Toyota to Recall 1.9 Million Priuses to Update Software

The Prius+ has a different final drive ratio, maybe that eases the load.
Here in the UK we also have the Li-Ion HV Battery, if that makes any difference ?

 

Prius+, or Prius v, here in the states, along with the Lexus CT200h, all supposedly use the same drivetrain as the Prius hatchback. Since neither the v or CT200h is part of the recall, it indeed must be software related.

Galaxy Nexus ? 2

 

The letter to the NHTSA (the United States National Highway, Transportation, and Safety Administration, a division of the executive branch of the U.S. Government in charge of recalls, airplane crashes, and the like) specifically says that the similar power trains on other cars made by Toyota aren't subject to the same problem. One second... Just re-read the thing. It says in there that the Prius V, which uses a different battery and control software, isn't subject to the problem.

What I noticed from reading the Department of Energy's analysis of the Prius's inverter is that they compared the battery voltage on the Prius to the battery voltages on other hybrid cars that Toyota was making in 2009/2010. I think that I remember that the Prius had the lower battery voltage, around 200V.

Since this is a boost converter that, based upon what I've read so far, runs normally at 450V output with occasional boosts to 650V, running a lower battery voltage into the boost means that the boost is going to have to work harder and dissipate more heat then if a higher battery voltage is used. Since the whole problem with the Prius appears to be a corner case (power dissipation, software, etc.), it may be that other battery systems with higher voltages may not approach the corner case that may potentially cause failure.

KBeck.

 

Same drive train doesn't exclude the use of different spec parts. The CT is a bit more performance oriented, and the v 200 pounds heavier than the Prius. It is possible a more robust part was used in expectation of more of this boost voltage being used. The use of lithium batteries for some markets could mean a different part used. Part numbers should be easy to look up.

On the cynical side of the coin, there could just not be enough complaints with the v and Lexus for Toyota to bother with fixing yet, or their recalls are planned but held off to keep the headline numbers as low as possible.

 

Bob,

Once again, speaking to lay people:

So, we have a choice about using silicon diodes, Schottkey diodes, or a MOSFET, with the MOSFET being most efficient. But, as usual in these things, there's Other Considerations.

One of those has to do with the voltages and currents being played with.

Suppose, for example, one is trying to build a buck converter that goes, say, from 5V down to 0.9V. (Given the core voltage of 25W high-performance, high power devices like graphics controllers and Other Devices this is not at all unusual.) If one is dropping 1V or so with a silicon diode or 0.7V with a Schottkey (the bulk resistance a high current raiseth the voltage up!), there's a heck of a lot of loss using a standard diode.
Bob, good post. I'd like to add a couple of points, if I may.

First: Funny things happen when selecting parts for a high voltage DC-DC converter as compared to a low voltage DC-DC converter.

Take, for example, what one would use when designing a buck converter that goes from 5V down to 0.9 V. (Very common for ASICs with high power, low voltage, high current core logic). A buck converter happens to have a diode connected between ground and the 0.9V in this example. Since a standard silicon diode with a few amps across it will easily have more than 1.0 V across it, one would be forced to use a Schottky diode; even with that, the losses in that diode would be a double-digit percentage of the power delivered to the device using that 0.9V! Therefore, it's not just a nicety for low power consumption/good efficiency in the buck converter, it's durn near a requirement that one put a FET in there and make it look like a diode by playing with its gate. When the FET is on, what voltage drop it does have is simply its on-resistance (usually, for something like this in the range of 0.005 to 0.01 ohms) times the current passing through the FET; even with 4 or 5 amps, the losses are less than a watt.

Now, take the case of a Prius boost converter. We are (a) playing with relatively high voltages here and (b) plenty big currents.

