DC IR VS AC 1kH IR measurements

Announcement - Help us fight the BOTS! Please report all spam including stuff in your inbox!

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,561
Redpacket said:
That's some awesome work there.
I guessing the DC IR results also followed the discharge current rating for the cell?
Ie it would confirm high current cells would have a lower DC IR than say your average laptop cell?
A 1ohm load would be quite a high load for some cells right?
Yes high discharge or high drain cells such as IMRINR NCA chemistry cells tend to have a low AC or DC IR to begin with.
ICR chemistry is of course low drain and henceforth inherently has a higher AC and DC IR.
Preliminary results show with my rudimentary tester based on ohms law that the difference between AC and DC IR on a good cell is about 1 to 2.
Meaning that if a cell has 12m? of AC IR it will more than likely have a 25m? DC IRfor a high drain cell.
The same holds true for low drain with an IR of say 50m? will have a DC IR of 100m?.
A constant 1? load would behigh for most cells yes but in my case it is only applied for 1 second to get a stable vdrop Voltage.
I have tried 100ms 500ms and all kinds of variables. but the 1 second load seems to be the sweet spot to get a repeatable result.
I can do this test 3 or 4 times on a fully charged good cell and get the same results within a few m? every time.
My tester has a 1? 100W resistor load that gets put on the battery at 4.1 or so V.Technically that should produce a 4.1 Amp currentbut it does not as you can see by my discharge chart it is in the neighborhood of 2 .xxxAmps. ( I have verified that with my Fluke meters) The load wire is 10AWG very capable of handlingthat current. The sense wires are 20AWG although that really doesn't matter as there is no current flow through them.
So the calculation is V(open) - V(load) = V(drop)/mA(draw) = R
Unfortunately as the second paper I linked to basically says that there are no DC IR standards set so how do we know we are on a the right track?
Article publishedAPR 15, 2020.https://www.electronicdesign.com/te...e/21128843/measuring-dcir-of-lithiumion-cells

image_uuwroz.jpg

So I took an educated guess and these are my results. Are they right? I guess so. We are in somewhat uncharted territory and till there is a standard set we just don't know.
One encouraging spec sheet for the 18650-30Q does give us a DC IR value of:

Type Spec. (Tentative) Typical INR18650-30Q
Initial IR (m? DC (10A-1A)) ? 30 19.94 2
But how was this determined?
My tester seems to hit that nail on the head does that mean it is right?
\_(?)_/

Wolf
 

gauss163

Member
Joined
Jun 28, 2020
Messages
278
Wolf said:
[...] So my conclusion follows this presentation done in February, 13, 2020
https://fhi.nl/app/uploads/sites/74/2020/02/Batenburg-Mechatronica.pdf
Hioki said:
Conclusions
Low frequency AC-IR is a valid alternative for DC-IR measurement [my emphasis - gauss]
AC-IR reduces measurement time dramatically
Possible to measure resistance of 1m? or less accurately
4-terminal pair test leads required to measure low resistance with AC-IR

I could not have said it better myself. Additional reading: https://www.electronicdesign.com/te...e/21128843/measuring-dcir-of-lithiumion-cells

Oops, it seems you've misread. That conclusion does not support your claim. Rather Hioki's (full) argument emphasizes the same points I've made above and elsewhere about the importance of DC IR as a much more refined measure of health vs. purely ohmic1 kHZ AC IR measurements. One of the dangers of reading only summary conclusions is that we can miss subtleties abbreviated in a summary. Above it seems you missed the crucial modifier "low frequency" in the key claim that "Low frequencyAC-IR is a valid alternative for DC-IR measurement".

If youread the entire slideshow (or better theirwhite paper)you'll seethat the primary point they make is that low frequency (1 Hz) AC IR measurement is closely correlated with DC IR measurement, so it may be used as an equivalent of DC IR (Hioki and others haveleveraged this point for many years to market high-end $5000 variable-frequency AC impedance meters). But your $50 AC IR meters cannot perform low-frequency (1 Hz) tests; rather they can only do a fixed high-frequency (1 kHz) test, which cannot measure the longer timescale non-ohmic components (R2 below)of the resistance (major components of degradation). Unfortunately the cell batches inyour tests above don't have a wide-enough distribution of health to make this point clear(but I suspect some of your older tests do).

