> Why not use a 555? It is simple, it is reliable, ... > Here is an interview question I often ask: "Which is more reliable? A > simpler system or a more complex system?" You'd be surprised how many people > answer "The more complex system". > > -Adam
Maybe they like Rube Goldberg. What's the general background of the people asked this question ? I don't see a hammer break too often. Elegant & simple. Nothing wrong with a 555 based supply, as long as its used within its limitations. My 1st 3 nixie clocks used 555 based supplies. I switched latter to MC34063 based units. The MK1.5 or MKII type, not the version 1 type. On my dekatron clocks, however, I use a pair of pins off my uC, just like Will is using. My implementation though is very simplistic. I use the AVR's voltage comparator feature. Algorithm is real simple. In a timer interrupt (20KHz or faster), I toggle the output bit, alternately ON or OFF. During the phase when its suppose to be flipped ON, the comparator is checked to see if the input is over the threshold. If it is, I do not turn ON the output (which drives the FET). Its turned ON only if its below the threshold. This simple scheme works quite well. I've used it in supplies that deliver more than 40mA, at nixie voltages. I also use a push-pull buffer between the uC's output bit, and the FET's gate. This serves two functions. One is level translation. The FET works better if the gate sees a voltage closer to 10V or more. That buffer is also cap coupled to the high side xstr (the one turning it ON). If the uC ever locks up, the gate drive to the FET, will eventually (within a few milliseconds) turn OFF, indifferent to the uC's output level. No smokey FET or coil. Coils have particularly bad stink, when they get toasted. Back to separate supplies. In this group, for self made supplies, there seems to be a hierarchy of performance. 555 types are at the bottom. I've played backwards and forwards with these, but I could never get their efficiencies over 65%. I though I had one upto 75% once, but could not repeat it. Most likely a measurement error. They still can be useful. Keep the circuit as simple as possible. Don't add that extra op-amp, to improve regulation. If all you need is 15mA or less, these are plenty good. Next up on the list are the MC34063 base designs. This is probably the most versatile switching supply chip ever made. You can use it for almost anything. If your 7805 is too hot, wire the MC34063 as a buck converter, and get your +5V that way. For powering nixies, you'll need an external FET. There are 'version one' circuits out there that drive the FET straight off the chip. These designs aren't good for anything over 10mA. If you add a pair of transistors (MKII), or a diode & transistor (MK1.5), the efficiency on these can be very good. I've been making these with efficiencies ~80%, regularly. Sometimes they exhibit stability issues, but these are usually quite easy to fix. I've run a few of these at 45mA (180V out, 12V in), for months, and they still work quite well. The top of the list are the MAX1771 designs. If made properly, they can deliver over 100mA, with efficiencies near 90%. Nick deSmith did an in-depth look at this chip a few years ago, and I believe he has a board layout available to anyone who wants to make it (for personal use). There are also 'off-the-shelf' supplies that can be had by Mr Taylor, or on eBay. Plus other chips have been used to make supplies, which include op-amps, voltage comparators, 4000 CMOS chips, and other converter chips. The 3 above, just seem to be the most common. Of course, all these designs require that you use 'GOOD' parts. Firstly, a good coil. This is key to boosting the voltage. It needs to hold and deliver the energy efficiently enough to get transfer it to the output. That means a high saturation current, and an inductance that doesn't drop too much as the current increases. All coils do this, but some are better than others. Second, a good power FET. It should be able to handle the maximum output voltage with some margin. Its ON resistance should be low. It should be able to pass the peak current that's also going thru the coil. Its gate charge should also be as low as possible, though chips like the MAX1771 can switch them pretty well indifferent to this spec. Your HV rectifier should also switch quickley. Ultrafast rectifiers are usually used. They switch at under 100nS. A common rectifier like a 1N4007 has slow switching characterstics. Often not even published, which means extra crappy, and can not be used for this kind of app. The electrolytics on both the HV and input sides should also be picked carefully. Electrolytics have this spec called ripple current. Back in the days of linear supplies, nobody really looked too carefully at it, since the caps used where usually quite high value, and had a correspondingly high ripple current spec. When switching supplies came along, the capacitance values went down. But ripple current handling usually follows capacitance, so now it became a real everyday concern. Low ESR caps are often specified, since their ripple current spec is usually higher than those of general purpose types. Using several lower value caps in parallel, as oppose to one big one, often also does the trick. I guess someones going to scream now, since I and a prior person, mentioned the 555. Signetics (long since absorbed by Philips) invented these back ~1968 (+/-). Still being used, and 3 of the 12 chips that Radio Shack still sells. -- You received this message because you are subscribed to the Google Groups "neonixie-l" group. To post to this group, send an email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/neonixie-l?hl=en-GB.