The point, and this hasn't been made up to now: A high voltage MOSFET has to have low doping levels (contaminants that make the silicon act like a transistor, rather than like a chuck of just pure silicon). If the doping levels are low, the internal resistance goes up; and that 0.005 ohm resistance and the like that I was citing for that low-voltage buck converter goes right out the window; instead of having at most a half a volt across a FET rated for 1000 V with full current, you'd have a couple of volts! Now, suppose one uses a real, true-blue diode. The diode equation is
Id = Io(e**(q*Vd/eta*K*T) - 1), where Io is in the range of 1e-10 amps or so, e is 2.71, "**" is the exponent operator, q=1.609e-19 (electron charge, in Columbs!), eta is a fudge factor between 1 and 2, K is Boltzman's constant, 1.38e-23, and T is the temperature in Kelvin. At room temperature (ha!) this simplifies to Id=Io*(e**(Vd/0.026) - 1).

What this means: As the current goes up linearly, the voltage goes up as the log of the current. Note that with a FET, as the current goes up linearly, the voltage goes up linearly, too. For those of you who are hanging on by your fingernails, it's like this:
If the current goes up by 2X in a MOSFET, the voltage goes up by 2X as well, and the power goes up by 2*2=4.

In a diode, if the current goes up by 2x, the voltage goes up by ln(2)=0.693, and the power goes up by 2*.693 = 1.38. You might note that 1.38 is less than 4! And, even if the MOSFET starts out lower than a diode, the diode will win with lower power dissipation when the current values get Up There. Which is likely the case with a Prius.

There is also a small bulk resistance, pretty blame small in a power diode. Finally, true-blue MOSFETs are planar devices where current goes along the surface and is therefore somewhat concentrated there; with power diodes, the anode (usually) is at the bottom of the chip and the cathode at the top, spreading out the current and making for a lower bulk resistance.

Two results of all this:

• If one is playing with high current in a high voltage environment, one may actually get less power dissipated using a true-blue high voltage silicon diode than with a high voltage MOSFET.
• Since the overall efficiency of a boost converter includes terms involving the absolute voltage level, when the voltage being switched is up around 200V to 650V, the power dropped in a diode dropping a volt to a volt and a half is a small percentage in the overall power being transferred and has a small effect on the overall efficiency.

What I'm saying: Using real, silicon diodes is probably not because somebody is trying to be cheap or is trying to avoid complex software; it may, instead, be the first choice in making a high-voltage, efficient system.

KBeck

 

Here is a drawing of the Hybrid System from the 2010 Repair Manual:

 

Hi jdcollins5,

Thanks for that. It appears to me to be electrically identical (with some additional control blocks added), to the schematic shown in the ORNL document previously posted here in this thread, so I guess we can assume that's how it's wired.

However, there is something about this whole Buck/Boost converter thing that just doesn't add up. I wonder if anyone can explain it to me? Here's my understanding, and where I don't understand it...

The motor generators used in all Prii are 3-phase PMSMs - Permanent Magnet Synchronous Motors. They are similar in construction to ordinary, 'common or garden' AC Induction motors, except for one important thing. In an AC Induction motor, the rotor is made of the same basic material as the stator (magnetic iron, "Mu metal"), and has no inherent magnetic field. The field required to make the motor (rotor) turn (by magnetic attraction with the rotating magnetic field in the stator) is induced in the rotor by the field in the stator. Hence the name; Induction motor. This kind of motor always has slip - the rotor doesn't turn as fast as the field, and the more load, the more slip, hence it's an asynchronous motor. This (slip) makes them less efficient, and because the rotor field is induced, it also makes them extremely poor generators - not generators at all, for practical purposes.

This problem is solved in the Prius by inserting powerful permanent magnets inside the rotor, so that the rotor is permanently surrounded by strong magnetic fields. (There's a lot of science involved in the design of the gaps inside the rotor to direct the magnetic fields, but we don't need to know about that.) This makes the resulting motor a highly effective generator as well. And it means that because the magnetic fields are always in the same places around the rotor, there is no "slipping" of the magnetic field around the rotor as the load or other variables change. So, in this kind of motor, the rotor always turns 'in step with' the rotating magnetic fields in the stator, hence the name synchronous motor.

