Standardizing the measurement of internal resistance

[BatteryMooch]

I saw the pictures of your mods Doctorbass...nice!

I also looked for a thermal sensor of one type or another and never found it. I'm guessing that they're using the change in voltage drop across the MOSFET as a temp sensor? Once calibrated, it can work pretty well though.

 

[steveo]

Hey

I have gone back to the irfb4110mosfet.Why? well i try discharging a single a123 @25 amps and the cba would not go past 2 amps with the P75nf75 mosfet. I assume the fet have a min operation voltage!..

Just so you guys know i was pulling in the 90watt range with 12v of a123's! and it worked flawlessly !!!! but for lower voltage i think from 6v and under from what i tried, it would not discharge battery properly.

-steveo

 

[BatteryMooch]

Yup, you're right. I just checked the datasheet for the P75NF75 and with a gate voltage of 3.9V (the max the CBA can rise to) and around 3.5V-4V for the cell, the current it will let pass is off the chart...in the wrong direction. I'm amazed you got 2A!

If you can wait a bit, your IPP048N06L is already in the mail. IMHO, putting a IRFB4110 back in will risk blowing another one very quickly due to its really limited ratings as a load with cells/packs above 5V. It might last for along time, but it might blow in seconds. And sometimes a CBA blows its op-amp and several other passive components along with the MOSFET.

 

[steveo]

Yup, you're right. I just checked the datasheet for the P75NF75 and with a gate voltage of 3.9V (the max the CBA can rise to) and around 3.5V-4V for the cell, the current it will let pass is off the chart...in the wrong direction. I'm amazed you got 2A!

If you can wait a bit, your IPP048N06L is already in the mail. IMHO, putting a IRFB4110 back in will risk blowing another one very quickly due to its really limited ratings as a load with cells/packs above 5V. It might last for along time, but it might blow in seconds. And sometimes a CBA blows its op-amp and several other passive components along with the MOSFET.

just tested 3 cells of a123 @ 25amps and all ok ..

I have a question though; with my cba set @ 25amps .. during the test is stayed at 22-23 amps .. and never hit 25 .. even with i set the test at 30 amps .. it still sat around 22-23 amps..

Is this the reason doc did the heat sink thing? .. or it maybe the mosfet for me .. hmm

-steveo

 

[BatteryMooch]

It's the MOSFET.
It's made to be switched on and off hard at a much higher voltage (10V at least) than the CBA can deliver (3.9V max). When it's turned on hard, the on-state resistance drops to that nice and low 11mOhms that's in the datasheet. But, as the voltage that's turning it on gets lower and lower, the on-state resistance gets higher and higher. At a certain point, it gets so high that it's preventing you from running the CBA at the current levels you want to.

[Edit] I should mention, for anyone else searching for a different MOSFET to use in their CBA, that the the IRL2910 used in the CBA is a "logic-level" MOSFET and is made to be turned on at lower voltages than other MOSFETs. All of the International Rectifier IRL MOSFETs are logic level. But, if you're not sure the one you're considering will work, check the Typical Output Characteristics graphs in the datasheet. If there aren't any Vgs plot lines for below 5V, it's not a logic-level MOSFET and probably won't be able to be turned on hard enough for high-amp discharges or for use with single cells.

 

[ZapPat]

...This isn't much of a problem in a switching application since the MOSFET spends so little time in its linear region, but it's why you need to switch a FET on/off as fast as possible because it spends time in this linear region, quickly heating up as it does. But when used as a load, always biased in its linear region, inevitably one part of the MOSFET's die will heat up more than another. This leads to a lowering of the threshold voltage for that part of the die and results in it hogging a bit more current, which heats it up more, etc. It's called "hotspotting" and can lead to thermal runaway and destruction of the MOSFET. The construction of linear MOSFETs (vs. a switching MOSFET's "trench" construction) helps to prevent this from happening but linear MOSFETs are typically much more expensive and often have very high voltage, but very low current ratings compared to switching MOSFETs.

Ahhh! Thanks for that explanation, Camlight! I was aware of the hotspoting issue that can give problems with paralleled FETs in switchers, but never thought that this would be such a huge limiting factor when working them in their linear region. Going down from 300W or so dissipation in hard switching to 17W or so in worst case linear... ouch!

 

[Mr. Mik]

It's the MOSFET.
It's made to be switched on and off hard at a much higher voltage (10V at least) than the CBA can deliver (3.9V max). When it's turned on hard, the on-state resistance drops to that nice and low 11mOhms that's in the datasheet. But, as the voltage that's turning it on gets lower and lower, the on-state resistance gets higher and higher. At a certain point, it gets so high that it's preventing you from running the CBA at the current levels you want to.

