eBike Master Switch Design

If I can't have them both, then I won't take any at this time. thanks anyway.

OK, no problem.

We should have more in a few weeks.

that would be fine.... i'm a few weeks out anyway.

please let me know when your good to go.

One left.

Why not just share the design via OSHpark and people can order them that way? You wouldn't need to deal with shipping or coordinating it.

I might do that with the older model. But then folks would have to want three boards. If I get a batch I can offer the boards individually at a reasonable price.

Edit - All boards are committed.

Still collecting parts for my Borg Switch. Can't find my stash of 4110's. May have to get more.

I did find my clear "Rescue Tape" which I may use for the "enclosure". This silicone self-amalgamating tape is stretchy and sticks only to itself not with an adhesive, but by melding with itself, and then makes a waterproof bond. It can be cut off later with little mess since it doesn't actually stick to anything but itself. The stretch allows it to be very tight and it stays tight.

http://www.amazon.com/RESCUE-TAPE-Self-Fusing-Silicone-Tape/dp/B00AZOK55G

Weather is improving (and motivation), but work and chores are eating up the calendar.

I'll mail out the last 2 bare boards on Saturday.

Thanks.

My first run board is still running solid. I ended up installing it into a customer bike because he wanted a master switch. Daily use for months, hasn't missed a beat.

johnrobholmes said:

My first run board is still running solid. I ended up installing it into a customer bike because he wanted a master switch. Daily use for months, hasn't missed a beat.

Click to expand...

Thanks for the update John. Excellent test and result. Nothing wrong with the first boards. They just turn off a bit slower.

Making some small progress here. Found my 4110's, and the Mouser order came. Also found my clear Rescue Tape, Silicone insulated wire, PowerPole 75's and resistor assortments. Nearly all the parts.

I think I'm going to try using SMT techniques to flat solder the 4110's to the board. Use board preheater, hot air pencil, flux and solder paste.

The two PC boards are "in the mail".

Got mine... hopefully I have all the necessary bits here and I can assemble soon.

Tonight I set up the SMT hot air rework station and used solder paste to solder the FETs to the board.

It wasn't easy, which I expected, but it may have worked. Three of the four FETs are "stuck" to the PCB so they must have at least partially soldered underneath. The one on the far end from where I was working didn't, which makes some sense as it probably received the least heat.

It is not required for the tabs to be soldered to the board but it is nice if it can be done. I may give it one more try later. I need to find my flux first. That would have helped a lot, this time I was relying on the flux in the paste which is probably not enough for this large an area, plus the paste is getting a bit old.

Progress Update

My switch build got stalled due to other commitments and events. As reported before, 3 of 4 FET tabs soldered to the board, the fourth tab did not solder.

I thought I had ordered more of these V1.1 PC boards, but I can't find any record of that, so that apparently did not happen.

A couple of days ago I decided to try a different design approach for this switch, and this morning I spice modelled the circuit. The models look very good, and I'm inclined to go forward with the new design. We could also make a few more of the 1.1 boards if folks want them, they will be quicker than the new design since the PCB layout will have to be adjusted.

Here are some of the differences between V1.1 and the new design, which I'm calling V2.0:

1) faster turn on and turn off (turn on about 110 milliseconds with 2,000uF), turn off is essentially instant)
2) no difficult to source parts
3) less critical of parts values
4) no hard to get high voltage capacitor to determine soft turn-on time
5) FETs not operated in linear mode - no potential violation of safe operating area
6) precharge resistor (built in) takes the capacitor precharge heating (rather than the FETs as before)
7) slightly higher parts count

I'm probably going to have to give up on the present plastic box, it is not clear that was really working out well with all the wire management and I'll need a bit more space. We can include mounting holes, though putting heatshrink over it as part of the battery cabling should work fine.

I'd like to hear some feedback on this - what would folks like to see here? How are the five V1.1 boards that are in the field now doing? Is there interest in a new V2.0 design? Is anyone interested in finishing off the V1.1 board build I started?

