Higher lowest HV battery voltage. Boosting? Need advice

[m.art.y]

Hi. So I have an issue - I've got some rather unfortunate hardware (inverter and dcdc) that will shut down below 330 VDC but will run up to 450 VDC. I wonder what would be my options to raise HV battery voltage? I've got a 96 cell (~80 Ah) battery pack like most. Adding more cells could be an option but would be tricky due to cost and number of cells needed. What voltage do you usually run your cells down? If I run my cells down to 3.2 V then I would need around 104S battery pack. That's 16 extra 40 Ah cells! As another option I wonder how reliable and how energy efficient would be to boost voltage a little with a boost converter from toyota gen3 prius inverter which I have. I know some has tried boosting HV voltage when testing MGR but how realistic would be to boost voltage in the long term? Yet another option to mess around with voltage sensing circuitry to trick the inverter and dcdc converter into thinking the voltage is higher than it really is. But would it still work? Does inverter care what voltage it is switching? Thanks

[johu]

I run my batteries down to 300V.
What sort of power do you need?

[bexander]

I think your best option is to try and modify the voltage sensing circuits in the inverter and DC/DC to run at lower voltage. Usually there are some kind of voltage division that can be modified.

[m.art.y]

I run my batteries down to 300V.
What sort of power do you need?

The inverter is rated for 35 kW continous and 50 kW peak. So my idea was to boost voltage a little only when it drops to 330 V to be able to use the range left in the battery as inverter/dcdc shuts down at 327 Vdc.

[m.art.y]

I think your best option is to try and modify the voltage sensing circuits in the inverter and DC/DC to run at lower voltage. Usually there are some kind of voltage division that can be modified.

I would prefer that too. So this would essentially be tricking inverter into thinking battery voltage is higher than it is? No adverse effect on how inverter works?

[bexander]

I would prefer that too. So this would essentially be tricking inverter into thinking battery voltage is higher than it is? No adverse effect on how inverter works?

Correct. I think the inverter should work fine regardless of the input voltage but the DC/DC might be more sensitive to voltage changes. However, changing the min voltage from 330V to 300V is a change of less than 10% and should not be a problem. I suspect that the min voltage level is selected to match the originally used battery voltage and not the minimum voltage the equipment can tolerate.

Have you looked inside the inverter?

Boosting the voltage will introduce more losses. I think the gen3 buck/boost is rated 27kW so there is a limiting factor as well.

[m.art.y]

Correct. I think the inverter should work fine regardless of the input voltage but the DC/DC might be more sensitive to voltage changes.

Turns out you're spot on. I've messed around with it a bit more and after some elaborate procedure I've managed to start the inverter at 298 Vdc! And it seems to work fine. However dcdc converter that is attached to the inverter does not start below 334 V so it was not working. I think dcdc is the culprit here as when it does not start it lets inverter know that it had not started and therefore inverter does not start. Any clues how to search for that voltage sensing circuitry?

[bexander]

Best guess is to look for a thin wire connection of HV to the PCB followed by some voltage dividing resistors.
If you can post a picture of the control PCB inside the DC/DC I might be able to help.

[m.art.y]

If you can post a picture of the control PCB inside the DC/DC I might be able to help.

It's a bit hard to take a good resolution photo of entire pcb but I uploaded a bunch of pictures here:

https://mega.nz/folder/p0gD3C5Q#RntCVUCWYZfwLnE3e8hBRw

If you need me to zoom in on something particular I will. I could not see any extra wires going to HV terminals but HV wires come directly into PCB where the big screw holes are. I marked which is HV+, HV- and grounding to the casing. I noticed that after voltage was too low for it and I raised voltage till it was happy I heard a click and it started working. I could trace something with the multimeter if you tell me what exactly. I don't really understand how voltage sensing works. I found high voltage comes to those 2 sets of rather large resistors on the first board?

[bexander]

Can you find out the resistor values for the three resistors above the IC circled in red? The marking is to fuzzy when I zoom in. Or take a picture of these resistors.
Also, can you see and component number on the IC circled in red?

[m.art.y]

Can you find out the resistor values for the three resistors above the IC circled in red?

Hi, thanks. It was a job to read the markings on that chip but it looks like ST chip with number 2903Y? Could also be 2903V as it is so hard to read. I attach a zoomed in photo.

[bexander]

It's hard to find a IC with 2903 and TSSOP14 packing but it is likely some kind of comparator. lm2901 makes more sens in that package.

On the connector is the left HV- and the right HV+ circled in red. Needs to be connected to the other board I guess.
There seem to be a voltage division going on into the assumed comparator IC, marked by the red line.
HV+ - 221k - 221k - 221k - 47,5k - 7,5k - sens - 10k - HV -??
Would result in close to 4,5V at 330V input, if the above is correct.

Can you read what IC I've circled? A voltage reference or voltage regulator perhaps?

[m.art.y]

Can you read what IC I've circled? A voltage reference or voltage regulator perhaps?

Yes, the former chip is definitely 2901Y when I looked again. And this other chip that you circled is 4264 G.
You're right all last 4 pins on the connector that I circled are HV+ as I can trace continuity to the HV+ terminal on the top PCB. I cannot trace any continuity to the HV- terminal and the connector pin on the left or in fact any of them. Perhaps HV- does not come in there directly and therefore multimeter does not pick it up? I attach a photo of how the top PCB connects to the bottom PCB.

[m.art.y]

All last 4 pins are HV+ that I circled on the top PCB.

