Yet another Charger - ESP120

[SpmP]

Good day all,
I am working on a charger based on the ESP120 3kW PSU, in my configuration (28cells) I am using two in series. But there is no reason not to go to eight in series.
The beauty of the ESP120 is that you can get them for $US20 delivered each! Unless you are somewhere far away, in which case shipping is $160...
There has already been really good research done on these by a german crew who have produced an analogue controller, which is probably already very awesome, for eu75.
I cannot post links yet, as I am not a 3l1t3 enough modder, but search hp_3kw_netzteil_als_lader_hp_3kw_netzteil_als_lade r
Its all there including I2C spec!

Anyway, the circuit I propose is very very simple, some of the bits are a tad expensive (60A Schottky for example), but all in all it should be a very easy and cheap build.

So, comments please before I begin ordering 8)

Jackbauer, if applicable, could I start with the firmware from your charger (assuming atmel and C) please. I suck at LCD's, and was going to use serial mostly 8)

I like I2C, so will be mostly I2C, especially as the PSU's can be turned on with I2C and will tell me their current usage 8)

I have seen a few different approaches to charging LiFePO4's, what are the opinion of the forum members here on the best life preserving cycle is. My batteries are Sinopoly, with a miserly datasheet, stating 3.6V charge. The supplier claims 3.65V charge. Recommend to charge at 0.3C(30A for 100Ah cells).

Questions questions 8)

There are many more imagers should anyone want them of the supplies.

Cheers

[JoeG]

Hi SpmP,
I've looked into using PSU's for a higher power charger for my 144V Lifepo4 pack in my Geo Metro Ev. The problem most of the server PSU's have is they don't current limit. If you have a depleted pack the PSU will shut down as it tries to put out more Amps than it is rated for. A discharged pack looks somewhat like a dead short at first. The RSP series meanwells for example are constant current limiting, that is they put out their max rated current and that no more, and could be used, but are not cheap.
Good luck with your charger project,
Joe

[thingstodo]

Originally Posted by SpmP

I am working on a charger based on the ESP120 3kW PSU, in my configuration (28cells) I am using two in series. But there is no reason not to go to eight in series.

It looks to me like the PSU does not allow you, at least via I2C, to set the maximum output current- just as JoeG mentioned. As I read it, you can enable or disable the output - but that's about it. Lots of information read back but minimal control.

So I think a couple of things would need to be changed in your design. The PSUs can still be chained together to get a higher bus voltage. I think an open source DC motor controller circuit would work to limit the current to a set value during the constant current phase of the charge.

The constant voltage phase, where the voltage stays constant and the current drops, is built into the PSU but that's only if the output voltage happens to match the battery pack. I didn't see an I2C adjustment for the output voltage. Is there an adjustment via sense resistors that I missed?

If the pack voltage does not match, it gets more complicated.

I have not looked at Jack's circuit - it may deal with all of this.

[SpmP]

I am sorry for not being clear(er)
The current control is done indirectly by modifying the output voltage of one of the supplies. This is done by applying a voltage (12 -- -0.6V) to the pad of the removed calibration potentiometer.


It turns out I can simplify this all greatly by using the shunt on the pcb through an lm358 (differential input) all with the internal 10bit ADC's on an AVR. 10bit gives enough resolution for this application.
The voltage output 'DAC' can of course be a simple PWM->rc network. Very similar it turns out, to the circuits on the aforementioned german website.

Also note that I am currently using this as the charger, but with a pot in place of the DAC and a cycle analyst as my meter(s)
So on I code...

[thingstodo]

Thanks for the clarification. I guess I should read *ALL* of the spec sheet before commenting.

To get a decent range on discharging your pack, I would have expected you to have to adjust the output on both power supplies. See my notes further below.

Regarding using an Op-amp to amplify millivolt signals, I have had noise problems doing that sort of thing (in the distant past). The commercial applications that I have seen (maybe 10 years ago) appear to use a variant of a wheatstone bridge to do that job. I expect that the instrumentation amplifiers are much better now. Let us know how it works out for you.

My notes - to see if I understand what you are doing now and trying to do

Do I understand?
- you are using a potentiometer in your present charging setup
- you plan to use a pwm signal (DAC) to the calibration input on the power supply to vary the output voltage +/- 5%
- you will stop using the I2C to get information from the power supply and use a differential input on the millivolt shunt for each power supply to determine output current, load sharing, etc

If that is the case, since you are making it work right now,
- you are charging your cells to 102.2V (3.65V each) plus your diode drop
- you've added several diodes so that your power supplies output maximum (108V) and your batteries receive 102.2V
- your charger can drop 10.3V to minimum voltage, making 91.9V at the pack terminals
- DIYelectriccar has a couple of postings listing Sinopoly charge and discharge curves. I don't know if they match your batteries, but using them you can discharge your pack down to about 20% - 3.28V. Sounds about right to me.

[rmay635703]

Although server PSUs do not limit current a PSU for an LED does.

Think of how the honda insight grid charger works, would be easy to ramp up into this application.

Then your charger would have a current limited rate between the upper and lower limits of the LED psu.

Cheers

[P-hack]

just curious on the hp, do you know what the bus voltage is? between the boost and buck sections?

[mcdanlj]

Originally Posted by SpmP

Good day all,
I like I2C, so will be mostly I2C, especially as the PSU's can be turned on with I2C and will tell me their current usage 8)

Have you actually succeeded in I²C commands and sensor readings?

I've tried with both a bus pirate 3.6 and bus pirate 4, and have run into a few issues.

I can't reliably talk to the ESP120 faster than 5KHz, even though the I²C spec for the power supply says that the supported clock range is 10-100KHz. 50KHz sees only the 0x00/0x01 pair and doesn't talk successfully to it, and 100KHz scan completely fails.

Using the 0x01 command to read analog sensors gives me garbage data. And when I tried to use 0x04 to put it into test mode, I was clearly instead writing to EEPROM, because the command sequence I used to try to go into test mode showed up as analog sensor data and survived a input full power cycle of the power supply.

Initially, max output current was 0xAFFFFF; after I tried to write [0xae 0x04 0x04 0x04 0x08] to go into test mode (without success) my max output current was reported as 0x080404 and when I ran the "bogus" command [0xae 0x04 0x05 0x06 0x07] my max output current was reported as 0x070605.

The firmware revision is reported as 0xAF.0xFF which also seems unlikely:
[0xae 0x06[0xaf r:2]
0xAF 0xFF