Re: VFD SmartSprockets
- --- In email@example.com, Quixotic Nixotic <nixotic1@...> wrote:
>Hi John, you were about to get a message about this today. Very briefly, a colleague broke her ankle a week after I spoke about beta testing and I had to go onto 6 day weeks which took up most of my time.
> You talked a while back about beta testing, prototyping your new
> sockets. I didn't hear anything back from you.
> Regards, John S
I thought I only had a small amount of work left to do on the sockets and have spent the first 3 days of this week getting to grips with it, now that my colleague has returned and I had 3 days off.
As you know, feature creep had taken over, and I had to make the decision yesterday to cut away the four letter word functions, and the RTCC, to leave a passive 4 tube smartsocket array.
I was starting to reach the limits of what
I could expect to achieve using the programming language, the particular processor and my ability. What I've learnt is that my programming style seems too inflexible and is not best suited to later modification and I need to make better use of variables and be more determined to make use of general subroutines and avoid spaghetti.(I'm being drawn towards object oriented methods , as well as overuse of the word 'and' !)
Another lesson has been about trying to create a spec for a device, with a clear idea of it's functions *before* writing code. This creativity thing is fabulous, but how are you supposed to know when it is just creativity for creativity's sake !
(I was only supposed to be making an IV-17 version of an existing product !)
A major rewrite could have helped, but, the more I write the more ideas I am coming up with so I felt it was time to stick a stake into the ground
I stripped back the code to bare essential passive socket code, with the latest functions which include single instruction scrolling, programmable delays and better effects than previous versions. This has meant another processor change to an 18f26k20 has been possible. (a forthcoming replacement, the 'k22, is 5v compatible - losing the 3.3V regulator)
The four lettter word/swearbox/clock alarm code can all be moved to a seperate processor now for development of much richer functions, given that the critical display and effect timing demands will be dealt with seperately. The SD card / USB programming feature could also become a possibility now - I was struggling to get that to work with the all-in-one Smartsocket. (So hang on to any 18f26J50's you may have already ordered)
I'ld like to bring I2C sensors into the picture too, as well as text message receipt, using SMS or Simon's new RSS reader. In other words there is plenty of scope for development now that the display overhead has been removed, which is actually the reason I made the sockets in the first place now that I come to think about it !
Now that I've reeled everything back in to a more manageable level all I need to do for the 4 digit IV-17 SS is make use of the eeprom for UDC's (no external eeprom required now) and tweak up a couple of effects. A 6 digit unit is not out of reach now too.
The 40V supply idea I had of using the pic's PWM output is working, but for the cost of the external FET it is possible to buy a dedicated 8 pin SMPS IC with an onboard 60V fet.. The thing which will tip the scales for me and direct my choice is if I can buy a discrete 100V FET capable of switching 1Amp (includes a margin of tolerance) for less than £1.20. If not, it makes better sense to me to use a custom chip with onboard FET running at a very high frequency and not requiring any processor intervention.
I'm only going to focus on the 4 digit Smartsocket for now with the intention of producing a beta version within a few weeks. I am heading to France to work in under 3 weeks and so I want to try to send something to you and Andrew (who has a pretty cool job, am slightly envious), before then, to get your feedback.
- On 3 Jun 2010, at 07:22, fixitsan2 wrote:
> if I can buy a discrete 100V FET capable of switching 1Amp
> (includes a margin of tolerance) for less than £1.20. If not, it
> makes better sense to me to use a custom chip with onboard FET
> running at a very high frequency and not requiring any processor
Let me know your address off-list and I will send you the 3 sample
PICs I have.
As for the FET, Farnell have FDPF14N30 at 31pence while stocks last.
Might this do? I can send one for you to try.
Also maybe BUZ73 at 89 pence.