Toyota to Recall 1.9 Million Priuses to Update Software

I read the number somewhere in a newspaper but after googling, I found the following links for the 450 number quoted by Toyota.

See links below. (funny links because I'm too new to PriusChat to post normal links)

news:drive:com:au /drive /motor-news /toyota-recalls-5500-prius-cars-20140212-32i7g :html
bizmology:hoovers:com /companies /toyota-motor-corporation-2/
smh:drive:com:au /motor-news /toyota-recalls-5500-prius-cars-20140212-32i7g :html

 

Toyota will not use it for US press, as they have been caught by the NHTSA before hiding incidents.
Toyota Recalls 1.9 Million Prius Hybrids Over Software Flaw - IEEE Spectrum

I would bet that most japanese incidents are not reported in this number, as they haven't been in the past. We also know that in 2010 and 2011 toyota wasn't reporting problems presented to dealers in the US unless the driver also reported to the NHTSA.

There shouldn't be any controversey here. Toyota now knows about the problem and has either a software fix or work around that should prevent it once they reprogram our cars.

That is a good thing.

 

I'm also doubting the accuracy of these figures. Toyota claim this issue has not affected any Australian owners but I'm certain that inverter failures have occurred here. The one I know of showed the same complete failure but was replaced under warranty.



iPhone ? - now Free

 

Exactly- why is reasonably civil discussion thread just completely eliminated from existence? Shomeshing really fishy ish going on...

 

Yes, I agree. I think the thread has been fully recovered - I'm sure it's linked somewhere in this thread, in the first page or two? I posted a zip file containing page 3 of the original thread, but as it turns out, I was able to recover pages 1-4 of 5, if anyone really wants them.

Anyway, as per my earlier comments, I'm surprised by the roughly made look of the inverter innards. Maybe it's a cost thing, but I would have expected them to use off-the-shelf commercial components, such as this IGBT package, used in a standard VSD. Not sure what drive this was out of, somewhere between 4 and 11kW, I think. The module is an infineon FP75R12KE3, 75 Amp, 1200 Volt unit. The datasheet is here.



There will be many like this available in different ratings to suit various applications. I don't know why they didn't go with parts like this, as most of the hard work is already done for you. Three of these should have been enough to do the whole job, and would probably have taken much less space. (If you want to see a real grunty 3-phase IGBT Module, check out this 1200 Amp, 3300 Volt item from Mitsubishi care of Wikipedia.)

 

Probably multiple reasons including legacy due to the 1995-96 design efforts for the first Prius. Having an in-house 'forge' meant Toyota could control resources dedicated to this critical part and changes to the design to meet requirements discovered during testing. It is functionally equivalent to the engine, a parallel power system.

Bob Wilson

 

Had mine in this morning at Smithtown Toyota, approx. 45 min in and out.

 

Hi!

My question: Are all 2010-2014 Liftback affected?

If not, why?

Thanks!

p.s. Still waiting to see if mine is affected, Toyota Canada's site says no recall as of yesterday

 

Dropped the car off at 7:30, done by 9:30 and freshly washed.

No problem.

 

I took a bit of a closer look at those documents from Oak Ridge. A couple of things:

1. That analysis was done on an early Gen III Prius. I do not know about you, but there's always cost reduction initiatives that take place after the first ship of a product; I'd be highly surprised if engineering changes (Rev 2, Rev 3, ...) weren't put in place to reduce the cost/increase the reliability of the product, especially as production ramps up. So, what they were shipping then may look substantially different from what they're shipping now.
2. I've had a bit of fun looking through that document, staring at the pictures, and looking at some IGBJT manufacturer web sites.
   1. That wasn't my imagination: Those fine little wires going to the transistors are bond wires.
   2. Those aren't transistors in a package - they're naked semiconductor die!

Back in the day I happen to have worked on a production line troubleshooting HICs, Hybrid Integrated Circuits. They aren't as popular as they used to be, but what they were was a ceramic (as in, white!) substrate with metals traces and resistive material printed onto the ceramic. Then, the manufacturer would solder naked semiconductor die to the substrate. These were operational amplifiers, transistors, diodes, and the like. The substrate would then be placed in a somewhat larger-than-usual IC package and glued down; finally, some ladies running a bond wire machine would connect all the ICs to the substrate and the substrate itself to the lead frame inside the package. The completed circuit would then be tested; my job was to take the failures, probe around with some really fine-tipped probes connected to standard test equipment (voltmeter, oscilloscope, frequency counter, etc.), with my eyes firmly planted on binocular microscope, and figure out the fault. Bad parts could be chipped off and new ones put down until it passed. Finally, they'd seal the whole thing up in a dry nitrogen atmosphere (after baking) and test again.

