something to switch off. correct?
 
changed R1 and R2 from 1m to 56k, as i coudln't find any 100k. result is the same as before.
voltage raises from 0 to 47v within a second(?) and drops from 47v to 3v within 3 seconds.
anything else i could try? at least i can confirm that there is no spark when you connect the battery with pin2 on battery plus - which starts the circuit when battery is connected.
 
I changed R3 to 3.3K as suggested to address the shut-off issue that resulted in periodic flashing of the Cycle Analyst screen after turning off the master switch. Before changing R3, I counted the number of screen flashes over 10 secs and determined that after turning off the master switch, the screen was flashing at about 2/sec. I also checked the value of the initial R3 and it was 1K, as intended. After changed R3 to 3.3K, there appeared to be no change, frequency cycle analyst flashing after power turn off was still about 2/sec. However, I am fairly sure that each "flash" was longer. I am guessing that this is not what you would expect. I could try going to a 10K R3 if you think that is worth trying. Or would it make more sense to try a lower valve for R3 hoping for a shorter flash duration?

It is probably also worth noting that with the JST-XH connector that I used for J1, the lid would not fit on the Hammond box (this is independent of the issue of the battery and controller leads). There may be a lower profile connector that could be used. Of course, wires could simply be soldered directly to the board and if a connector is desired, an in-line connector like a jst-sm could then be used outside of the box.

Rich
 
I have a Fluke 117 which although I have never paid attention, I think has a bargraph which is supposed to respond relatively rapidly to voltage changes. So if I just monitor voltage at the connection to the controller using the Fluke 117, it should yield the kind data that could be helpful?

Rich
 
You must log in or register to reply here.

Similar threads

X
Converting a dual motor scooter from 2 controllers to 1 temporarily. (kyesang ESCs)
Replies
1
Views
250
chuyskywalker
chuyskywalker
thorlancaster328
Designing the ultimate (400A+) motor controller - what features should it have?
Replies
11
Views
2,980
SammyMc
S
J
Integrated, Scalable Solar/Battery/AC Charging/Inverting/EV DC Charging and More
Replies
1
Views
1,942
borrelan
B
O
EBiCS Firmware on a M365 STM32 Controller
2
Replies
46
Views
8,902
M0g13r
M
Beachcruzer
Craigslist Yescom - Did I score or get screwed?
Replies
13
Views
1,525
999zip999
999zip999
Back
Top