Panel Meter Lost Cal?

[metrologist]

I have this panel meter connected to my solar panel.

http://www.mpja.com/download/32541medata.pdf

It was working great, but I noticed the voltage reading seemed too high. It was fine when new. I installed a 2nd panel meter on another solar panel and it too was fine, but the voltage reading became suspiciously high.

I too the panel meter to power supply and inject 12V and it reads ~18V. I pulled the back cover and noticed 3 ICs, a RTC crystal, and a handful of passives.

So I wonder, since this happened to two panel meters, if the solar panel starts to act as a load and wipes out calibration data on an IC? This seems unlikely though because I would expect the entire code to become corrupt - the device seems too simple to have separate memory just for cal data.

So, without any pots, I guess accuracy is either designed or calibrated in SW and there is nothing that can fix these?

I can post a pic of the PCB later.

[Seekonk]

I have higher voltage meters like that and the voltage dropping SMD resistors are not rated for near the operating voltage. That shouldn't be your case at less than 100V.  At what voltage do your panels operate at? Operating these at anywhere near their rating is iffy.  If I had to guess, I'd say the voltage regulator got hit with a spike and it is now putting out a lower voltage which the chip may use as a reference. Does amps have a similar problem?  Some versions have a trace you can cut to operate the micro  from a separate lower voltage supply. Guess the lesson here is to buy something different.

[metrologist]

Thanks. They were 12V/24V panels. The current readings are still accurate. Just voltage and the power readings are just about right at 50% high.

I can check the regulator voltages. I also played with that one button to see if there was a setting. I saw one YT video where it seemed the guy got the voltage reading to blink like a setting. It was a different range or the AC version though. I was able to set the alarms, clear the readings, and set the current shunt (the current shunt is smd R002) between 50A/100A. No effect of those settings.

[metrologist]

It seems all the power stuff is done on a for-purpose IC. I found this info:

Inside it has a Holtek HT2621B IC (LCD driver), a 24C02 memory to save the cumulative kWh values ??and an SD8103. From what you can understand from the datasheet the SD8103 is a microcontroller specific to meters (multimeters?) With an AD converter that seemed very good (16 bits effective with 0.01% of linearity).

http://www.pakequis.com.br/2016/10/medidor-peacefair-pzem-031.html

Also some hacking the data: https://www.webx.dk/oz2cpu/energy-meter/energy-meter.htm

Perhaps a mfg: http://www.sdicmicro.com/application.html?apply=DC%20Energy%20Meter

I was wondering if cal data could have been stored on Atmel eeprom and have been wiped out? I know the panel meter would loose it's accumulated energy once in a while, and I assumed it was because of the solar panel either presenting a load and wiping the memory or due to operation just below low voltage drop-out.


(one more image: http://tucnak.nagano.cz/pokus/index.php?title=File:Paem-031s.jpg
and datasheet: http://tucnak.nagano.cz/wiki/images/c/ca/Sd8103.pdf)

[Seekonk]

I'd be changing a resistor and calling it a day.  I use those Turnigy GT Power meters and change the resistor so they are 600V for a times ten reading since they stop at about 65V.  I do think 50% is just a little to big for a cal amount.  I would look for a transistor or FET that switches an extra resistor out on the divider so it has better low range values.  Just a thought.

[metrologist]

I'm thinking the input to the ADC is on pins 5, 6, 7, 8... Maybe I can figure out the actual circuit and work from there.

[metrologist]

In the photo I posted above, lower right corner, there is R12 and R11 that make the voltage divider: 220k/1k

So, about 45mV to cover the 6-100V range. That goes from R11 to R1, then to U1 and C1.

I figure about 2k on top of R11 will get it close.