Thanks! That is the explanation for the problems I experienced whilst testing the fuses for the Vectux BMS:

The current limitation to 18.5A experienced with the CBA2 was just an artefact caused by the internal resistance of the CBA2. Without the CBA2 the resistance is low enough to allow a single (charged) cell to blow one of the fuses.
There is only a small risk that a short could occur when the cells are already very empty and unable to melt the fuse wire. The cells would then remain shorted and get damaged, probably severely and permanently.

http://www.endless-sphere.com/forums/viewtopic.php?f=14&t=6853&p=118086#p116754

Now I understand much better why it happens.

 

[steveo]

Hey All

I've done some experimenting today with my cba,

I don't know if it is because i'm using a 4110 mosfet & not the oem one but the values measured by the cba software are very inaccurate.

just doing a 200mah drain test with the cba software showing approx at 3.3v @ 2.3 amps; my multimeter & watts up meter would measure approx 3.22v & 2.13 amps.

I even Upgrade my wires on the cba to 8 AWG thinking it was a resistance problem..Is this good or bad?? Maybe cba is setup to be accurate with the exact lenght & wire size it comes with .. hmm i doubt it...

Is there anything at all ... ANYTHING!! that could be done to increase the accuracy ???.....Doc did a similar sence mod with the Megapower charger which i've done and is working great!!

-steveo

 

[BatteryMooch]

Hi steveo,
I think it's still a resistance problem, but with the IRFB4110 and not the wiring. Try the same 2.3A current setting in your CBA but with a higher battery voltage. At 3.3V there's barely enough voltage to push current through the 4110's very high resistance (when only driven at the CBA's max gate voltage of 3.9V). If you put a, let's say, 12V pack on there, I'm guessing that the current level will be correct.

If not, your CBA might be damaged.

 

[Mr. Mik]

I tested my CBA2 at 10A and at 20A today.

10A was done with one DMM in series, 20A with 2 DMM's parallel to each other, in series with the battery.

I used 3 of the Vectrix cells (NiMH) in series, about 3.6V. Voltage at the CBA2 would have been lower than that.

Voltage was not measured (or rather not taken much notice of), I know it is very incorrect anyway due to the multiple cables and fuses I have in the loop in addition to the CBA2 cables. I can look up the CBA2 voltages in the .txt files if anyone needs them, but I did not measure the voltage at the battery terminals.

At 10A (set) the DMM showed 10.15A and the CBA2 showed 10.03A. That's 1.1% difference.

At 20A (set) the two DMMs added together showed 20.34A when the CBA2 showed 20.0A. That's 1.7% difference.

I do not know how accurate these DMMs are.

 

[BatteryMooch]

[Edit] I goofed on my accuracy numbers, now corrected. The results you got were worse than mine.

With one of my CBAs and a discharge at 20A, it was off 1.0%.
About -0.2% at the start and then drifting to -1.0% within a couple of minutes.

At 10A I got about a 0.4% difference.
About -0.2% off at start, drifting to -0.4%.

Both were measured with a 0.25% shunt and a Fluke 8846A meter so I'm pretty confident of the numbers.

A big chunk of why the CBA is off was due to the drifting of the current sense resistors as they heated up at these higher current levels. They heat up, the resistance goes up, the CBA thinks the current has gone up, the CBA adjust the current downward to "compensate".

You can see if that's what you're experiencing by cooling the CBA completely and carefully measuring the current immediately after the start of the discharge and then comparing that to the actual current reading a few minutes later. The CBA can take up to several seconds to reach its set current level but that should be quick enough to catch the resistors before they start drifting too much.

 

[Doctorbass]

I tested my CBA2 at 10A and at 20A today.

10A was done with one DMM in series, 20A with 2 DMM's parallel to each other, in series with the battery.

I used 3 of the Vectrix cells (NiMH) in series, about 3.6V. Voltage at the CBA2 would have been lower than that.

Voltage was not measured (or rather not taken much notice of), I know it is very incorrect anyway due to the multiple cables and fuses I have in the loop in addition to the CBA2 cables. I can look up the CBA2 voltages in the .txt files if anyone needs them, but I did not measure the voltage at the battery terminals.

At 10A (set) the DMM showed 10.15A and the CBA2 showed 10.03A. That's 1.1% difference.

At 20A (set) the two DMMs added together showed 20.34A when the CBA2 showed 20.0A. That's 1.7% difference.

I do not know how accurate these DMMs are.

Mik,

Usually, the multumeter have around 5% error on A scale.. the mA scale os more acurate.

Also they guaranty this acuraci if the measurement take no more than 10 or 5 sec depending on the multimeter. that's becasue of the temp rise of the internal 50Wshunt in the DMM that add error.

If you use a Fluke or a precise Kithley lab current meter, you will get the best acurate value to compare..

My personal reference is a 0.1ohm 1% 250W DALE resistor I get 1.00 +/-0.01V per 10A and it is sealed so perfectly insulated agaist moisture.