It turns out that a fair number of transistors and other analog integrated circuits are available in naked die format. I keep on saying "naked die", I know the EE's around here know what that means, but, just to make this clear: When one makes an integrated circuit, transistor, diode, or other silicon product, one does this on 6" to 12" diameter wafers of exceedingly pure silicon, with multiple trips through ridiculously hot ovens with arsenic atmospheres under pressure, trips through hydroflouric acid baths, and the like. The final product has tiny aluminum contacts (bigger when there's high current going through) and where there's not aluminum there's a layer of silicon dioxide (i.e., glass), all very pretty under a microscope. They don't make just one of these per wafer; one may have (depending upon the size of the device) anywhere between thousands to hundreds of devices per wafer. And these wafers are pretty blame thin, too.

The cost of processing per wafer is more-or-less fixed, but if you can get 10,000 devices on a wafer, the cost per device can get down into the pennies, one of the wonders of cheap electronics these days, and why wafers keep on getting bigger over time.

After the wafer has finished processing, automated machinery runs probes over each of the devices on a wafer, looking for defects, of which there are always a certain number, usually around 100 or so. Bad devices typically get marked with a dot of paint. After this step, diamond saws are used that slice the wafer into tiny little pieces, hence the names "chip" or "die". A low power, less complicated device may be just bigger than the letters you're reading at the moment, with most of the area taken up with metal contacts around the periphery of the chip; complicated or high power devices may be as big as the fingernail on your pinkie or maybe your forefinger.

At this point, most of the time, these devices are bonded into the center pit of some kind of package, usually plastic, with an integrated metal lead frame that protrudes metal leads out of the side of the package. A special wire bonding machine cold-welds tiny little metal wires between the lead frame and the aluminum metal contacts on the die; finally, a lid is attached to the package, usually with an inert atmosphere (no water vapor, please!) inside the package. The completed device can then be soldered to circuit boards and the like to the leads of the lead frame that protrude out the sides (or bottom) of the package.

There's good reason to seal these dies away: the protective layer of silicon dioxide on top of the die will rot in the presence of water vapor and other contaminants, killing the device, and those tiny bond wires, while strong in the presence of vibration and whatnot would not survive a finger hitting them.

However, there can be reasons why one would want to play with naked die. The HIC case above, where one is building specialized circuits than have to be small, nonetheless, is one example. In the high power transistor case, having the die on top of bonding material, which is then on top of a slab of metal, which is finally planted on the next electrical/thermal layer is obviously not as good, thermally, as soldering the die directly to a layer of aluminum, which is what Toyota has done.

This means that the whole transistor array area (of which we have pictures) is very, very thoroughly sealed and assembled in a clean room. That sideways picture of the board with semi-crushed fine wires on the right side has those semi-crushed bond wires there because Somebody crushed them in the process of opening the thing up. You can bet that these boards aren't being repaired by somebody at a Toyota dealership; it's even questionable whether it's worth Toyota's time to repair these transistor assemblies in the first place, given the cost of shipping them back to Japan (the likely place of manufacture) and getting them out of the inverter. I'm not sure about this, but I think it likely that just getting the whole thing exposed to go delving into the vagaries of broken solder (on the bottom of the die!) wrecks the board for future use, at least without cleaning out all the bond wires, baking the assembly, and reassembling it in the Approved Fashion.

Fun.

KBeck

 

How did you get the appointment this early? I just called them and was told I need to wait for the mail.

 

For these types of product, it is almost always better to replace instead of repair. There is additional risks of failure with a repair, which raises the cost to the automaker. This is based on my past experience with clean room manufacture, and automotive test.

AFAIK very few have failed, likely less than 1000 in the US. That isn't enough to even test a repair protocol. If the software change works, and I expect it to work, no reason to monitor repairs. I'm sure the engineers want to dead units to test and see if they can create a better part for future cars.

 

Toyota already knows that the booster/inverter module is defective either with poor solder joints or under rated IGBT transistors, the software fix is just a band aid to limit power or better SOA. This will fix probable 80% of the units.
Toyota should replace the defective units with new improved modules.

 

Without actually knowing the design, we can't determine that, and I am sure toyota doesn't want to provide them nor should they. The software needs to enforce the hardware limits, so the only way to tell if its hardware or software is to see what the spec of the units was. There is always a trade off between a higher spec part and cost. Software needs to protect the lower cost units. Since the failure rate appears to be lower than 0.1%/year, I would expect that we will not see anywhere near 20% fail for the life of the car after the software campaign. We will have to wait until more people have the fix, before we can determine if mpg or acceleration is impacted in a major way. I don't expect it, but have a wait and see attitude. I would expect that toyota should increase the waranty period for these inverters.

