CCS charging using Prius buck-boost converter

[MoonUnit]

Thanks to @MattsAwesomeStuff I've been able to edit the wiki and add more flesh to what I've been doing.

Link is here https://openinverter.org/wiki/Operating ... pplication.

Updates to follow, apologies as progress is a bit slow at my end.

[MattsAwesomeStuff]

Hmm, not sure if you intended to do this, but...

You didn't use the page I created. You created a really long page called "Operating the buck/boost converter for a low voltage CCS charging application"

And as a standalone article, it's not clear what "the" buck/boost converter is, as there's no Prius context in the title or early in the page.

Content-wise, otherwise, great research.

[MoonUnit]

Hi Matt - I sent you a PM on this. You are correct, I didn't mean to do that, so I can either leave as is or move the content to your page which I think is currently orphaned. I apologise ...

[MattsAwesomeStuff]

Hi Matt - I sent you a PM on this.

*checks*

... You, did not.

And no worries. Just copy/paste the code into the page I linked above. It'll be done.

[MoonUnit]

I have been trying to get an accurate measure of the DC bus voltage.

The buck/boost converter outputs 'VL' which I thought was a 100:1 scaled voltage measurement of whatever the orange sense wire is connected to. By default, it's connected to the positve post of the battery input and hence measures battery voltage. According to Johannes' charger video, he saw 0.9ish volts when applying 90ish volts in. I already know the battery voltage being input, so I wanted to try using the sense wire to measure output, boosted voltage on the DC bus, but I am not getting a simple 100:1 reading.

When the sense wire is not connected, the VL line is about 0.54V. When it is connected to a range of DC bus voltages from 10-100V, I get a linear relationship between DC bus voltage on my meter and the voltage on VL but it's not 100:1. I can plot and correlate the two, and get an equation for the line but I'm not happy about this - small errors lead to big differences (c.10%) in predicted vs measured voltage.

I'd like to find some information on how the VL value is officially interpreted. Does anyone know? Maybe it's not supposed to be very accurate?

[uhi22]

Or it is intentionally with an offset. To make it possible that the controller can distinguish between the cases "sense circuit broken" (0V) and "no HV" (0.54V). Just find the equation, and find out whether the result is stable enough.

I had a look to the wiki, quite well described
I think you need two contactors, between the converter and the CCS inlet. Because during the cableCheck phase, the EVSE will apply 500V to the DC pins, and expects that nothing is connected in this case. Without the contactors, there would be the capacitors, the discharge resistor, this may make the cableCheck fail. And even if not, could the 500V kill something?
For a first test, you could just boost to 230V, and configure clara to use standalone mode with 235V, this should give a first charge experience. And in a second step to measure the boosted voltage and feed this via CAN to clara as battery voltage, together with a 250V voltage limit, and a charge current of some amps. The charger will provide this current, so you do not need to care for adjusting the PWM ratio.

[MoonUnit]

Or it is intentionally with an offset. To make it possible that the controller can distinguish between the cases "sense circuit broken" (0V) and "no HV" (0.54V). Just find the equation, and find out whether the result is stable enough.

I think the problem was a poor ground reference for the measurement which was giving an offset. It's to do with poor prototype board development. If I correct for that, I can get pretty close to 1:100 with zero intercept. Close, but not exact, and the more I boost the less accurate it is. I have decided that I will use an ISA shunt, so I've ordered a cheap used one off Fleabay and will see if it works.

I had a look to the wiki, quite well described

Thanks. I hope to have some progress to add to it.

I think you need two contactors, between the converter and the CCS inlet. Because during the cableCheck phase, the EVSE will apply 500V to the DC pins, and expects that nothing is connected in this case. Without the contactors, there would be the capacitors, the discharge resistor, this may make the cableCheck fail. And even if not, could the 500V kill something?

Completely agree, and the contactors are part of the plan. I don't think they would need to be able to pre-charge the DC bus, it would only be if I could not reliably put boost voltage on the DC bus, hence the ISA shunt to be sure of the voltage first!

For a first test, you could just boost to 230V, and configure clara to use standalone mode with 235V, this should give a first charge experience. And in a second step to measure the boosted voltage and feed this via CAN to clara as battery voltage, together with a 250V voltage limit, and a charge current of some amps. The charger will provide this current, so you do not need to care for adjusting the PWM ratio.

This is really helpful, thank you. I assume that in the second step you describe, the charger will hold the voltage at 250V and a constant current?

[uhi22]

The charger will limit both, the current and the voltage, to the values we request. This means, if we request 5A and 300V, the 5A will be consumed by the buck converter, it will feed eg 10A into the battery, the battery voltage will rise from 120V to 121V, and the voltage on the CCS from 240V to 242V. (just assuming 50%PWM for the example). The voltage on battery and charger will slowly further increase, until the 300V are reached, 150 on battery. Now the charger is in constant voltage mode.

So the goal is to send a healthy current limit (to not overstress the battery, cables and inductor), and a voltage limit to not overshoot the 4.1V cell voltage. Both values could be fix.

[MoonUnit]

I decided not to use the isolated voltage level that the buck/boost module can supply as I wasn't confident I could read it accurately. It's normally connected to the Prius battery input voltage. There is a high voltage signal available which measures the boosted DC bus voltage, but for that you need to power the inverter stage electronics which I am going to ditch, and I had similar inability to decode it accurately.

So I opted for an Isabellenhutte ISA shunt as used by others here. They are expensive to buy new so I cut a corner and ordered one from Ebay which was shipped internationally to me and cost 200 EUR and took a fortnight. They sent me the wrong one, so I had to wait another week for the right one to come but they did not request the other one back so I now have two, happy days. It took me a full week to get them working and I will link to my solution below, but in a nutshell, don't trust cheap CAN transceivers from Amazon, and read other people's code fully before you decide to use it.

My plan is to use the ISA shunt to measure both input voltage (my pack voltage) and boosted voltage (which I will 'present' to the EVSE). I can already measure the current into the pack with an existing current sensor. I will probably use the second ISA shunt to measure current from the EVSE at boosted voltage. This may not be needed in the future but given I have the bonus sensor, I'll probably use it.

Link to my sensor initialisation solution viewtopic.php?t=2117&start=35