Alan B said:

How are the five V1.1 boards that are in the field now doing? Is there interest in a new V2.0 design? Is anyone interested in finishing off the V1.1 board build I started?

Click to expand...

hi alan
i now did the first test ride with a cut-in-half board. so two mosfets for 2.5kw@48v. everything working fine so far. even with the easier to source transistor. nice turn-on voltage up ramp and almost instant off - fast enough that the CA won't log the low voltage of the cut off.
totally happy as is.

Writing to simply let you know of my interest in getting few of them as they are finalized.
Thanks for the great and continued efforts.

Thanks for the update, and the interest.

I would be interested in a few as well. If that helps. Thanks for your efforts. This looks really cool.

Alan B said:

Here are some of the differences between V1.1 and the new design, which I'm calling V2.0:

1) faster turn on and turn off (turn on about 110 milliseconds with 2,000uF), turn off is essentially instant)
2) no difficult to source parts
3) less critical of parts values
4) no hard to get high voltage capacitor to determine soft turn-on time
5) FETs not operated in linear mode - no potential violation of safe operating area
6) precharge resistor (built in) takes the capacitor precharge heating (rather than the FETs as before)
7) slightly higher parts count

Click to expand...

That's the typical approach for most large EVs and hybrid cars.

One possible issue is if you're using a resistor to precharge, you can't have any load other than the main capacitors or you will never get it to fully charge. This means you can't have the controller 'always on' and just turned of by disconnecting the main power. You also can't have a DC-DC or lighting across the main controller power without a separate switch. In real life, if the resistor value is low enough, you could probably get close enough to full charge that nothing blows up when the switch turns on even if you have some load always on. One of the advantages of the linear rate ramp up was you could still precharge with some amount of constant load.

fechter said:

...

That's the typical approach for most large EVs and hybrid cars.

One possible issue is if you're using a resistor to precharge, you can't have any load other than the main capacitors or you will never get it to fully charge. This means you can't have the controller 'always on' and just turned of by disconnecting the main power. You also can't have a DC-DC or lighting across the main controller power without a separate switch. In real life, if the resistor value is low enough, you could probably get close enough to full charge that nothing blows up when the switch turns on even if you have some load always on. One of the advantages of the linear rate ramp up was you could still precharge with some amount of constant load.

Click to expand...

Precisely. I've been modelling different loads to investigate this effect. By adjusting the charge resistance, and the voltage threshold for switching this can be adjusted to handle light loads. I've also been looking at adding an accessory high side FET to handle these loads and turn them on after charging the capacitor bank. One could also combine the two processes, use the resistor for the majority of the charging, then ramp the FETs on linearly for the small remainder.

At this point I'm planning to make a small run of V1.1 pc boards. That will give folks something to work with sooner.

yes it will not charge the caps up to full battery voltage if a load is attached, but the most DCDC converter for our ebike usage (for example i use a normal 12V 3A 110V ac power supply) will only draw a few mA not much more. And it is better for the FET's to charge for example from 40V to 60V instead of 0 to 60V right?

Alan B said:

I've also been looking at adding an accessory high side FET to handle these loads and turn them on after charging the capacitor bank.

Click to expand...

would this mean there will be an extra connection point for a DCDC on the master switch board?

im also interested in the V2 board. no difficult to source parts sounds good.

If I add a high side FET to the V2 design it would present a switched positive output for the controller logic power and other battery voltage low current loads. This output would not come on until the controller capacitor bank had charged to within a volt or so of the full voltage at which point the FET switches would come on and both ground the controller hard (bypassing the charging resistor) and supplying +battery for controller logic, CA etc. One nice thing about this is the CA would not see the charge/discharge swings, it would snap on and off.

Ill take one Alan if you still have any for sale.

sacko said:

Ill take one Alan if you still have any for sale.

Click to expand...

i can highly recommend alan's board. they work great here.

I need to send out for another run of boards.

Alan B said:

I need to send out for another run of boards.

Click to expand...

Have you continued with improvements? Boards currently available?

Thanks.