Cell Internal Resistance Meter - more useful than 'C' ratings - Now available.
Now at http://nqrc.com/?vp=PLD-IRM-004 for $59.00 AUD (about $47 USD)
This is a product development thread. The final device will typically go on commercial sale on my site ( http://nqrc.com ) but I like to show the community what's happening during the development process which is useful because invariably it results in a superior end-product.
Have you ever noticed that the 'C' ratings for lipo cells (and Ni*) are generally meaningless thumbsuck values in most cases?
Your 20C from one brand is probably a 15C or 12C in another brand, really there's just no standard and it's mostly luck that you get what you need.
So, being the "scratch the itch" sort of person, I've started working on my internal resistance meter - and yes, it will be commercially produced.
What is "internal resistance" you might ask, well, it's the resistance value of the actual battery cell, this is a very useful figure because from it you can work out things like your voltage-drop under load and how much power is being burned up inside the cell as heat, also it finally lets you -really- compare between packs, rather than just hoping the 'C' values are the same.
As an example, if you have a cell with a 50mR resistance inside, it's a 2S pack, so that's 100mR for the whole pack, we'll ignore the lead and connector resistance for now (though they can an DO cause problems at times - witness people using JST connectors at 10~15A... meltdown).
Okay, so we've got our 100mR pack, we want to draw 9A from it... what's the voltage real voltage (from 8.4V) and how hot will that pack get?
V = I * R
V = 9A * 0.1R
V = 0.9V drop at 9A, so that means your 8.4V pack only gives you 8.4 -0.9V, 7.5V at the motor/ESC
Now, how about how much power/heat is being lost in the pack itself...
P = I * I * R
P = 9A * 9A * 0.1R
P = 81 * 0.1R
P = 8.1W ... that's quite a considerable amount of heat for a small pack. Won't take long to push it over 45~55'C at that rate.
So, that maths is all very good - certainly shows you why knowing the internal resistance is a lot more useful than the 'C' rating doesn't it!
Here's a couple of photos of the first prototype I've built.
(please note, to budding electronics people out there - I know there's a lot of ways of testing cell resistances, including techniques such as impedence spectroscopy, however for now I'm keeping things simple with a static load for a fixed period ... at least in the first prototype ;) ).
Features (thus far)
* Single cell test, all cell types that are up to 5V (per cell).
* Externally powered by 7V+ (9V battery etc)
* Performs 3 different tests (1kHz, and two separate static loads)
* Displays the internal resistance computed from each test
* Very simple to operate, just place the cell to be tested onto the 2-pin connector.
Looks to me like you'll be a millionaire shortly after these go to market. Please keep us up to date.
Interesting project, thanks for sharing.
Can you explain the fields on the display ? I presume that "R" is the measured resistance. My guess is that the others might be the unloaded voltage, loaded voltage and current but the numbers don't seem to work out.
More info on your measurement algorithm would be interesting.
How many cells do you plan to support and is it only overall pack resistance or per-cell resistance you will be reporting ?
Hi there everyone,
icrashrc, sure hope I am a millionaire soon, been trying for a long time... so long as it's not Zimbabwean dollars.
kgfly, nice to see someone from Australia (I'm a bit further north :) ). The fields are just the raw ADC (analogue-to-digital-converter) values (10-bit, 5V reference, so something like '809' means V = 809/1024 * 5 ~= 3.95V ). I've since updated the software to convert them back to proper voltage values. I was somewhat "excited" to get things out to the public to see so I just took the very first LCD outputs up.
The algorithm isn't anything profoundly exciting, I simply first test the open-circuit voltage and store, then I connect the load and check the voltage again to get the loaded voltage. There's also a sample taken on the other side of the load to negate any effects of the switching MOSFET involved.
Essentially you end up with the following data (examples);
Voc = 4.2V
Vload = 4.1V
Vfet = 0.01V
Since the resistor value is fairly constant and known, you can then derive the internal resistance. The very nice thing is that you don't even have to convert the ADC values to get the result, which saves a lot of rounding/conversion errors creeping in.
Right now it's only showing a single cell resistance as picked off from the balance plug.
Moving to multiple cells brings in a lot more challenges and that's something I need to tackle in a benchtop/precision unit, the cost of such a version would be quite a lot more.
