DC IR VS AC 1kH IR measurements

gauss163 said:
................................... DC IR testing into the 80% to 30% SOC range......................................

This is the info I found about SOC.

image_iijeyc.jpg

If that is correct I should be able to test these cells between3.8V and 3.9V.
Guess I need to addsome more columns to my sheet.
After liberating the cells either discharge to 3.85 or charge to 3.85 depending on what SOC the cells werefound in.
Lots of work but I will rise to the challenge.
In the long run it will be interesting to see the differences in readings between AC IR and DC IR in all these stages.
I think 1000 cells should give us a pretty good picture.

Wolf
 
^^^ Thenumbers in that table seem strange. It looks to be voltage under some load, not resting voltage. Where did you find the table?

3.80V - 3.90V resting should be ok, i.e. between 30%-80% for most common chemistries, e.g. see here, which shows we can extend that interval by about 30mV on each end if need be (for all chemistries there).
 
^^^ The LHS values in your table are "info he picked up" somewhere. Those are either voltages with current, or an ancient chemistry, or erroneous. The RHS numbers are resting voltages measured on his cell, and they correspond very closely to typical modern LiCo resting voltages such as the Sanyo in the lygte link I gave above. So his "picked up" numbers are not relevant for (modern) common resting voltage (OCV) vs SOC.

Note that you can also see the voltages under 1A and 3A loads in the lygte tables, which is sometimes useful to know.
 
Yep I got that hence I chose 3.85V.
Goodnumber for testing purposes for multiplechemesries.
Wolf
 
Btw, beware that info from that highly-active CPF user cannot generally be trusted, e.g. a few posts laterhe claims the following:

SilverfoxonCPF said:
Li-Ion cells thrive on frequent top offs. If you treat your flashlight batteries similar to how you treat your cell phone batteries (toping them up after each day of use) you will get very long cycle life from them.

Of course this is the exact opposite of the truth, i.e. frequently topping off to high SOC is one of the most unhealthy ways to manage Li-ion batteriesbecause high voltage/SOCand higher tempsare the primaryculprits driving accelerated degradation.

Alas, that user is also a highlyactive moderator in the battery section ofCPF, and he (and other mods)frequently shutdown threads and deleted posts that contradicttheirwrong beliefs. Obviously suchactions pose serious problems for accuracy and vetting of knowledge.Due to such (and related matters), many of the most knowledgeable users have left CPF. But that's just one of many mismanagement tumors that often lead to the demise of websites.
 
I don't want to be cavalier about the above statements but I really don't care about that "user" on CPF. I found some numbers that seemed to be correct with my preconceived educated assumption of 3.8 to 3.9 V being in the 50% to 60% range to meet the 80% to 30% criteria set for DC IR testing.
I split it in the middle and 3.85 was the answer.
That number seems to meet your criteria also.
We are here to discuss AC IR and DC IR please.

Thank you

Wolf
 
Last edited:
^^ It's certainly helpful to know should you encounter analogous info on that site (assuming you value facts vs. hearsay). That was the point of the warning. I'm here to help you learn the facts. Part of that process involves pointing out what is wrong, and leading you towards good sources, and away from unreliable sources (the web is full of old wives tales - even for Li-ion batteries).
 
oliverthom707 said:
I have read a bunch of studies on this subject till my eyes have glazed over and I have to get back to reality and say to myself KISS. (Keep it simple stupid)

Yes KISS.
I agree sometimes we get wrapped up in the complexity of it all.
For me it's more about learning how to build a piece of test equipment that can be duplicated easely enough by others if they so desire.
Learning how to write the code to accomplish the task at hand. It is kinda satisfying when you upload the sketch and it works.
As this DC IR study matures I hope to find some way to incorporate it into a testing procedure that is simple and effective for the average person.
Lord knows most of the users here won't even test for AC IR let alone DC IR. Also who am I to tell them otherwise. Some of the DIY builders have packs that where put together without any of the testing other than capacity and running just fine.
My goal in this exercise is to give everyone the option of using the tools that are available to verify their cells SOH. To be able to build battery packs that are balanced to the best of their ability and hopefully be trouble free for many years.
The rest is up to the individual.

Wolf
 
OK on a lighter note I have some interesting information regarding AC vs DC IR.
I have tested 226 cells so far with varying degrees of SOH.
Some of these cells where pulled out of newly acquired packs, some have been siting in my storage area because they were cells I wouldn't use in my power wall as they were less than my 2250mAh criteria.
All voltages were recorded as the cell was found. AC IR was tested and DC IR was also tested if there was enough V in the battery.
Out of all of the cells 4 were below 0.5V and 2 had no V at all with a tripped CID suspected as the AC IR was ∞.
The 4 cells that were < 0.5V had a respectable AC IR hence were gently recovery charged to 3V at 20mA and then finish charged and tested at full 1A on the OPUS. No Explosion, No rupture, No fire.
Some pictures of the various packs these cells came from.

image_iyvpko.jpg

image_syjbfd.jpg

image_wcrdve.jpg


image_nsdrpg.jpg

image_aaehnl.jpg


I also pulled some selected cells from my close by stock to try to round out the selection. I am committed to testing 1000 cells with this study so I guess I will have to go to my cell stock in my storage trailer this week to get some more.

image_axsnzg.jpg

image_ixgktp.jpg


So now to the good part.
The whole procedure for each cell is as follows:
Note: checked" implies recorded
AC IR checked, Voltage checked, DC IR checked, Cells inserted into Charger/Testers and the IR that the Tester showed checked, C/D/C at 1A except for the Vapecell which only does 1A charge and .5A discharge. The mAh results checked, the post C/D/C V and AC IR checked, and the post DC IR checked twice.
All this info can be found on this sheet hereor downloaded below as an attachment.