 

Called on Friday and told them I entered my vin on the Toyota web site and it was included in the recall.
The made the appt. with no problem.
John

 

Toyota's German website claims (Toyota ruft Fahrzeuge vom Typ Prius III in die Werkstätten) that: about 400 cases worldwide have been reported, of which 11 in Europe (GrumpyCabbie's is one of them I suppose...). No accidents or injuries reported. They say the update takes 30mins and 136816 Prius III (which means no other model), up to those produced in Feb. 2014, will be recalled.

This means that: 1) apparently Gen 3s built recently have the same software of their sisters built in 2009 (like mine) - so apparently no improved software after the facelift model came to the market; 2) the problem can arise in all revisions of hardware (if any such revisions ever took place).

Moreover, this might mean that the Prius v/Alpha/+ as well as the Auris (that supposedly has the *same* drivetrain of the Prius, but clearly it hasn't since it is not recalled) have a different software for whatever reason, and hence either the problem still has not shown up at any customer (due also to the lower number of vehicles sold - I suppose the 1.9M is referring to the standard liftback version), or Toyota decided to not install the software also on newly built Prius either because the electronics are different (different micro controllers with different programming) or because you never change a running system - remember that the automotive industry is *extremely* conservative. If it works, you don't touch it. So on the Prius you keep the old software even on new revisions of the car, and you install a new software only on a new car (e.g. Auris). Please note that also the Lexus CT200h has the *exact* drivetrain of the Prius. If you think it is different, it isn't - it is all a marketing gag.

So when Toyota had evidence they needed to change the software (which means 400-450 or 0.2ppm for them is a problem, and that is a good thing), they had to do it, and likely quite not too much happy about it.
So no extra MPG or HPs or torque from this update, just an added failsafe mode added.
The more Prius they sell, the more corner cases will be discovered, and the more cases reported, the higher the chances of a software update.

PS: I would like to remind everybody here that the Prius is not a clutch transmission car of last century. It's a car with a *lot* of software running it and many systems interacting with one another. I find it very positive that we get a software update to fix a "bug". They might even fix other bugs we will never know about. Who knows.
Other manufacturers reprogram likely their cars without any recall or without so much noise from the press, instead Toyota is Toyota and the Prius is the Prius. Who knows how much software has been updated in a Golf? or better yet the Ford Focus, the most sold car in the world in 2013? Maybe there is no software to update because the Ford Focus is after all a car from the 19th century after all.

Nobody knows what the technicians of any car manufacturer is doing to our cars when they are serviced. And if the HQ tells them to change the software, they change it without you knowing it. So I find this open initiative from Toyota quite positive.

 

Hi there bedrock8x,

Cool handle.

However, I'm very curious. HOW do you know this?

I see you are a Senior Member with many posts to your name. Do you know something the rest of us don't?

In my logic, if the same hardware (apparently, so we are informed... FWIW) doesn't have a problem when run by different software, doesn't that suggest that the hardware might actually be ok? And that the software is bad?

Of course, it's possible for good software to protect weak hardware, but that usually results in apparent effects, derated, poor performance, etc. The Prius in general seems to be walking a difficult tightrope between people's expectations, power, acceleration, reliability, longevity and economy, and, of course, Cost! Not to mention tall poppy syndrome!

Seems to me that the Gen3 might have "pushed the envelope" (a la SOA) just a fraction too far. Or, not been tested quite enough. It's such a fine balance, but we certainly don't want Toyota to become disheartened. Let's not knock them too much!

 

Hello All,

Forgive me for quoting my own post, but you'll see why in a sec...

Now that we've seen what's inside the Inverter and Power Converter, I can say that both KBeck and I were both fairly wide of the mark, as the Boost Converter doesn't employ IGBTs but uses plain diodes to conduct the Boost current (and IGBTs for the Regen (Buck) current.)

Not that many of you probably care, it's all a bit like rocket science. I'm just very grateful to be able to buy a complete, working, fairly solid Gen2 Prius for so little money (compared to rockets that is, and at a price favourably comparable with so many very "ordinary" cars.)

Anyway, all this:

... just ain't relevant!

Toyota opted to KISS (Keep it Simple Son - or whatever), for which I am now grateful. Using simple diodes, instead of trying to use IGBTs to conduct the forward current, while it may cost a fraction more in extra power / heat, makes it SOOOO much simpler and more reliable. IF the inductor is producing more volts than there is in the capacitor at that moment, the current will simply flow into the capacitor. When the inductor "runs out of puff", the flow simply stops. End of story!

This design has many benefits, by (1) protecting M1 against high voltage spikes that would occur if M2 wasn't turned on at precisely the right instant, (2) maximising the charge transfer from the inductor to the capacitor, (3) eliminating the need for a whole MESS of complex software such as was hypothesised about at some length.