[bexander]

Ok so the 4264G seem to be a 5V regulator.
My conclusion from this is that these resistors form a voltage division of the HV+ down to some kind of gnd (close to HV-).
HV+ - 221k - 221k - 221k - 47,5k - 7,5k - sens - 10k - GND1. Sens connected to 2901 pin 8 or 3IN-. 330V HV+ input results in 4,53V at pin 8.
There is a 01D (100k) and a 01C (10k) resistor to right of the 2901 which likely form a voltage division from +5V - ref - GND1. Ref connected to 2901 pin 9 or 3IN+. Results i 4,53V at pin 9.

Can you measure connection as marked in this picture to verify my guesses?

The ultimate test would be to decrease the resistance of one of these resistors to 150kohm (1503). Or you can connect a 470kohm resistor in parallel with one of them for a result of 150kohm.

This should then decrease the min voltage threshold to just below 300V.

[m.art.y]

Can you measure connection as marked in this picture to verify my guesses?

Ok I traced the continuity of 01D and 01C and I attach a photo. If I missunderstood you please clarify and I will try again. I could not trace any continuity to the other paths that you had marked.

[bexander]

Ok, so my guess was incorrect.
Can you get any continuity between colour matching points?

[m.art.y]

Can you get any continuity between colour matching points?

Sadly I could not trace continuity between all of the marked points. The board is sprayed with protective varnish that makes it more difficult. I attach a photo where I found continuity.

[bexander]

At the moment I have no other idea then to test to reduce the resistance in one of the 221kohm resistors (marked 2213) and see if the low voltage level decreases. Make sure that any resistor placed there can withstand at least 150V. Easiest way is to parallel with another resistor.

[m.art.y]

Make sure that any resistor placed there can withstand at least 150V.

Ok I just put 470k resistor in parallel with the 221k one and nothing exploded. And dcdc started at 298 Vdc! You must have some training to be able to calculate that as I had no idea.
Inverter also started. Well inverter didn't like going under 300 Vdc too much but just above it seemed to be fine. Dcdc gave ~13.53 V as usual.
I only had a through hole 470k which was 0.5W I think, not sure for what voltage it is rated. Do you think it would last or do I need to look for a higher voltage one?
Also what does this mod essentially do? Does it trick the dcdc into thinking the voltage is higher than it really is? And how does this affect the upper voltage limit?
Thanks a lot!

[bexander]

That is great news!
I think a through hole resistor will be just fine voltage wise as the capsule usually is quite large. For a more permanent fix I would use very short connections and maybe some glue to make it more vibration resistant. Hot melt glue usually works fine.

The mod lowers the input voltage threshold. Will probably lower the high voltage threshold as well. I guess in some sense you can call it to trick the DC/DC into thinking the input voltage is higher than what it really is.

The calculations is a follows:
3pc of 221k in series with 1pc 47,5k and 1pc 7,5k sums up to a total of 718kohm. There is also a 10k to gnd so the voltage division output voltage into the 2901 IC is
330V*10k/(10k+718k) = 4,53V
This tells me the input threshold is 4,53V and so one need to change the voltage division to end up with 4,53V at 300V.
To get the new needed resistor value
300V*10k/(10k+R) = 4,53V
300V*10k = 4,53(10k+R)
300V*10k = 4,53*10k + 4,53*R
R = (300V*10k - 4,53*10k) / 4,53
R = 652k
R was 718k and now needs to decrease to 652k, thats a difference of 66k.
To do this one needs to reduce one of the series resistor values by 66k and the simplest is to reduce one of the 221k resistors by 66k to 155k, however this is not a standard value so 150k is the prefered choise as this small (5k) difference does not matter so much.
To just connect another resistor in parallel for a total of 150k the calculation is as follow:
1/150k = 1/221k + 1/R
1/150k-1/221k = 1/R
R = 1/(1/150k-1/221k)
R = 467k
Closest standard value is 470k.
Note that by paralleling one could get closer to 155k but now it is possible to replace the 221k//470k to one 150k resistor without changing the threshold level, if desired.

[m.art.y]

Note that by paralleling one could get closer to 155k but now it is possible to replace the 221k//470k to one 150k resistor without changing the threshold level, if desired.

So I tested a bit more and dcdc stops working at 295 Vdc. Also I tested upper voltage limit and I got to 415 Vdc that dcdc still worked and I did not try any higher that should be enough for me. To start the dcdc I used a light bulb in line with HV for first few seconds just in case and I noticed that after the mod from 380 V to 415 V the bulb lit up maybe twice as bright at those voltages than before the mod. So I was a bit cautious if something might be wrong? But nothing exploded and dcdc worked.

Do the 221k resistors see the full HV voltage? I checked and the resistor I had purchesed is metal film and is rated for 250V. Do you think this would be fine in this application? To later replace it if it blows would be a pain as so hard to access. Thanks again.

[bexander]

Do the 221k resistors see the full HV voltage? I checked and the resistor I had purchesed is metal film and is rated for 250V. Do you think this would be fine in this application? To later replace it if it blows would be a pain as so hard to access. Thanks again.

One of the 221K resistor gets less than 1/3 of the total voltage.
By OEM: 450V * 221k / 728k = 137V.
After modification: 450V * 150k / 657k = 103V max. The remaining 221k resistors will get a slight voltage increase: 450V * 221k / 657k = 152V.
If you never go higher than 415V then the remaining 221k resistors get: 415 * 221k / 657k = 140V. So really no change to the stock configuration.

Using a resistor with 250V rating gives a good margin, and should work well.

FYI: This is most likely why there are 3pcs of 221k resistors in series instead of just one 663k resistor, to increase the tolerated voltage.