NO contact to TSOL M1600 with serial 12*

[MuckelMarko]

Is your feature request related to a problem? Please describe.

Hi what a cool reverse engineering project.
Unfortunately I got a TSOL M1600 Micro Inverter with serial number 12 *.
There is no communication between TSOL and openDTU. See console logging in attached file (console.txt).
Does someone have a idea hast wrong?
The Serial contains characters (TSOL-serial.jpg). Is this the problem?

console.txt

thank you for this cool project!

Describe the solution you'd like

Is there a possibility to update OpenDTU to be able to communicate with a TSOL M1660 with probably new firmware?

Describe alternatives you've considered

No response

Additional context

No response

[tbnobody]

If you look at the inverter settings. I would assume your inverter was detected as HM-1500 etc?
It is currently sending packets. That means that the serial number is known for a specific device. Is the NRF Module correctly detected? (Info --> System)

[MuckelMarko]

Due the Installation it tells me someting about HM-1500 Type. I will check it.

The NRF Module seems correct detected:

Radio Information

Chip Status	connected
nRF24L01+

[UweSchneiderGmbH]

Hmm,

I've a TSOL with a serial number beginning with 11.. but my serial number sticker looks complete different (sorry for the bad quality of the picture):

My difficulty with the connection was, that the TSOL needs a connected solar panel (or DC source) with enough power (~10W) before the inverter activates and is answering to the connection requests. With less power the LED blinks red and green, but no communication. With only AC connected the inverter does nothing.

The inverter was bought in January 2023 and is reported as HM-1500. The bootloader version is 0.1.0 and the firmware is 1.0.12 with the date 2020-06-24.

[stefan123t]

TSOL and some OEM / so called SubBrands of Hoymiles may require a different Serial Number calculation which rovo89 found out on Discord.

rovo89: schau dir ggf. mal ProductSNGeneUtils.sn92long() an.

public static long sn92long(String str) {
  if (str.length() != 9) {
    throw new IllegalArgumentException("Illegal SN");
  }
  int[] iArr = _FASTDICT;
  return ((((iArr[str.charAt(0)] << 10) | (iArr[str.charAt(1)] << 5)) | iArr[str.charAt(2)]) << 32)
          | iArr[str.charAt(8)] | (iArr[str.charAt(3)] << 25) | (iArr[str.charAt(4)] << 20)
          | (iArr[str.charAt(5)] << 15) | (iArr[str.charAt(6)] << 10) | (iArr[str.charAt(7)] << 5);
}

Die 9 Stellen dürften die vor dem Bindestrich sein, also A11001TW3, die 193 ist ein "Extend code" (Prüfnummer?).
_FASTDICT ist ein Mapping für die möglichen Zeichen zu ihrer Position in folgender Liste: 0123456789ABCDEFGHJKLMNPRSTUVWXY

Da fehlen ein paar Zeichen, die leicht zu verwechseln sind: I, O, Q, Z. So ergeben sich 32 Zeichen, d.h. 0 = 0, A = 10, Y = 31
Mit dem Code oben wird ein 64-Bit-Wert erzeugt. Da der größte Wert 31 ist, passt er in 5 Bit, und so ergeben die Bitverschiebungen mehr Sinn. Wenn man Seriennummer als ABCDEFGHI darstellt, finden sich die jeweiligen Werte nachher so wieder: 0AAA AABB BBBC CCCC 00DD DDDE EEEE FFFF FGGG GGHH HHHI IIII. Jetzt könnte man es von Hand ausrechnen, aber ich würde es lieber heute Abend mal in ein Python-Skript packen und damit ausrechnen. Kontrollmöglichkeiten gibt es vermutlich nicht?
Wenn man das Ergebnis in Hex darstellt, ergibt sich auf jeden Fall eine 12-stellige Zahl, was zu den klassischen Seriennummern passt.

sn = 'A11001TW4'
CHARS32 = '0123456789ABCDEFGHJKLMNPRSTUVWXY'
CHARS64 = 'HMFLGW5XC301234567899Z67YRT2S8ABCDEFGHJKDVEJ4KQPUALMNPRSTUVWXYNB'

first_char = '1'
i = CHARS32.find(first_char)
sum1 = sum(ord(c) for c in sn) & 31
sum2 = sum(CHARS32.find(c) for c in sn) & 31
ext = first_char + CHARS64[sum1 + i] + CHARS64[sum2 + i]
print(ext)

rovo89: Übrigens sind die ersten vier Stellen wohl der Gerätetyp (A=MI, B=ES, C=RSD, D=DTU), dann zwei Stellen für den OEM, dann eine Stelle für die Fabriknummer. Die restlichen 5 Stellen sind eine Sequenznummer. Plus dann halt ein Bindestrich und der "Extend Code".
Für die "klassischen", 12-stelligen Seriennummern gibt es übrigens auch einen Extend Code:

sum1 = sum(ord(c) for c in sn) & 31
sum2 = sum(CHARS32.find(c) for c in sn) & 31
ext = CHARS64[sum1] + CHARS64[sum2]
print(ext)

Ich glaube das Prefix der Seriennummer entscheidet idR noch darüber über welches RF-Modul (NRF24L01+, CMT2300A und in Zukunft zweites WLAN Interface für WiFi-Series) die OpenDTU/AhoyDTU mit dem WR (bzw in Zukunft dessen DTUBI) kommuniziert.

Wie rovo89 richtig vermutet wird das Prefix 4412 bzw 2821 mW noch nicht von der OpenDTU erwartet / akzeptiert bzw der bisherige Code weiss nicht was er mit dem WR anfangen soll. Da fehlen bestimmt noch ein paar Anpassungen für die HERF und andere “SubBrand” Inverter.

[Birkenstab]

Is there any information on whether OpenDTU supports the Herf-800 inverter or whether somebody has already figured some stuff out on how to support it?

[Birkenstab]

Nevermind I figured something out: #1751

[tbnobody]

I also added some remarks etc. in #1751

[tbnobody]

But as far as I see it, it's still unknown how to decode the RE12508005262?

[stefan123t]

@tbnobody is there an issue with the above code for sn92long() decoding the RE12508005262 ?

I did try that Serial ID with the above sn92long code from #1751 (comment) and #1751 (comment)

SN92 Long: RE12508005262-1X5
Serial ID: 61C104502000

@MuckelMarko this Serial Number is from you.

Please follow the description by @Birkenstab in the first link above.

Does it work with the Serial ID above, e.g. 116104502000 or similar 1161508005262 with a different 4-in-1, 2-in1 or 1-in1 model serial prefix ?

I tried it with several sequences of digits / characters from the Serial Number you have given:

Serial Number: RE12508005262

SN92 Long: RE12508005262-1X5
Serial ID: 61C1 04502000
-> 1161 04502000

SN92 Long: E12508005262-1Z4
Serial ID: 3822 0A040005
-> 1161 0A040005

SN92 Long: 12508005262-16C
Serial ID: 4450 08000A2
-> 1161 08000A2

SN92 Long: 508005262-138
Serial ID: 1408 000288C2
-> 1161 000288C2

SN92 Long: E12508005-12G
Serial ID: 3822 0A040005
-> 1161 0A040005

SN92 Long: 125080052-1CY
Serial ID: 4450 08000A2
-> 1161 08000A2 (most likely one digit too short!)

[stefan123t]

@MuckelMarko I also tried to scan / detect the bar code given in your image, though it is not sharp enough to yield a tangible result.

May I ask you to make a better mugshot of your inverter Serial ID ?

zbarimg HERF-M1600_barcode_levels.jpg 
scanned 0 barcode symbols from 1 images in 0,01 seconds