The test-desk. Power supply hooked up to an old gel tray with TBE in it.

I did some testing on the boost converter today.  My naive methodology to learn what to expect was to first RTFM on the IOR site, then read the links they provided in their documentation on boost/buck converters.  Some links I found handy on that page/that I found were:

• A maxim appnote
• A nixie tube PSU*
• A series of videos about boost converters
• Adafruit on boost converters

*this design uses almost exclusively (expensive looking) SMD components, so it is a good reference

Once I read those documents and took a look at the BOM, it looked like I would be able to figure out SMD replacements for most of the components.  This is good, in case I want to make 10-20 of these, since they can be batched in an oven, or even by a PCB assembler instead of done one lead at a time by hand.  The next task, since I have the parts, was to build the power supply with as few parts as possible.  A gel power supply does not need to have low ripple, or be particularly “clean”.  This means that a lot of the decoupling caps, and associated assembly cost, can be left out.

I eventually settled on adding in all of the capacitors except the input capacitors, C2, C3, and C4.  This didn’t work very well, as I suspect it tripped the over-current protection on my 12 V 10A input from a computer power supply.  Unfortunately, I don’t have a very good scope in my room for testing this- I am using a xprotolab on the feedback pins (FB on the max1771) to take my fast measurements, and a ‘harbor fright’ multimeter for my slow higher-voltage readings.  With no input capacitors I could reach about 40V, which is about half way.  Adding in C4 allowed me to easily reach 100 V.

The next step is just to get equivalent SMD parts where I can.  Some things like the MAX1771 chip are cheaper in DIP-8 than SOIC-8, so it might be worth looking into re-flowing DIPs, but I would want to make sure that that did not also impact pick-and-place-ability.  I could look at other controller chips, or simpler types of control, but I really want to get this out fast, and with as few iterations as I can manage- after all, each iteration is a pretty big cost on my end, which reduces the other cool things I can spend time and money on to revamp the gel box.