In fact Hiokigraphically emphasizes this in the slide preceding the conclusion you quoted -which I have copied below. Just as I showed in prior posts on Nyquistplots,note how as the cell agesthe ohmic component R1 (= 1 kHz AC IR)changes very little, but the non-ohmic component R2 changes much more greatly (here DC IR = R1 + R2).

image_eruiio.jpg
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,561
I stand corrected.
1Hz AC IR measurement is equivalent to DC IR my mistake.
Time to figure out how to build a 1Hz AC IR meter.
Although I already have a DC IR meter that is equivalent. I only need to have a fully charged battery to use it.
The other question still needs to be answered though and that is what are the acceptable specifications for individual batteries
that do not have a 1Hz AC IR spec and or a DC IR spec?
We can look at charts on how to measure it all day long but what is a good reading and what is not?
Since there are no standards how do we even know what is right?
There are 2 spec sheets that list DC IR
One is for a Samsung 18650-30Q and the other is for a Samsung INR18650-25R. I have not found any others.
https://eu.nkon.nl/sk/k/30q.pdf
https://www.pdf-archive.com/2015/11/09/datasheet/datasheet.pdf
Now that begs the question.
How are we to test batteries from used packs and know that they are reasonably good enough for us to charge andto reuse in our builds?
As far as I am concerned, andfrom my experience (my cheat sheet) if a high frequency AC IR test can give me an indication of the (somewhat) SOH of a cell
I can at least test the cell and find out if it is any good with an of the shelf charger/tester.
One observationthat cannot be disputed though is that if you follow my cheat sheet (measuring cells with a $50.00 1kHz AC IR meter)your success rate with SOH is extremely elevated.

I will continue to try to find other cells ( I have over 3000 cells that have been tested but didn't make the grade)to round out the test sheet with more marginal cells so that we can get a clearer picture of the relationship between high frequency AC IR and DC IR.
I do understand that R1+R2 gives us a moretotal picture of the SOH of the battery given thatAC IR is atthe same starting point.

Best I can do.

Wolf
 

Redpacket

Member
Joined
Feb 28, 2018
Messages
1,288
Wolf, just thinking about the current flow & the 2A current (better than 4A IMHO), this could be from the "round trip resistance" of the leads, cell contacts, mosfets switch board, etc, etc?
Might be another 1-2ohms there given the results?
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,561
Redpacket said:
Wolf, just thinking about the current flow & the 2A current (better than 4A IMHO), this could be from the "round trip resistance" of the leads, cell contacts, mosfets switch board, etc, etc?
Might be another 1-2ohms there given the results?
Good Point!
So I did a resistance check with no battery in the holder and the MOSFET on and the resistance was 1.444?
I then did a resistance check with my dummy battery installed and came up with 1.432?
Also standing resistance with MOSFET offis 30.701k?

4.1 V /1.443? = 2.863A getting closer to what the sensor shows.
Wouldn't take a lot of cell contact resistance ~ 0.4? to get to 2.2A
Nice observation.
Good thing I didn't take the face value of the resistor as my calculation constant but rather the actual amps.

Wolf

image_knecre.jpg

image_hkfghy.jpg

image_wunsin.jpg
 

Redpacket

Member
Joined
Feb 28, 2018
Messages
1,288
Curious about the MOSFET resistance, depending on the exact device, I would be looking for "nearly copper wire" on resistance!
I'm thinking this should be < 50 milli ohms.

Off reading is typically higher that 30k (if you fully ground the gate to 0V, doesn't take much Vgs to start a trickle of current). Not that this matters much for these tests!
What's the device part #?

PS, love the work bench pics (active project!) Nice gear too, I'm green with envy!
 

Redpacket

Member
Joined
Feb 28, 2018
Messages
1,288
be sure to drive the gate with 10-12Vdc to ensure it's on properly :)
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,561
Redpacket said:
be sure to drive the gate with 10-12Vdc to ensure it's on properly :)
Hmm I can't do that.Not with esp32 max is 3.3V.
I guess I could use aN/P MOSFET one drives the other but that is just making it more complicated than it needs to be.
Also I would then need a seperate 12V source. Also not a show stopper but makes all a bit more complicated and not self contained.

