300W Adjustable / constant current discharger (again?)

[T i m]

Hi all,

First post here so please be gentle with me. ;-)

I have been playing with electronics / electro-mechanical 'stuff' since I was a kid but I'm no designer. I spent 5 years with British Telecom in their electronics repair workshop, a year with Kodak on microfiche and film gear and another 15 years with a local datacomms Co installing and (field / bench) repairing their own products.

I also built and ran a mobile disco for a few years (for fun), where the building included building all the speaker cabinets, amplifiers, and light sequencers etc (that was my real interest). When I say 'building' I really mean 'assembling from a plan' or (ideally), a kit of parts. I also co-designed and completely built some of this stuff with a mate (I provided the requirement, he (mostly) provided the design and I built it). Why did I prefer building rather than just buying ready made? Well, partly price and availability (in those days) and partly with a d-i-y solution you also got the circuit diagrams so self repair was also possible. ;-)

Anyway, along with the thousands of things I'm supposed to be doing on my ever growing 'To Do' list, I have a couple of projects on the go, the first of which I'l mention here. It's all very close to this thread:

https://www.electro-tech-online.com/threads/automated-load-tester.125069/page-5

Basically, it's to have the ability to log the voltage on a 12V lead acid battery whilst discharging it down to a specific voltage and logging it via a USB interfaced DMM. So as to not to have to also log the current (I can do so via a shunt and second logging DMM, which could end up easier for straight capacity tests?) I'd like the load to provide a constant current.

Now, I think I have the circuit that would allow me to automatically monitor the voltage, start the process with a pushbutton and permenantly disconnect when said threshold is reached. The bit I don't have (in a form I can cope with anyway) is the (max) 25A @ 12V constant current load.

Now, when I was racing RC electric cars years ago I built (again, with the help of said mate) a constant current charger with the range of 0-5A using an 2N3055 to do the work. After loads and loads of Googling it looks like MOSFETS are now often used for such things but they seem to come with their own issues? The requirement to be adjustable is down to the desire to test all sorts of 12V LA batteries from small motorcycle / personal UPS to car / leisure batteries.

So, could anyone be so kind as to point me towards either a kit that would give me what I require, or some pretty concise circuit diagrams / component lists that I could just assemble please?

I'm happy with all the heatsinking / cooling. ATM I have tried using a 2R2 / 100W wirewound resistor clamped to the bottom of a PC CPU heatsink and fan and that works fine (just runs warm) but of course, isn't constant current. I have also tested it with 2 'loads' and it is also ok, it just gets quite a bit hotter. ;-)

**broken link removed**

I was thinking of having the whole thing enabled via a h/d relay, such that when it 'trips' there is no load on the battery under test at all (so it could be left unattended).

All the best, sorry for the long (initial) post and thanks for your time in any case. ;-)

T i m

 

[JimB]

sorry for the long (initial) post

Welcome to ETO.
There is no need to apologise for the long post, it gives lots of information about your experience, what you are trying to do, and your problem is.
If only a few more posters here would give a good description, helping them would be much easier.

Having said that, I am going to disappear and cook myself some lunch!
Hopefully someone will be along soon to give some technical info.

JimB

 

[T i m]

Welcome to ETO.

Thanks Jim. ;-)

There is no need to apologise for the long post, it gives lots of information about your experience, what you are trying to do, and your problem is.
If only a few more posters here would give a good description, helping them would be much easier.

Ok and thanks for that. I'm new here so didn't like to assume.

Having said that, I am going to disappear and cook myself some lunch!

Enjoy!

Hopefully someone will be along soon to give some technical info.

JimB

I hope so too (but not too technical <g>), whilst I do like to have an idea what's going on, I'm a bit too old to learn too many new tricks and so it's solutions I'm really looking for. I don't mind getting my 30 year old, 50W Weller soldering station out though and joining bits together. ;-)

All the best, T i m

 

[T i m]

I was just thinking (it was ok, I was sitting down). ;-)

Could I 'split' the discharge load between some passive stuff and some active / constant current stuff?