Right now I consider being able to pick off per-cell resistances from the balance plug as being the most 'useful' starting point, since it won't be hard for people to add these up for the 'pack' resistance, it also lets you pick out dud cells. You can furthermore compute the wiring loom resistance with a trick (measure between cell-1's +ve terminal and the main battery -ve lead (eg, deans). If you already know the cell resistance is 40mR and then the battery->lead reads 45mR, then you know your lead/plug has ~5mR resistance.
The resolution/precision of this device is +/-5mR at the moment (that's my personal aim) without calibration, the biggest limits are the load resistor, of which I've got special 1% tollerance ones arriving tomorrow, the resistance of the terminals used to probe the balance plug and the built in error factor of the ADC conversions.
Got a rough idea of price? At least a neighborhood - $30? $50? $100?
Any thoughts on comparing your values to those obtained from chargers that measure IR? Would be a good selling point if your measuring produced comparable values to say, the cellpro charger.
Two examples to look at
Here's two batteries as an example.
The first one is a 3S450-15C pack that's long since given up the ghost, take note of the position of the white balance plug on the 2-pins of the meter circuit board (notice how they shift for each cell).
Cell resistances are: 78mR, 125, 103mR ... this pack is useless ( I keep it for testing things like this ).
For the second pack, it's newer 2S850-20C pack, though it's been through about 30 cycles and starting to 'sag' a bit
Cell resistances are: 29mR and 29mR, great - still useful but the cells aren't as "lively" as before.
Now, time to compute our power/voltage wastage due to the resistance, this pack 'claims' to be 20C, so that means we should be able to do 20 x 0.85A = 17A, let's see what really happens...
V = IR = 17 * 0.058 = 0.986V drop (8.4V pack becomes 7.41V)
P = IIR = 17 * 17 * 0.058 = 16.762W (burned up in the pack)
I'd certainly say that the pack is 'reasonably' specced but the power (16W+) burned up inside is a bit of a worry, 16W is more than some soldering irons! Good reason to let your lipos have as much cooling as possible.
If we drop to a more conservative 12C, the results are a lot nicer
V = 10.2A * 0.058R = 0.5916V drop
P = IIR = 10.2 * 10.2 * 0.058 = 6.03W, much nicer, still gets warm but easier to keep cool.
Hope that's illustrated a bit more about how useful/important internal resistance is, even if it's only within 5mR of the true value (remember you're going to see another 10~20mR in the wire/connectors too - hence why it's important to have good soldering and suitable gauge wire).
Of course if you have a watt meter like "Wattsup" and i suspect most do, then plug it in to the battery and record the no load battery voltage. Next connect this now to your ESC and motor. Run the motor up to record a reasonable current then record the current and new voltage. The difference in battery voltage is the voltage lost due to the internal resistance (IR) (and of course connectors etc) Using the same maths above you can calculate the IR. eg using the same setup as the first example you measure unloaded battery as 8.4V and when supplying 9 amps it has dropped to 7.5 volts.
Voltage drop = 8.4-7.5 = 0.9 V
R=E/I = 0.9/ 9 = 0.1 Ohm
Same result but with a meter I already have and accurate enough for the exercise.
Yes, that's certainly another way of doing it - of course that does only give you a 'pack' resistance inclusive of the leads (as you mentioned) and that's excellent for that job. Of course, things are a bit tricker when you want to do a per-cell test, though yes, you can perform the same task via the balance tap one cell at a time and it'll work. One could also do all the same things with a suitable multimeter too if one is handy.
As with everything, there's no absolute need for a device, however a custom built IRM certainly makes a handy tool, especially when it saves you having to rig up your motor/esc/battery/TX and calculator.
I guess for me (and others) this kind of tool represents another piece of precision kit that helps us maximise what we've got with minimal effort :) It's a reality, we could all buy a cheaper charger and a cheap/decent multimeter (i.e. to do Watt tests etc.) and still achieve some measurements but we move into the hassle zone, where a lot of folks would rather not go but instead we pay a bit more for something that gives us additional information without effort required to calculate etc.
