DC/DC converter in Auris/Yaris and Prius Gen3 inverter

[bexander]

In my testing today, I ran into problems with the precharge vs DC/DC converter in my Yaris inverter.
I'm using a 140V pack voltage connected to the HV-bus side. Using a 40ohm precharge resistor.
I also have a short program in the 328p to set the high-side IGBT ON in the buck/boost, to supply power to the DC/DC side (input DC bus). This is done at startup so the precharge will include the capacitor here as well.

Problem is that when using a jumper on the CONN3, the DC/DC starts immediately and starts drawing current thorugh the precharge resistor and the main contactor is never set due to to low bus voltage.

To fix this I removed the jumper (CONN3) and instead placed a P-FET over the CONN3 pins and connected the gate to the Main contactor control IC (IC8). And also added some RC-delay at around 100ms so that when the main contactor is triggerd there is a further 100ms delay before the DC/DC is allowed to start, to ensure that the DC/DC does not start before the actual main contactor contacts are closed.

How have you solved this problem? Or have I missed something and solved a problem not needed to be solved? Or should something be implemented on the logic boards in the future? Or some other means of controlling the DC/DC?

[Jack Bauer]

Thats a good idea. I'll add it to the design for the next rev.

[bexander]

Thats a good idea. I'll add it to the design for the next rev.

You might figure out something better but here is my solution.

DCDC_delay.pdf

(27.15 KiB) Downloaded 270 times

[180jacob]

Going to try your idea out on some logic boards. As you say there are probably countless ways of doing it but to reduce the chance of error I have copied exactly as you have drawn. Shout if you can see anything wrong or would recommend doing it differently.

[Bigpie]

Just added this info to the wiki, will update the small section when there's a new revision.

[bexander]

Going to try your idea out on some logic boards. As you say there are probably countless ways of doing it but to reduce the chance of error I have copied exactly as you have drawn. Shout if you can see anything wrong or would recommend doing it differently.

The P-fet you have selected have a lower VGSth of typically -0,9V as compared to the one I used which, if I remember correctly, is typically at -3V.
This will result in a shorter delay-time. Depending on what main contactor you are using it might be a good idea to adjust the RC-delay or select another p-fet with higher threshold voltage.

[180jacob]

Going to try your idea out on some logic boards. As you say there are probably countless ways of doing it but to reduce the chance of error I have copied exactly as you have drawn. Shout if you can see anything wrong or would recommend doing it differently.

The P-fet you have selected have a lower VGSth of typically -0,9V as compared to the one I used which, if I remember correctly, is typically at -3V.
This will result in a shorter delay-time. Depending on what main contactor you are using it might be a good idea to adjust the RC-delay or select another p-fet with higher threshold voltage.

Yes I did think it was going to be a bit brief but it was getting too late at night to think about it and that was the only P-fet in JLC's basic parts list in a SOT-23 package. I will have a think what some more preferable values would be or squeeze a different package in there, would rather not use extended parts for a feature not everyone will use.
Thanks for the advise.

[180jacob]

Ok I've decided to stick with the sot-23 p-fet, and have changed the resistors to 22k and 100k giving a VGS of -2.1V and delay to the -0.9V VGSth of about 100ms, could swap the cap for 22µF to be extra sure its long enough I suppose. But 10µF, 22k and 100k are already on the BOM so keeps the cost down.

[bexander]

Ok I've decided to stick with the sot-23 p-fet, and have changed the resistors to 22k and 100k giving a VGS of -2.1V and delay to the -0.9V VGSth of about 100ms, could swap the cap for 22µF to be extra sure its long enough I suppose. But 10µF, 22k and 100k are already on the BOM so keeps the cost down.

Adding a useful delay without increasing BOM, to me, that sounds great!
I think 100ms is quite enough. The conntactor datasheets I have looked in (Kilovac LEV200 and Panasonic AEV) have a set (on) time of 50ms or less.
If someone have a very slow conntactor it is possible for them to adjust the component values.

[m.art.y]

I also have a short program in the 328p to set the high-side IGBT ON in the buck/boost, to supply power to the DC/DC side (input DC bus).

Hi, would you share more in detail how you got DCDC going and your code? Do we still need to do a resistor swap to be able to DCDC from 360 V battery or so? Regarding a delay I could just place a relay across the DCDC enable jumper that would be wired to precharge contactor. When precharge contactor would switch off the relay would then connect the jumper to enable DCDC as a temporary measure to just get it going. I suppose you are modifying your boards or adding some add on circuitry in there?

[bexander]

This was done when I feed the battery directly to the MG DC bus. If you feed the battery to the battery side DC bus there is no need for a any code in the 328p.

