Discussion in 'Hobby Engineering' started by rainwulf, Jul 15, 2018.
Yep, I found the same issues. And they still wont handle discharge.
Where the cap switching would. Mmm.
Just for reference, I've tracked down 5mOhm P Channels MOSFETS and I'd need 4 per cell (back to back pairs to block their body diode). Similarly I can get high current 1F supercaps with 0.2ohms resistance, 10 in parallel would be 20mOhms. Total loop resistance of 40mOhms and total capacity of 10F gives me my 0.4 second time constant. Based on worst case 0.1V difference, peak current 2.5A, after 1 second it would drop to 82mA and 92% charge transfer. Measuring the cell voltage at the end of the cycle would be reasonable, the 80mA wouldn't cause too much disturbance.
Keeping things simple and eliminating special drivers, the gates for the upper cell MOSFET can be pulled down to mid-pack to avoid excessive gate voltage, eg cell 12 MOSFET gate pulled to cell 6, cell 11 MOSFET gate pulled to cell 5, cell 10 MOSFET gate pulled to cell 4 etc. Then the complementary arrangement for N MOSFETs on the lower cells. I recon this is doable!
Its that pesky gate drive that concerns me.
What about using pulse transformers for isolation?
Nah I dont recon it needs it and they arent cheap either.
The cell terminals give a convenient set of multi-voltage tap points. Top cell might be 26C and mid cell around 13V.
A pull up resistor on the upper P Channel MOSFET keeps it off. To switch it on, an optocoupler or pnp transistor pulls the gate down to mid pack 13V. Over the range of charge/discharge voltages you'll still get 9 to 13 V of gate drive which is sufficient. I know opto's arent super fast but the switching speeds here are low anyway.
I'll try to draw the circuit out this weekend, I think it would be reasonably elegant.
I have been more busy. Arduino mega hosted with ebay ethernet, rtc, and OLED.
Initial circuit schematic with just the upper and lower cell balancers. The rest are just repeats of those.
Click to view full size!
To keep things practical, I think I can make each balancer module on a smaller daughterboard with 3 input connections (cell+, cell-, mid-pack), 2 outputs (cap+, cap-) and 2 signals (opto+, opto-). Could possibly then solder these vertically onto a main board.
Hard part now is just component selection, there are some gate drive optocouplers not far off the price of regular ones.
i would buy that.
I have only had a moment to look at this but I am pretty sure it won't work as intended.
In the depicted arrangement, if we take for example the lower MOSFET's when the coupler is on, the right MOSFET will be off, and the left MOSFET will be on, meaning if the flying capacitor is flown to a higher potential (assuming the other switches actually worked correctly), a current would conduct through the body diode of one MOSFET and straight through the other. In the other case - the optocoupler is off - a voltage large enough will cause current to flow through the body diode of one MOSFET and switch on the other in either direction. This just indicates that you have the MOSFET's paired the wrong way around, however even if they were the other away around you couldn't switch on the SSR as intended (similar thing as the first statement will happen) with the depicted drive method.
Rod Elliot of ESP has a good overview article of MOSFET based SSR's and ways to drive them. In summary, the traditional same gender MOSFET SSR relies on a floating charge relative to their sources in order to function correctly.
I did a fair bit of simulation on the upper P-Channel MOSFETS and they worked fine, not so much on the lower ones. I'll go back and review.
P.S you are dead right. I thought I had the sources tied together on the N Channel MOSFETS but I made a silly assumption based on the diode direction. I'll correct it and do some proper testing/simulation.
Also, just a further update. As much as I'd like to fuse these balancing boards, the fuse resistance (20 - 60mOhm) is as much or greater than the MOSFETS and supercapacitors and would limit balance current. Considering running "commando" on these - if something shorted, the current from the batteries would pop the semiconductors pretty quick anyway.
Would just run a fusible link, a specific part of the PCB that is a bit narrower then the rest.
Not a bad idea. First revision of the upper module PCB layout in the image, there's space for a fusable link if need be. Single layer board with a mix of surface mount and through-hole parts.
Currently 26mm wide and designed to site side by side, with some wires daisy chained between them. 12 of them will take up quite a bit of space!
Click to view full size!
Impressive!!! Throw on a TL431 and a Schmidt trigger to indicate cell voltage at a glance?
All of this tech talk is meaningless to me, but as a (non DIY) consumer, I'd certainly be interested in a 24v, 3.75A (90w continuous) alternative to a RESMED POWER STATION II. These things retail, with charger, at about 600 bucks, and provide about 13 hrs of use at max drain (probably more than double that for me as I don't use the heating element so its basically just an air pump), with a 4 hr recharge time.
It seems to me that these things will provide a much longer usage time, and maybe cost less, once you get your design sorted.
If you get the details sorted, sign me up !! I'm not bothered by physical size, but a properly packaged, safe and sorted system into which I can simply plug in some kind of charger, and either sort my own connectors or have them provided, is something I'd definitely look at buying.
I'm also interested in a 12V solution as a car stereo power supply, to be set up on a dual battery system and charged via a car alternator.
Again, if you get it sorted, I'm a customer .
Problem is each LTO cell is 1.4 odd kg Thats the 40ah ones.
If you need 12 volts, thats 5 cells, so not exactly light, but thats real 40ah. What ah is that power station device?
rainwulf he's talking about a power supply for an (expensive) medical device... Just thinking you would want to be super careful...
I know, i gathered that from the data sheet. Its why more information is required. That thing is pretty narrow, but not sure of the actual dimensions, and its battery capacity.
Ok talked to Jazper about those devices. LTO really wouldn't suit you i dont think.
LTO's strengths are their amazing charge and discharge currents, and their longevity/cycle life. They aren't as power dense as lipo, and they are relatively heavy.
So they don't suit the Resmed, that's a shame. Those Resmed things are 600 bucks each, provide a max of 3.75A, and can be drained by the machine in 12-13 hrs.
How about the car battery version? Your's was for a motorbike was it not? How close are you to perfecting a marketable car battery, specifically a deep cycle, high demand second battery for my car stereo? Any guesses as to how much a finished product would be?
90W for 13 hours is nearly 1.2kWh, to get that for $600 retail is an absolute bargain and is cheaper than these LTO's. Im guessing that must be a medicare subsidised price?