NiCad Battery Testing

[ADWSystems]

As I previously posted, I'm interested in building a circuit to test NiCad batteries. The testing will be focused cell by cell to weed out or combine like power cells, and junk the bad ones. From the information that I have been able to find (especially since I was able to find several battery tester threads when I didn't search for NiCad), the battery should be discharge at 1/10 mAHr Capacity to 1.0 and never below 0.8. This applies to a single cell. Can this be directly extrpolated to the multi-cell packs, such that one should discharge a 9.6V 8cell pack to 8.0V and never below 6.4V, etc.?

From this thread, https://www.electro-tech-online.com/threads/battery-tester-help-needed-thanks.7232/ I have come up with this circuit and appears to simulate quite well.

 

[3v0]

Given that cells do not always loose capacity at the same rate you need to test each cell.

I have no problem with LM317. But IFF you are using a micro controller maybe do away with the LM317 for discharge.

Use a fixed resistor and the ADC input of the micro controller to determine the current at any given time. This info can be used to determine how much mAh has been discharged.

 

[Sceadwian]

You should raise the lower floor a bit for series and parallel packs to avoid cell inversion as the cells age. So stick with 1.0 volt per cell and every now and then when you fully discharge the pack check each individual cells voltage to make sure they're above the absolute minimum for a single cell, if you start getting to that point you'll want to raise the lower floor even more to 1.1 volts and check the lowest cell voltage on a discharge pack again. You'll lose capacity compared to allowing a deeper discharge but if a cell goes bad you could end up with a fire, or at the very least it'll kill the rest of the cells in the pack.

If you're trying to match random cells that's a little more complicated, generally speaking you wouldn't use random cells in a pack you'd purchase the cells as a pack and for rechargeable cells it's preferable for them to have come off the assembly line in direct sequence to be closely as matched as possible chemistry wise.

I'm not sure what other users would recommend, but fingerprinting a batteries state isn't something I've ever seen before. I would personally discharge them all to the same voltage, charge them singly at a low rate with a fixed current until the battery should be around 80-90% according to the amount of current you've put into it for it's capacity. Let the cells sit until they're room temperature and then use variable load resistor to sweep from 1/10c discharge rate all the way up to a 10C discharge rate in about 20 seconds, hold that 10C discharge for another 5 seconds, you can graph that with a storage scope, ADC (a micro controller would do this easily) Then match up cells that have discharge graphs that are very close to each other. Since you're interested in a pretty narrow voltage range you could use an opamp to spread out the voltage range you're interested in to give the ADC a little more resolution in the range you want.

 

[audioguru]

Does anybody use old Ni-Cads anymore? My son's RC cars used them 18 years ago. I used them for wireless microphones and my beach audio system 20 years ago. Ni-MH cells are used today and my RC airplanes use Li-Po cells.

 

[Sceadwian]

They're still used quiet a bit AG.
Lithium is definitely taking over, but NiCad's still work.

 

[audioguru]

They're still used quiet a bit AG.
Lithium is definitely taking over, but NiCad's still work.

I know. My extremely cheap Chinese solar garden lights (a dollar each) still use very toxic Ni-Cad battery cells but their life is usually only a few months.
I replace them with American Energizer (made in Japan) Ni-MH cells that last for many years.
Does Sanyo still make old Ni-Cad cells for little RC cars?

 

[Sceadwian]

AG, that's because those cheap solar garden lights you got don't have proper charge control for NiCad's, there's nothing wrong with the chemistry, for heavy load applications it's superior to NiMh because the peak current you can get out of a NiCad cell is significantly higher than you can get out of a NiMh. This is why you almost never see NiMH packs in battery powered RC cars. NiCad's are still used in cars because their extra size and weight over a Lithium pack is actually a net benefit because it keeps the center of gravity low just by making a flat pack as close to the skid plate as possible, a lithium car you might actually have to add weight to get enough traction, since NiCad's are cheaper than lithium cells it's a no brainer to use them in some situations. Cadmium being toxic is a non issue as long as the cells are disposed of properly, you'll see the same warning lable on NiMh and Lithium batteries because even if they aren't as toxic as Cadmium there nothing environmentally friendly about the chemistry used in ANY battery.