If we were to assume say a starting voltage of 12.00 (and a fuly discharged voltage of 10.5) and the 'passive' bit to handle say 20A , if we used a load resistor of R=V/I = 12/20 = .6 ohms, when the voltage is down to 10.5 the current will drop to I=V/R = 10.5/0.6 = 17.5A, leaving only 7.5A for the 'automatic' bit? I think I have seen designs for such out there and potentially that would be easier to build? If the load resistor was also the shunt resistor for any feedback, could that then provide a 17.5 - 27.5A semi-variable constant-current load for 12V batteries? What if we were to 'switch out' some of the passive load (if it was made of say 4 x 2R2 in parallel). Could we then get what is required, eg, a 0-25A cc variable load?

I'd need to draw it out and do some rough sums but I'm not sure that would help *me* much even if I did. ;-(

Cheers, T i m

 

[JimB]

Enjoy!

I did, thank you.
Rump steak with a mushroom and onion gravy, chips and some mixed vegetables.
Jamie Oliver could probably do it better, but he cannot do electronics!

So, to your constant current problem.
There are several ways to approach it, either one big in your face adjustable 0 to 25 amp thing.
Or several lower current "modules" that could be switched into circuit as required.
Your idea of a fixed resistor and a constant current circuit in parallel is OK, but adds complication in that you need to measure the current in the resistor (which varies as the battery discharges) and compensate for that in the adjustable constant current circuit.

I thought that by using five separate constant current modules, nominally identical in design, built for 1, 2, 5, 8 and 10 amps, you have something that is switchable from 1 to 26 amps.
This could be simplified by making the 2 amp module adjustable from 1 to 2 amp and not building the 1 amp module.

What do you think?
How accurately do you want to set the current?
How much money do you want to spend on it?

JimB

 

[Nigel Goodwin]

Why 'constant current ' at all? - loads in practice will never be constant current, so testing with a simple resistive load is simpler and makes more sense.

Going back a fair number of years now I did discharge tests on different AA batteries, and considered doing constant current - but in the end opted for the more common load of a simple torch bulb.

As you're wanting to do non-destructive testing you just need a comparator and BIG relay to disconnect the loads when the battery reaches your require lower limit.

You could easily have different switched loads to provide higher or lower currents, bearing in mind they will have to dissipate LOT of heat - so think safety to prevent burns. Electric fire elements could be used for a nice easy solution, unwinding the wire to get lower values if required.

 

[T i m]

I did, thank you.
Rump steak with a mushroom and onion gravy, chips and some mixed vegetables.

Hmmmm, g r a v y ... <drool>

Jamie Oliver could probably do it better, but he cannot do electronics!

Hey, if you enjoyed it then how much better could it get. ;-)

So, to your constant current problem.
There are several ways to approach it, either one big in your face adjustable 0 to 25 amp thing.

Hehe.

Or several lower current "modules" that could be switched into circuit as required.

Ok, that sounds like it could be a plan.

Your idea of a fixed resistor and a constant current circuit in parallel is OK, but adds complication in that you need to measure the current in the resistor (which varies as the battery discharges) and compensate for that in the adjustable constant current circuit.

Yup, that was my idea, the common load resistor doubling up as a straight 'load resistor' and also being the source of the feedback for the constant current bit or even splitting the load into two series sections (forming a potential devider) and taking the feedback of of (half of) that (although I'm not sure what advantage that latter bit would be). ;-(

I thought that by using five separate constant current modules, nominally identical in design, built for 1, 2, 5, 8 and 10 amps, you have something that is switchable from 1 to 26 amps.

Ah, now there is and idea. I'm guessing the 'fixed' / constant current module would be simpler by design than an adjustable one?

This could be simplified by making the 2 amp module adjustable from 1 to 2 amp and not building the 1 amp module.

Well, that would be one less module and still give me some fine tuning (by a couple of amps) and the oppertunity of adaptin the rigs use it other / smaller batteries (like Nicad packs etc).

What do you think?

I think that's a jolly good idea Jim. ;-)

How accurately do you want to set the current?

I don't really, as long as it is reasonably constant and outside maybe the 25A being fairly close for 'Reserve capacity' tests ... but I'm not in any lab or testing house so 'closeish' would be more than good enough. ;-)

How much money do you want to spend on it?