Doc

 

[BatteryMooch]

Good points and good advice Doc!
A huge resistor is a great idea to keep temps (and drift) down. You can also look up its TCR (temperature coefficient of resistance) and power vs. temp graphs and calculate what the resistance drift will be at various current levels.

Since I made a mistake earlier in remembering what one of my CBA's accuracy and drift numbers were, here are the correct ones. All tests done with a regulated power supply to keep the power levels steady.

10A
-0.2% off at start.
-0.4% off after 2 minutes due to drift.

20A
-0.2% off at start.
-1.0% off after 2 minutes due to drift.

29A (highest I could do without exceeding the CBA's max)
-0.35% off at start.
-1.7% off after 2 minutes.
Ran for one more minute without change and then the fuseholders fell out of the circuit board.

After soldering the fuseholders back in, I decided that the tests were over.

 

[steveo]

Hey Everyone,

Seems like everyone is playing with a cba 2 tonight lol ..

Anyways i though i might post an update;

Mods to CBA

-4110 (test mosfet)
-Wires changed to 8AWG & Huge alligator clips ( Like Docs)
-Added copper sinks to fet legs; (Like Docs)
-Tested at 8amps @ 12v = 96watts with dual fans for 10mins (successful!!!)

I had to calibrate the current at 100mah; but even still; the software mahs are different from the watts up meter at approx 14mahs per ah based on my test for 8amps @ 12v;

I will do some more testing; @ 15 c if possible tommorw (limited with my 30 amp fuse ( i think i've gone through half a dozen by now)

-steveo

 

[steveo]

Hey Everyone,

Seems like everyone is playing with a cba 2 tonight lol ..

Anyways i though i might post an update;

Mods to CBA

-4110 (test mosfet)
-Wires changed to 8AWG & Huge alligator clips ( Like Docs)
-Added copper sinks to fet legs; (Like Docs)
-Tested at 8amps @ 12v = 96watts with dual fans for 10mins (successful!!!)

I had to calibrate the current at 100mah; but even still; the software mahs are different from the watts up meter at approx 14mahs per ah based on my test for 8amps @ 12v;

I will do some more testing; @ 15 c if possible tommorw (limited with my 30 amp fuse ( i think i've gone through half a dozen by now)

-steveo

I just ran another test at 10amps @ 12v and ran fine for 5 mins; However i stopped the test because it was running really really hot.

Max i'd say based on the cooling is 9amps @ 12v which i can live with

However i've hit a 15 - 16 amp barrier @ 3.3v; I can't seem to discharge at any higher rate, I presume its something to do with the 4110 specs.

-steveo

 

[BatteryMooch]

Hi steveo,
Yea, the IRFB4110 just can't pass much current when used in the CBA.

Running a couple of numbers from the data sheet shows that the 4110 runs at about 178C when in the CBA and a discharge of 9A @ 12V. That's pretty darn hot. Actually, wayyyyy too hot. The absolute max temp rating is 175C.

But, the max rating when used as a load (like in the CBA) at 12V is only 4A! Even though the temperature might be around its max with that load, the 4110 must be severely derated at above 8V or so to prevent hotspotting and thermal runaway...very bad. It'll run for a while, maybe a long while, but that MOSFET is not happy right now.

I recommend keeping it at 75W or lower when discharging cells/pack of 8V or lower.
Keep it at 4A or less at 12V
Keep it at 1A or less at 24V
Forget about > 24V

[Edit] Here's the Safe Operating Area graph for the IRFB4110. Discharge current is the vertical axis, battery voltage is the horizontal axis. You can see how severely it must be derated as the time spent as a load increases from 100usec to continuous and as the battery voltage rises. You should never operate above or to the right of the DC plot line to prevent the MOSFET from burning out.

 

[steveo]

Hi steveo,
Yea, the IRFB4110 just can't pass much current when used in the CBA.

Running a couple of numbers from the data sheet shows that the 4110 runs at about 178C when in the CBA and a discharge of 9A @ 12V. That's pretty darn hot. Actually, wayyyyy too hot. The absolute max temp rating is 175C.

But, the max rating when used as a load (like in the CBA) at 12V is only 4A! Even though the temperature might be around its max with that load, the 4110 must be severely derated at above 8V or so to prevent hotspotting and thermal runaway...very bad. It'll run for a while, maybe a long while, but that MOSFET is not happy right now.

I recommend keeping it at 75W or lower when discharging cells/pack of 8V or lower.
Keep it at 4A or less at 12V
Keep it at 1A or less at 24V
Forget about > 24V

[Edit] Here's the Safe Operating Area graph for the IRFB4110. Discharge current is the vertical axis, battery voltage is the horizontal axis. You can see how severely it must be derated as the time spent as a load increases from 100usec to continuous and as the battery voltage rises. You should never operate above or to the right of the DC plot line to prevent the MOSFET from burning out.

