gs450h - normal inverter "2" temperature range?

Lwerewolf · August 24, 2019

Hello,

I was driving with hybridassistant on as of late to keep an eye on the engine temperature (had a suspect tiny headgasket leak, so far according to pressure testing & watching the cleaned "dip" area with a boroscope + a few days of generally stupidly hard accelerations & harsh driving... nothing to worry about - spilled coolant dip in the valley during inverter bleeding in the past), and I noticed that the "inv. 2" temperature sensor (which I'm guessing is the IGBTs for MG2) was hitting 140c+ on sustained "hard" (100kw+) accelerations.

First thought - air lock in the inverter cooling loop, so I bled it according to the service instructions (and tinkered a bit with a vacuum pump). The radiator appears to be fine, the inverter pump works (turbulence + it pumps through hoses attached to the bleed ports), the electric fans both turn on. The current daytime outside temperatures are ~32 celsius, so that might be something.

At any rate, went testing today, and the inverter generally stays below 90 except when accelerating hard at low speeds in second gear, until MG2 comes to revs. 0-140kph doesn't see temps higher than 90 or so (if they ever peak that high, usually they float around 78 during acceleration), 65-140kph can see up to 134 (probably momentarily higher due to OBD2 readout periods), generally only on an uphill and generally only in the range of, say... 65-100kph. Night time testing with 15deg. ambient leads to a max of 110 or so, same highway, same conditions. Steady 150kph (GPS speed) cruise = inv. 1 temperatures ~90, inv. 2 sub-60, even on uphills, so I'm not really worried about normal day-to-day driving conditions.

Now the bleeding did improve things, but I still find a 134+ celsius worst-case (full throtttle, high gear & low revs mg2, uphill slope) to be worrying.

One last note - it's a LHD car, so the inverter cooling loop is slightly different due to the relocated inverter - there's an additional bleed port right next to it. I'm starting to wonder if that's why I mostly read about LHD cars blowing inverters >_<

At any rate, does anybody have an idea if the temperatures that I'm looking at are ok?

Cheers!

Mihanicos · August 25, 2019

Hi,

From the data of the engine active test, the coolant temperature range is between -40C and 140C. Normal range is between 80C and 105C after warm up.

A note about the temperature sensor states that if the displayed temperature is 140C, then the sensor is short circuited. If it displays -40C, then it is open circuited. It may be worth checking the sensor although from your description it seems to be ok. Sometimes a short circuit occurs at higher temperatures than normal.

Another point to consider which is critical, is the radiator cap, and if it opens at a lower than specified pressure, this leads to coolant boiling at a lower temperature. If this happens, then the rate of heat rejected through the radiator is reduced proportionally. Hence a rise in the coolant temperature.

I would suggest that you check the inverter cap and make sure it is within specifications. [ Range : 15 Kpa to 44Kpa ] Minimum 12 Kpa. If it opens at a lower pressure, replace the radiator cap. Check that its vacuum valve is not stuck and free to move.

Also make certain that everything is clean a free from any rust deposits. The coolant should changed initially at 100, 000 miles or 10 years of life whichever comes first.

I hope the foregoing will be of help to you.

Chris.

Mihanicos · August 25, 2019

Coolant replacement is at 100,000 miles irrespective of time, please accept my apologies for my mistake.

Make sure that you have no leaks at all because this would reduce the system operating pressure and this in turn would lower the boiling point of the coolant. Once the coolant boils within the system, more coolant will be released from the radiator cap leading to less mass of coolant which in turn raises its temperature causing further boiling. At standard atmospheric pressure pure water boils at 100C.

Chris.

