Oct 012018
 

So, I’ve done the driver board.  This is bigger than I thought it would be, at first I thought it’d just be the LVDS receiver, MOSFET driver, and a few capacitors/resistors, and the connectors.  Ideally I wanted something that could be slipped over the pins of the MOSFETs, allowing the drain and source to be connected to other connections which could take the current.

I had just laid everything out on a 5×3.5cm board, two-layer (so dirt cheap).  Nice and tidy.  For the receiver I ended up using the DS90C402: it was already in Kicad.  All looked good, until I saw this in the datasheet (highlighting mine):

In short, if the cable gets unplugged, the receiver will effectively drive both MOSFETs hard on 100%.  Kaboom!

So, I had to introduce an inverter into the circuit.  A bit more propagation delay, and another component, it’s the biggest part on the board (they don’t make SOIC-8 inverters).  I’ve chosen a 74AHC family part like I did for the driver board so it should have the speed needed.

I’m not sure if this is needed for LVDS, but I’ve added a number of pull-up and pull-down resistors as well as the 100R terminations.  These are underneath.

Likewise, I realised I had omitted doing the same on the controller.  There were some for I²C, but I’ve re-located these to the bottom.  So I’ve made some room for them.  Better to do it now than find out I need them later.

Like the driver board, I’ve documented resistors used for pull-up, pull-down and termination.  I’m not exactly sure which ones are needed or what values they should be, but fixing a silk-screen isn’t a big issue.

The LVDS outputs have resistors too, you can see those near the relevant sockets.  I suspect the answer is they are needed at the receiver, not the driver.

Sep 292018
 

So, I was busy routing a board having come up with a basic schematic.  I wasn’t going to order the board yet, I wanted to just play around with the design, see how compact I could make this.

One thing that was niggling in the back of my mind, was how the traces would cope with the current.  I use 6AWG cable from the solar panels, and 8AWG from the batteries.  How wide should I make the traces? One calculator reckoned I should make them about 7cm wide!  Another option was to use heavier gauge traces, maybe 3oz copper.  A 5cm×5cm 2-layer board would cost a staggering AU$263 for just the PCB!

Okay, so I can work around this by fiddling the solder mask in Kicad and just solder some copper wire along the trace.  Not a show stopper.  I’ll just make wide traces so I know where to lay the wire and have plenty of area to solder it.

I was making the traces as wide as Kicad would let me, but something didn’t seem right.  The inductor, just seemed so, small…

When I did the search on Mouser, their interface allows you to pick a value, then hit the ≥ button to select everything “greater than”.  What I missed, is the option right down the bottom:

The “-” option, better known as “we couldn’t be stuffed looking up what the real value is”, is seen as “greater than everything else”.

A check of the datasheet itself, revealed the truth.

In short, there is no way that little tiddler is going to manage the current I was contemplating throwing at it!

What’s the biggest I can get that will handle that current?  Well if I take the “-” option out of the equation, they suggest this monster .  It’s 10uH instead of 33, so my ripple voltage will increase.  At $837.84, it is also a rare exception to the free shipping over $60 offer.

I might need to go play with some numbers to see what I can get away with.  The good news is that discontinuous output is not a show stopper for a battery charger.  I might have to make do with nanohenries of inductance instead of microhenries.