Hey

I will take your advise on this ; I've unmoded my cba for now .. and am waiting on your fet; I want to make sure the cba works 100% accurately or very close .. so far after about a 2000-2300mah test its about 20-30mah off from my watts up meter which is not to bad.. and that's with the current adjustment at 100mah (max adjustment setting)

-steveo
-steveo

 

[BatteryMooch]

Hey

I will take your advise on this ; I've unmoded my cba for now .. and am waiting on your fet; I want to make sure the cba works 100% accurately or very close .. so far after about a 2000-2300mah test its about 20-30mah off from my watts up meter which is not to bad.. and that's with the current adjustment at 100mah (max adjustment setting)

-steveo
-steveo

The MOSFET should be arriving any day now.
Your CBA and the WattsUp agreeing to within 1% is darn good IMHO!

 

[steveo]

Hey

I will take your advise on this ; I've unmoded my cba for now .. and am waiting on your fet; I want to make sure the cba works 100% accurately or very close .. so far after about a 2000-2300mah test its about 20-30mah off from my watts up meter which is not to bad.. and that's with the current adjustment at 100mah (max adjustment setting)

-steveo
-steveo

The MOSFET should be arriving any day now.
Your CBA and the WattsUp agreeing to within 1% is darn good IMHO!

I still personally think the watts up metter is more accurate .. however i could be wrong.. I bet with the new mosfet I should have very more accurate results; this difference has to do with the resistance of the fet IMO ..

hopefully with x-mas here that will get here soon .. i'm in no rush so its cool

-steveo

 

[steveo]

Hey

I will take your advise on this ; I've unmoded my cba for now .. and am waiting on your fet; I want to make sure the cba works 100% accurately or very close .. so far after about a 2000-2300mah test its about 20-30mah off from my watts up meter which is not to bad.. and that's with the current adjustment at 100mah (max adjustment setting)

-steveo
-steveo

The MOSFET should be arriving any day now.
Your CBA and the WattsUp agreeing to within 1% is darn good IMHO!

Finally got those mosfets this week; thanks again;

I will try to calabrate it this week (current) and see how that goes, so far it works really well and alot less heat i think!

-steveo

 

[BatteryMooch]

Finally got those mosfets this week; thanks again;
I will try to calabrate it this week (current) and see how that goes, so far it works really well and alot less heat i think!
-steveo

Wow, they took a long time to get to you!
Looking forward to hearing how the mod goes.

 

[steveo]

Finally got those mosfets this week; thanks again;
I will try to calabrate it this week (current) and see how that goes, so far it works really well and alot less heat i think!
-steveo

Wow, they took a long time to get to you!
Looking forward to hearing how the mod goes.

Hey Camlight,

Finally have gotten around to doing some hard testing @ 20amps on the cba 2 & it work perfectly; It runs fairly cool! I can easily crank out 30amps out of a 3.3v a123 cell Thanks again!!

-steveo

 

[BatteryMooch]

Great news!
Thanks for the update.

 

[bobmcree]

i have not been in on this thread until now, but i just went through it and appreciate all the info about the CBA. I have never trusted the CBA to make measurements of cell impedance, or above 20A, for many of the reasons cited, but also because the standard method for measurement of cell impedance is to do it at 1 kHz and thus avoid many of the inherent errors that are not just specific to the CBA.

i have found that if i measure the cell voltage with the CBA2 at 10A and then again at 20A and subtract the difference that the measurements are fairly consistent and in line with the advertised specs and the measurements made at 1 kHz with my other test setups.

i had a big argument with batteryspace years ago about their cells not meeting the advertised spec on impedance, and they insisted the test must be done at 1 kHz. i guess they figured i would shut up and go away. i took a .005 ohm .1% 250w resistor i have laying around, and put it in series with a fully charged cell. i then connected my sine wave generator set to 1 kHz to a stereo amp, and i drove the 1 khz signal through the battery, the resistor, and a load. the ratio of the AC voltage across the resistor to the AC voltage across the cell can be measured easily with inexpensive equipment to well below 1% measurement error. the resistor is physically very large and its temperature does not vary significantly during the test. this is a very accurate way to measure cell impedance. any time a ratiometric measurement like this can be made the errors are usually reduced significantly and make it more accurate than a direct measurement, which is still really a ratiometric measurement involving a calibration source that is not available at the site.

 

[Doctorbass]

Thanks Bob for your exemple of good measurement of impedence using 1000hz. Some manufactur will show two RI value ex: like the A123 or Saft or Konion.. this is great.

Your method to test it seems intreresting! maybe with a NI-DAQ and few external component that could become a faster measurement for big battery batch!


Doc