PSAS/ avionics/ av3-battery-board

Overview

The battery board interfaces the Lithium Ion Polymer battery pack to the avionics power system, which is the power system of the rocket. The battery board does safety functions (overvoltage, undervoltage, overcurrent) and measures the battery's state (voltage, current, state of charge, temperature), and does cell balancing.

Here's a block diagram of how this board fits into the rest of the system:

There's a bit of history behind this board:

Requirements

Other requirements

Preliminary Design

Choosing the "maximum current allowed" from the batteries is the most important target to hit here, since this effects the rating and thus expense and avaiablilty of most of the components. We've chosen:

We spent a lot of time looking for a chipset that would do all of this. After a lot of research, we found the TI bq3060 "Smart Battery" chip. It does everything we want, and way, way, way, way more. We should all be very scared of this chip, since it has a RISC core, and comes with flash memory that you can actually upgrade :) So, as long as the complexity of this chip doesn't bite us, it's going to be great. From the data sheet: "The bq3060 provides software 1st level and 2nd level safety protection on overvoltage, undervoltage, overtemperature, and overcharge, as well as hardware-overcurrent in discharge, short circuit in charge and discharge protection." Not too shabby.

Unfortunatately, the evaluation module (EVM) isn't available for this part. But all of the documents, as well as the reference design, are, which is invaluable. We can essentially copy the EVM design, with a few tweaks, and we'll have something usable.

Here's the information on the chip and EVM:

A few notes on the differences between our design and theirs:

Components

F100 main fuse [Littlefuse 0501015.WR, DK F2919CT-ND]

15A fuse in a 1206 package. It's rated at 15 A continuous @ 32 V. It has a nominal R of 0.0025 ohms, and and I^2T rating of 36.100. The data sheets says that it can go for up to 4 hours at its rated 15 A, 10 seconds at 35 A, 1 second at 50 A, and 1 ms at 200 A.

R123 shunt resistor [Vishary WSH2818R0100FEA, DK WSHA-.01CT-ND]

There's no place in the various data sheets that tells you how to pick a resistance for the shunt resistors, but there are clues:

So let's choose 10 mOhm. We could go less, but then we drop voltage gain. We could go higher, but then we power disappation is a problem. 10 mOhm seems like a good engineering comprimise. Now, what power?

Here's a comparison of available > 3 W shunt resistors:

Digi-Key Part Number P (W) Tc (ppm/°C) W L Area W/in^2 5 sec over (A)
WSRB-.01CT-ND 3 75 45 27 0.1215 24.6913580246914 -
WSLI-.01CT-ND 3 50 36 27 0.0972 30.8641975308642 -
WSRC-.01CT-ND 5 75 45 27 0.1215 41.1522633744856 -
CSSH27280.01FRCT-ND 4 25 27 28 0.0756 52.9100529100529 44.7213595499958
KRL11C.010CT-ND 5 50 43 20 0.086 58.1395348837209 ??
FCSL90R010FERCT-ND 4 50 35 18 0.063 63.4920634920635 ??
KRL76C.010CT-ND 3 50 30 15 0.045 66.6666666666667 ??
FCSL76R010FERCT-ND 3 50 30 15 0.045 66.6666666666667 ??
WSHA-.01CT-ND 5 75 28 18 0.0504 99.2063492063492 44.7213595499958
CRA2512-FZ-R010ELFCT-ND 3 75 25 12 0.03 100 38.7298334620742

Direct data download

We choose the WSHA-.01CT-ND (Vishary WSH2818R0100FEA) because it's 5 W but still has a 99 W/in^2 rating, and has 45 A of overrating curent for 5 seconds. The only problem is that the pads are fairly tight together, and there's not a lot of room for getting the Kelvin connections out, but that's OK.

TH100, TH101

Layout

Copper widths

Sunstone uses 0.5 oz/ft^2 Cu on external layers and 1 oz/ft^2 Cu on internal layers. Then they plate the external layers with 0.8 mil Cu, bringing it up another 0.5 oz/ft^2. So we have 1 oz/ft^2 on all layers.

Assuming 2 oz/ft^2 (that is, we can keep the current on two layers at all times), and 300 mm of trace length (connector to ground, B1 to B2, B2 to B3, B3 to B4, and B4 to connector was roughly 270 mm).

When charging, which is our highest current activity, we'll be pumping 8 A through the lines. Most importantly, we don't want that area to heat up very much. So we should burn less than 0.5 W of power during 8A. That's about 0.6% wasted power given a 10V pack, so that seems OK.

Using an on-line calculator, I played around with various trace widths and settled on

TODO