Friday, February 10, 2012

Battling a Frame Leak

This is a god-awful long, and most likely boring story.  Once again I'm hoping I can serve as an example to all you fine readers,  for what not to do.

About two weeks ago I pulled the Z3 into the garage and plugged it in.  When I flipped on the charger I heard, what sounded like, a popping noise coming from the front of the car.  It sounded like the noise you hear when you plug a speakers into the audio port on your computer, deep and quick.  But there were some kids playing out front and I wasn't sure exactly where the noise came from.  But the charger continued to charge and everything on the car checked out.

The following day I went to plug it in after another day's driving and I listened closely, this time with the garage door closed and the hood up.  Sure enough, I heard the same pop, only this time the charger turned itself off.  You may remember about a year and half ago, I was measuring one of the batteries during a charging cycle when one of the probes slipped and made contact with the chassis and the terminal at the same time.  There was a pop, the end of my probe was vaporized, the charger shut off and the breaker for the 240 volt AC outlet tripped.  That little mistake cost me $150 because I had to send the charger back to Manzanita to repair the AC rectifier which I blew up in the process.

After this recent pop I figured 1. I must have a high voltage leak to ground and 2. I'm going to have to send my charger along with another $150 to Manzanita.  I decided to check for the leak and got out my multimeter.  Sure enough, if I measured at the most positive terminal on the battery pack, and the chassis, I could measure the full 160 volts of the pack.  Damn!  I went to the fourth battery from the end of the pack and measured there and got the expected 13.2 volts.  I wondered how much current could get through this leak, so I clipped a regular automotive tail light bulb to the positive terminal of that battery and the chassis and it glowed nice and bright.  If there were no leak, or no path for current, that bulb would not light up.

OK, it was time to start disconnecting things to isolate where the leak was.  This should be a pretty straight forward task, and it seemed to be at first, but it wasn't long before it got weird.  I'll explain.  I started by unhooking all the peripherals, one by one, using the multimeter after each to see if I had voltage to ground.  I disconnected the DC to DC converter, the charger, the Masterflux A/C system, the ceramic heating element and finally the Link-Pro meter.  By this time only thing hooked to the high voltage system where the Zilla controller, the motor and the batteries them selves, but still the leak was present.  I unhooked the cables leading from the Zilla to the motor and the leak was gone.  Ah ha, found it!

My guess was that through poor design, I'd placed a cable close to something sharp and it had rubbed through and shorted to the chassis.  In deed, I found what looked like a suspicious wear mark on top of one of the rubber boots, under which is one of the terminals of the motor.  But as it happened, it had not worn through, though it would have eventually.  But then I noticed that I'd done something else that was remarkably stupid.  The main high voltage fuse of the car, which is held in a special holder which does not keep the fuse from sort of sliding one direction or another, had migrated one direction and looked like it was touching a part of the chassis.  Well, there you go!!!

I fixed that problem and insulated the connection so that it can creep all it wants and will never short against anything again.  I checked all the other cables and connections and found them good, so I started putting everything together.  I got everything back together, put the multimeter on the car and... I could still read the pack voltage.  Ah crap!  But this time, there was an important difference.  When I performed the light bulb test, there was not enough current flowing through the chassis to make the bulb light up.  At this point, I'm thinking there must have been two leaks.  I'm assuming I fixed the more serious one, but there's still a smaller one.

Now's where things get strange.  I started testing again, trying to be as methodical as possible, but I kept getting strange results, or results I didn't expect and couldn't explain.  For a while I was certain there was a leak through the Zilla, but that wasn't the case.  You see the problem is that I've run up to the end of my knowledge at this point.  When it comes to electrical stuff and electronics, if I can't SEE it, I'm probably not going to understand it.  I can SEE a wire touching the frame.  But once that wire enters a device, as far as I'm concerned what goes on in there is simply magic.

I'll save you some of the wretched details, but suffice it to say after disconnecting everything one at a time, again, I was left with only the batteries connected to the Zilla and I still had a leak.  But then I saw that I still had the current sensing leads from the Link-Pro connected to the shunt.  I thought well that can't be the problem, but I'll disconnect them to be sure.  Lo and behold, the leak went away.  What the hell!!  Maybe that line got pinched or abraded.  Nope, it looked fine.  The meter is supposed to be completely isolated from the chassis because I've used an isolating DC to DC converter specifically to power the meter.  I measured the 12 V output of that DC to DC converter and found that I could read 12 Volts if I grounded to the chassis.  Well that's not supposed to happen!  There's my problem!!  Or so I thought.