Well, 'as little as reasonable' really as this isn't a 'need' as much as a 'want', however, if say there were some ready made PCB's or kits available at reasonable prices that save me some time then I'd be happy to pay a bit?

Cheers and thanks again ... ;-)

T i m

 

[KeepItSimpleStupid]

Hey, why not a bunch of relatively matched transistors in parallel (e.g. the vernerable 2N3055 or something else ), a bunch of emitter resistors and a hall effect (HE) sensor from here https://www.pololu.com/product/2453 and a good ole OP amp and reference to tie things together. OK maybe a few other parts.

The NICE part of the HE sensor is that the sensor is isolated.

 

[crutschow]

Below is a CC circuit that uses power resistors in the N-MOSFET source to absorb much of the power. The input voltage to the op amp will vary the current with the maximum being 5V (which gives 10V across the resistors due to the op amp gain of 2). The current is thus equal to twice the input voltage (2 x Vcntl) divided by the parallel value of the source resistors.

The simulation shows the maximum power dissipated in the MOSFET is about 80W at 12A with a value of 54W at the maximum current of 20A (2nd graph). To reduce the peak MOSFET power dissipation you can switch out some of the power resistors at the lower currents with the reduction proportional to the number removed. Thus two resistors will give a peak MOSFET power of 40W and a maximum current of 10A.

The resistors will dissipate a maximum of 50W each so should probably be rated for a least 75W.

The N-MOSFET should be a logic-level type that can handle at least 30A and obviously needs to be well heat sunk to dissipate the power. If you use a standard (non logic-level) MOSFET then you will need to increase the op amp supply voltage to about 20v.

 

[T i m]

Why 'constant current ' at all? - loads in practice will never be constant current, so testing with a simple resistive load is simpler and makes more sense.

Hi Nigel. I'm not sure what you are including in your 'loads' here? I will only be testing batteries with this particular one?

Going back a fair number of years now I did discharge tests on different AA batteries, and considered doing constant current - but in the end opted for the more common load of a simple torch bulb.

Yes, I've seen lamps being suggested as loads as I think they offer a form of current control but that means having quite a few 12V or an inverter and 240V lamps on whilst the discharge process is running. I could do that of course but would prefer something dedicated to the role and more durable etc.

As you're wanting to do non-destructive testing you just need a comparator and BIG relay to disconnect the loads when the battery reaches your require lower limit.

Yup, I have that side covered thanks. As an off_the_shef solution I've used a 12V / 10A commercial Low voltage battery protector, the sort that you might plug a Peltier cooler into in your car, if your outlet isn't switched with the ignition. That disconnects at 11.2V but (automatically) reconnects at 11.5, however I could use that for the sensing and to drive another realay nearer the load that could then be self holding. That could be replaced with the twin cutoff voltage and non self resetting I'm building (with help of a designer friend). ;-)

You could easily have different switched loads to provide higher or lower currents, bearing in mind they will have to dissipate LOT of heat - so think safety to prevent burns.

Yup, I have aready got the switches lined up and a 50A shunt resistor to feed my (Optically isolated) datalogging DMM. I've already mocked up a cooling solution and with one load resistor it only runs warm and even with two it''s only noticably 'hot' (but still not too hot to hold). A mate gave me a bunch of new heatsinks and fans and I was considering fitting them and the resistors (with thermal compound) to a reasonably substantial ally baseplate / sink, pooling all the heatsinks (and fans if required) between all the resistors.

**broken link removed**

Electric fire elements could be used for a nice easy solution, unwinding the wire to get lower values if required.

Yup. I needed a 'soft start' switch when I designed, built and raced an electric 'motorcycle' some years ago. Being as it was endurance race, wasting energy was only a real issue as you first pulled away in 1st gear and even over that short period, many strands of 'Eureka wire' wound in parallel were glowing dull red. ;-)

As I have two datalogging DMMs, even if I am drawing more current than the DMM could theoretically carry (~10A) I have calibrated it with a trim-pot to read 'Amps' when on the DMM millivolt range. As the resultant Volts and virtual Amps spreadsheets would be timestamped, they could be merged and the power drawn over time then calculated.

I'm still now sure how I would be able to get an accurate measure of a batteries 'Reserve Capacity' without a 25A constant current load?