So far, the only correlation I see between AC IR and DC IR (of cells I would consider using because the AC IR was acceptable) is that DC IR is somewhat higher.
In the filtered graph of cells, I would use, you will notice that the precharge DC IR is very high on some of the cells. That is because they are < 2.8V and the DC IR is high due to the V drop. The cells recover to a respectable DC IR after their charge.

image_kpqevp.jpg


Here is a graph of all the cells with precharge and postcharge voltage.
If you notice the AC IR does not change much at all and the DC IR follows it in a more exaggerated fashion depending on the voltage of the cell.
The current observation is that AC IR so far has been a very good indicator of a cells SOH with DC IR confirming it.
AC IR can indicate a cells SOH even at low V i.e. a cells cutoff V of ≈2.8V to 2.5V and even below that, where DC IR needs the cell to be at least ≈3.2V to get a reasonably respectable reading.

image_ypksvh.jpg


Here is a chart of the SOH compared to AC and DC IR.

image_fqadgj.jpg


These are just preliminary findings and by no means support a conclusion yet.
If someone wants to dig into the data on the sheet and extrapolate more info that would be great.
Stay tuned
Wolf
Excel Sheet of 246 cells tested with AC IR and DC IR
 
Last edited:
Hi Wolf,
So this device has a reasonably low gate threshold.
https://www.sparkfun.com/datasheets/Components/General/RFP30N06LE.pdf
Another way is add bit of circuitry to get a boosted rail from the cell if you wanted, but complexity...
This chip could drive the existing MOSFET gate properly via 10ohm resistor with only a few parts:
https://www.maximintegrated.com/en/products/interface/transceivers/MAX3380E.html approx $8
it creates it's own rails for RS232 drive from 3.3V
(note you'd need to invert the drive from the esp32 - maybe in do in s/w?)

Another way is with capacitor voltage doubler circuit like this:
https://www.maximintegrated.com/en/products/power/charge-pumps/MAX1682.html
Then feed a driver circuit...
 
Redpacket
So yea the IRFZ34PBF didn't seem to work out to well it is actually worse than the IRF520N.
I did a bit more searching and came up with the FQP30N06L so I got some of those coming.
This AM I also ordered the Spark Fun RFP30N06LE logic level breakout board.
I think that may be the final solution 3.3 to 5V lets see what happens.
Thanks for all your help. :)

Wolf
 
Last edited:
Here is an update what is happening with the DC IR tester.
So I was struggling to find a way to fully drive the gate of the MOS with 5V and I played around with a bunch of different potential solutions some worked some did not.
One solution was to use a NPN transistor to drive the gate with 5V which worked but unfortunate the solution drove the gatehigh if the GPIO pin was low.
I rewrote the code to keep the GPIO pin high till I wanted the MOS to turn on and it worked quite well but it was clumsy and dangerous. Can't be sure the GPIO pin will always be high to keep the MOS off. Especially if there is a battery in the holder.
I have not tried a gate driver yet but I may not need to.
I did a little experiment just to see how much different loads would affect DC IR so I pulled a 3.3V relay from my stock and hooked it up. Worked flawlessly and I was finally pulling almost 4A as ohm's law would dictate 4V, 1Ω, 4A. What I did discover was that DC IR didn't change very much at all. I was more or less not surprised as my tester calculates DC IR from mA and Vdrop.
So I did a test using the MOS and the relay with a 1Ω and 4Ω resistor.
The delta between the readings was at max ~ 10mΩ. So as far as I am concerned going with a 1Ω resistor and the MOS which produces ~2A load on the cell is just fine. Not too much and not too little.
The cell is a UR18650E and it was tested 4 times so that can account for a little difference in the reading due to usage.
Here are the results.
4Ω Relay

image_auepdy.jpg

4Ω MOS

image_fdeldy.jpg

1Ω Relay

image_xhuelc.jpg

1Ω MOS

image_wwzpkp.jpg


Now while testing some NCR18650A cell I found a funny IR creep on the charts. It is not Voltage or IR specific
Here are 2 NCR18650A cells with similar IR but 1 cell at 4.142V and the other at 3.764V
Look how the IR slowly rises and the Vdrop does the same. this is not how most cells react. The generally react like the UR18650E above.

image_sipptm.jpg

image_qdzjzk.jpg

Now here is a NCR1865 A with a quite normal trace.

image_fbwdtm.jpg

And here a CGR18650E

image_ryysij.jpg


So many questions so little time I will keep plugging along.
That's it for now.