However, unlike the asynchronous 3-phase induction motor, it is not possible to simply throw the switch and directly apply 50 or 60Hz mains frequency, and stand back while the motor rapidly accelerates up to full speed. The slipping of the field in the induction motor's rotor is what allows the motor to accelerate 'safely'. If you tried to do this with a synchronous motor from a standstill, the motor would not be able to accelerate fast enough to make it to the next pole in time, so it would sit there growling angrily without turning, drawing enormous locked rotor stator current, until something burned out. Of course there are various methods to safely and successfully start Synchronous Motors, but we need not go into all of those.

The only relevant method is the one used in the Prius - by gradual acceleration from rest, using a Variable Speed Drive, in which the AC frequency supplied to the motor is increased steadily as the motor speeds up (synchronously with the motor speed). These synchronous motors and their associated Variable Speed Drives (VSDs aka Inverters and VVVF [Variable Voltage, Variable Frequency] Drives) are at the very heart of the Prius HSD system, and, considering all aspects of this system, the word Synergy is very apt.

Ok, now let's consider the generator aspect of the synchronous motor. Who can recall, way back in the day, a bicycle dynamo? This was a simple AC generator that was usually spring loaded and ran its toothed pinion against a serrated track molded into the sidewall of the bicycle tire. These dynamos consisted of a permanent magnet attached to the end of the pinion shaft and a wound 2 or 4-pole stator. These dynamos were very inefficient, with a lot of frictional losses against the tyre, and seemed to absorb more rider effort than pushing the bike itself.

Perhaps the most memorable feature of these old dynamos was the dismal amount of light they produced, especially at low speed. One had to pedal flat out (exhausting) to get a worthy amount of light out of the headlight. I remember, because I really wanted one, but when I finally got one, it was a bitter disappointment. The idea of being able to generate one's own power without needing a battery appealed to me, but it wasn't worth the effort. But I guess I can trace my love of the Prius back to my childhood.

Anyway, the point is that the voltage output of the AC dynamo is proportional to the speed of rotation.

You may think that a bicycle dynamo has no resemblance to the Prius Motor Generator, and you'd be half right. In fact, when operating as a generator, the two have very similar characteristics. They both produce an AC output voltage, the frequencies and voltages of which are directly proportional to the shaft RPM.

That this is true of the Prius MG when running as a generator is confirmed by the following graph from the Oak Ridge NL document. Notice that the lines are dead straight. Back EMF is another term for generated voltage when referring to a motor, and it is the existence of back EMF that causes the motor current to drop as the motor speed increases (for the same applied voltage).



My point is that when the back EMF falls below 200V (the graph is of Phase to Neutral voltage so doesn't relate precisely but you get the idea), then NO REGENERATION POWER will be consumed. The battery requires more than 200V to charge, so at motor speeds below the point at which 200V is output will not produce any Regenerative Braking.

This is in fact an argument in favour of a lower DC battery voltage (rather than higher as proposed by other posters), but the question remains - how is the Regen Braking achieved at low road speeds?

 

The DC-DC Boost converter is bi-directional (ie. For both drive and regen)

 

KiwiAI,

The Prius motor/generators remind me a lot of the PMG (permanent magnet) cooling fans I muck with regularly. First point: Those fans do have a non-slipjoint method of reporting axle position back to the controller running said fan. Second point: When these fans run, a medium-sized FET turns on at the appropriate point, causing current to flow through a particular coil on the stator; this drags the rotor around, at which point the control software notices and turns off the first FET and turns on the next in line, further dragging things around, and so on. DC voltage in: Rotating fan out, and no arcs and sparks, although I do admit that the current into the fan can get pretty interesting looking as the fan "commutes" around.

The Prius diagrams show three pairs of transistors per motor, so it's pretty clear that the motor controller does something like this, turning on pairs of transistors at a time to make ye motor go 'round and round. Further, it's pretty much a gimmee that the pairs of transistors that are turned on at a time are likely PWM controlled in order to control the amount of torque/power delivered.

