Battery Calibration does not help

[GoogleCodeExporter]

Board: Turnigy 9x stock

Opentx-r2491

EEPR: 231-0

What steps will reproduce the problem?
1. Discharge your tx's battery down to -say- 3.5V/cell. For a 2S total V is 7V.
2. Turn the tx ON. Go to battery calibration, change the figure to show 7V 
3. Charge the battery. The 2S should read ~8.4V
4. Turn the tx ON. Voltage shown is not 8.4V. Calibrate the reading to show 8.4V
5. Use the tx until (real) voltage becomes 7V again. The tx will show the wrong 
value again...

What is the expected output? What do you see instead?
One should need to calibrate the voltage once and the reading should be 
accurate thereafter.

Please provide any additional information below.
I'm using a 1600mah 2S on a Turnigy 9x. Other mods: backlight, ASP taps.

Original issue reported on code.google.com by cma...@gmail.com on 30 Jul 2013 at 6:36

[GoogleCodeExporter]

You are not suppose to go so down on that mainboard:

The dropdown of the main stabilizer can be more than 2V.
That means that the voltage reference that is supposed to be 5V is indeed less. 
All ADC conversion will be wrong. 
No calibration can help if reference voltage is wrong.
It will change as soon battery will be charged.
You are running a piece of hardware out of specs and you are asking software to 
solve hardware problems. Minimum voltage is around 7.4V

Original comment by romolo.m...@gmail.com on 30 Jul 2013 at 6:45

[GoogleCodeExporter]

Thanks for your reply :)
I'm actually not running it as low as 7V. Usually around 7.5V. I just gave that 
as an example, but I guess I should have used real-life ones.

I've heard many others running 2S, and that's why I'm doing it. I didn't know 
of the 7.4V min.

Anyway, the same thing happens when going from 7.7V to 8.4V. Should it be 
happening?

Actually, I've just calibrated it at 7.7V and will charge it. I'll let you know 
what the reading is when fully charged.

Original comment by cma...@gmail.com on 30 Jul 2013 at 6:49

[GoogleCodeExporter]

There is people using LiIon batteries.
Here is a typical discharge curve:
http://www.mpoweruk.com/images/discharge-C-rate.gif

Especially at the very low current used by the radio (< 1C) the battery stays 
at a voltage > 7.4 until it is fully discharged, ant that explains why people 
can use it even with a dropdown of the main stabilizer > 2V
The battery will be almost fully discharged at 7.4V.

Also you need to verify the battery voltage under usage so comparing the 
voltage read from the radio ADC with the voltage read by a multimeter in 
exactly the same moment, so while the radio is running, you need to measure 
while calibrating.

Original comment by romolo.m...@gmail.com on 30 Jul 2013 at 7:03

[GoogleCodeExporter]

Ok, understood. I'm using a lipo, but I guess the discharge curve is pretty 
much the same.

So, I AM measuring at the same time that I'm calibrating it. I use the 
battery's balance port to measure voltage.

I calibrated at 7.7V. Then charged it fully. It is now measuring 8.4V on the 
balance port, but the radio shows 7.8V!

Before installing opentx, I had done this mod 
(http://www.rcmodelreviews.com/tendollarbattery3.shtml), but eventually I undid 
it. I have removed the extra resistor so I don't think it interferes.

Thanks,
Chris

Original comment by cma...@gmail.com on 30 Jul 2013 at 7:21

[GoogleCodeExporter]

Are you sure you didn't damaged the original resistor by overheating it ?

Anyway this issue is invalid it's an hardware one!!!

Original comment by romolo.m...@gmail.com on 30 Jul 2013 at 7:25

• Changed state: Invalid

[GoogleCodeExporter]

Ok, so it should not do that... That's what I needed to know. 

It's a hardware problem, ok, I'll look into it. Maybe resolder the resistor, 
but it didn't work with that either (with opentx). With stock firmware and the 
resistor installed the alarm worked fine (same battery) for years...

Original comment by cma...@gmail.com on 30 Jul 2013 at 7:28

[GoogleCodeExporter]

Wait if you have still the additional resistor installed you need to remove 
it...

Original comment by romolo.m...@gmail.com on 30 Jul 2013 at 7:36

[GoogleCodeExporter]

No, of course not! I had removed it soon after I installed opentx. But since 
it's not working like this, I'll put it back on. It worked with the original 
fw, it should work with opentx too....

Original comment by cma...@gmail.com on 31 Jul 2013 at 12:24

[GoogleCodeExporter]

Ok, I did an all night test: with the extra resistor connected:

I fully charged it, calibrated it and let it discharge overnight.

3:40am: Real V: 8.39. Tx shows: 10.5 (calibrated it)
10:40am: Real V: 7.45. Tx shows: 10.5.

When default FW was used, the battery indication went from 10.5 to 8.9V, while 
real voltage went from 8.4 to 7.2. I am sure about these values cause I had 
done extensive tests and have a chart :)

So, I think there's something different with the way opentx handles 
calibration. 
Clue: when clicking through values (in calibration), if I click all the way to 
the left, it goes to 10.1 and then just beeps if I keep pressing. If I go right 
it goes up to 12.5 and then if I click some more (right) it goes to 9.2v. If I 
keep clicking the right button, it goes up to 12.5 and then 9.2 and then up to 
12.5 and so on. After a few cycles (5-6) it "stops" and beeps like it does when 
I go left. It stops at 11.6V for me.

The question is: should I be using a value near the ends of the spectrum or one 
in between?

Original comment by cma...@gmail.com on 31 Jul 2013 at 10:03

[GoogleCodeExporter]

Ok, I slided the calibration all the way to the right, and then set the voltage 
to 11.2 when real voltage is 8.4V.

Now, when the battery is low (7.73V) the Tx shows 9.9V, which is where I set my 
alarm.

So it works like this!

I don't know why it doesn't work when I calibrate it at the far left of the 
calibration spectrum... Maybe a S/W bug?

Original comment by cma...@gmail.com on 1 Aug 2013 at 8:08

[GoogleCodeExporter]

I think there is a calculation overflow problem there.

Original comment by bson...@gmail.com on 2 Aug 2013 at 7:29

• Changed state: Accepted

[GoogleCodeExporter]

Overflow fixed on stock board. Will be in next minor release.

Original comment by bson...@gmail.com on 4 Aug 2013 at 10:19

• Changed state: Fixed

[GoogleCodeExporter]

Thanks a lot for looking at this :)

Original comment by cma...@gmail.com on 5 Aug 2013 at 7:43