Here is the circuit diagram. Suggestions and enhancements are welcome.
Wolf

image_qvznsx.jpg
 

gauss163

Member
Joined
Jun 28, 2020
Messages
278
Wolf said:
How do we know we are on the right track, sinceunfortunatelythe second paper[size=small]([size=small][size=small]publishedAPR 15, 2020)[/size][/size][/size]I linkedbasically says that there are no DC IR standards, i.e.
BobZolloatKeysight said:
Unlike ACIR, dc internal resistance measurement lacks standardization and thus tends to be somewhat enigmatic.

So I took an educated guess and these are my results. Are they right? I guess so. We are in somewhat uncharted territory and till there is a standard set we just don't know.

Like the other Hioki slideshow you cited, that document toois essentially anadvertisement (here for$5000Keysight power supplies). There aKeysight employee argues that we can employtheir fast accurate (18b ADC)high-end power supplies to measure the ohmic part of the DC IR. Butwe already know that we can use $50 AC IR meters to measure that ohmiccomponent.

Rather than peruse articles biased by marketing hype, recall that acouple pages back in post #47I gave linksto more comprehensive introductions to various methods for DC IR measurement. Theyare much better starting points to begin learning about these techniques.

There is no (single) DC IR measurement standard just like there is no single number for measuring human health, or a car's mpg (depends on your particular mix of city and highway driving, etc). Similarly the type of AC/DC IR measurement that is most pertinent will depend on battery usage conditions, viz.the type of load (e.g. constant current vs. short large pulses as in vaping/jump starting,or some specific mixture), and ontemperature ranges, etc. Only by knowingall of the components of the DC IR will will have any hope of derivinggood estimates of how abattery performs under diverse conditions. Armed with such we have a much betterchanceof achieving near-optimal battery performance and lifetime - whether in powerwalls or other devices.

Wolf said:
Although I already have a DC IR meter that is equivalent. I only need to have a fully charged battery to use it.

Generally DC IR is near constant vs. SOC except at low & high SOC, so it's essential to measure it in this constant region, typically between 80%-30% SOC, e.g. see the graphs I posted here. Otherwise the measurements will not be comparable so will be of little use.
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,561
gauss163 said:
There is no (single) DC IR measurement standard just like there is no single number for measuring human health, or a car's mpg (depends on your particular mix of city and highway driving, etc). Similarly the type of AC/DC IR measurement that is most pertinent will depend on battery usage conditions, viz.the type of load (e.g. constant current vs. short large pulses as in vaping/jump starting,or some specific mixture), and ontemperature ranges, etc. Only by knowingall of the components of the DC IR will will have any hope of derivinggood estimates of how abattery performs under diverse conditions. Armed with such we have a much betterchanceof achieving near-optimal battery performance and lifetime - whether in powerwalls or other devices.


Generally DC IR is near constant vs. SOC except at low & high SOC, so it's essential to measure it in this constant region, typically between 80%-30% SOC, e.g. see the graphs I posted here. Otherwise the measurements will not be comparable so will be of little use.

Interesting.

So if we do not have a (single) DC IR standard we don't have a chance in getting it right. We may have an idea where it should be but with temperature, SOC and many other factors we are always trying to hit a moving target.
We do have numbers that measure human health. Blood pressure, body mass index, cholesterol levels, an array of blood chemistry tests, ocularpressure test for glaucoma, bone density, alsoof late body temperatureand many more. Some are more subjective to interpretation others are not. Most of these tests have a low limit, a high limit and of course an average that has been determined by experience. Fuel mpg is a guess at the average also whetherit is city or highway or both.

So for DC IR we are left with a guess. How close can we get to the guess being right? I don't know There has to be a standard to go by.
I.e. Cell at 20 C, 3.6V, SOC 50%, 1? load, 1s pulse, and for ICR chemistry DC IR should be ?120m? and ?80m?.
It could even include the same parameters as above and change the ? load value and or the time the load is applied.

Looking at one of the graphs out ofA study of the infuence of measurement timescale on internal resistance characterisation methodologies for lithium-ion cells
the graphs that my system produces is incredibly similar. Am I on the right track? Maybe.

image_qsdxhx.jpg

Posthttps://secondlifestorage.com/showthread.php?tid=9374&pid=64352#pid64352
You wrote
"Second, be aware that DC IR depends on SOC (State Of Charge) and temperature - at extreme SOC and temp it can increase by 10x or more, e.g. see the graphs below. So don't measure it in these extreme ranges (unless testingthoseextremes)."

