Nevilleh's BMW

Post up a thread for your EV. Progress pics, description and assorted alliteration
HeadsUp
Groupie
Posts: 281
Joined: Sat, 28 Feb 2009, 09:13
Real Name: Mark W
Location: Sydney

Nevilleh's BMW

Post by HeadsUp »

weber wrote:

You really shouldn't leave scorch marks (carbonised epoxy) bridging across high voltage areas. Do you have a small drill that can take a grinding type bit to grind it away, or you could maybe just drill a row of overlapping holes. It only needs a line a few mm wide cutting through the black. Maybe even sandpaper would get it off.


good point weber

didnt think of that Image

the sharing in this place tis good
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Nevilleh's BMW

Post by Nevilleh »

The busbars are just Aluminium. Be quite difficult to drill more holes, even though it is only a double sided board. There's not a lot of room between screws, but it could be done with a bit of care. I'd have to cut away the B- copper plane to clear the screws. I had thought to shift it to the other side of the board, but the caps get in the way.

I don't know where the thermistor was mounted originally as they have not fitted one. There is a 2-pin header marked on the overlay at the end of the board and it just has a 50K resistor soldered in there instead. I wonder why they did away with it?

There a quite a few odd things! Another example is the first regulator. It comprised an IRF mosfet (200V/16A) with a 26V zener connected to the gate. It is only supplying the 15V regulator (gate drive volts) and the 12v regulator for the controller board, only about 20 mA in total. I couldn't get a replacement mosfet (even IR had never heard of it and the alternative they suggested was some $6 plus $10 postage!) so I changed it to a 200V bipolar transistor and it works very well.
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Nevilleh's BMW

Post by Nevilleh »

HeadsUp wrote:
weber wrote:

You really shouldn't leave scorch marks (carbonised epoxy) bridging across high voltage areas. Do you have a small drill that can take a grinding type bit to grind it away, or you could maybe just drill a row of overlapping holes. It only needs a line a few mm wide cutting through the black. Maybe even sandpaper would get it off.


good point weber

didnt think of that Image

the sharing in this place tis good


Yes it really is!
I had a go at the scorch marks with a rotary wire brush - incidentally, it took all the silcone rubber off beatifully - but it wouldn't remove any more than you see now. Certainly my trusty ohm-meter doesn't show any conductivity. I think it is more surface marking than anything.

I have a couple of little things to finish off on it and then I'll try powering it up and see waht happens.
User avatar
Johny
Senior Member
Posts: 3749
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Nevilleh's BMW

Post by Johny »

Nevilleh wrote: The busbars are just Aluminium. Be quite difficult to drill more holes, even though it is only a double sided board. There's not a lot of room between screws, but it could be done with a bit of care.
That kind of rules out brass screws.
How about replacing them with bigger ones?

Don't underestimate the power of the carbonised PCB - if it becomes conductive at higher voltage or high humidity it may avalanche. It really is worth drilling it out and plugging the holes with something non-conductive.

Again, easy for me to type...
HeadsUp
Groupie
Posts: 281
Joined: Sat, 28 Feb 2009, 09:13
Real Name: Mark W
Location: Sydney

Nevilleh's BMW

Post by HeadsUp »

Nevilleh wrote:

I don't know where the thermistor was mounted originally as they have not fitted one. There is a 2-pin header marked on the overlay at the end of the board and it just has a 50K resistor soldered in there instead. I wonder why they did away with it?

.


if you were to put a switch in series with that 50 k resistor then power up the controller at lowest possible input voltage , will controller output switch on or off as you flick the switch ?

wondering if the thermistor / sensor influences the logic board powered on state directly or if there was a secondary circuit with relay to do it. ?


Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Nevilleh's BMW

Post by Nevilleh »

Bigger screws is hard too! I would still have to cut away the copper around the holes to clear the ground plane and there is not a great deal of room alongside the caps. Here's the board with just the capacitors and busbars fitted so you can see how it works.