Anyway, the upshot of this is that I for one, cannot really see what they could have been doing wrong in the software driving the Boost converter. It's actually a pretty simple thing. (More or Less) just turn on M1 for a while, then turn it off for a while. All the rest will take care of itself. It's only the speed and duration of the On and Off times that determine how much voltage and power is transferred to the output, and how efficiently it runs.

This leads me to wonder if in fact the whole "Boost Converter problem" isn't just a big red herring.

However, as we did surmise correctly, the Boost Converter is a dual function device, really a Buck & Boost converter. In Motoring mode, it converts the 200-odd Volts DC in the traction battery, to (up to) 650 V DC to drive the Motor Generators. In Regen mode, (in a highly controlled manner) it (down-)converts the excess voltage coming back out of the motors (now acting as generators) into current to recharge the traction battery.

That's about as much as I've figured out so far. There has to be a bit more to it, because from driving my G2, it seems that the regen process continues until the car is almost stopped. I imagine that at those low speeds, it will be unlikely that there is enough power generated by the slow turning motors to actually achieve any significant battery charging and thereby braking (remembering that in order to actually charge the battery, you have to produce over 200 Volts and a certain amount of current). So I'm suspecting that the Buck/Boost converter may actually go into "Reverse Boost Mode" to maintain regenerative braking at low speeds. Not sure about that yet.

But, if there was a software problem such as we discussed earlier, wherein two transistors are turned on together, causing extreme (damaging) currents to flow, or not sequenced correctly so as to generate (damaging) voltage spikes, then this is probably where it would be. Maybe it's a bit of sly marketing misinformation to lead us to think it's the (Forward) Boost converter at fault, when maybe, really, it's something much more tricky. (Maybe) they are really referring to the bit I haven't figured out yet, where it (maybe) goes into (Reverse) Boost mode to increase the voltage out of the regenerating motors, so that regen braking continues down to almost stopped. (Maybe) they didn't want the competition to figure that out either...

And maybe I'm all wrong again, but as I said before:

Whatever the problem is, it doesn't really matter, so long as the fix is correct, there is some kind of testing to determine that the IGBTs haven't been degraded, and there is some kind of Warranty Extension to cover failures in this area, then it's gotta be All Good![/quote]

Long Live Toyota!

 

For our lay friends, there are several diode/rectification design restrictions:

• silicon diodes - require ~0.65V before passing current, least efficient
• Schottky diodes - ~0.30V before passing current, more efficient, less heat
• MOSFET switch - even lower, most efficient

To use a MOSFET switch like a perfect rectifier requires the control voltage be turned on at the 'right time.' This is what the software does in the Prius inverter, the big-brother to MOSFET, IGBT devices.

Over the years, I've enjoyed writing devices drivers in part because I get to read the schematics and print sets of the devices my software interfaces. Eventually you realize there is a gray area between what is done in hardware and software and designs can be optimized where some functions are done in hardware (i.e., dumb diodes) or software (i.e., hashed or index table lookup.) Our Prius paved a path with elevating software to a new level of vehicle integration:

• accelerator is an input to the HV ECU - this computer operates everything else
• inverter, traction battery, HV ECU and software - this is the Prius transmission

You can see this progression best in the accessories:

• air conditioner compressor
  • 2001-03 - traditional electronic clutch
  • 2004-* - electric motor operated compressor
• engine coolant pump
  • 2001-09 - traditional belt driven pump
  • 2010-* - electric motor operated pump

Each time the less efficient, mechanical systems were replaced with more efficient electrical, often software controlled systems. So the air conditioner not only uses just the amount of power needed for cooling but it also runs when the engine is off. The engine coolant motor also only runs as needed with a much lower overhead at all engine power levels than the belt driven pump.

I appreciate the simplicity of a basic mechanical solution but when married with a control computer and quality software, truly amazing efficiencies are possible. But such software is not trivial.

Bob Wilson

 

I haven't looked at the Oak Ridge documents that carefully nor have I personally seen the innards of the Gen 3 Prius inverter, at least not w/the IGBT's visible, that I recall.

However, on the Gen 2 Prius (yeah, yeah, the design is very different), at the teardown presentation I was present at. The speaker commented that he felt the wire bonding was very good. And he talked about the goop/gel around the many of the components. I did feel the nearly transparent goop. I think it was sticky.

You can make it out at Embedded Systems Conference Prius Teardown | PriusChat and some of the pics at Embedded Systems Conference Prius Teardown | PriusChat. Article at Special Issue: Inside the Toyota Prius: Part 5 - Inverter/converter is Prius' power broker | EE Times. And, if memory serves, he believed the IGBTs were actually manufactured by Toyota and not a 3rd-party supplier.