Anyway, here is the code I used:

Code: Select all

// I/O-PINS
const uint8_t boostLoDrivePIN = 9; // D9 (PB1)
const uint8_t boostHiDrivePIN = 10; // D10 (PB2)


/********
* SETUP *
********/
void setup()
{
  pinMode(boostHiDrivePIN, OUTPUT);
  digitalWrite(boostHiDrivePIN, HIGH); // To set high drive ON
  
  pinMode(boostLoDrivePIN, OUTPUT);
  digitalWrite(boostLoDrivePIN, LOW); // To set low drive OFF
}


/*******
* LOOP *
*******/
void loop()
{
  
}

Still needs the resistor swap to run DC/DC at 360V battery.

There are some other code created by JB that will run the buck/boost as a buck to be able to run DC/DC unmodified at 360V but that will require battery connected to MG DC bus.

EDIT: JackBauer's code can be found here:
viewtopic.php?f=14&t=825&hilit=Runs+atm ... =25#p12720

[m.art.y]

This was done when I feed the battery directly to the MG DC bus. If you feed the battery to the battery side DC bus there is no need for a any code in the 328p.

Sorry I am confused which bus is which. Do you mean if I connect 12 V battery to this post and enable DCDC jumper on the board it will work without the code?

https://openinverter.org/wiki/File:20200705_190706.jpg

And also do you mean that for DCDC to work without resistor mod 12 V battery needs to be connected to MG1 terminals (which ones)?

[arber333]

This was done when I feed the battery directly to the MG DC bus. If you feed the battery to the battery side DC bus there is no need for a any code in the 328p.

Sorry I am confused which bus is which. Do you mean if I connect 12 V battery to this post and enable DCDC jumper on the board it will work without the code?

https://openinverter.org/wiki/File:20200705_190706.jpg

And also do you mean that for DCDC to work without resistor mod 12 V battery needs to be connected to MG1 terminals (which ones)?

Look here and read some posts forward: viewtopic.php?f=14&t=51&start=25#p939

[Bigpie]

There's 2 ways to enable the DCDC (to output ~14v)

1/ One way is to change the resistors in the DCDC and bridge the battery + to bypass the buck boost. So on the above image, your battery + will also connect to the first of the 3 if you want to do it externally without any code.

2/ Another way that Damien has used in one of the videos (viewtopic.php?f=14&t=825&hilit=Runs+atm ... =25#p12720) is to buck down the battery voltage to an unmodified DCDC using the Atmega. So battery + only to the last of the 3 as shown above.

[bexander]

Here is an overview of the toyota inverter.

If connecting the battery to the "HV Battery", at the bottom in the picture, you will feed the DC/DC directly and resistor modifications are required if you are to run 360V.
If connecting the battery to the "Inverter", center of the picture, you will need either to set Hi-side IGBT on in the boost converter by using my code in the 328p and modding resistors in the DC/DC OR using a bridge connection between the unmarked screw (the left one) and HV+ in Bigpie:s picture to bypass the boost converter and modding the resistors in the DC/DC OR use JackBauer:s code to use the boost converter as an buck to reduce the voltage feed to the DC/DC, no mods needed to the resistors in he DC/DC.

[m.art.y]

Hello, I chose to change out the resistors and connected the HV battery to the DC/DC tab as per above picture. However it does not want to work. Does it need to have 12V connected to give any output? On the V1c I only bridged dc/dc on/off connector and not the dc/dc vsel. Does dc/dc vsel has to be bridged too? I got the dc/dc to work once with 12V battery connected with jumpleads outside the car it was outputting ~15 V. But that made encoder and throttle signal very noisy. Now I put the battery inside the car and dc/dc does not start again. Any ideas please? I got 280 V on the HV battery

EDIT: dcdc does need 12 V battery connected to work. I can see it drawing current through the precharge resistor and it outputed ~14 V to a separately connected battery (through a precharge resistor still). I tried to engage the main contactor and I heard some sort of ongoing clicking noises from the inverter and I measured 18V at the connected battery! I guess the above described delay is crucial to get it going.

EDIT2: No something is wrong. I have only enabled dcdc now after main contactor has been closed and dcdc seem to start I can see it briefly show just over 15V and then it goes off again. When I also attach another battery (that is not connected to the car) just to the inverter case and dcdc output I get over 18 V output on the connected battery and I can hear a ticking noise that seem to come from dcdc converter. Does it matter to which part of the inverter case negative is connected? Maybe I accidentaly bridged something when replacing resistors I don't know. I had not tried to run it while resistors were not yet changed. Sad.

[MattsAwesomeStuff]

I tried to look this up on the Wiki, but it (currently) only briefly explains it and then says to follow details below, and below, it just has a note for someone to add details.