In that statement I am not sure what you mean "at extreme SOC and temp" (in my minds eye) is fully charged. DC IR if anything is lower according to my tests, an average of 24% lower on all my cells good or bad. If you meant a extreme SOC (fully charged) and high temp,with thisother variable then I do not know as my ambient temp is (house temp) 20C to 25C. Additionally if you meant SOC (meaning discharged) then yes the DC IR would definitely be higheras the V drop would be significantly increased. Temperature is again a variable I have not tested with.
The chart you are referring to does indicate DOD of 80% to100% for those high readings if I am not mistaken.

image_nlnnrm.jpg


As with1kHz AC measurements we do have a clue as to what it should be. Mostmanufacturers actually publish the 1kHz AC IR measurements as being
? some value. At least we have something to go by. I understand that it is a value assigned to a new cell and it is just to quickly verify that the cell after production is valid. Nevertheless it is a value even if it only represents a small portion of the SOH of a cell. The rest of the SOH of a cell I believe can be judged byhow well it performs in a C/D/C test.

One more thing.
In the conclusion of theComparison of Several Methods for Determining the Internal Resistance of Lithium Ion Cells
paper it states:
(Just an excerpt)

The method of measuring AC resistance at fixed frequency gives fast results. Because each cell type shows individual frequency dependence of impedance, AC resistance measured at different type of cells cannot be generally used for cell comparison and benchmarking. Therefore this method is only suitable for measuring and comparing internal resistance of the same type of cell, e.g., for quality screening.However, at frequencies about 1 kHz AC-resistance for different cells (which have comparable size) has very similar frequency dependence. It also could be used for comparing different cells (relatively). It may be expected that cells with higher small signal resistance also have higher large signal resistance. Despite the fact that the impedance spectrum of the cell was recorded with higher damping than the measurement at fixed frequency, the same result for 1 kHz was found.

I have read this over and over again and it certainly does not say that it is the wrong way but just another way.I am not a scientist or claim to be.
What I do know, is that through experimentationyou can find things out and if the experiment is repeatable every time there must be something to it.
I will continue to play around with DC IR as I find it a fascinatingstudy. Maybe I can find some correlation with the randomness of it all.

As far as my cell testing is concerned I will stick to AC IR for now as I know it has proven itself time and time again with the outcome shown on the testers mAH results.
Wolf
 

gauss163

Member
Joined
Jun 28, 2020
Messages
278
Wolf said:
[...] the graphs that my system produces is incredibly similar. Am I on the right track? Maybe.

All CC discharge graphs have that shape. Though your method is not quite CC (it's CR = Constant Resistance), it should be very close to CC over such a short discharge, with likely very little effect on the calculated resistance.

Wolf said:
I am not sure what you mean "at extreme SOC and temp" [...]

See my prior post, viz "measure it in this constant region, typically between 80%-30% SOC".

Wolf said:
[...] I have read this over and over again and it certainly does not say that it is the wrong way but just another way.I am not a scientist or claim to be.What I do know, is that through experimentationyou can find things out and if the experiment is repeatable every time there must be something to it.

AC IR is generally only useful as a very coarse first-level screening tool, to weed out very unhealthy or damaged cells. It cannot be used formore refined analyses because it does not incorporate thenon-ohmic components of IR. As I hinted before, likely you don't yet have enough experience to appreciate thisbecause your cells don't have a large enough distribution of health and, further, because your analysis is static, i.e. you don't yet have any cycling data to see the results of non-optimal IR matching. It is known that poor (DC) IR matching can lead to significant reduction in life, e.g. below

BruenandMarco said:
Gogoana et al. [13] cycle-aged two cylindrical lithium iron phosphate (LFP) cells connected in parallel. They found that a 20% difference in internal resistance resulted in a 40% reduction in the useful life of the pair of cells compared to if the cells had approximately equal internal resistances. The authors attribute this to the uneven current distribution between the cells. Their results highlight that each cell will go through periods where it experiences high currents that will in turn age the cells more quickly.

Gong et al. [1] drew similar conclusions from their experimental work with 32 Ah cells. When two cells with a 20% impedance difference were connected in parallel, the peak current experienced was 40% higher than if the cells were identical. The authors also performed simulation studies, using the Mathwork's Simscape extension to Simulink to connect two equivalent circuit models (ECMs) in parallel. This is one of the few examples of parallel cell modelling within the literature from Modelling and experimental evaluation of parallel connected lithium ion cells for an electric vehicle battery system

[excerpted from p. 92 of Modelling and experimental evaluation of parallel connected lithium ion cells for an electric vehicle battery system (2016)].

