Regarding using the 74HC595s, it's true that they are technically out of spec if too many segments are on, but I believe the failure cause would be whole chip heating rather than the single output transistors, and especially with the PWM dimming going on (the segments are never on 100%, since that would be out of spec for *them* - they're actually rated for 4 volts, not 5), everything stays nice and cool. I haven't had any problems with ground bounce screwing up the logic states.
The cold-filament current could be a real deal, and in fact happens when power is first applied to the clock. I have a start-up routine that sequentially warms up each filament to just below visibly glowing, so that condition only exists for less than a second, and only 100 mSec per segment. After that, the filaments are all kept warm. Actually, the original RCA data sheets recommend doing something like that to limit the current surge of the cold filaments. I have a couple of old clocks that don't have this algorithm, using my precious DTF-104B tubes, and it doesn't seem to have unduly affected them either, though. Regarding the timing of the bit-banging, the particular processor I'm using avoids all of those problems, mostly by virtue of it being rather archaic. Firstly, the port pins in an 8051 are in an area of memory that's bit-addressable, so that glitch when the the entire 8-bit port is written doesn't exist. It's also a couple of orders of magnitude slower enough so that it won't outpace the specs of the '595s. (I did put spacing instructions in when bit-banging some slow SPI peripherals on another project, though.) This old, tiny processor certainly doesn't have any built-in SPI peripheral, though. I'll have to look into the TPIC6b595s - those sound useful... ~~ Mark Moulding On Wednesday, August 17, 2022 at 1:29:36 PM UTC-7 LB wrote: > I’ve had great experience with the TPIC6B595s, I used them to drive a > whole bunch of miniature bayonet style bulbs for a clock using IEE one > plane readouts, no problems with them so far. Probably a little over 120 mA > per IC. > > On Aug 17, 2022, at 8:12 PM, Christian Riise Wagner <c.riis...@gmail.com> > wrote: > > Even running more than 3 segments @20mA is actually out of spec as the > maximum ground and source current for the 74HC595 is 70mA. I found that it > didn't work satisfactory for an IV-19. Too high a voltage drop. Might be > okay for smaller numitrons though. I've ordered some TPIC6C595 instead to > try out instead. > > > > Regarding the software, if the microcontroller has an SPI peripheral (most > do), you can use that to send out the serial data instead of bit banging. > It's how I've done it on all my microcontroller-based clocks so far. Works > on all the 3 different architectures I've tried. It's slightly more > efficient than bit-banging. The performance increase mostly won't matter, > but it does make sense when you're doing cross-fading of the digits where > you might be updating the registers hundreds of times per second. > >> -- > > 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+...@googlegroups.com. > > To view this discussion on the web, visit > https://groups.google.com/d/msgid/neonixie-l/26e03d85-8222-4436-88c7-9db9489081abn%40googlegroups.com > > <https://groups.google.com/d/msgid/neonixie-l/26e03d85-8222-4436-88c7-9db9489081abn%40googlegroups.com?utm_medium=email&utm_source=footer> > . > > -- 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 view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/3aea45ca-5928-4a57-8f2a-3f4b8e8770d3n%40googlegroups.com.