If anyone does know the answer, please update the wiki too (or I'll try to, if I notice this thread again).

https://openinverter.org/wiki/Toyota_Pr ... _Converter

[Bigpie]

I've not changed the resistors yet, so haven't taken photos and documented it. I'll trawl some threads and add some details for the options.

EDIT
I've stolen some screenshots from DM's video and added some details. I'll take more detailed photos when I get to it.

[m.art.y]

I'm not entirely sure what was wrong with it, I resoldered the resistors, cleaned some left over dried brownish flux from some of the pins, grounded the inverter case better and it works now. On start up it jumps to 16 V but then drops to ~15 V. My plan with a relay seems to work well. I used a relay that is closed when not energized and connected it to conn3 (dcdc enable). I feed the relay from the precharge contactor. When precharge contactor is energized my relay disconnects conn3 allowing inverter to precharge properly. When I put the inverter in run mode precharge relay switch off together with the relay enabling dcdc and main HV contactor closes also - I have not had it not closing yet.

[m.art.y]

I don't know what is wrong with that dcdc converter it stopped working again. It seems to start but immediatelly cuts out. This time it is burried deep inside the car. Is there a safety feature or something maybe it does not like something or is just faulty?

EDIT: Thought I will update that the dc/dc converter in this Prius gen3 actually works. It seems like it does not like if the 12 V battery is too low, if the battery is kept charged up it starts every time. Using an always closed relay I don't allow dcdc to start during precharge - relay is powered off a precharge contactor and it breaks the circuit when precharge contactor is triggered, then when precharge contactor is switched off and main contactor is engaged power to the relay is cut and it closes enabling dcdc.

[bexander]

When using the DC/DC without the control-pin "VLO" CONN10:pin5 on my logic pcb the DC/DC does a overshoot of up to 15,7V for approx. 2s at startup and then stabilizes at 14,6V. I don't like this behavior and I also would like a bit lower voltage so I decided to use what is done here: viewtopic.php?p=8485#p8485.
I'm using a Prius gen3 inverter and I don't get the same results concerning duty-cycle and adjustability. I can't get the voltage below 14,0V and also a lot less duty-cycle swing to get different output voltage. Anyway, when I control the "VLO" there are no longer an overshoot of the output voltage during start and I can also get 14,0V output, which is what I want, so now I'm looking into ways to implement this more permanently.

For my own purpose I can use the AC-precharge pin of the 328p micro controller but a more universal solution would be to use the STM32. On my logic board pin3 and 4 of the STM32 are not in use as X3 is not mounted. Would any of these pins be usable to create a square wave signal with a variable duty-cycle that can be adjusted from the web interface, with a frequency anywhere between 100 and 1000Hz?
It will also require adding a transistor and a couple of resistors to get a open collector/drain output for the "VLO".
Does anyone else have interest in being able to control the DC/DC output voltage?

[m.art.y]

When using the DC/DC without the control-pin "VLO" CONN10:pin5 on my logic pcb the DC/DC does a overshoot of up to 15,7V for approx. 2s at startup and then stabilizes at 14,6V.

That's exactly what I've seen on my prius gen3 inverter and wrote in the post above. I think even when the voltage stabilizes it is still a bit high? If there would be a chance to adjust it that would be great.

[bexander]

I now know why the DC/DC creates an overshoot when it is turned off. The CONN10:pin8 is a voltage sens pin and this is by default connected to 12V_IN on the logic board. This results in the sens pin gets open circuit when the ignition key is turned off resulting in that the DC/DC tries to increase the voltage as much as it can with the remaning energy stored in its internal capacitors. I have measured close to 24V for about 20ms duration when turning key off.

I decided to test and connect this via a separate cable to the 12V battery instead, a true sens cable, and the result is that there are no overshoot when turning key off.

With the above sens cable connection I can also set the DC/DC to produce as low as 13,7V, just as Martin1775 got.
I have also put the AC-precharge output PIN from the 328p to use by letting it control the "VLO" for the DC/DC and by outputing a 19,5Hz signal with a duty-cycle of 52% I get 14,0V at the 12V battery and no overshoot at startup nor shutdown.

[Jack Bauer]

Can we please document this in the wiki and ideally submit on github so I can get it added to the next rev of boards.

[bexander]

There are already a note and a link to this thread on the wiki regarding the DC/DC. Should I try and extend the information given on the wiki regarding the DC/DC? Might be better to wait until a permanent fix is in place?
What do you want me to submit? I think it is better if we can use the STM32 to generate the VLO signal but that will require some work from johu or someone else as I only can write code for the 328p.