Inductance can be a real pest; V=Ldi/dt , so switching a bunch of segments on or off at the same time (even if they are lit dimly) can cause problematic voltage-drops across a relatively small inductance. One trick I've done on IC's is to stagger the turn-on and turn-off of current-consuming circuits by a few microseconds; can also be done in software.
On Wednesday, March 29, 2023 at 7:28:42 PM UTC-7 Moses wrote: > I got around to adding a "keep warm" function for unlit segments on my > IV-9 clock kit. My experiences and a short video for those interested. > > I thought this would be a quick and easy modification.. it was anything > but! I added a hardware interrupt at 1600hz, and added code to keep unlit > segments warm with about a 1/10 duty cycle. Worked well.. but seemingly > random reboots of the ESP32 micro drove me nuts, I chased a non-existing > software bug for more then an hour (much more). Later (much later!) I > realized that the load imposed by largish changes in brightness, either on > power up or a commanded brightness change, combined with the keep warm > function was a pretty big load on the power supply. Yes, even higher then > just turning all tubes and segments on all at once. I tried different power > cables, different clocks, half a dozen different power supplies from boat > anchor linears to modern switchers, I was able to trip up all of them. I > even have a very low ESR polymer cap on the clock as well, but nope, the > spike load was just too much. > > My conclusion is the first few moments of the keep warm feature at 1/10 > duty cycle, on 6 tubes and all segments, is a greater load then just > turning on all segments normally. I'll have to hook up some proper > equipment to quantify this soon. I could have brute force fixed this in > hardware by throwing a bunch of caps at it.. but I went with the more > elegant software solution. The brightness is no longer directly controlled, > but ramped up and down (on bootup too) to match the requested brightness, > this fixed the problem entirely. > > With a normal lit segment being driven at 20mA, the 1/10th duty cycle > gives the unlit segments approx 2.5mA current, which is barely visible. I > provided a user adjustable setting for the duty cycle, but having no > official manufacturer guidance to go on I think 2.5mA is as good a default > value as any. > > I hope this saves someone 4 hours of profanity ridden shaking of fists at > a clock. > > Side note, while I was in there.. I added cross fading on the digits, > which I must admit looks better then I had imagined on a numitron clock. > See the video and be well! > > https://youtu.be/dkOwaZvVQrg > > Regards, > -Moses > On Wednesday, March 8, 2023 at 3:08:56 AM UTC-8 gregebert wrote: > >> Very interesting. Others who have multiplexed nixie tubes, which run at a >> lower current, have reported audible noise. I would expect more noise at >> higher currents due to magnetism effects, but that isn't happening here >> with your numitrons. I've heard it myself with incandescent bulbs; I havn't >> done any work with numitorns. >> >> On Tuesday, March 7, 2023 at 6:00:25 PM UTC-3 Moses wrote: >> >>> Gregebert, >>> >>> I've not heard these tubes at all. I was initially running them at >>> 100Hz, then changed to 400Hz for a while before settling at 1600Hz. >>> I've got half a dozen prototypes running for several months, so if they >>> were going to make noise I would have heard them by now. >>> >>> Neat idea on measuring the time-constant, I'll give it a try. >>> >>> Regards, >>> -Moses >>> >>> On Monday, March 6, 2023 at 5:12:55 AM UTC-8 gregebert wrote: >>> >>>> Very good info, Moses. >>>> >>>> From that, I would target the operating current well-below 20mA to >>>> prolong the life of the display. You could use an NPN current-limiter >>>> (driver) driven from 5V TTL logic. Let me know if you need circuit >>>> details, >>>> but it's a very simple design (1 NPN + 1 emitter resistor). As a starting >>>> point, a 270 ohm emitter resistor will limit current to 16mA. >>>> >>>> Do you hear any noise with your PWM running at 1600Hz ? If not, I >>>> would stick with PWM and not use the NPN driver, as it's not necessary. As >>>> long as the PWM cycle-time is shorter than the thermal time-constant of >>>> the >>>> filament (probably a few milliseconds), you should be fine. I think you >>>> can >>>> measure the thermal time-constant of the filament with a phototransistor >>>> and a scope. Starting at a low PWM frequency, the phototransistor will >>>> show >>>> fluctuations in the bulb's intensity. As the frequency is increased, you >>>> should see the amplitude decrease towards zero. I've never actually done >>>> this but I'm pretty sure it will work. You have to use a phototransistor >>>> (or photodiode); a CdS photosensor