diy solar

diy solar

Load tester calibration

NMNeil

Solar Enthusiast
Joined
Mar 14, 2020
Messages
441
I received one of these a few days ago.
As a test I charged an old overcharged and bloated 200Ah cell to 3.65, waited an hour then plugged it into the tester with the cutoff set at 2.5 volts.
The problem I'm having is that when the unit is showing 2.6 volts, just before the shutdown, the voltage at the battery terminals is still around 3 volts, so the tester is shutting down prematurely.
The multimeters used are Fluke and BK precision, so I believe their readings are accurate, and there is only about a 50 millivolt or less voltage drop in the 10G wires between the cell and the tester.
The included instructions. and the font of all knowledge; YouTube, are of no help, so does anyone know how to calibrate these testers to read the correct voltage?
 
Low current/high voltage = wiring losses.

It is likely reading the correct voltage. The problem is it's reading the voltage RESPONSE to the test current, where your meters are measuring open circuit voltage.

It can't be "calibrated," since this isn't a calibration issue.

If the 0.4V delta is reliable, simply set cut-off to 2.1V.
 
Low current/high voltage = wiring losses.

It is likely reading the correct voltage. The problem is it's reading the voltage RESPONSE to the test current, where your meters are measuring open circuit voltage.

It can't be "calibrated," since this isn't a calibration issue.

If the 0.4V delta is reliable, simply set cut-off to 2.1V.
At the moment I have the cell back on charge with a set voltage limit of 3.65 volts. Current is around 7 to 8 amps and the tester is hooked up, but not discharging.
Turning off the charger and waiting a few minutes and the Fluke is showing 3.4 volts at the battery but the tester display is only showing 3.1 volts.
Not sure about setting the cutoff to 2.1V before testing my new cells. As I said I'm using an old bloated cell for the testing just to make sure it's safe.
 
I'm not sure about your model, but I'm using a DL24 tester. It has separate inputs for voltage reading, which you should connect with thinner wires to the battery/cell in addition to the load wires.
That allows it to read voltage more accurately.

I still see some voltage difference when under load, but only when it drops under 3v.
For example I set voltage cutoff at 2.6v but when the test is done, I see 2.95v on the cell.
I'm not sure if this isn't because of the Lifepo4 voltage curve: when in the low "knee" the voltage drop might be higher when under load, as apposed to the flat curve in the 80% if SOC.
 
The newer DL24P units have the four wire test terminals that separates out the voltage sense pair. This allows you to run small gauge voltage sensing pair directly to battery terminals so they are not effected by heavy load current wire pair voltage drop.

If you don't have the separate voltage sensing inputs your only option is to reduce the voltage drop on common wires by using larger gauge (#8 or #6) wire and keeping its length as short as possible.

DL24P 4 wire test.jpg
 

Attachments

  • DL24P Schematic diagram.pdf
    736.5 KB · Views: 36
  • DL24 PCB layout.pdf
    1.4 MB · Views: 28
Last edited:
The tester is the same one that Will uses, which does not have a 4 wire option that I'm aware of.
And as I said the tester is not discharging the battery, It's just acting as another voltmeter.
The voltage reading at the battery terminals, and at the tester connections are within a few millivolts of each other, so minimal voltage drop, but the voltage readout on the tester is off by a significant amount.
Unfortunately Will's advice is to buy 2 or 3 of them and if one's inaccurate, throw it away o_O
 
The newer DL24P units have the four wire test terminals that separates out the voltage sense pair. This allows you to run small gauge voltage sensing pair directly to battery terminals so they are not effected by heavy load current wire pair voltage drop.

If you don't have the separate voltage sensing inputs your only option is to reduce the voltage drop on common wires by using larger gauge (#8 or #6) wire and keeping its length as short as possible.

View attachment 77905
I'm using 10AWG, which should be more than enough for a 15A single cell discharge.
Attach a cell to the tester and my Fluke reads 3.22 volts at the cell terminals, check the voltage at the terminals on the tester and it's 3.22 volts, but the voltage displayed on the screen is 3.19 volts, so something needs adjusting. Which is the crux of the problem; how do you adjust it?
(I initially put 3.49 volts, but I have a hangover and shaky hand, so please make allowances ):giggle:
 
And as I said the tester is not discharging the battery, It's just acting as another voltmeter.
The voltage reading at the battery terminals, and at the tester connections are within a few millivolts of each other, so minimal voltage drop, but the voltage readout
Sounds like the load MOSFET or Darlington bipolar device is blown out.

If the load device is blown out then there is no load current through the wires so little to no voltage drop in wires..

I'm using 10AWG, which should be more than enough for a 15A single cell discharge.
Attach a cell to the tester and my Fluke reads 3.22 volts at the cell terminals, check the voltage at the terminals on the tester and it's 3.22 volts, but the voltage displayed on the screen is 3.19 volts, so something needs adjusting. Which is the crux of the problem; how do you adjust it?
(I initially put 3.49 volts, but I have a hangover and shaky hand, so please make allowances ):giggle:

You can use smaller than #10 since there is no current, so no wire voltage drop. Another unit for the trash.

If the load is working and you are just turning off or lowering load current to battery to read voltage without wire voltage drop then you will be getting cell kinetic overpotential recovery voltage effect. The battery will jump up in voltage when load removed then rise more over the next couple of minutes with no load on battery as the exponential recovery occurs.

You can also measure the wire voltage drop with load current and add that to unit's voltage reading to get battery voltage.

