fault reading of Watt Meter

[menash]

Hello everybody,
I have this watt meter http://www.dx.com/p/ts-68691-0-100a-0-60v-power-battery-tester-watt-meter-dynamometer-414035/
I tried to measure the capacity of 18650 li-ion battery (3.7v) with it but I found out that when this watt meter show 1 amps current my multimeter says 0.8 what means 200 mA fault!!
I  had disconnected the screen to look the chip and tried to figure it out but it unmarked chip.
I attached picture of the board. any suggestion how to diagnose it and try to fix it or it just a garbage?
thanks!

[mmagin]

What was the load?  Did you measure with both this and your multimeter simultaneously, e.g. placing them both in series between the battery and the load?

[tronde]

You must use external power supply to the watt meter if the battery voltage is below 4V. The 3-pin connector is for external power. It's meant to be fed from a receiver battery pack (for RC equipment) of minimum 4.8V. 0V is closest to the wires, + in the middle. The pin most far away from the wires is meant as a reset for the hour / energy counter. Connect to 0V for reset.

It's Chinese, so check that your meter is connected with + in the middle.

I don't know if the external power must be floating, or if it can share 0V with the input.

I have seen different types of this meter. Some of them will release smoke if used without external power, and the voltage on the input terminal is about 30V. Search for 60A 100V watt meter, and you will find a lot of information.
 

[menash]

What was the load?  Did you measure with both this and your multimeter simultaneously, e.g. placing them both in series between the battery and the load?

the load was 4Ohm resistor . and yes i did connected both of them in series in the same time while measuring.

You must use external power supply to the watt meter if the battery voltage is below 4V. The 3-pin connector is for external power. It's meant to be fed from a receiver battery pack (for RC equipment) of minimum 4.8V. 0V is closest to the wires, + in the middle. The pin most far away from the wires is meant as a reset for the hour / energy counter. Connect to 0V for reset.

It's Chinese, so check that your meter is connected with + in the middle.

I don't know if the external power must be floating, or if it can share 0V with the input.

I have seen different types of this meter. Some of them will release smoke if used without external power, and the voltage on the input terminal is about 30V. Search for 60A 100V watt meter, and you will find a lot of information.
 

I do have an external regulated power supply that I made for it, and the + in the middle (the reset never worked for me BTW).
I did tried to found information about something similar but nothings. I will try again.

[Seekonk]

I have an old Turnigy and that is pretty accurate.  It had internal pots to calibrate V and A.  These knockoffs have no calibration, but still a pretty useful instrument even when one I have says 1.5W with nothing connected. I have six different makes of these.

For Li testing I found this very accurate, down to even 30ma it was within 1ma.  Rated up to 3A it has a tint FET and I wouldn't do any where near that. End turn off is .5 to 12V, but can only be adjusted to .1V.

http://www.ebay.com/itm/302159402311?_trksid=p2057872.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

[menash]

I have an old Turnigy and that is pretty accurate.  It had internal pots to calibrate V and A.  These knockoffs have no calibration, but still a pretty useful instrument even when one I have says 1.5W with nothing connected. I have six different makes of these.

I also don't need very accurate meter I bought knockoffs because it cheap. I guess I got bad luck with it

[menash]

Ok, so I did some more measurements to see what going on. I connect it to a reliable 5V 3A phone charger and loaded it with an adjustable load that I have (of course on series to my multimeter) and I compared the result while taking up the current from 0-3A.

here is the interesting results: (all unit in Amps)

multimeter |  faulty wattmeter
-------------------------------
0.00-0.15  |  read 0
0.15-0.20  |  accurate
0.20-0.55  |  wrong reading
0.55-0.70  |  accurate
0.70-1.10  |  wrong reading
1.10-1.20  |  accurate
1.20-1.60  |  wrong reading
1.60-1.75  |  accurate
1.75-2.20  |  wrong reading
2.20-2.35  |  accurate
2.35-2.75  |  wrong reading
2.75-2.85  |  accurate
2.85-3.00  |  wrong reading

as wrong reading means 0.05-0.2 A difference (mostly 0.15-0.2 and always up).

Is it gives anyone any idea? means something?

[mmagin]

That looks like the measurement is quantitized to approximately half-amp values.  Maybe they are using an 8 bit ADC as built into a cheap microcontroller and since it's spec'ed to measure 0 to 100 amps, it's going to presumably split that range into 256 (2^8) different levels.

[menash]

I do think I saw somewhere that it is a 8 bit microcontroller but I can't remember where.

Ok, so what the chance that it behave like that because some bad filter or another component on the board, or it is probably problem with the chip and then there is nothing to do but burn it up? 

[mmagin]

It seems like its just a design limitation.  If you want more accuracy, either don't cover the whole 0-100A in a single range, or use an ADC with greater bit depth.

[menash]

either don't cover the whole 0-100A

Is it something I can modify in my meter or just a tip for next shopping?

thanks for your answers, I must learn more about how those things works

[macboy]

I have something similar, but the PCB definitely looks different to yours. I added trimpots for calibration, with OK results, refer to this post, where I posted a schematic matching mine. I don't have the apparent linearity issue that you have; when carefully tweaked, my readings were accurate within a couple of counts over a wide range. That's quite amazing considering a 130.00 A full scale reading which is 13000 counts (15 bits of resolution). I don't think there is a 15 bit ADC in there, so they are likely doing averaging of many samples to get there, but the linearity on mine seems quite good. Maybe (highly likely?) yours uses a clone/knockoff microcontroller that has a crap ADC with poor linearity.

Also, don't fool yourself into thinking that this can measure 100 A continuously; that poor little 0.001 ohm shunt R will be burning off 10 W of power and will quickly desolder itself and/or damage the board. It can certainly measure 25 A all day long, and maybe 50 A would be a reasonable limit (then 2.5 W of power in the shunt).

[mmagin]

either don't cover the whole 0-100A

Is it something I can modify in my meter or just a tip for next shopping?

thanks for your answers, I must learn more about how those things works

Well, you might be able to change the shunt resistor and reprogram the microcontroller, but probably easier to just shop for something better.