Lwerewolf · August 25, 2019

Hi Chris,

Coolant temps >65c. for the inverter don't appear to be considered normal, at least according to everything that I've read. The inverter temperatures spike, but they are measured at the IGBTs, and electronics can put out ridiculous amounts of heat in a short time frame 🙂

No leaks found while presure testing. The cap appears to hold pressure, having attached a vacuum/pressure pump to the inverter-side bleed port I can force the cap to hold ~0.3 bar and it also draws fluid from the expansion tank without holding vacuum... on the other hand, I'm wondering if it might be drawing air from the expansion tank line - it's quite a long line, and given that the inverter coolant doesn't get hot at all... maybe it doesn't expand enough to "bleed" that line. I've been attaching it with the expansion tank high above in the air, I should've pushed fluid to the expansion tank by attaching the pressure pump to the inverter bleed port and forcing coolant through it.

I'm waiting for a replacement cap to arrive regardless - preventative maintenance can't hurt, and it's probably the original one 🙂

Coolant is pretty much fresh - I did blow an inverter about a year ago when I got the car... let's say that some maintenance hadn't been done properly and I wasn't quite aware of it at the time.

At any rate, tested today again and... 27deg. ambient, high gear, 65KPH acceleration - inv temp 2 goes to 125 almost immediately and then stays there until about 110kph or so. Uphill or level doesn't matter. Low gear acceleration in the same speed range doesn't see inverter temperatures higher than 100, and only for a split second - usually it peaks to 90 for one reading (1/4sec refresh intervals or so) and then holds 75 to 80. I'm not surprised that the temperatures are higher in the "high gear" scenario, but I'm wondering if they are supposed to go that high (hitting thermal throttling) on a properly functioning cooling system (and other components, of course).

Mihanicos · August 26, 2019

Hi,

I agree, inverters do generate a lot of heat instantly and the coolant should be able to absorb it.

Upon further reflection, may be the coolant is not circulating fast enough to remove the heat from the inverters. At the same time, the engine and the inverters share the same cooling system, and if the engine has developed gas blow by from the combustion chambers due to piston wear, this would contribute to more heat added to the coolant. Gas blow by is more intense under hard and severe acceleration. You can check this by removing the engine oil filler cap momentarily and see if there are any gases coming out. Engine oil consumption is another indication. Beware of oil coming out usually from a nearby camshaft. However, this heat from gas blow by should not cause an immediate coolant temperature rise as that due to inverter heat generation.

Whenever the cooling system is opened for any repairs, replacements, you must make sure the heater is fully on and bleed the air from the heater radiator by squeezing the flow hose into the heater radiator while the engine is running, with the engine thermostat open and the radiator cap off. As you squeeze the hose the coolant level in the radiator should fluctuate. keep doing so until the coolant level changes to the slightest squeeze on the heater hose. Any trapped air in the heater radiator will compromise the coolant flow and this will hinder efficient coolant flow.

Another point to consider is the water pump impeller being worn out, a very remote possibility but I have seen some cases totally unbelievable.

As long as the expansion tank has coolant in it, any air in the tube from the radiator to it should bleed out. The volume of the coolant when it expands is more than the internal volume of the tube from the radiator to the expansion tank.

Check the water temperature sensor by immersing it in water and reading its resistance.

At 20C, the resistance between its terminals should be 2.32 to 2.59 K Ohms.

At 80C " " " " " 310 to 326 Ohms.

Replace the sensor if the readings are not within specified range.

Chris.

Lwerewolf · August 27, 2019

They don't share the same cooling system - the expansion tank is the same on the 3GS, but the pressurized part (i.e. the actual cooling loop) isn't... Getting combustion gasses from the engine in the inverter coolant loop is highly unlikely.

Engine-wise - no oil consumption, no antifreeze consumption, can't hit 100deg. on the engine (OBD2 monitor) to save my life - 95c at 60kph on a 7km long uphill requiring 50kw to maintain speed, that was from yesterday, highway accelerations to stupid speeds can't get more than that either... it doesn't really see coolant temps higher than 87 in daily driving scenarios.

Mihanicos · August 28, 2019

I have been doing some research on the Internet about Inverter applications for cars. Researchgate.net and Freescale have some papers on high temperature electronics, such as Insulated Gate BipolarTransistor [ IGBT ] modules for electric and hybrid vehicles.

