DIY Electric Car Forums banner

1 - 15 of 15 Posts

·
Registered
Joined
·
3,219 Posts
Discussion Starter #1
We got a surprise when we placed an order for another 100 IGBT modules back in October - that was when the manufacturer decided to inform us that the lead time had jumped from the usual 4-6 weeks to 20-24 weeks... Nice.

And thus began a long, arduous process of finding some way around this fiasco because my partner, Seb, felt that we needed to do everything possible to get product to our dealers. I, on the other hand, argued that we could end up spending lots of time and money chasing down "genuine" modules from shady excess inventory brokers, or modifying the guts to accommodate different size modules only to find out the "genuine" parts weren't or that the substitutes didn't allow for the same performance envelope of the controller.

Well, some $15K in modules and modifications later, we found a substitute that works as well as the original modules except in voltage drop and EMI (they ring more during turn on). Oh, and they are more expensive, too...

However, the substitute modules do reliably deliver over 1000A into our dyno with a ~300V pack and, more importantly, they survived both operating into an existing short circuit as well as a short applied while running. These two tests are especially critical because I had to tweak all of the desaturation detection parameters to get these modules to work and I was very concerned that I might have rendered desaturation detection useless at that point. Nope. As the next couple of posts with pics will show (with some log file excerpts) the controller still happily regulates current into a direct short (ramping up from zero current, in other words) and happily survives a short across the motor terminals even when it is already delivering 1150A to the motor (the calibration was a little off :rolleyes: ).

BTW - Qer made the testing process a lot easier for me by modifying the code to help me discriminate between legitimate desats (because the modules just sucked) and fake ones, because the modules were just a little slower or rang a little more, or had a higher saturation voltage (or all of the three...). It is very possible I would have dismissed this last batch of modules as unsuitable were it not for him. So, thanks Qer. We'll be sending you more Skittles as a bonus! :D

The first graph shows the Soliton1 ramping current up just under 1000A (967A, the default maximum if throttle isn't calibrated) into a dead short across the motor terminals. The supply voltage is 165V. Oh, and kids - don't try this at home.
 

Attachments

·
Registered
Joined
·
3,219 Posts
Discussion Starter #2 (Edited)
This pic shows the rather uneventful result of applying the short while running. The first is at an indicated motor current of just under 600A (actually 650A because of calibration error) and the second is at an indicated current of just under 1000A (but actually 1150A due to the same calibration error).

We used a single Gigavac contactor "protected" by a 600A Class-T fuse to short the motor terminals together. The fuse never blew but the contactors welded shut on two occasions (destroying the contactor each time, of course). This is not a bad reflection on the Gigavac contactor - it was being grossly overloaded, after all - rather, it should illustrate just how valuable good desaturation detection can be when there is a fatal short in your motor circuit (either because the cables were damaged or the internal insulation failed or a zorch, etc...)

The actual peak current is unknown because logger is about 10,000 times too slow to catch a desat, and our 2kA clamp meter doesn't have a peak-hold function (my Fluke 337 does, but it maxes out at 1kA). Desat is calculated to trip around 1800A both with the original IGBT modules and the substitutes, but not verified in production because this is considered abusive testing which does shorten the life of the module - 100 desats is the lifetime rating for most IGBT modules, btw - so any IGBTs that get tested for desat are then discarded.
 

Attachments

·
Registered
Joined
·
542 Posts
The actual peak current is unknown because logger is about 10,000 times too slow to catch a desat, and our 2kA clamp meter doesn't have a peak-hold function (my Fluke 337 does, but it maxes out at 1kA). Desat is calculated to trip around 1800A both with the original IGBT modules
Divide your cable into two or three identical runs and put the Fluke on one of them.
 

·
Registered
Joined
·
3,219 Posts
Discussion Starter #4
Divide your cable into two or three identical runs and put the Fluke on one of them.
D'oh! :eek:

Can't believe I missed something that obvious! Thanks for pointing that out, GerhardRP, even if I feel like an idiot as a result :D
 

·
Registered
Joined
·
220 Posts
Hi to all,

Has anyone raced with Soliton1 yet?

We need racing controller. There are so many controller that you can use for road use and almost no controller for racing beside Zilla.

We need exposure to create awareness.

Albano
 

·
Registered
Joined
·
363 Posts
I'm currently putting together an MX-5 with a Kostov 11" and a Soliton 1. Though I will be taking the car to the drag strip in spring, just for fun and to verify perfomance, I'm fairly sure that the same motor with a Zilla 2k would beat it.
However I may also manage to get it on the proper race track since I have lots of tie's to the electric racing scene here. I suspect a 1k controller would do as well as a 2k in a 15-20 minite race, where the extra power will just end up heating the motor/controller until it is forced to derate. I'm curious as to how well a series motor can handle this kind of abuse (rather than the usual 10 second power run), and there's only one way to find out!
I also have access to a decent car dyno, and will endeavor to get some results to publish here.

Steve
 

·
Registered
Joined
·
220 Posts
I'm currently putting together an MX-5 with a Kostov 11" and a Soliton 1. Though I will be taking the car to the drag strip in spring, just for fun and to verify perfomance, I'm fairly sure that the same motor with a Zilla 2k would beat it.
However I may also manage to get it on the proper race track since I have lots of tie's to the electric racing scene here. I suspect a 1k controller would do as well as a 2k in a 15-20 minite race, where the extra power will just end up heating the motor/controller until it is forced to derate. I'm curious as to how well a series motor can handle this kind of abuse (rather than the usual 10 second power run), and there's only one way to find out!
I also have access to a decent car dyno, and will endeavor to get some results to publish here.

Steve
Wish you all the best and hope for some good result.