Finally, as far as regeneration goes: I suspect that if all the transistors around a motor/gen were turned off, the voltages generated as the motor/gen shaft turns would get well above 200V without much trouble at relatively low motor rpms. All those Really Strong rare-earth magnets have a purpose, you know. (And for you lay people: A moving magnetic field hitting a bunch of electrons in the motor/gen windings makes EMF, you betcha. The stronger the magnetic field, the bigger the EMF, and the Prius is reported to have magnets in there that you Really Don't Want To Get Your Fingers Between, Lest You Have Mush and Not Fingers. )

KBeck

 

Hello myPriusAcct,

Yeah..... Nah! Knew that, thanks.

Actually, as far as I can work out so far, it's only Boost in the Drive direction - from Battery, to Motors. Most of the time in Regen it's actually Buck, which effectively means "losslessly" (ok, with low losses) dropping the generated voltage down to a bit over 200V to recharge the battery. My point is that there doesn't seem to be any way to capture the energy (and have any regenerative braking), when the car and the motor/gens slow down, and produce less than 200V+. And yet, from driving my G2, that doesn't seem to be the case. My question then, is How?

Maybe I didn't explain it too well, kinda rushed and tired last night. I'll try again later.

The graph of motor/generator Back EMF in my previous post is the important data from which the minimum effective regen speed can be estimated.

 

Watching the Boost Voltage with SGII it definitely boosts the voltage to 650V during regen braking.

 

G'Day KBeck,

For what you say, I suspect those probably fall under the category of "Brushless DC motors". Similar, but I think not so relevant. As I described, the Prius Motors are 3-phase PMSMs. Similar to ordinary AC induction motors, but with Permanent Magnet rotors (like the bicycle dynamo I mentioned - when acting as generators, that is).

Yep! Standard 3-phase AC motor control, pretty much identical to what is found in an ordinary 3-phase VSD. Even the parallel diodes are there, except they perform an additional, completely different function in the motor/generator setup. They create a standard, full wave, 3-phase Bridge Rectifier to convert the AC from the motor windings into HV DC, to recharge the battery.

Check out the circuit diagram for the std, off the shelf, 3-phase VSD IGBT package I posted earlier. All Toyota have done, as you have described earlier, is build the IGBT/Diode package themselves, and gone to extreme lengths to make the Thermal Resistance from junction to coolant as low as economically possible. Knowing this, I don't buy the idea that the IGBT dies are overheating due to "normal" full load currents or voltages. There has to be a software bug in the original Gen3 code that infrequently causes abnormal currents or voltages to damage the devices.

Anyway, another likely reason (haven't looked into it) that they had to build these inverters themselves is that the "freewheeling" diodes in the standard IGBT packages are not big enough to withstand full-on Regen current. So they possibly decided, well, let's do it ourselves.

Yeah, I know... Check out the BackEMF graph I posted above.

Sure, the motors generate well over 200V, over the upper 2/3 of their speed range. I recall reading somewhere that the max speed (not sure which model) with the ICE off is 62MPH. On that basis, at speeds below about 20MPH, no regenerative battery charging is possible.

So what's happening below 20MPH? Is it a trick which just looks like regen braking/ battery charging, but really, it isn't?

 

Hi jd,

Depends on what the SGII is measuring? If it's measuring the Inverter DC Rail voltage, then I'm guessing that the MGs are outputting 650V onto the rail, and the Buck/Boost Converter will be dropping some of that down to 200-odd volts and feeding it back into the battery. Otherwise, it's driving, not regenning?

Anyway, glad you have an SGII that you can take out on the road with you. I'd be very curious to know at what speed you're seeing 650V? Can you let us know what happens to the voltage you are reading when (a) freewheeling (gliding) at 20MPH, and when in Regen at 20MPH, downhill say? That would be very revealing!

