I had four things on my side.
One was Dave Gault of Kuriwa Observatory, who was also interested in occultations and was keen to solve the same problem. Dave became the Quality Assurance person on the team, which is an ugly task for anyone.
One other was the Atmel MEGA 1280, an absolutely rip-snorting microcontroller with enough SRAM and clock cycles to solve this problem.
The third was the Maxim MAX7456 video titling IC, a chip I had previously used to good effect which made the whole "write to video" so much easier.
The fourth was the very good GPS module I chose by accident - the EM406A from US Global Sat. The battery backup on this unit was defective, and it would forget its NVRAM after about 3 hours of no-power. It also had old firmware, which had been installed before the most recent UT second addition, so when the NVRAM collapsed, the unit would start up one second fast, and after a period of time (a few minutes) it would download the correct GPS-UT offset from the satellites, and output the same second twice as it adjusted itself to correct time.
This initially caused me no end of headaches. It made no sense to have a unit that would sometimes start up with correct time (because the NVRAM was still alive) and sometimes start wrong. But it alerted me to a serious deficiency of GPS, that can't be easily resolved - that GPS units will happily output the wrong time and not tell you; that when they correct their output they don't tell you either, and all that gives the game away is the non-sequential output of time points.
I ended up putting in a good deal of code to test the time stamp integrity every second while the unit was operational, with the idea that it would noisily advertise any failure and we could look at the GPS sentences and try to work out what was going on.
I was highly embarrassed by these failures, but Dave thought it was great. As a seasoned user of the previous VTI device, he was aware that errors did occur, and in the previous device they were not advertised till the end of the session. IOTA-VTI would (by contrast) check everything it could every second, and keep the user informed of issues as they might occur.
About the time I started getting ready to consider production of the device, Dave had a conversation with me about how it should be presented to the occultation observing community. He rightly pointed out that most folk do not build their own devices; and the number of people in the community who would be OK with building and programming a microcontroller was pretty limited. He also advised me of the unfortunate demise of the previous VTI, which had left a lot of people with nowhere to go for spares or new devices.
We felt that the best thing we could do would be to provide IOTA, the governing body for occultation timing, with the source code, design, and a permanent gratuitous licence to make the device. Here in Australia, there aren't a lot of people doing occultations. In the States, there are heaps more people and a lot better infrastructure to make such devices quickly and cheaply. IOTA was happy to accept the licence, and arranged for the device to be built by VideoTimers Inc. in California.