As far as I could make, IGBT are used in conjunction with other semiconductors and the operating temperature limit of their circuits is 150C mainly due to solder considerations. Research is going on to improve the operating temperatures.

Lwerewolf · October 19, 2019

Well, here's some food for thought...

The following applies for the 3rd gen GS, 4th gen hybrids use 3.266 diffs across the board.

JDM & USDM gs450h's have a 3.769 final drive diff - the exact same unit that the GS350 (& Mark X 133).

EUDM (& General) gs450h's have a 3.266 final drive diff, the exact same unit that the IS220D uses.

This explains the shift point & shift speed discrepancies that I've noticed in acceleration & track day videos, and since the rest of the drivetrain is the same (engine, transmission, driveshaft, inverter)... it might just be a simple case of the EU version being overgeared for full throttle below certain speeds.

I'm looking into diff options, it'll probably be a gs350 diff with an eventual LSD upgrade.

Britprius · October 19, 2019

The GS450H has a LSD as standard equipment. This I would think would be a direct fit to the GS350 diff.

John.

Lwerewolf · October 19, 2019

2 hours ago, Britprius said:

The GS450H has a LSD as standard equipment. This I would think would be a direct fit to the GS350 diff.

John.

Where are you getting this information from?

These are the torsen cassettes:

IS-F: https://www.toyodiy.com/parts/xref?s=41301-53050&mE=on

IS-F, RC-F, GS-F, ls460 f-sport, gs200t f-sport: https://www.toyodiy.com/parts/xref?s=41301-53051&mE=on

gt86 & 1st gen IS: https://www.toyodiy.com/parts/xref?s=41301-53020 &mE=on

And now the gs450h cassette:

https://www.toyodiy.com/parts/xref?s=41301-50030&mE=on

https://www.toyodiy.com/parts/xref?s=41301-50031&mE=on

Inside view (from a uzs190 gs430, same cassette as shown on toyodiy):

https://www.drive2.com/l/533520976021815931/ - the thing being given a certain rude hand gesture, halfway through the article 😄

50031 seems to be a revision of the same basic part, looked up images of it just in case... same open diff.

At any rate, the gs430 diff does seem to fit (not my car & not my video):

Britprius · October 19, 2019

The 3gen UK GS450H definitely has a LSD as standard equipment.

John.

Lwerewolf · October 25, 2019

On 10/19/2019 at 8:53 PM, Britprius said:

The 3gen UK GS450H definitely has a LSD as standard equipment.

John.

https://www.lexusy.pl/LS430/ncf/ncf260e/m_02_0061.pdf

https://st.club-lexus(blocked word)/files/m_dr_0006_142634.pdf

Most likely something like this. If it was a clutch pack or a helical, we would've known for sure 😉

At any rate, found out about these guys:

https://www.facebook.com/valracing1/

I've seen several mentions of them doing custom diffs according to measurements, so that's an option for now... unless I find a leftover quaife (discontinued years ago) or decide to go for a clutch pack (probably a death wish on a daily for icy mountain roads, even if it's 1-way... not that a torsen isn't dangerous).

On a sidenote, talk about derailing the topic... sorry 😕

Lwerewolf · May 15, 2020

Well, it's this season again... Over here temperatures are quite high, 35 celsius today. Inverter IGBT temperatures were approaching 80 degrees while cruising, the rest is mostly still the same - replaced the oil pump's bearings and the upshifts/downshifts are noticeably faster.

3.769 diff from Japan (nengun, yahoo auctions) soon-ish.

Steam cleaned radiators, no change. The inverter coolant temperature is consistently 20-25 celsius over ambient, and doesn't seem to be affected - doesn't matter if you're doing 60-140-60-140 10 times in a row or you're just cruising along. Prime suspects - IGBT cold plate solder deterioration (pretty much means new inverter), inverter cooling channels blockage, insufficient pump flow/pressure, air lock, "car is performing to spec".