I suspect you'll find the prior paper of interest given your prior posts here on the study of paralleledmismatched cells.
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,561
gauss163 said:
AC IR is generally only useful as a very coarse first-level screening tool, to weed out very unhealthy or damaged cells. It cannot be used formore refined analyses because it does not incorporate thenon-ohmic components of IR. As I hinted before, likely you don't yet have enough experience to appreciate thisbecause your cells don't have a large enough distribution of health and, further, because your analysis is static, i.e. you don't yet have any cycling data to see the results of non-optimal IR matching. It is known that poor (DC) IR matching can lead to significant reduction in life, e.g. below

I suspect you'll find the prior paper of interest given your prior posts here on the study of paralleledmismatched cells.
I am working on the larger distribution of cells with different health levels. It will take me some time as I have pretty much reduced my inventory to cells with acceptable IR (AC IR) so that is unfortunate. I do think I have a box of 1000 or so though that may prove to be substandard. I will see.
That should round the sheet out.
Well I did the paralleled mismatch cell study with AC IR and capacity in mind. Itwill be interesting to see how cells with just similar DC IR respond.

Here is an Interesting chart of the 135 cells tested so far.
The after the C/D/C cycle tested AC IR and DC IR coloration.

image_yvxulx.jpg



Wolf
 

gauss163

Member
Joined
Jun 28, 2020
Messages
278
^^^ It'd be interesting to include in the prior plot the pretest DC IR, esp. if that was done in the non-extreme region 80%-30% SOC.
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,561
gauss163 said:
^^^ It'd be interesting to include in the prior plot the pretest DC IR, esp. if that was done in the non-extreme region 80%-30% SOC.

Best I can do for now.

Voltage filtered between 2.7V and 3.8V
Wolf


image_mknqwu.jpg
 

Redpacket

Member
Joined
Feb 28, 2018
Messages
1,288
Wolf said:
Redpacket said:
be sure to drive the gate with 10-12Vdc to ensure it's on properly :)
Hmm I can't do that.Not with esp32 max is 3.3V.
I guess I could use aN/P MOSFET one drives the other but that is just making it more complicated than it needs to be.
Also I would then need a seperate 12V source. Also not a show stopper but makes all a bit more complicated and not self contained.

Here is the circuit diagram. Suggestions and enhancements are welcome.
Wolf

image_qvznsx.jpg

There's a logic level part IRFZ34BPF (I should have suggested one like this first up, my bad, sorry!)
Or there's drive modules but extra complexity like you say.
Maybe just check the voltage drop across the device when on, if it's low, all good.
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,561
Redpacket said:
There's a logic level part IRFZ34BPF (I should have suggested one like this first up, my bad, sorry!)
Maybe just check the voltage drop across the device when on, if it's low, all good.
No worries I should have looked at the specs myself. Besides having some spare MOSFETS in stock is never a bad thing. :D
The IRFZ34BPF are ordered should be in Wednesday.

image_flmgbr.jpg

You know sometimes I wonder where my brain is. :sOut to lunch apparently.Of course checking the V drop across the MOSFET would give me a good indication if its open all the way.
Well guess what with 3.24V on the gate the V drop is 1.1V ouch. With 4.69V its .29V so obviously the gate is not open all the way. with 3.24V coming out of the ESP32
Resistance is .296? at 3.24V and .233? at 4.7V.
So there is our missing amperage I did a test with the MOSFET at 4.69V (manual) and the amperage was ~3452mA instead of the2248mA I am now getting.
The neat thing though is that the IR calculation still maintains very close The amperage is higher so the vdrop is higher and it comes out darn close.
Also it seems to be a bit more stable. We will see what this new MOSFET does.

All in all though I think I am on the right track with a DC IR tester. Now if we could only get some standards to compare,
Guess we will just have to make our own......
Wolf

image_fdmork.jpg
 

gauss163

Member
Joined
Jun 28, 2020
Messages
278
^^ Good to see that you got that debugged. Don't forget to shift the DC IR testing into the 80% to 30% SOC range, else the variations at the extremes will greatly complicate matters. And they're already complicated enough as it is...
 
Top