Image

The thermistor looks like the bottom leg of a voltage divider connected to the - input of an LM258 section. The + input goes to the 4.7v regulated supply. I'd say when the resistance drops far enough, the output of the 258 toggles and this wanders off to what is probably the current limit control.
I'll try your suggestion of connecting a switch or something to it and see what it does. I think I can power it up initially from my 35V adjustable current limit bench power supply and check out all the controls. I hope to try that tomorrow. (Its nearly beer o'clock here).
HeadsUp
Groupie
Posts: 281
Joined: Sat, 28 Feb 2009, 09:13
Real Name: Mark W
Location: Sydney

Nevilleh's BMW

Post by HeadsUp »

Nevilleh wrote:

The thermistor looks like the bottom leg of a voltage divider connected to the - input of an LM258 section. The + input goes to the 4.7v regulated supply. I'd say when the resistance drops far enough, the output of the 258 toggles and this wanders off to what is probably the current limit control.


you reminded me that thermistors / thermocouples come in two and three wire configuration ( 4 even ) so wasnt sure how that would be set up

cheers.
User avatar
Johny
Senior Member
Posts: 3749
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Nevilleh's BMW

Post by Johny »

Nevilleh wrote:... Here's the board with just the capacitors and busbars fitted so you can see how it works.
Thanks - that's a good pic. I see what you mean.
User avatar
avolt
Noobie
Posts: 19
Joined: Thu, 22 May 2008, 23:29
Real Name: Andrew
Location: N.S.W Australia

Nevilleh's BMW

Post by avolt »

Here are some photos of my near finished power board for the Revolt controller,thought you might like a look!

Image

Image
HeadsUp
Groupie
Posts: 281
Joined: Sat, 28 Feb 2009, 09:13
Real Name: Mark W
Location: Sydney

Nevilleh's BMW

Post by HeadsUp »


looks sexy enough

now if only somebody would do an AC one
Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Nevilleh's BMW

Post by Tritium_James »

HeadsUp wrote:now if only somebody would do an AC one

http://www.tritium.com.au/products/TRI74/index.html   Image
HeadsUp
Groupie
Posts: 281
Joined: Sat, 28 Feb 2009, 09:13
Real Name: Mark W
Location: Sydney

Nevilleh's BMW

Post by HeadsUp »

Tritium_James wrote:
HeadsUp wrote:now if only somebody would do an AC one

http://www.tritium.com.au/products/TRI74/index.html   Image


tell em the price son ;)

$6600 AUD

released to market as of today ?

congratulations

hate to think how many 1000's of hours work in those .
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Nevilleh's BMW

Post by Nevilleh »

That Cougar board looks very nice, I have one on order.

James's ac controller looks even better! My next ev will definitely be ac powered, but got to get this one finished and reliable first. Maybe the price will have come down a bit by the time I want one......

edit: It works!
Have just now powered up my rebuilt controller, running from my 35V bench power supply (with the primary regulator fiddled so it'll work down to this voltage - normally needs over 100V). I have a couple of 12V quartz halogen lamps connected in series for a dummy load. Throttle pot varies the brightness very nicely. Now to fiddle with things like the thermistor input and maybe play around with the drive resistors.

Here's the gate drive waveform (top trace) and the resultant load waveform. The thing is drawing just 1 amp from the supply. I'm wondering why the load waveform has such rounded corners? I stuck a 100 ohm resistor across one gate drive resistor and it sharpened up the falling edge corner very noticeably. Also, shorting out the thermistor input dropped the drive frequency from 13.6 kHz to 1 kHz. Not at all what I expected. The current drain from the power supply didn't change, nor did the lamp brightness by any appreciable amount.

Image
Last edited by Nevilleh on Tue, 13 Apr 2010, 05:46, edited 1 time in total.
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Nevilleh's BMW

Post by Nevilleh »

Here's a nice picture of the delay caused by the long pcb traces on the LS power board. The top trace is right at the output of the driver chip, the bottom one at the drive end of the gate resistor on the IGBT furthest away from the driver chip. I'd be interested to know what anyone thinjs of this. Particularly T-James, who seems to have no small experience with these things!



Image
User avatar
Johny
Senior Member
Posts: 3749
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Nevilleh's BMW

Post by Johny »

I have a theory. The drive to the IGBTs is so slow, caused by the 200 Ohm resistors, that the delay doesn't matter almost at all. If you speed up the gate drive by lowering the resistor values then you really will have IGBTs turning on 10 to 20nS before other ones rather than them all "ramping up" together.

Now, which is worse?
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Nevilleh's BMW

Post by Nevilleh »

Yes, I like that theory and had thought that myself. The percentage difference - ie 15 or 20 nS in a couple of microseconds is really very small.

This is the load current waveform (bottom) referenced to the drive pulse.
As you can see, it is just about 2 uS after the leading edge of the drive pules before the current reaches maximum and the turn off time is 3 uS after the end of the drive pulse.


