4

The vast majority of existing ancillary protection IC's available (such as Seiko) are not compatible with LiFePO4 chemistry due to the lower operating voltage. And the few vendors selling iron-phosphates often have a datasheet which is unclear about any built-in OVP/UVP circuitry. After contacting mine, the 18650's do not have internal protection, so must have it externally added. The vendor also does not have any solutions to recommend.

There are a metric ton of "BMS" and related cell and pack protectors on eBay, some claiming to work with LiFePO4. Which made me ask what these components were, and what I could use for my LiFePO4's...

Surprisingly, google knows very little about protection IC's for these batteries. After much searching and finding only lithium-ion protectors, I decided to look more closely at these "BMS" boards to try and determine which chips they were using. Most have some FETs on them, and a SOT-6 package labeled "3P20" or similar. I couldn't find any match to this part code, but I was able to get the manufacturer of the MOSFET from it's symbol, Alpha & Omega. So researching this company's products, they do have something like the AOC2806 which has a really low gate-drive voltage. Neat... so perhaps these are being used directly to "switch off" a cell when it's voltage goes below the gate threshold voltage of the FET? So I decided to simulate it. (Right-click and view in new page for higher resolution.)

enter image description here

The red trace is an imaginary "cell" voltage, starting at -1v for half a second, then ramping up to 3.2v, staying there for 2s, then ramping down again. Orange trace is the voltage response into load of the left-most circuit, yellow the 2nd circuit, green the 3rd, and blue the 4th.

The ideal response would be a sharp cut-off at 2.00v to zero, but as can be seen, achieving this is easier said than done. I am also very dubious about the real-world accuracy of the green and blue traces. Even if it did perform like that in reality, it still would not work well, as high-impedance loads would still drain the battery beyond 2.00v (due to the curve at the bottom) down to about 1.6v.

So my question is... does anyone know of a LiFePO4 UVP device? Or can suggest refinements to my simulation, or any other solution whatsoever? The only requirement I have (and it's a doozie), is that the solution must use as little quiescent current as possible. (Another reason for exploring the MOSFET-only route: gate leakage below VgsON should be very very low.)

rdtsc
  • 15,913
  • 4
  • 30
  • 67
  • Are you sure Seiko does not have protection circuits that cut off at 2.0V? Anyway, all you really need is a micro-power reference and a comparator and a PFET. You can also use a voltage monitor IC. You are definitely going to need a precision reference rather than trying to use Vbe or Vgs(th) as a reference. Do you need OVP? – user57037 Oct 29 '15 at 06:51
  • The charging circuitry is off-board due to space limits, and a resettable polyfuse is used on-board for over-current protection. The whole design uses very little current even when active, so the goal of course is longest battery life. (Cannot use LiSOCl2 in this application, and Li-ion have horrible self-discharge.) – rdtsc Oct 29 '15 at 12:51
  • @Fahad asks a somewhat related [question](http://electronics.stackexchange.com/questions/42559/under-voltage-protection-for-lipo-battery?rq=1) which uses a Microchip voltage sensor. Researching, the [TC54](http://www.microchip.com/wwwproducts/Devices.aspx?product=TC54) at 2.06v would work, however it draws 1uA, while the rest of my circuit draws 50nA 99.99% of the time. The SII [S-1009](http://www.sii-ic.com/en/semicon/datasheets/power-management-ic/voltage-detector-reset-ic/s-1009/) has Iq of 270nA, and [MAX16057](https://datasheets.maximintegrated.com/en/ds/MAX16056-MAX16059.pdf) at 125nA... – rdtsc Oct 29 '15 at 13:32
  • 1
    So you are trying/hoping to get way below 1 uA, but still have a protection circuit? If I may ask, are you sure you need this? It is going to make it difficult to find a solution. Have you calculated the estimated life of the battery under typical useage with and without the extra 1uA load? In other words, during the 0.01% of the time that the battery is being drained, what is the current? – user57037 Oct 29 '15 at 16:12
  • I have indeed calculated it using the Microchip [XLP BLE](http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=2680&dDocName=en545243) and my cell. At these low nA ranges, battery life is almost completely dominated by the self-discharge rate. The 0.01% current is around 200mA average. Looking into LiFeS2 cells, those would be nice but no rechargeability. Might have to think about this further. – rdtsc Oct 29 '15 at 17:54
  • 2
    It sounds like the battery is going to last for years between charges. Usually, for things like that, primary cells are acceptable. But I am sure you know your requirements. I am just throwing this out there. – user57037 Oct 29 '15 at 18:40
  • not sure if something like [this](http://www.aliexpress.com/item/1S-2A-lifepo4-BMS-PCM-battery-protection-board-bms-pcm-for-lifepo4-battery-cell-pack/32368923014.html?spm=2114.01010208.3.2.My3Up5&ws_ab_test=searchweb201556_7,searchweb201602_4_10037_10017_10032_10040,searchweb201603_1&btsid=008ee7a6-1c2e-4151-b591-f3420bd3b221) would suit your needs. <5uA current consumption. Several version are available on aliexpress. – jmaturner Jun 22 '16 at 15:37

1 Answers1

1

UV- and OV-limits do vary due to LiFePO4 chemistries (yes there are quite remarkable differences). If not sure what limits you need try by charging/discharging at realistic temperatures and measuring the dU/Ah (steep at beginning an end of SOC). Maybe you could make the (seiko) chips compatible with your (under)voltage level by adopting the measured voltage (between Vdd and Vss) so that the measured voltage is reduced ( that means protection level starts higher as needed with LFPs). Beware of changes of Idd and temperature dependent effects.

WindFritz
  • 69
  • 2