Valence u27-36xp compatibility

[BEMSman]

Hello,
Is there any plans to modify this project to communicate with the Valence u27-36xp battery modules?

[seb303]

I don't personally have plans to do this, no. But I guess it wouldn't be too big a change.

[BEMSman]

Interesting.... I do have the valence software (version 12.12) which is on various sites however I believe it is for the 12/18V batteries, however it does not bring any cell info back from the 36v xp module, the only info it brings back is the Modbus ID which you can change, it also brings serial number and firmware version/date but not the battery type.

[seb303]

Would be interesting to use a serial monitor to watch the data returned and see if you can reverse engineer where the values are in the data. That's how I did it for the 12V version. The command bytes to send the batteries are all documented in the source code (e.g. readVolts, etc.) Quite likely they will be the same even for the batteries with more cells, just the returned data structure will be different.

[BEMSman]

Tried that, but one issue is that the valence software is the wrong version, it will not pull back much in the way of battery data except for an address and serial number. I did try compiling your project for an ESP32 but it errored and wouldn't compile, I did however purchase a teeny4.0 and that pulls back data:

However I guess I would have 12 cells and the voltages are wrong, my battery terminal voltage is 39.57V which I assume each cell should be around 3.29V. would you have any pointers on the changes I would need to make to obtain realistic values and number of cells?

[seb303]

You'd have to dump out the raw binary data returned from the battery and try to figure out which bytes correspond to which data. The cell voltages V1 to V4 come from byte pairs at indices 9->16 (see lines 407->410 in the code). Maybe look at the next bytes from index 17?

The VT figure is just the total of V1 to V4, it's not read from the battery.

[BEMSman]

Thanks,
I changed the byte pairs from 9->16 to 17>24, at first I was impressed, then looking closer the first two pairs was the same value as your original code 3rd & 4th pairs (2.450). the new 3rd & 4th pairs was way off the mark with a value around 19.xx
This is quite a challenge as I'm by no means good at coding, however hopefully I will try and learn something.
I'm not so sure how to pull and work out valid responses without the correct software for these batteries, I have contacted the manufacturers and they are not interested in talking to end users.
One thing I may notice is that it would appear that possibly 6 cells can be read 9->20, however the values are 2.512, 2.353, 2.450, 2.450, 2.450 & 2.450, I also don't know if it means anything to you but if those voltages are multiplied by 1.35 that gets you in the correct ballpark???

[seb303]

I doubt you should have to multiply any of the values to get the correct numbers. In general they are returned as 16bit integers with big-endian byte order. Voltages are in mV units, temperatures in 0.01C, current in 0.01A, SOC in 0.1%

Try modifying the code to dump the raw values to the console connection ...
e.g. stick this around line 406:
logBytes(res, readVoltsResLen);

Or you could install the demo of this:
https://www.eltima.com/products/serial-port-monitor/
This will enable you to examine the serial bus comms.

[seb303]

Regarding the voltage values you are seeing: ~2.5V is low, but possible if the battery is almost discharged. What does the total battery voltage measure?

[cheezemanrich]

BEMSman,

Do you have the diagnostic tool that will read the U27-36xp? If so, I can get you the registers you need. Without it, it's only guesses. My best guess is that they either extended the the response so it includes the extra cells or they added a new register that contains the extended cell values.

Regarding ESP32, I have done a lot of work to try and make it work. The biggest issue is that the batt communication uses Mark Parity with 1 Stop Bit. The ESP32 hardware does not support it so I had to play with different settings and found it very unreliable and occasionally causing watchdog reboots. I can say that the Teensy code works very well and is reliable so I ended up going back to it.

[BEMSman]

Regarding the voltage values you are seeing: ~2.5V is low, but possible if the battery is almost discharged. What does the total battery voltage measure?

my battery terminal voltage is 39.57V which I assume each cell should be around 3.29V. Led is flashing green.
I have Wireshark. Is there a benefit of using the serial port monitor? I have now installed it so will try it out in the next day or two.
Also I wonder if there is any byte swapping going on? otherwise why do the Valence software not work correctly for a different battery?

Update, I have used the monitor you suggested and have captured the attached, don't mean too much to me. you will need to remove the .txt extension to be able to load it in.
COM9 Monitoring Session.spm.txt

Just done a further capture whilst using the Valence tool, (COM4) the queries were, 1, locate battery, 2, start read, and 3 was "batt system info" scan.
COM4 Monitoring Session.spm.txt

[BEMSman]

BEMSman,

Do you have the diagnostic tool that will read the U27-36xp? If so, I can get you the registers you need. Without it, it's only guesses. My best guess is that they either extended the the response so it includes the extra cells or they added a new register that contains the extended cell values.

Regarding ESP32, I have done a lot of work to try and make it work. The biggest issue is that the batt communication uses Mark Parity with 1 Stop Bit. The ESP32 hardware does not support it so I had to play with different settings and found it very unreliable and occasionally causing watchdog reboots. I can say that the Teensy code works very well and is reliable so I ended up going back to it.

I really wished I had the correct diagnostic tool for the U27-36xp battery, I could then do a Wireshark capture which I'm sure with some help would provide many answers. Truth be known, there are probably several peeps that would like to know

[seb303]

Update, I have used the monitor you suggested and have captured the attached, don't mean too much to me. you will need to remove the .txt extension to be able to load it in. COM9 Monitoring Session.spm.txt

Just done a further capture whilst using the Valence tool, (COM4) the queries were, 1, locate battery, 2, start read, and 3 was "batt system info" scan. COM4 Monitoring Session.spm.txt

Unfortunately my demo of the software expired now so it won't let me open these files. But essentially what you need to do is look at the data returned when you send a readVolts command, and look for some byte pairs which roughly match what your cell voltages would be (approx 3298mV going by the Vt you measured). Look for the MSbyte which will be 0x0C or possible 0x0D if you have a cell >= 3328mW. Hopefully there will be 12 of these, probably each 1 address apart.