A battery that has been overcharged and bloated will have decomposed some of its electrolyte. The CO2 gas from the electrolyte decomposition will bloat the cell. Other decomposition components are gummy tars that clog the electrode pores reducing ion diffusion through cell which increases cell impedance. This increases the terminal voltage slump with current and corresponding rise in terminal voltage when load removed.

Electrolyte decomposition get very accelerated at a cell voltage above 4.3v. That amount of overcharge voltage will bloat a cell in 10 to 20 minutes.
 
Last edited:
I'm measuring all the voltages with no current draw.
Voltage at cell measured by Fluke. 3.22 volts. Amperage draw 0.
Voltage at testers terminals measured by Fluke. 3.22 volts. Amperage draw 0.
Voltage displayed on the tester screen. 3.19 volts. Amperage draw 0.
 
I just measured across my battery currently being tested, same tester, and it shows 3.2v, but the tester says 2.83... ? Also if I try to pull more than 15A the tester voltage drops below the 2.6v cutoff rather quickly... can anyone explain to me this effect?

Thanks for your guidance.

Shane
 
I'm not sure about your model, but I'm using a DL24 tester. It has separate inputs for voltage reading, which you should connect with thinner wires to the battery/cell in addition to the load wires.
That allows it to read voltage more accurately.

I still see some voltage difference when under load, but only when it drops under 3v.
For example I set voltage cutoff at 2.6v but when the test is done, I see 2.95v on the cell.
I'm not sure if this isn't because of the Lifepo4 voltage curve: when in the low "knee" the voltage drop might be higher when under load, as apposed to the flat curve in the 80% if SOC.
With laboratory tools discharging down to 2.6 after cut off it will go up to 2.800- 2.890 V . We have to understand that even if you discharge this chemistry down to 2.5 V over 24 hours it will have enough charge to pull more Ah but I will not advise on this . To explain on chemical level what happens that would be long discussion that will make many bored to death trying to understand. If you see 2.95 raise there is something not done right while doing discharge test. It is possible but we are talking about single cells that are 1000 Ah . 300+ Ah could get to 2.95 but this are cells designed for military or EV purposes Grade A+++
 
The weak link on the DL24 is the reverse current protection diode above the terminal strip slightly under the MOSFET heatsink. It gets very hot with high current and, just mounted on PCB, there is not enough heat sinking for it.

I solder a 35A automotive fuse on PCB across the protection diode. The body diode within the load MOSFET should have no problem blowing the 35 amp fuse if battery is accidentally connected in reverse without damaging anything, other than fuse.

With the protection diode bypassed by fuse, the MOSFET with its fan cooled heat sink is able to take 30 amp load current from a 3.4v LFP cell. (102 watts of heating). The next weakest link is the two 10 milliohm shunt resistors. I would not run greater than 20 amps for a full cell capacity discharge test due to the shunt resistors heating. 30 amps is okay for a three minute load test.

On a 280 AH cell, the 30 amp load test is much better at evaluating quality of cell by measuring overpotential terminal voltage drop after 3 minutes of load. Matched cells will have the same overpotential terminal voltage slump for same load current so the 3 min. load test is also a good way to verify cell matching. Since cell impedance increases at cold temp you need to be 25 degsC,+/-5 degsC to make the tests.

The sampling rate on DL24 is not fast enough to follow the initial conduction resistance slump (what is mostly measured by battery impedance meter) so taking OCV before starting test and taking voltage initial slump at top of exponential discharge gives the conduction resistance. It matches what I measure on YR1035+ fairly well. It is a little difficult to distinguish the starting point voltage accurately for the exponential discharge due to the low sampling rate.

Series conduction resistance measured by battery impedance meter like YR1035 has very little to do with cell matching. It will show up any delamination of electrode graphite/LFP material to copper/aluminum foil layers. Three minute load test is best way to test cell matching. For same model cell, it also correlates well to the cell capacity. Three minute load test is a lot easier than a full discharge capacity test.

Over the life of cell the overpotential voltage slump can increase 3x to 5x what it was when new. Anything above 3x is a pretty poor condition cell, only capable of lower load current usage. The same aging processes that produces the increase in overpotential voltage slump also consumed free lithium within cell resulting in less AH capacity from cell. This is why the 3 minute load test correlates well to a full capacity test.

Battery Impedance.png

LF280 overpotiential curve.png
 
The tester is the same one that Will uses, which does not have a 4 wire option that I'm aware of.
And as I said the tester is not discharging the battery, It's just acting as another voltmeter.
The voltage reading at the battery terminals, and at the tester connections are within a few millivolts of each other, so minimal voltage drop, but the voltage readout on the tester is off by a significant amount.
Unfortunately Will's advice is to buy 2 or 3 of them and if one's inaccurate, throw it away o_O
Tag for later
 
The tester is the same one that Will uses, which does not have a 4 wire option that I'm aware of.
And as I said the tester is not discharging the battery, It's just acting as another voltmeter.
The voltage reading at the battery terminals, and at the tester connections are within a few millivolts of each other, so minimal voltage drop, but the voltage readout on the tester is off by a significant amount.
Unfortunately Will's advice is to buy 2 or 3 of them and if one's inaccurate, throw it away o_O
I have just had something similar with the new (4 pin model). I had 4 4mm cables about 50cm long, one at each terminal. I changed to 2.5mm for current and speaker wire for voltage and all 4 are now about 15cm long. I am now getting same voltage on tester as I am on 2 multimeters. Hope this helps someone.
 
Back
Top