Sanyo still makes NiCad's and still markets them, because of the applications they fill they still work great. NiMh's are more for consumer electronics because they tend to power handheld devices where energy density isn't critical, Lithium is taking over much of the mobile device market but NiMh's will probably always be used for some device, the chemistry is cheaper than Lithiums.

They do still use them in those mini RC cars, I'm, not sure why, although peak discharge current might have something to do with it. Those little cars are putting those little batteries under relatively speaking major loads. Try getting 10 or 20C discharge rate out of a NiMH, your voltage will drop through the floor. The current good Lithium chemistries aren't as good discharge wise as NiCad's though you can cheat with Lithium's easier and add packs in parallel to increase max current, but pack management is more critical because 2s2p or more complex packs have to be taken care of properly with an intelligent charger. With a lower voltage hobby NiCad pack it's a straight series pack so easier to manage. Depends on if you need that energy density and weight savings that Lithiums provide. If you don't NiCad's are more practical where NiMh's have their place as well. I don't see any of the chemistry going away anytime soon.

Just to give you some perspective you can still buy "Heavy Duty" batteries, which are Zinc Chloride based and was patented in 1886. The reason they're still used is simple... They DIRT cheap and still work, even if their energy density is horrid.

 

[ADWSystems]

Yep. I am resurrecting an old thread, because I am resurrecting an old project.

Finally I am moving to actually prototype the above circuit. I have come across two items that I was wondering if the experts here could validate and/or expand upon.

For clarity, off I'm not using the opamp and fet as pictured, but instead I'm using a MC17332 opamp and a BUZ101S fet.

First, the OP amp output seems to operate in three regions ~2.4V, 2.5-4V and 5V (rail).

As the battery voltage changes from 1.2 to 1.05 volts, I don't see any change in the opamp output. Once the voltage starts to get below 1.05 then the opamp output starts to climb, reaching 4V at around battaery voltage 0.99, then jumps to 4V.

If you have followed me so far, the change in the opamp output once the battery gets close 1V I understand. What I don't understand is why the opamp output stays at 2.4V while the battery voltage is dropping from 1.20 to 1.05. I expected there to be more change in the opamp output during the battery voltage change from 1.20 to 1.05.

What am I missing here?

 

[JimB]

A couple of quick comments,

I cannot find a datasheet for MC17332, is that number correct, or, is my google-foo failing?

The BUZ101S does not seem to start to turn on until the Vgs is of the order of 4 volts.
This may be rather high if your supply is only 5 volts as implied by the schematic in post#1

JimB

 

[ADWSystems]

My bad. That should be MC33172. I was told it was one of the "next gen" 741 opamp. Good general purpose opamp, without the nasties found in the '741s.

The Vgs(th) is 2.1-4. If I understand correctly, the one in my circuit must start around 2.4V and is fully on by 4V.

 

[MikeMl]

...The Vgs(th) is 2.1-4. If I understand correctly, the one in my circuit must start around 2.4V and is fully on by 4V.

To be fully on at 4V, an NFet will have to have a Vgs threshold of <1.5V

 

[Flyback]

Nicad is very robust so you dont need to worry too much about the lower threshold.....yes they will wear out quicker if you over discharge them but nicad is the most resilient to overdischarge.....nicads are also the best when used without cell balancing........this is why nicads are still legal in emergency lighting for fire escapes...because nicad can get abused in circuit and not catch fire. Lithiums are almost exclusively NOT used in emergency lighting cases.......in such cases, when a buildings electricity goes off for ages, the batteries self discharge right down to nothing, and then they have to be brought back into service...so lithiums are a no-no...but nicads no problem.

 

[audioguru]

I have used them. The MC33172 is nothing like an old 741 opamp, instead it is an improved LM358 opamp because it has the same low supply voltage and current but it has no crossover distortion and instead of poor performance above 2kHz it works well to 35kHz. It has the same inputs that do not work if they are within 1.8V from the positive supply and the same output problem that does not go higher than 1.2V less than the positive supply.