In addition I see it as a valuable troubleshooting tool. I've had situations where the AMP output is not what it should be given the C rating notably at the start of the current draw and over time. Where do I go next? Is it bad joints/leads/cells/speedie etc etc. Coming from an IT background I always like troubleshooting at the source - in this case the Cell.
If Paul can provide a tool (with a decent reference sheet) that gives us some clues as to the performance of the starting point of our power without having to go through calculations etc. then I see this as a good buy as I KNOW that a lot of the RC flyers out there don't feel comfortable entering the maths zone :) and we know how many *what the hell* threads there are out there re doing something as simple as calculating the right prop!.
Maybe some info would include comparisons between IR of the Cell and what is being seen at the Deans plug (in the documentation?) would also be helpful.
Actually now I think of it the PLD 'End to End Diagnostic Tool' - EEDT *abreviation is a bit lame...use COOL for now :) - which would allow you to diagnose issues starting from the Cell and then going all the way up to the prop would be great.
I'd love to pull out one kit (whether it be a little satchel of goodies) that allow you to setup your planes electrical system with a minium of fuss.
I know there are different products that handle each of the bits, but for me the important bits are:
1 - Is the prop/motor combo right for the job (a combined RPM/Watt draw meter?)
2 - Is the speedie right for the job (can we see if the speedie is holding back i.e. rated @ 20 amps but only delivering 16amps?)
3 - Is the battery delivering the AMPs required - steadily and consistently - to manage item 1 and if so or not is there an IR or other issue?
Call me lazy - but I would rather be building or flying than pulling electronics out and hooking up to numerous testing items.
An end-2-end tool would be a rather massive undertaking. You'd want/need something that basically intercepts every connection and processes that information as you test, eg, you'd have the following plugs;
*receiver (in, for ESC/Ch1)
*ESC signal (from receiver)
*ESC power (from battery)
*ESC motor (2 lead for brushed, 3 lead for brushless)
*motor (2 lead or 3 lead)
And with all that - a few cans of magic-smoke on the inside ;)
Such a device would be incredibly complex and possibly very useful. You'd probably need 2 LCD panels (battery->ESC, ESC->motor) and maybe a 3rd for seeing what the receiver is doing with the ESC.
... I think my brain is getting ahead of my hands ... might need to get this one sorted out first :)
tis interesting that when I was racing RC cars using Ni-MH and brushed set-ups, ALL the serious racers has balanced, resistance tested battery packs where each cell was tested. Was considered the norm and those packs were consistently re-tested and when they started to lose efficiency they were out.
We flyers are a relaxed bunch :)
Well, I have to say... I'm a bit dissapointed so far with the community... NONE of you have pointed out the very obvious flaw in my design...
There's no power or status LED on the board... tsk tsk tsk.
Will be fixed in the next revision which will include the following;
* Reverse polarity alarm
* Dual loads
* LED's :D
I like the idea and if you can keep the cost around $60, I suspect they will sell. The primary advantage of your device over a whattmeter is that the user gains visibility into the condition of each cell rather than the whole pack.
A couple of suggestions:
Your symboloy is a bit confusing:
a. O: stands for open circuit voltage. 'Vo' might be better.
b. C: stands for loaded voltage, maybe Vl is better (i.e. C is recognized as capacity in the context of batteries)
c. resistance has the units of ohms so you might replace 'mR' with 'mohms'
In the final product, I suspect you will have to wire the cell balance plug to accept all cells and then cycle through all cells (either by user prompt or automatically), showing the condition of one cell at a time. Moving the cell balance plug is a pain.
In addition, not all cell adapter plugs are wired the same. Some are wired with increasing voltage on pins in a 'linear' fashion, others separate the most-positive tap from the others.And then, not all cell balance adaptor pins have the same spacing and physical size. Really a mess!
A really neat feature might be to have the device automatically choose the pins to use for each cell based on the voltage difference between pins so you don't need an adapter. Cell voltage for each chemistry (Lipos, A123s and NiCD/NiMH) is different so you'll need some way to determine what cell type is being used to make this method work.
If you don't have the device choose pin-pairs automatically, I suspect you will have to sell balance plug adapters (separate from the device to give the purchaser a choice) to make this device more attractive. After all, it will be ease of use that will make this device sell.
|All times are GMT -5. The time now is 01:56 AM.|