How critical is the inverter pump?

This was also posted to the Prius_Technical_Stuff yahoogroup.
.
I've been investigating some inverter temperature stuff. I don't
think anyone's determined the right incantation to make a Scangauge
display inverter or motor temps yet, so I've brought out taps from
the GIVT and MIVT temperature-monitor ECU leads for the MG1 and MG2
inverters respectively which produce a voltage somewhere between
4.5 at "cold" and 2.something at "fairly hot". Similar inverse
indication as a NTC thermistor with a pullup, but linear rather
than having the odd curve that the ECT lead does, and it's an actual
electronically-driven output rather than a raw sensor. I assume
they're derived from those extra leads right into the transistor dies
that sense heat. I can't really tell where either xIVT lead goes on
the inverter board as they come off the connector and quickly dive
into the inner layers.
.
I divide them by 3 and switch them into the same ECT linearizer
nonetheless, figuring it's a convenient way to just get a relative
reading even if it's not dead accurate. I want something I can
check as an overheat warning on hot drives. But I thought it
would have slightly better range than I've been seeing so I went to
investigate more closely this evening. I set up to monitor temps
about three different ways: the AE laptop doing active queries to the
hybrid ECU, a voltmeter hung straight off the outputs, and my little
panel circuit. I wanted to see a larger range, and thought okay, how
does one get the inverter decently warm fairly quickly? Obvious
thing to try: pull the coolant pump power and then carefully send
some current through the transistors.
.
We've heard plenty of stories of peoples' Priuses dying on the side
of the road after losing the inverter pump. Well, a careful listen
to my five year and 100K-plus old unit tells me that my own pump is
starting to get a bit of bearing noise and probably won't last
forever either, and it's time to spring for a new one to keep ready
to swap in. But in the meantime, there's some playing to be done.
After pulling the power and hearing it go silent, I tried a very
careful little bit of force-charging while hawking GIVT. From a
little more than 40C, mostly coming from existing underhood temps,
it bounced up a little but still settled back down quickly after I
stopped. I tried a little more. I tried watching MG2 and dragging
the car back and forth a bit against the parking brake, sending 20
amps through the big guns for the length of the driveway, back and
forth. Those temps crawled up a little but even with the coolant
loop still *not* running, it was clear that transistor heat was
rapidly dissipating into the big surrounding block of metal inverter
case. In fact the rear brake drums were getting considerably warmer
than the inverter from all that.
.
I went for a drive. MG1 works the hardest under normal street travel
if you aren't demanding a lot of electric, and MG2 gets occasional
bursts on braking events, but while both were pushing 70 C after a
while [with the engine right next to it up at full running 85 C
temp by now], neither figure seemed to be getting to truly dangerous
levels. Under normal driving in ~ 70F ambient temps, it seems that
the coolant pump is less important than any of us thought. Not to
the nonessential level -- it's still important to have the cooling
capacity for those switching elements... but having it fail does
*not* spell instant inverter death and crispy-fried transistors,
and one could probably tool along in not-too-hot ambient weather
for quite a while without ever realizing the thing wasn't running.
.
I wound up stopping by a local cruise-in, whose organizers reassured
me that there was certainly a place for a new and interesting car among
the old and loud ones [phrased as: "if it's got four wheels and rolls,
bring it in!"] ... so I pulled in and entertained people with tours
of the ripped-apart dash and such. Lots of interest, in fact, which
I found sort of surprising from folks who had come out to celebrate
muscle-car mystique. Left the system powered up the whole time I
was there, inverter pump still disconnected, with the hood up so
people could see the stuff and ask questions if they'd never really
looked at a Prius before. Even with periodic HV-battery topoffs by
the engine, the inverter temps drifted down to around 50C and stayed.
The inverter case was only mildly warm to the touch.
.
And to think of all those car and energy shows and other events I've
been to where I had the system powered up all day, putting needless
hours on the inverter pump when really, I didn't need it just to run
the DC/DC converter for my power feed. The metal case can radiate quite
a few watts on its own without help, and that's apparently sufficient
to keep it in safe temp ranges in powered-up-but-quiescent operation.
It's probably time to put in a switch for the pump, since its connector
is actually not that easy to reach in and unhook.
.
I did get a few funny stares as I silently slid my way along the rows
at the cruise-in, at less than a walking pace, looking for a spot.
I looked back at some of these folks and said "vroom, vroom!" and
they thought that was hilarious. It's especially appropriate because
I'm intending to get into Madison early and make the pre-Hybridfest
cruise-in next Thursday. Wires hanging out of the dash and a laptop
with bargraphs in the shotgun seat does attract a certain interest.
.
_H*

 

An additional piece of info: Toyota's "operation history data" logging
facility of the HV ECU flags occurrences of either inverter temp rising
above 111 deg C. That's quite hot, and possibly also the point at which
the triangle goes on and the car takes steps to protects itself. The
80 C I was able to only momentarily achieve through various forms of
abuse yesterday was still well under that.
.
I also observed in the "driveway tests" that when MG2 is pulling hard
against a load but going VERY slowly, its inverter temp actually jiggles
upward a little, maybe about 5C worth, about 1/3 of the time. Apparently
some source for the temp readings is actually affected by whether its
associated transistor is actively passing current at the time or not.
That's fascinating. I'm not sure if that really means the die's
temp is really going up and down on such a fractions-of-a-second
basis, or the presence of all that current simply affects the sensing
element directly. Maybe some combination of both.
.
_H*