Cheers,
Al

 

The boost converter is not used during regen braking when the back EMF is less than the battery voltage, instead the winding drive transistors are used and functions like a boost converter.

See diagram below.

 

Howdy bedrock8x!

Now we're talking! I get the concept - VERY cool! I can't read that diagram, too low res - but I get the idea! I couldn't figure before where they would get the extra inductor they'd need, but they must be using the idle one inside the motor. That is VERY, VERY clever. But I'm not yet convinced of it in practice.

Anyway, if that is what they are doing, then, well... It's pretty easy to see how a timing error could result in momentarily shorting two of the IGBTs across the 650V rail under certain regen conditions...

Need to think about it. But many thanks!

Al

 

Thanks, I am good.

I don't believe the whole thing is a timing issue because there are sensors all over the place for rotor position and current magnitude. The motor drive algorithm is no longer black magic and very matured. The same algorithm is used by other models with no problem. As noted in the bulletin, it is a thermal issue that causes failure or thermal design issue with this module used in gen3 Prius only.

The software fix I assume is to limit the boost voltage in consequence reduce drive current to the inductive winding,
and the lower voltage and current will let the transistor be in lower region SOA.
There is no current control by the MCU but only voltage control.

 

The SGII description is Voltage After Boost which would be the Inverter DC input voltage. The only time that I see 650 VDC is on heavy power demand and on regen braking. The other times it is same as HV battery volts at low speeds or 500 VDC above about 30 to 35 mph or normal acceleration. I have not seen 650V at normal highway speeds unless accelerating. Normal at highway speeds is 500V.

The change from HV battery volts to 500V and then 650V is always a step change. Except from 650V to HV battery volts on regen braking.

What I have seen during regen braking is 650V initially and then declining fairly linearly to HV battery volts somewhere prior to 7 mph when friction brakes kick in.

I will pay attention to volts at 20 mph gliding and downhill and let you know.

 

Hi again,

Sorry, I was referring to the timing of the ON pulses to the IGBTs, being generated by the software, rather than position errors. It's quite possible for the sensors to be telling the software the exact truth, but due to an error in the software, it outputs a signal at the wrong time.

Fair comment. But this very cunning regen system may be less mature? Besides, what's to stop the Toyota programmers simply making a mistake? Presumably, they would have had to modify it from the Industry Standard algorithm, to make it regulate (reduce) battery charge rate when the battery starts to warm up. Funny that they seemed to have got it right in the G2, but messed up in the G3. But Hey, just look at Microsoft. Every new release of Windows recreates half of the old bugs all over again, plus a whole swathe of new ones! Doesn't seem to have stopped Microsoft dominating the Desktop for many years.

Ok, I'm outvoted. ;-) But I don't tend to believe what the spin doctors say. Whether the bulletin contains spin, I can't know, but I think we can say that "the Boost Converter" could be a misleading term, since it appears that there are effectively two of them. One of them, very simple, and, I submit, very easy to "Test to Destruction" under extreme overload, while the other one is much more "tricky", embedded within the motor as it is. I'd liken it to the Knife-Finger game. Get it slightly wrong...

My vote is still with the Regen Boost Converter...

I guess we'll just have to wait and see! I mean, it seems odd, given how much they improved the thermal design over the G2, that they would get the SOA of the standalone boost converter wrong. Here's hoping we one day find out...

 

Um. Sorry, but I believe the proper answer is: Mechanical brakes. My understanding from reading Prius Chat and the car manual is that's precisely what happens at low speeds and is part of the Magic Sauce that Toyota puts into their brake control system. (And why it's so blame expensive if anything brake-related lets go under the hood.)

KBeck

 

If that were the case (Mechanical brakes from 20MPH down), the car would not be able to achieve around 4 l/100km in city driving. Wasting too much precious energy. As above, thanks to bedrock8x, we know it's much trickier than that. Remember E=mv2 (m.v.v). Having to (mech) brake below 5MPH is not a big deal.