Lwerewolf · May 21, 2020

Well, change of plans.

I tore down the inverter that I blew up two years ago - not too difficult to do, at least on the IPM (IGBTs) side. Boost converter is trickier, but unnecessary.

You can separate the coolant flow plates if you want to - you need to remove the IGBT board first, as the boost converter case is bolted to the IGBT case by hex bolts under the IGBT board. Before you do that - let me tell you straight away, I simply don't see how these passages can get clogged.

Here's the rx400h TSB for the IPM replacement:

https://static.nhtsa.gov/odi/rcl/2013/RCRIT-13V396-9944.pdf

Key points - thermal grease. My car was from Italy, manufactured in August 2008 and had approx. 245000km at the time of the inverter failure. Long time + decent mileage + a relatively hot region + usage of thermal grease...

You can probably guess what I'm ordering 🙂

LHD and RHD inverters appear to use the same IGBT & control boards. I'm guessing the main reason for most failures is people actually connecting MG1 in the MG2 slot & vice-versa. Just a thought for anybody in need of a LHD inverter, since they are extremely rare nowadays... and RHD inverters are cheap and plentiful. At any rate, if this turns out to be the primary reason for inverters blowing up (thermal grease degradation + rapid heat buildup = blown IGBTs, given enough degradation and heat), I'd just recommend that people reapply the grease every few years.

I'll let you know how the TIM replacement (thermal interface material - thermal paste - thermal grease - etc) goes - I'm quite hopeful... so far this makes the most sense - spiking inverter mg1/mg2(IGBTs) temperatures, no apparent problems in inverter coolant temperature/flow/condition.

EDIT:

If anybody's tinkering with his electronics, the grease part should have been pretty obvious. Otherwise (well, still a good read):
https://www.nrel.gov/transportation/assets/pdfs/42972.pdf

I wonder if the primary reason for not seeing inverter failures in the ls600h, 4th gen gs450h and so on is the new cooling method for the IGBTs (and them being of the SiC variety) - look up the ls600h oak ridge national laboratory article for a teardown of its inverter. The gs450h one is pretty much the same as the 1st gen camry hybrid, alphard hybrid, nissan altima hybrid and a few others - key point is, shares IGBT cooling design with earlier hybrids incl. the rx400h.

Lwerewolf · June 18, 2020

So.. the grease has turned mostly translucent, and pretty much all the thermally conductive filler material (if it's shin-etsu g747 like in the rx400h, it should be zinc oxide) has been pushed out of the hot zones.

I maaaay have found my problem 😄

Lwerewolf · June 20, 2020

Well, either that wasn't the problem, or I messed something up (IGBT torque as per rx400h tsb might not be enough, bad thermal paste layer, etc). Waiting for some tools to make life easier when removing/installing the inverter, then I'll take a look again. Have some other thoughts in the meantime, still reading.

Lwerewolf · July 25, 2020

So far - not the thermal grease, not the capacitor bank (at least according to what I've managed to measure via the DE5000 LCR meter that I got - all HV caps look fine). At this point I have to assume that it's the IGBTs - thermal fatigue (cracks) in the solder layer, or something else.

The plan going forward is to attempt to replace the inverter with one from an ls600h - reading this thread should give you a clue as to why I think it's quite possible with the stock ECU:

https://www.diyelectriccar.com/threads/reverse-engineering-the-lexus-gs450h-inverter-converter.200883/

This is the precursor thread:

https://www.diyelectriccar.com/threads/bmw-330ci-conversion.142946/

And here is where I embarrass myself publicly:

https://openinverter.org/forum/viewtopic.php?f=14&t=776

🤣

As for the "why an ls600h inverter" questions:

-I haven't found a single reference to a Toyota inverter with double-side-cooling IGBTs failing. These include the ls600h (first application), rx450h (found it listed as such in toyota australia marketing materials), 4th gen gs450h & is/gs/rc300h (visibly so in Damien Maguire's is300h inverter teardowns), 4th gen Prius (weber state university teardown, toyota media material, so on). On the other hand, plenty of stories about failed inverters of Prius 2/3 (and their derivatives - auris/priusV/so on), rx400h, gs450h (camry with double the MG1 IGBTs/diodes by the looks of it) - basically anything with the classic IGBT packaging. Not a Toyota-exclusive problem by any stretch, would've been nice if they'd sell only the IPM (the power module - basically the IGBTs) for, say, 1-2k euros (I know, overpriced but that's most OEM parts for you) instead of the whole inverter only (and for 4k+)... Actually, they do sell the IPM for the rx400h due to the service campaign, but the prices that I've gotten are downright bonkers - a new inverter costs about the same.

-The 3rd gen gs450h and ls600h are very similar. The transmissions are the same (well, apart from one having a transfer case at the end), the electrical wiring diagram to the inverter looks pretty much the same, the communication protocols are supposedly the same (according to what I've read in the above-mentioned threads), the wiring harness connectors are the same (shuffled pin numbers).

Mihanicos · July 25, 2020

Hi

I admire and congratulate your perseverence to find the cause of your issue because you are a perfectionist.

One point worth considering is the fact that all capacitors degrade over time and this causes their series internal direct current resistance to increase. This in turn increases their operating temperature depending on the current passing through them.

An ideal capacitor has zero internal resistance and blocks all direct currents through it. Alternating currents are conducted and the resistance of the capacitor depends on the frequency of the current through it.

The Toyota/Lexus inverters have six such capacitors and they are at the bottom of the inverter hausing, if my memory serves me right.

Under hard acceleration the inverter dc current is at its maximum hence the generated heat from the capacitors, due to their internal series resistance, is at maximum too and this heat is dissipated into the coolant.

Chris.

steve2006 · July 25, 2020

I think I’m going to enjoy this thread now we are going into the realms of ESR, Capacitive and Inductive reactance and Z impedance.

I still say most of the problems with electronics today is the use of “Lead Free Solder” thankfully I still have a good supply of MIL grade leaded which does a proper job of jointing.

Mihanicos · July 25, 2020

Hi Steve

" It takes one to know one " .

Liberace, Monte Carlo concert, 1982 !

Chris.

Lwerewolf · July 25, 2020

Thanks for the kind words 🙂

Also keep in mind that I'm very inexperienced in this field, but then again - what's better than a learning experience 😄

The high voltage capacitors in all Toyota inverters are SH film caps - don't know which kind exactly, but the ORNL articles have graphs of capacitance, ESR & dissipation factor over different frequencies and temperatures, and the different film caps (PP/PET/PEN/whatever the film material happens to be) appear, at least to me, to be easily differentiated by those responses - so we can find out.

Otherwise, there's the main DC link capacitor bank (750v/2855uf, ORNL x-rays show 20+ capacitors in parallel, potted in black, opaque epoxy or something), the DC link snubber capacitor (750v/0.9uf, very low dissipation factor) and the boost converter capacitor (500v/378uf). I've seen basically no deviations in capacitance, ESR, dissipation factor & phase shift compared to what was listed over at ORNL - keep in mind that I'm limited to 100/120hz for most measurements on the big caps (2855uf&378uf), but ESR isn't limited. I don't have access to a bench LCR meter, so this will have to suffice.

At any rate - there are no thermal probes in the capacitor banks, and they are not liquid cooled either. Their location varies - in the 3rd gen 450h all banks are separate modules, whereas the ls600h combines the 750v caps (main bank & snubber cap) into one potted assembly. There are various teardown videos & articles, you can see for yourself. The 1st gen prius' inverter is the most obvious - the DC link bank caps are screw-terminal cylindrical panasonics, exposed for your eyes to see 🙂

Re: solder - there are other ways to get around it:

https://www.semikron.com/innovation-technology/packaging-technology/skin-technology.html