will have too-slow of a response time. >>>> >>>> On Monday, March 6, 2023 at 2:37:45 AM UTC-3 Moses wrote: >>>> >>>>> The TI CD4511B datasheet lists the lists the *minimum* high level >>>>> output (at 5v VDD) of 4.1v but typical is 4.55v, maybe others are >>>>> different? >>>>> >>>>> https://www.ti.com/lit/ds/symlink/cd4511b.pdf >>>>> >>>>> At 4.55v the segment current on the IV-9 is going to be about 25ma. >>>>> Typical IV-9 segment current is about 20ma I believe, so that would be a >>>>> 25% overdrive. >>>>> >>>>> I've put up a drive voltage vs segment current table on my site, maybe >>>>> it's useful to someone: http://www.neonixie.com/IV-9-6D-RR/#tech >>>>> >>>>> I settled with powering the tubes in my designs in one of two ways.. >>>>> 5.0v direct with a recommended PWM of about 65% (the 165 (out of 255) >>>>> value >>>>> in the table) OR 5v through a regular silicon diode, dropping the voltage >>>>> to about 4.2v at the tube and PWM that to 76%. Both result in a segment >>>>> current of 20ma. Don't know if one or the other is better in regards to >>>>> lifetime, I was not able to find any official sources of information. I >>>>> experimented with driving them using a 3.3v regulator.. but at that >>>>> voltage >>>>> they were a bit dim. >>>>> >>>>> NOTE: The IV-9 datasheet does mention to avoid running the tubes at a >>>>> frequency between 105 and 1000 Hz, presumably to avoid mechanical >>>>> resonance. I elected to run them at 1600 Hz. >>>>> >>>>> I looked through some of my testing data.. at a segment current of 3mA >>>>> the segments are visible, 2mA they are "barely visible" and at 1mA I >>>>> noted >>>>> them as not visible. Keeping the unlit segments warm with a slight >>>>> current >>>>> seems to make the most sense to me. It's all software on my clock so >>>>> maybe >>>>> an option for the next revision. >>>>> >>>>> That's all I know so far. >>>>> >>>>> Regards, >>>>> -Moses >>>>> On Saturday, March 4, 2023 at 5:52:05 AM UTC-8 [email protected] >>>>> wrote: >>>>> >>>>>> Hi Chris, >>>>>> >>>>>> >>>>>> >>>>>> The numitrons are driven by a 4511 decoder, and according to their >>>>>> data sheets the max. output at that voltage is 4.1 V, so that is well >>>>>> within specification for the tube. Yes, I agree, lowering the voltage >>>>>> even >>>>>> further should increase lifespan. >>>>>> >>>>>> >>>>>> >>>>>> Bill v >>>>>> >>>>>> >>>>>> >>>>>> *From:* [email protected] <[email protected]> *On >>>>>> Behalf Of *Chris >>>>>> *Sent:* Saturday, March 04, 2023 8:19 AM >>>>>> *To:* neonixie-l <[email protected]> >>>>>> *Subject:* [neonixie-l] Re: IV-9 numitron lifespan >>>>>> >>>>>> >>>>>> >>>>>> Hi, >>>>>> >>>>>> While i am not sure for the IV9 numitrons have a look at the DA2300 >>>>>> lifetime expectancy, >>>>>> >>>>>> >>>>>> >>>>>> Basically reducing the voltage increases the life expectancy >>>>>> exponentially. >>>>>> >>>>>> So i usually run my numitrons at 4.5V or lower and have had no issues >>>>>> so far. >>>>>> >>>>>> On Friday, March 3, 2023 at 8:05:48 PM UTC+1 [email protected] >>>>>> wrote: >>>>>> >>>>>> I seem to recall there was some discussion on the life expectancy of >>>>>> IV-9 numitrons some time ago. >>>>>> >>>>>> >>>>>> >>>>>> In 2015 I built a number of clocks with these tubes, and they started >>>>>> failing last year. The clock in the picture is on my bench now, the >>>>>> three >>>>>> good tubes were replaced in November last year. So today all 6 tubes >>>>>> will >>>>>> be replaced. I have replaced all 6 tubes on a few other clocks also. One >>>>>> or >>>>>> more segments will no longer light up, and I do not see any blackening >>>>>> of >>>>>> the glass tube (The black you see in the picture is my permanent marker, >>>>>> s >>>>>> I will not replace the wrong tube when I disconnect the power). >>>>>> >>>>>> >>>>>> >>>>>> The tubes are driven directly from a 4511 chip on a 5V power line. >>>>>> >>>>>> >>>>>> >>>>>> There does not seem to be any order in which they fail, so I do not >>>>>> have the feeling that any of the tubes are significantly impacted by >>>>>> thermal stresses from flashing on and off. I would say based on my >>>>>> experience with them, the expected lifespan is about 7 years of >>>>>> continues >>>>>> use. >>>>>> >>>>>> >>>>>> >>>>>> Bill v >>>>>> >>>>>> -- >>>>>> 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 [email protected]. >>>>>> To view this discussion on the web, visit >>>>>> https://groups.google.com/d/msgid/neonixie-l/81b5b7e6-d476-4b83-9a8e-c9d6cc468db9n%40googlegroups.com >>>>>> >>>>>> <https://groups.google.com/d/msgid/neonixie-l/81b5b7e6-d476-4b83-9a8e-c9d6cc468db9n%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 [email protected]. To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/128fb4cf-5156-4cf2-aa40-0c43b76f6e47n%40googlegroups.com.