Cheers, T i m

 

[T i m]

Hey, why not a bunch of relatively matched transistors in parallel (e.g. the vernerable 2N3055 or something else )

Well, I could, just most of the solutions offered 'today' seem to use MOSFETS?

a bunch of emitter resistors and a hall effect (HE) sensor from here https://www.pololu.com/product/2453 and a good ole OP amp and reference to tie things together. OK maybe a few other parts.

The NICE part of the HE sensor is that the sensor is isolated.

That's a neat device, cheers. As it happens I aready have a 50A shunt and my datalogging DMM is optically isolated but that does open up possibilities for other solutions.

Lots of interesting ideas guys, thanks. ;-)

Cheers, T i m

 

[KeepItSimpleStupid]

MOSFETS are a bit easier to parallel. I'm old school. 2n3055's ar probably < $1.00 each. In general, your concerned about looses and in this case your not, but you might need a significant amount of base current/

 

[crutschow]

I noticed my circuit is designed for a maximum of 20A instead of the desired 25A. For that just add another 2Ω in parallel . That will increase the maximum MOSFET power dissipation to 100W and 68W at the maximum current.

 

[pilko]

@ crutschow ---- nice solution.
Can you run the simm. with a Vbatt dropping from 12.7V to 7V?

Thanks

 

[T i m]

MOSFETS are a bit easier to parallel. I'm old school. 2n3055's ar probably < $1.00 each. In general, your concerned about looses and in this case your not, but you might need a significant amount of base current/

All the merrier (as you say, in this case), as long as it can be included in the overall load count. ;-)

Cheers, T i m

 

[T i m]

I noticed my circuit is designed for a maximum of 20A instead of the desired 25A. For that just add another 2Ω in parallel . That will increase the maximum MOSFET power dissipation to 100W and 68W at the maximum current.

Thanks for that (and the update). I must admit I don't fully understand the circuit diagram when presented like that and for my simple head. ;-(

Also, it seems the MOSFET mentioned is a surface mount device, but I may be looking at the incorrect variant?

http://www.irf.com/part/_/A~IRLHM620#tab-tab1

Cheers, T i m

 

[geko]

I'm going to jump in here with a question.

I'd been helping Tim outside of the forum with another project and he found this design for a CC load.

https://ianjohnston.com/images/stories/IanJ/DummyLoad2/schematic.JPG

When I looked at it I can't see how it balances the load between the two MOSFETs. I know that the specific MOSFETs shown are designed for linear power operation with extended FBSOA (and very expensive). However without individual ballast resistors in the source circuit of each MOSFET and the gates connected together, if one MOSFET carries more current, it gets hotter, its Rds goes up, so the current drops, but the constant current feedback drives the gates harder to try and bring the current back up.

I'd be interested to know if I've got the operation correct and it doesn't really balance the load or I've misunderstood how it is working. The guy who designed it is clearly very good based on the other projects on his website so I'm thinking I've not understood it.

It is outside my knowledge and is one of the reason I suggested Tim join the forum.

Pete

 

[Nigel Goodwin]

When I looked at it I can't see how it balances the load between the two MOSFETs.

As long as you get the correct type of MOSFET's they are self balancing - they have a positive temperature coefficient instead of a negative one, so don't require balancing resistors.

 

[Nigel Goodwin]

I'm still now sure how I would be able to get an accurate measure of a batteries 'Reserve Capacity' without a 25A constant current load?

I'm not sure how you could with a constant current load? - as it bears no relation to reality.

The only way to do it accurately is to do it with the exact load that it's going to be used with, as it will vary considerably depending on the exact load conditions.

 

[geko]

As long as you get the correct type of MOSFET's they are self balancing - they have a positive temperature coefficient instead of a negative one, so don't require balancing resistors.

I understand how that would work with a fixed gate voltage, but as configured in the previously linked schematic, when the current drops due to the PTC effect the feedback circuit increases the gate voltage so both device see the same increase in Vgs. Also in the specific example they're bolted to the same heatsink so they're tracking thermally too. I don't see how in the posted schematic it balances the load current.

Even this appnote from the IXYS shows ballast resistors with those devices.
https://www.ixys.com/Documents/AppNotes/IXAN0068.pdf