There are still bond wires used in the prius gen3 inverter ipm, and all new IPMs seem to be of the double-sided cooling variety, so I guess Toyota hasn't used such packaging in any of their production units - I'm quite sure that they've experimented internally, though... Conservative products don't mean a lack of innovation 😄

Again - the temperatures on the IGBTs (I assume the "inverter 1" and "inverter 2" temperatures are taken directly from the temperature sense probes, which are present for all IGBTs) spike when they're driving a motor under low RPM conditions - under high RPMs there are no spikes whatsoever. Above 100kph I can keep it floored for as long as I'd like and the reported IGBT junction temperatures (again, assuming "inv1" and "inv2" are those) stay very close to the inverter coolant temperatures, which is (as far as I know) measured at the coolant inlet - might as well be "inverter case/cold plate temperature", I guess. Still, looking at hybridassistant clips (mostly of 3rd gen prius & derivatives, a few rav4 hybrids as well) and comparing them to what I have - my inverter temperatures "jump around" a lot more, whereas the others stay relatively flat. There are a couple of clips with a 2010 camry (which incidentally shares an almost identical inverter with the GS) and the temperatures spike a lot more there as well - my best guess is still age-related solder fatigue leading to very high thermal resistance and resp. heat buildup. It's listed as the most common failure mode of IGBTs, along with bond wire lift-off - mostly due to the same reasons. Double-side cooled packages & sintered packages tend to not have these problems, hence the ls600h idea 🙂

EDIT:

Links to ORNL articles - posted earlier, but anyways:

https://www.osti.gov/biblio/928684-evaluation-toyota-camry-hybrid-synergy-drive-system - the GS inverter is almost the same, different wire harness + bigger DC-link cap + more power devices for MG1 and that's it

https://www.osti.gov/biblio/947393-evaluation-lexus-ls-hybrid-synergy-drive-system - The LS transmission is essentially the same, barring the transfer case. The 10,230RPM limit on MG2 is false, unless the transfer case has a reduction ratio that Toyota haven't told anybody about - MG2 max RPM should be 14k, just like the GS. Incidentally, MG1 max RPM is 10,230 off the top of my head, I'll go rev it outside now to confirm 😄

Lwerewolf · October 24, 2020

Somewhat of an update, detailed here (different thread on openinverter):

https://openinverter.org/forum/viewtopic.php?f=14&t=931#p18441

tl:dr so far:

gs450h and ls600h inverters share data, HV Battery and AC connectors. MG harness connectors are different, data connector pinout is different.

I repinned the ls600h inverter that I got to the gs450h wiring diagram (easier plug-and-play). The car didn't complain in "accessory" mode (with just the inverter data connector connected, also inserted a clip in the Battery connector interlock pin to fool it. I then connected everything else (custom wiring harness to the MGs, nonshielded) - attempted to crank, immediate p0a78-306 and p0a7a-344 - "Motor/Generator torque execution monitoring malfunction", more specifically:

If the difference between the requested (MG1/MG2) torque and the actual (MG1/MG2) torque exceeds a predetermined value, the MG ECU determines that there is a malfunction in the execution or monitoring of the (MG1/MG2) torque. Then, the hybrid vehicle control ECU will illuminate the MIL and set a DTC.

The MG ECU is essentially the inverter. The car did "crank" the engine once (well, you know how the hybrids crank their engines 😄 ), and the codes were set at that exact point in time. Freeze frame data has nothing useful (no inverter-related parameters whatsoever), everything on live data seems OK (believable values) but the refresh rate is way too abysmal (2-3 times per second) in order to catch something as instant as this - I attempted to capture a snapshot a few times, and only one of the frames was of the time during which the high voltage system was on (detectable by the inverter's pre-boost voltage being just about the same as the reported HV Battery voltage).

Prime suspects - no shielding on the MG wiring harness extension + the two MG harnesses essentially interfering with each other, didn't repin the inverter properly (MG resolvers, maybe), missed something on the diagram, etc, etc...

Still looks promising. Car is obviously back to stock for now 🙂