I started putting everything back together, one thing at a time, taking measurements with each connection made.  For kicks I connected the Link-Pro's shunt lines, expecting to see the leak and it wasn't there.  Ah, for crying out loud!  In fact, I measured the output of the Link-Pro's DC to DC converter power supply and it now no longer has any connection to ground.  Still scratching my head, wondering what the hell is going on, I connect the motor back up and the leak reappears.  Right about now, I'm prepared to burn the car to the ground and walk away.

Ignoring the Link-Pro issues for now, I focus on the motor.  I took the Zilla out of the loop and had simply a small pack of 12 cells and the motor ready for testing.  I hooked the negative line from the battery to the negative input of the motor, which happens to be labeled 'A1'.  Then I put the multimeter on the positive end of the battery and the car's chassis and I could read the battery's voltage.  That means there is a path for current from the battery, through the motor, into the chassis and back to the battery.  I measured the resistance and found that it was 1.19K ohms.  Now that's not much, and if I plug that into an Ohm's law calculator taking into account the full 160V pack voltage, that means that leak can pass only 0.135 amps, or 135 milliamps. 

I don't think that's normal.  I believe that the high voltage fields are supposed to be completely isolated from the case of the motor, but I could be wrong.  I know that leak hasn't existed in the past.  Truth is, I have no idea what might cause that, but I have emails in to some experts who hopefully will help me with that.

The good news through all this is that at one point I had the battery pack isolated from everything else and was able to connect just the charger to it.  I crossed my fingers and threw the switch.  It came on, stayed on and was charging the batteries perfectly.  No need for repairs there.

As for the Link-Pro and it's DC to DC converter, I have no idea why it would pass through the high voltage at one point and not another.  I'm wondering if that little converter has capacitors that were storing the power.  I really have no idea.  Again, as far as I'm concerned, that thing is magic.  But I did order a replacement DC to DC converter for a whopping $8.50 that I'll keep handy in case I need to swap out the other one.

If you made it though this and are still awake, congratulations!  If you have any words of wisdom you'd like to share or guidance to offer, I'm keen to hear it.  Just leave it in the comments.

11 comments:

Alex said...

Hi Tim,

I don't have any solutions to offer you, but don't get discouraged. Keep at it, and with enough fresh looks, the answer will come to you.

In other news, Jack seems to have found a gadget that may be better than the Xantrex, though the company he bought it from (Lightobject, maybe?) isn't supplying that model at the moment. Sorry, don't know many of the details about it, but Jack does mention it on the blog and in the latest show.

Wish I could be there in sunny Chandler to help you figger this one out! Hang in there!

Alex

Alex said...

Hi Tim,

I don't have any solutions to offer you, but don't get discouraged. Keep at it, and with enough fresh looks, the answer will come to you.

In other news, Jack seems to have found a gadget that may be better than the Xantrex, though the company he bought it from (Lightobject, maybe?) isn't supplying that model at the moment. Sorry, don't know many of the details about it, but Jack does mention it on the blog and in the latest show.

Wish I could be there in sunny Chandler to help you figger this one out! Hang in there!

Alex

Tim Catellier said...

Thanks for the encouragement Alex. I'm sure I'll get it sorted out eventually, but it's frustrating to have my primary means of transportation of line. It would be great if you were here to help out. Having the extra brain in the garage would prove quite useful as mine seems to fall just short of the task at hand.

I did see that meter Jack was talking about and it looks quite promising. Hopefully someone can find out if they're still available from somewhere.

I did order one of the other meters he talked about a couple weeks back that allows you to control a relay based on voltage input. From the moment I saw that I thought it would be perfect to use as a backup monitor to the charger. Meaning, if something goes wrong with the charger and it doesn't kick off when it should, the meter would act as a watchdog, and shut down the charger if the voltage gets too high.

I'm going to have to figure out how to program it and how to work it into the car, but that should be fun.

I'll keep everyone posted.

Steven said...

Hi Tim,

Depending on the voltmeter you are using it may have an rather high internal impedance. High impedance is nice not to cause disturbances in electronic circuits when measuring voltages, but it may also be rather difficult to asses if there's a current leak to chassis by just probing with a voltmeter alone. Next time you may find it handy to add a parallel resistor over the multimeter inputs. Make sure the resistor voltage & power rating is suited for the full battery voltage. An initial guess for me would be to use a 220kOhm 2watt resistor might help to isolate current leaks. Some multimeters (usually expensive models) also include this option to dissipate small quantities of stray capacitance or minute current sources.