Albano
 

·
Registered
Joined
·
333 Posts
We got a surprise when we placed an order for another 100 IGBT modules back in October - that was when the manufacturer decided to inform us that the lead time had jumped from the usual 4-6 weeks to 20-24 weeks... Nice.

And thus began a long, arduous process of finding some way around this fiasco because my partner, Seb, felt that we needed to do everything possible to get product to our dealers. I, on the other hand, argued that we could end up spending lots of time and money chasing down "genuine" modules from shady excess inventory brokers, or modifying the guts to accommodate different size modules only to find out the "genuine" parts weren't or that the substitutes didn't allow for the same performance envelope of the controller.

Well, some $15K in modules and modifications later, we found a substitute that works as well as the original modules except in voltage drop and EMI (they ring more during turn on). Oh, and they are more expensive, too...

I'm impressed with all your testing and searching for substitute parts. Probably not a bad idea anyway, given any volatility in the electronics market. So what does all this do to your production schedule for S1 or Jr? Will this affect short term availability?

In short, what does this mean for your customers?
 

·
Registered
Joined
·
3,219 Posts
Discussion Starter #9
..So what does all this do to your production schedule for S1 or Jr? Will this affect short term availability? ...
Well, we obviously couldn't make controllers once we ran out of IGBT modules (which occurred in early December) so there was a period where you couldn't get a Soliton1. Thanks to the new modules, production did resume this week. The backlog in orders that built up over that time will take about 4-6 weeks to clear, however.

Junior wasn't affected by this fiasco as it uses a completely different IGBT module (which we bought all of the manufacturer's remaining stock, since it does use the same IGBT chips inside, and it is those chips that are in short supply). We still need to do a little tweaking of the temperature derating and maybe push the amperage envelope a little more, but Juniors are being assembled this week. I'll post some controller pr0n in the Junior's first scream thread.
 

·
Registered
Joined
·
1,956 Posts
That's good news, I had to send my partner "Doug" my Soliton 1 to finish his ev conversion. I guess I will be getting one of these newer models when shipping starts. My Soliton is for my Mitsubishi 3000GT with a Netgain 11" HV motor, contsruction will not start until the Camaro is completed. I am glad you have things rolling again. We are still waiting for the Big Sol.:p

Well, we obviously couldn't make controllers once we ran out of IGBT modules (which occurred in early December) so there was a period where you couldn't get a Soliton1. Thanks to the new modules, production did resume this week. The backlog in orders that built up over that time will take about 4-6 weeks to clear, however.

Junior wasn't affected by this fiasco as it uses a completely different IGBT module (which we bought all of the manufacturer's remaining stock, since it does use the same IGBT chips inside, and it is those chips that are in short supply). We still need to do a little tweaking of the temperature derating and maybe push the amperage envelope a little more, but Juniors are being assembled this week. I'll post some controller pr0n in the Junior's first scream thread.
 

·
Registered
Joined
·
225 Posts
....BTW - Qer made the testing process a lot easier for me by modifying the code to help me discriminate between legitimate desats (because the modules just sucked) and fake ones, because the modules were just a little slower or rang a little more, or had a higher saturation voltage (or all of the three...)....
Does this mean there is a firmware fix coming for the cold weather related desat errors that some of us are experiencing?
 

·
Registered
Joined
·
3,219 Posts
Discussion Starter #13
No, Darren, this thread is about finding a suitable replacement for an IGBT module that won't be available again until sometime in May. If we didn't find a substitute it's probable we would no longer be in business so that was a pretty high priority for me these last few months.

We just learned about the sub-zero desat issue ourselves (literally, all three people reported it within a few days of each other) and are taking it very seriously, but it will take some effort on our part to be able to duplicate the conditions under which it seems to occur. We've bought a deep freezer and modified our dyno setup to be able to run the controller inside the freezer, chilled down to below 0C.

So, like I said in my official Evnetics response to your inquiry of a few days ago, we are working on it.
 

·
Registered
Joined
·
75 Posts
No, Darren, this thread is about finding a suitable replacement for an IGBT module that won't be available again until sometime in May. If we didn't find a substitute it's probable we would no longer be in business so that was a pretty high priority for me these last few months.

We just learned about the sub-zero desat issue ourselves (literally, all three people reported it within a few days of each other) and are taking it very seriously, but it will take some effort on our part to be able to duplicate the conditions under which it seems to occur. We've bought a deep freezer and modified our dyno setup to be able to run the controller inside the freezer, chilled down to below 0C.

So, like I said in my official Evnetics response to your inquiry of a few days ago, we are working on it.
I drove up from Sanibel to Syracuse on Saturday; I should've stopped by St. Pete and offered to bring a Soliton1 up here for testing :)
 

·
Registered
Joined
·
3,219 Posts
Discussion Starter #15
Resurrecting this thread because someone just asked me a couple of questions out of the blue about the short-circuit testing we did:

Q1: How much current can your battery pack [on our dyno] supply?

A1: Based on the dV/dI method of determining total battery pack "equivalent series resistance", our dyno pack can deliver a short circuit current of 2800A.


Q2: Is it possible that your battery pack couldn't deliver enough current to really stress the controller when the output was shorted?

A2: Nope. A few controllers were built with too high a desat threshold and they blew up during short-circuit testing. IGBTs self-limit the amount of current going through them during a short-circuit current (ie - through desaturation) but you still have to turn off the gate drive within a certain period of time or they will blow up anyway (note: MOSFETs do not self-limit during shorts, and that's the main reason why IGBTs are more rugged).

Oh, and Albano... we're making the racing controller you said you needed so badly in this thread:

...
We need racing controller. There are so many controller that you can use for road use and almost no controller for racing beside Zilla.

We need exposure to create awareness.

Albano
See the recent thread referring to "Big Sol".
 
1 - 15 of 15 Posts
Top