Two MG1 Questions

Bob,
There are times when you just want to do your job as best you know how.

I would point out that this choice never failed him in the movie, he let his expert make the wise choice.

 

Hi Vincent

Sorry for being late with the answer, but I have been very busy last weeks at work.

There are two PIDS that indicate the MG1 torque:
7E2 2167 AB => MG1 torque
7E2 2167 CD => MG1 torque execution value

As far as I know, the first one is the MG1 torque target calculated by the Hybrid ECU 7E2 and the second one is the actual MG1 torque value obtained by the whole system (PSD and car kinematics). In the case of regenerative braking, the torque execution values of MG1 and MG2 are compared with the target ones and the difference is applied to the friction brake electric actuator so the whole car behaves as expected by the driver. You can see target and actual ICE RPM in the PIDS 7E0 2149 C (target ICE RPM) and 7E0 2101 JK (actual ICE RPM).

When the car is in EV mode (or the ICE is stationary) the first PID (7E2 2167 AB) reads zero all the time; but the second PID (7E2 2167 CD) reads a small negative amount (-0.13Nm to -0.63Nm) that is applied to negative MG1 RPM (-2000RPM to -3000RPM at 30-40 Km/h) so it provides a small traction power. MG2 torque for these speeds and on a level road is in the range of 15Nm-35Nm.

Now you know the reason why I said MG1 works as a motor when the ICE is stationary. If I try to find a mechanical reason, I believe that MG1 pushes (to help MG2) a small amount using the ICE rotation friction when it is stationary as a fulcrum. If MG1 "pushes" too much in this situation the ICE would rotate; that happens when the ICE starts up, MG1 is requested to apply some +45Nm of target torque although it reaches some +36-+40Nm. Note that in this situation the MG1 torque is positive and therefore reversed to the case when ICE is stationary. To compensate that "braking" torque (to the whole car) MG2 increases its target torque in the same 45Nm so the whole car does not notice that the ICE is starting up and needs a lot of torque to do that.

Big hugs from Frank

 

Hi Frank,

Thanks for your replies!

Yes, there are 2 PIDs for MG1 Torque. One for calculated torque (MG1 torq) and the other for actual torque (MG1 Trq Exec Val) as mentioned by you. Calculated torque is the desired torque and actual torque is the measured torque.



There are indeed some small torque (sometimes -ve, sometimes +ve) as seen by the small ripples in the diagram.

However, I do not think you can use measured torque as a basis that MG1 acts as a motor. The measured values only serves to be feedback to HV ECU and it will then recalculate how much off in it's initial calculation and then compensate for it by controlling a new torque command to MG1 and MG2. This new torque is the desired torque and that is why you see MG1 is always controlled to have zero torque when ICE is stationary.

Let me quote a description from the NCF (Pg 52 ~ 53):

Vincent

 

Hi Vincent and other Prius owners

Just to add some information of the ICE starting process. Below are the charts of the hybrid battery current, ICE RPM, MG1 power and MG2 power during ICE start-up process. Readings are the average of 100 ICE start-ups in two city trips and are taking every 0.75 seconds.





I want to remark that 58A (12.5kW) are drained from the battery at the critical moment; 2kW go to MG1 and 9.4kW go to MG2 despite the common saying that MG1 starts the ICE. The remaining power (1.1kW) is lost in the inverters and in the 12V electric system.

I hope you like it

Big hugs from Frank

 

I'm impressed!!! Thank you, Frank!
So, from this incredible logging, we can assume that every ICE start draws about 5kw average during 2 sec?
Then the energy spent to start the ICE is roughly 10KJoule, or 2.7Wh... So little energy?!

 

Hi Frank,

My experience is a warm-start takes less than 0.25 seconds. Your sampling interval is so large relative to the speed of engine starting, the 'average' of multiple samples can distort what is going on. Good Prius Friend Hobbit has used an oscilloscope, +10 MHz, to take metrics and I've used two audio channel of a laptop, 40 KHz, to make some high resolution metrics. These are fast enough to observe fast events such as engine start and stop events:


More details here:


I have dealt with a much slower, miniscanner, about six OBD samples per second. I take pains to avoid report things happening faster than the Nyquist limit. My rule is to drop data when there are not at least two consecutive points with the same value. Otherwise, sampling errors can lead to some impossible values (i.e., ICE turning in reverse.) In particular, the 'smear' of data for the current leading before the start event is not quite accurate for such rapid events.

Let me suggest there may be high-speed, bluetooth or WiFi data capture devices that could be connected to some of the faster signals and snap-shot or buffer these rapid events. The amount of data grows by orders of magnitude but so too does the accuracy and insights gained. Nyquist keeps me from seeing things that are not there or distorted.

Bob Wilson

 

Hi Bob

Nyquist rule must be observed for a wave-like parameter, however ICE RPM in the first non-zero reading at 0.75 sec sampling may be assumed to be a monotonically increasing function. The histogram of the 100 first non-zero readings (after a zero reading) of ICE RPM are:



Now comes the magic. If we assume than the ICE RPM during start-up is a monotonically increasing function of time, the samplings were at random times (not related with ICE state) and all ICE start-ups had the same RPM=f(time) pattern we can plot them in order as this:



Obviously if ICE RPM were a wave-like function with 0.25sec period as you suggested, the graph would be totally wrong as Nyquist said. However, I think it is fairly good; as the earliest second readings (each one associated with the corresponding first reading) are close to the latest first readings. I think that the HyBattAmps graph of my previous post is good. MG1 and MG2 Power graphs seems to be also good enough in points other than 9 and 10.

Big hugs from Frank