You need to look at the datasheet for the BUZ101 Mosfet and almost all other Mosfets. The input threshold voltage is when it barely begins to conduct, only 1mA. With an input as high as 4V then some of them conduct only 1mA. With an input of 2.1V then some of them conduct 1mA. The datasheet shows that all of them are fully turned on when the input voltage is 10V.

 

[ADWSystems]

I have to admit I have not worked with FETs much, mostly BJTs. I understood the max Vgs(th) was the point where the FET was guaranteed to be fully on. Not correct? Where does the 10V come from, on the datasheet?

This also partly leads into my other question of understanding. The circuit drawn hows one cell with a control setpoint of 1 volt (R1-R2 resistor divider). To test an 8 cell battery pack, the setpoint would be raised to 8V and the load raised from 4 to 32 ohms.

If Vload is now 8V, then Vg would need to be 10.1-12 V to achieve a Vgs(th) of 2.1-4V, is that correct? And possibly, based on audioguru's post above, 18V to fully turn on the fet?

 

[audioguru]

I have to admit I have not worked with FETs much, mostly BJTs. I understood the max Vgs(th) was the point where the FET was guaranteed to be fully on. Not correct? Where does the 10V come from, on the datasheet?

The datasheet says that with a 2V to 4V gate-source voltage then the Mosfet current is a minimum of only 1mA. Duh, is that fully turned on?? No, it is the threshold where it is barely on or almost off.
The datasheet also shows the very low on-resistance and high current when the gate-source voltage is 10V.

This also partly leads into my other question of understanding. The circuit drawn hows one cell with a control setpoint of 1 volt (R1-R2 resistor divider). To test an 8 cell battery pack, the setpoint would be raised to 8V and the load raised from 4 to 32 ohms.

If Vload is now 8V, then Vg would need to be 10.1-12 V to achieve a Vgs(th) of 2.1-4V, is that correct? And possibly, based on audioguru's post above, 18V to fully turn on the fet?

Do not even think about Vth (when the Mosfet is almost turned off). For the Mosfet to be fully turned on its gate must be 10V higher than its source. Some Mosfets might be turned on enough when the gate is 7V more than the source.

 

[ADWSystems]

Not that I'm disagreeing with the 7-10V Vgs, but where is the 7V and 10V coming from on the datasheet? Is there any fets that are lower?

 

[audioguru]

Not that I'm disagreeing with the 7-10V Vgs, but where is the 7V and 10V coming from on the datasheet? Is there any fets that are lower?

It is clearly shown on the datasheets. "Logic level" Mosfets designed to be driven by 5V-powered logic circuits work well when the gate-source is only 5V.
Here are some snips from datasheets:

 

[ADWSystems]

Why am I getting the feeling this only works for 1 or 2 cells, and I'm not going to be able to test multi-cell battery packs? For a 2 cell battery pack, I would need a gate voltage of 12.8V (10Vgs+2*1.4V/per cell charged) and for an 8 cell battery pack, I would need a gate voltage close to 22V (10Vgs + 8*1.4V/cell charged).

 

[audioguru]

That is how Mosfets work. If you use bipolar transistors then the supply voltage can be 10V less.

 

[MikeMl]

Why am I getting the feeling this only works for 1 or 2 cells, and I'm not going to be able to test multi-cell battery packs? For a 2 cell battery pack, I would need a gate voltage of 12.8V (10Vgs+2*1.4V/per cell charged) and for an 8 cell battery pack, I would need a gate voltage close to 22V (10Vgs + 8*1.4V/cell charged).

Are you referring to the circuit you posted back in post #1 of this thread. If so, it is a constant-current sink circuit which will work with any number of cells (within reason). Since the NFET is always used as a linear amplifier, none of the comments about NFETs driven as a saturated switch posted by Audio and me apply. We all need to read the TS's post more carefully....