I ran the temperature test with 120°C hot air aimed directly at the JFET. Conclusion - the higher the temperature of JFET, the lower the current - so it is self stopping rather than increasing current, which is good for tubes (slight undercurrenting shouldn't damage them like overcurrenting would). Because of nature of the circuit (10k resistor + JFET limiter + Z567M as load), lower current means higher voltage drop on JFET, so power dissipation increases a bit, because voltage drop on the resistor decreases. I'm looking for a transistor with similar parameters as BF256B, but with higher breakdown voltage (at least 50V), so I could run the test without additional resistor. If I find one and it will pass the tests, I think that it would be it - a small circuit limiting anode current, allowing to conserve power by using HV just above striking voltage, while still having current under control. Next step would be some kind of a constant current step up converter.
If this is too unstable due to differences in Vg curves between transistors, then only an active current sensing and driving circuit is needed. I have an idea to use an isolated 5V supply, which would allow to use some simple circuits near HV voltage - an opamp amplifying voltage on current sensing resistor and driving a transistor, all connected somehow to HV. I'm afraid it is only a loose idea - I am not willing to test it myself, for me the JFET circuit is good enough, if I ever decide to use current limiting instead of good old resistors. W dniu sobota, 8 kwietnia 2017 14:52:13 UTC+2 użytkownik Tomasz Kowalczyk napisał: > > I'm afraid all active current limiters will change with temperature, just > some will be affected more and some less. But small changes (0,1mA or less) > won't be critical in a nixie clock, I think. So if a circuit will be stable > enough to provide desired current +-0,1mA in temperature range of 0°C to > about 70°C, with supply voltage changing 20V max - I'd call it good enough. > About variations of Vgs curves - within one production batch of > transistors differences are usually minimal. So after ordering a batch of > transistors you have to check just one of them and you can safely assume > that all other will behave almost identically. Vbe differences in current > mirrors are much more critical, as the difference gets amplified. > > If those two factors are still too big, then you have to use a circuit > using a current sensing resistor, an amplifier of Rsense voltage drop and a > drive transistor. Multiply that by 15 segments and by number of tubes and > you end up with a large and costly circuit. I think that that level of > current stability is not needed here. > > I'll test J-FET current limitter with a hot air gun - how it behaves in > temperature range of 20-150°C, both with 5V and 30V dropping on it. > If I find some time, I'll do the same with a very similar circuit using > LM317 (similar - both use only one resistor and one cheap active component). > > W dniu sobota, 8 kwietnia 2017 05:02:25 UTC+2 użytkownik gregebert napisał: >> >> My concern with current limiters that rely heavily upon the datasheet >> specs (Vgs for Depletion-mode regulator; Vbe for current-mirror) is that >> variations due to process & temperature will have significant impact on the >> actual current. Using a slightly more complex+costly design will mitigate >> this; well-worth it in my opinion when you consider the value of the tubes >> you are protecting. >> >> >> -- You received this message because you are subscribed to the Google Groups "neonixie-l" group. To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+unsubscr...@googlegroups.com. To post to this group, send an email to neonixie-l@googlegroups.com. To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/88eb74eb-45ca-465c-97f6-e991df2c1238%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.