//Steven

Tim Catellier said...

Steven, That is just beyond the grasp of my understanding, but I think I get the gist of what you're saying.

So far I haven't measured any current flow to the chassis directly, other than the first day of being able to light a light bulb with it. And even then, I didn't put a meter on it to see how much current it was drawing. But now that I've fixed the main leak, there's not enough current to light the bulb.

What I've done is to locate where the leak is occurring and measure the resistance of that leak. Then taking that number, and plugging it into Ohm's law formula I'm getting a value that should match how much current that connection can pass.

There's no question that I can measure the pack voltage on the chassis, when the motor is attached, and it goes away when I disconnect the motor. What I've been testing is the resistance between the motor's case and one of the terminals. Normally they are isolated from one another, but if I test them now with the ohm scale, I get that 1.18K ohm reading instead of an "OL".

In theory, that could pass 136 milliamps, but I haven't measured that directly. What I do know is however much current is being passed is enough to disrupt the charger and cause it to trip it's breaker.

Thanks for the advice.

Steven said...

Hello Tim,

If the motor truely is conducting to chassis it has become a safety hazard! Please make certain all your ground points for the EVSE, charger & charger port are in well working order. The 'manzi' charger effective puts the 240VAC mains un-isolated onto the batteries. (you already knew that I think) This leaves NO room for further fault mitigation.

Rinsing the motor may help to remove carbon deposits that are causing the current path. Ask Netgain what solution can be used to removing the deposits.

Hope you'll get the Zed on the road asap.
//Steven

achims E-Umbau said...

Hello Tim,

i have the same problem that you have!
And i did the same you did. First i changed from the 1:5 prescaler to the 1:10 prescaler. After that i have had a short circuit and the fuse of the 240Volt line was gone. I written the circumstances in my blog. At this time i bought a new DC-DC-converter (100% galvanic isolated) instead of the IOTA DC-DC converter wich i had used until that happens. Than i´ll put a second, smaller DC-DC converter for the supply of my e-xpert-pro, this converter sucks in two or three days my 12 Volt battery empty.
At Jack Rickart Blogg i found another solution to this problem.
But back to our MAIN problem, the leak.
In my Audi all connections are in save, certificated electric boxes. But i measured also 150 Volt between chassis- ground and Battery. I can´t believe it.
Zilla controller should be ok. Between B+ and B- is on closed connection onnly 12 to 14 Volt. Same on M+ and M-.
But what Steven said here in the comments, it can be that the motor have some rinsing through carbon deposits in the motor, is very interesting. I will try to clean the next days my motor with compressed air through every hole i found on the motor, and measure again.
Achim

Tim Catellier said...

Achim, I'm sorry to hear you're going through the same thing. Hopefully we can learn something from each other.

I tried blowing the motor out with compressed air and was surprised to see how much carbon dust was blown out into the air. I have a fan set up to constantly blow air into the motor at a rate of about 110 CFM, and my assumption all along was that would be sufficient to blow most of the carbon dust that wore off the brushes out of the motor.

Unfortunately, blowing the motor out doesn't seem to have had any effect at all. The resistance is still the same. It's possible that I simply can't get the air nozzle in the proper angle or place to make it work. Hopefully you'll have better luck. Keep us informed!

Tim

achims E-Umbau said...

Hi Tim,
today i tryed to clean my motor with air.
And it helps, now i have only 10,56 Volt and 0,98 Amps measured between Battery plus and chassis ground. The resistance these point is now 589K-Ohms.
And at second i found that the SSR-relais was leaking too. So it is always not one piece alone wich is faulty.
What kind off fan do you use for blowing air into the motor, is it the cooling kit from EV Source?

Achim

Tim Catellier said...

Achim,

Glad to hear the situation has improved for you. I also had two leaks, which made it a little difficult to track down, but mostly it was just disappointing. When I fixed and corrected the first one, I thought I was done, but it turned out there was one more.

The kit from EVSource is what I'm using for motor cooling. Well part of it at least. When I was building the car Ryan offered the fan by itself. I custom built all the shrouding and duct work I'm using.

Hopefully it won't be long before both of are cars are back on the road!

Unknown said...

Warm Greetings!




Today, I visit your website and after reading your blog i realize that it is very informative. I'm highly impressed to see the comprehensive resources being offered by your site.


Thanks and Regards



electric motors