Wolf
 
Last edited:
Wolf said:
[...] Now while testing some NCR18650A cell I found a funny IR creep on the charts. It is not Voltage or IR specific.
Here are 2 NCR18650A cells with similar IR but 1 cell at 4.142V and the other at 3.764V
Look how the IR slowly rises and the Vdrop does the same. this is not how most cells react.

Presumably this is due to those cells having much higher non-ohmic resistance components (RCT = Charge Transfer resistance, and Rp = polarization/diffusion resistance), cf. post #47.

As I emphasized above, for cells so aged (with higher non-ohmic IR)the DC IR yields much more info about how the cells will perform under nontrivial loads since it reflects the actualvoltage drops caused by these longer timescale components.
 
Wolf said:
Redpacket
So yea theIRFZ34PBF didn't seem to work out to well it is actually worse than the IRF520N.
I did a bit more searching and came up with theFQP30N06L so I got some of those coming.
This AM I also ordered the Spark FunRFP30N06LE logic levelbreakout board.
I think that may be the final solution 3.3 to 5V lets see what happens.
Thanks for all your help. :)

Wolf

Try driving the IRF520F with 14v gate voltage :)

I have a push pull circuit that will drive 6A at 10khz no problems :)

It has a nice linear response zone for accurate current control.

Computing the impendance needs 1khz with at least 16ksps (64x imp even better) , more the better. Have to sum those quadrants of V and I :)

Parallel synchronised adc even better.
 
boom...
Found a ESR difference of AC vs DC (in red rectangle) in spec for some super-capacitors. I have a two of SAMXON 50F/2.7V.
By spec ESR-AC max 30, ESR-DC max 50 (in green rectangle):
Selection_1074.png
Each of them measured by YR1035+ had ~9mR ESR, so real ESR is ~ 3times less than by spec.

I discharged them to 0V (balancing :) ), connected them sequentially, charged to 4.25V. Charged capacity is ~37mAh (calculated based on time).
Total ESR is 20mR, as expected:
IMG_20210911_211405.jpg

Then I tried to measure ESR by DC method by iCharger X8, as it's quite good equipment as a "reference" for DC method, right?

I used a proper "4 wires" method:
IMG_20210911_204832.jpg
You should consider value 24.8mR - that's based on Vdrop on sense wires, while 45.7mR is based on power wires.
Then I test other Li-Ion cells (low ESR, high drain ones), the iCharger X8 (DC) with 4 wires often shows results similar to YR1035+ (AC)

Each ESR-DC test was eating 0.1v of the super-capacitors set. I did it more than 15 times to discharge supercaps to 0.5v or so.
Interesting that during all these tests the ESR-DC was very consistent, so it did not depend on SOC of supercaps.

What I wanted to say here? Looking on the spec and based on my results, it indeed looks like that ESR by DC method might be ~2 times higher than AC method. Yes, it's all relative, but probably still DC method, when established by "your own standard" could be used if you need to test many identical cells to spot some wrong cell.

At the end some screenshot from some presentation mentioned earlier in this thread.

Selection_1076.png
 
Last edited:
Each ESR-DC test was eating 0.1v of the super-capacitors set. I did it more than 15 times to discharge supercaps to 0.5v or so.
Interesting that during all these tests the ESR-DC was very consistent, so it did not depend on SOC of supercaps.

What I wanted to say here? Looking on the spec and based on my results, it indeed looks like that ESR by DC method might be ~2 times higher than AC method.
Yes that the ESR of a capacitor would not change much makes perfect sense as it is not tied to a chemical reaction to release electrons such as a Li-Ion battery is. A Li-Ion battery has to ramp up to deliver the current and the less of a SOC it has the the "slower" it is and the more of a vdrop occurs hence the rise in "DC IR". Capacitors are not hampered by the ramp up, hence the stability. Indeed DC IR being ≈ 1.5X to 2X of AC IR given that the SOC of the Li-Ion cell is at least ≈ 80% was my findings also.
Chart of typical cell analyzed with several loads and compared to AC IR
Wolf
1631545719872.png
Also the sheet of 246 cells tested.
 
Last edited:
@Wolf,

if you are still looking for some logic-level Mosfet that will switch completely on at low Vgs (gate-source voltage) you have to look for the least Vgsth (gate-to-source threshold voltage) which is the voltage the channel starts to open (but still is not fully open). I collected a few parts numbers as i was in need for such Mosfets as well to securely switch open at typical Li-ion cell voltage of 3.7 V. Here are some suggestions:

FQP30N06L: 30V, 32A, 2,5Vgsth

IRL2910: 100V, 55A, 2Vgsth

IRLML2502: N-Ch_20V_4A_1Vgsth

Si2302DS: 20V_2,8A_Vgsth_1V SMD

Si8424db: 8V_9A_1Vgsth SMD, quite exotic, hard to find

They all should be able to pretty much switch completely open at least at 5Vgs, probably 90% open at 3.5V
I would go for the IRL(M)-types or the FQP30N06L, which i personelly used in some projects. Make sure you dont get fake ones (often sold by ebay-china sellers for the "best" prices).
 
Back
Top