Image
Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Nevilleh's BMW

Post by Tritium_James »

I'm actually kind of surprised at the gate drive signals, it's got a 40ns transition at the gate, which isn't really that slow. It's probably OK.

The rise/fall time of the output current probably has as much to do with the inductance of your wires/light bulbs as it does with the gate drive signal, so I'm not sure you can read too much into that one.

Dropping the switching frequency when over temp makes some kind of sense - especially if (due to the slow switched edges) most of the loss in the controller is switching loss. 15kHz -> 1kHz = 15x less switching loss in the controller, which isn't trivial!
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Nevilleh's BMW

Post by Nevilleh »

I can get a significant improvement in switching times by dropping the gate resistor values - both turn on and turn off and I can also see faster rise and fall time in the load current. I have quite long wires connecting the load and I tried coiling them up into about 10 turns while it is running to see if I could see any difference from the increased inductance, but the waveform didn't change at all. I can't really decide whether to modify the gate drive circuitry or just put it back in the car as-is and see what it does. They are supposed to have improved the design since the spate of blow-ups and I think that improvement might be going from MOSFETs to IGBTs in the power unit. They are now 600V parts, including the diodes. Incidentally, I didn't need to change those, they stood up to the abuse OK. I tested them with a 50V power supply and they did their diode thing quite happily.
User avatar
Johny
Senior Member
Posts: 3749
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Nevilleh's BMW

Post by Johny »

The rise times of IGBTs increase a lot at full load.
The IGBT that you are using has a near 10nS rise time at 15A going up to 40nS at 60A. The delay time is also dependent on current.

I can understand your wanting to change the gate resistors since the spec. sheet only shows data from 10 to 100 Ohms.
How about a compromise at 100 Ohms?

When you checked the controller after your 20 minute drive, was it very hot?
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Nevilleh's BMW

Post by Nevilleh »

I can't puch much current through the hings on the bench here, I just don't have a big power supply, nor a proper load, so its hard to get realistic measurements. I like the 100 ohm compromise, I can't see it doing any harm and it a do a lot of good.

It wasn't very hot after my 20 minute drive. I could keep my hand on it, so probably around 40 - 50 deg or so. It had the fans running the whole time.
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Nevilleh's BMW

Post by Nevilleh »

The thing is now re-assembled into its case, heatsink and all. The end caps were moulded/poured resin and the bottom plate was also stuck on with resin. It can go back with 4 screws but I basically had to destroy the end plates to remove them, so the only job left is to fabricate some new ones - maybe a bit of pvc sheet. I will definitely make them removable and Mr Murphy will see that and ensure that I never have to take them off again.

edit: I have changed all the gate drive resistors to 100 ohms and I have 24 IGBTs instead of the original 20 - mostly because I ordered 24 thinking that a couple of other transistors were just more IGBTs and there is space there, so the more the merrier. Should handle 1200A at 600V now.

Doesn't it look pretty? Neville's repair shop for dc motor controllers is now open for business Image

2nd edit: just hold that last until I see that it works in the car!!!
Image
Last edited by Nevilleh on Wed, 14 Apr 2010, 10:53, edited 1 time in total.
User avatar
Johny
Senior Member
Posts: 3749
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Nevilleh's BMW

Post by Johny »

Nevilleh wrote:Doesn't it look pretty?
Yes, good work.
Neville's repair shop for dc motor controllers is now open for business
Good, let's see if it's going to be profitable.
1. Numbers of hours to repair controller x hourly rate
2. Parts

Oh, oh Image

And where is your sense of adventure? Fill the thing full of expanding foam!
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Nevilleh's BMW

Post by Nevilleh »

Johny wrote: [
Good, let's see if it's going to be profitable.
1. Numbers of hours to repair controller x hourly rate
2. Parts

Oh, oh Image

And where is your sense of adventure? Fill the thing full of expanding foam!


Damn Killjoy! If I could find someone to pay for 1 and 2 above, I'd only need the one job and I could retire (again).
I am definitely not confident enough to use expanding foam though.
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Nevilleh's BMW

Post by Nevilleh »

Fitted the controller back in the car today and it works. Didn't have time to fiddle with the adjustments and it is producing a fairly violent, abrupt takeoff that spins the wheels, but it is looking quite hopeful. I'll get it up on the jack and see if I can set it for something a little more moderate.
User avatar
Johny
Senior Member
Posts: 3749
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Nevilleh's BMW

Post by Johny »

Good news. What did you do about the end caps?
Post Reply