Very many thanks for the tips - a short blanking interval made no difference but the switch to DC showed that the multiplexing was working well.
I need to look at the setup of my current A/C filament drive. As for the grid/segment voltages, I am currently using 12V - perhaps I need to change that? - Richard On Thursday, 2 April 2026 at 12:47:39 UTC+1 Grahame Marsh wrote: > I've always had success using a LM9022 VFD filament driver. It produces a > square wave AC on two pins at 5V which you connect across filament. So the > filament gets biased to about +2.5V. Go look for the datasheet and you'll > see how easy it is to use. You can also feed the outputs into a Dickson > charge pump to generate 25V for the grids and anodes. > > This chip is now unobtanium but the LM4871 (about £1 from Mouser) is > identical although not listed as a VFD filament supply. > > Hope this helps > > Graham > On 02/04/2026 12:03, David Pye wrote: > > Does a short blanking interval fix the issue? > > On Thu, 2 Apr 2026, 07:20 'Richard Scales' via neonixie-l, < > [email protected]> wrote: > > Many thanks indeed. >> As part of my debugging/research, I tried using 5V DC for the filament >> and got this: >> >> >> >> Here you can see that the brighter digits are at the end nearest GND and >> the less bright digits are at the end which is at 5V. I can understand that >> - at the 5V end there is only 12-5=7V between the filament and the >> grid/segments whilst at the GND end there is a full 12V between them - I am >> guessing that this is entirely relevant. >> >> - Richard >> >> >> On Thursday, 2 April 2026 at 07:06:21 UTC+1 gregebert wrote: >> >>> I think you want to drive the filament from a center-tapped transformer, >>> and ground the center-tap. This minimizes the voltage gradient across the >>> display, basically half of what it would be if single-ended. >>> >>> With your current setup (single-ended), one side is grounded, and the >>> other going positive/negative via AC, you are still setting up a voltage >>> gradient across the display. >>> >>> The other thing to consider is applying either a negative bias onto the >>> grids to turn them off better, OR...if you follow the above suggestion with >>> the center-tap transformer, apply a small positive DC bias to the >>> center-tap. This will effectively make your grids negative relative to the >>> filament (cathode) and improve the cutoff. You will also want to increase >>> the segment voltage roughly by the same amount of the DC bias at the >>> filament transformer center-tap, otherwise the brightness will be reduced. >>> >>> Be careful about driving the grid positive relative to the filament; it >>> will definitely increase the brightness as well as causing current thru the >>> grid. Normally, grid-current is close to zero, because it's intent is to >>> cause electrons to be repelled (blocked) from reaching the anode >>> (segments). The datasheet should have details about this. >>> >>> NIMO tubes behave similarly to VFDs, though the anode voltage is quite a >>> bit higher. >>> >>> BTW, I dont see any ghosting so that's a good sign. >>> On Wednesday, April 1, 2026 at 10:41:09 PM UTC-7 Richard Scales wrote: >>> >>>> >>>> >>>> This is supposed to show 0-9 twice - it is there but mostly masked by >>>> this 'effect' on the left. >>>> - Richard >>>> >>>> On Thursday, 2 April 2026 at 06:35:52 UTC+1 Richard Scales wrote: >>>> >>>>> I acquired a bunch of Itron FG2013A1 VFD displays. These are 16 >>>>> segments (plus DP/comma) 20 character displays. >>>>> >>>>> I wanted to use them as a reason to try and grasp the nettle of >>>>> multiplexing (which has eluded me for some time). I hooked the filament >>>>> up >>>>> to 4.5V and applied 12V to a grid and 12V to some segments - all good so >>>>> far. >>>>> I had a bunch of HV5812 to hand and tried driving the display with >>>>> these to turn on grids and segments - still good. >>>>> >>>>> I made up a board with 2 x HV5812 drivers cascaded to give me a total >>>>> of 18 segment drives and 20 grid drives. All I had to do then was to >>>>> crack >>>>> the multiplexing thing. >>>>> >>>>> After some time and with assistance I got an interrupt routine firing >>>>> every 1mS which would set the segments and grid I wanted to turn on and >>>>> an >>>>> SPI.transfer later - the display showed what I wanted. 1mS x 20 >>>>> characters >>>>> gives a refresh of 50Hz (if my math is right) which looks just fine. >>>>> >>>>> .. however, when I want to show all 20 characters (in my test case 0-9 >>>>> twice) I see some 'interesting' results. >>>>> >>>>> The A/C filament drive is connected at the left hand end of filament, >>>>> the other end to GND. >>>>> >>>>> I am using 12V for VPP to the HV5812 drivers so I expect this to be >>>>> what is going to each grid and segment that I want to activate. >>>>> >>>>> The ISR just sets the segments for the current grid and turns them on >>>>> in one hit (the SPI.Transfer is sending 40 bits - 20 for the grids, 18 >>>>> for >>>>> the segments and two are wasted) >>>>> >>>>> Everything checks out using a logic analyser and there is plenty of >>>>> time left to do other stuff - I am using an ESP-32 from AZ_Delivery for >>>>> the >>>>> testing and am using a CD40109 to convert the 3V3 logic to 5V. >>>>> >>>>> I can see that it really is showing the right segments on each digit - >>>>> it's just that there is a whole bunch more illumination of other unwanted >>>>> segments which is far worse at the A/C supply end of the filament and >>>>> completely gone ant the GND end. >>>>> >>>>> I call upon those more experienced in these things to see if anyone >>>>> can point me in the right direction? >>>>> >>>>> I plan to try a DC filament drive next just to rule that out. >>>>> >>>>> - Richard >>>>> >>>>> -- >> 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, visit >> https://groups.google.com/d/msgid/neonixie-l/323902ee-1ed3-4235-8979-b38e7e63cd2bn%40googlegroups.com >> >> <https://groups.google.com/d/msgid/neonixie-l/323902ee-1ed3-4235-8979-b38e7e63cd2bn%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, visit > https://groups.google.com/d/msgid/neonixie-l/CAOQ6x0GwffP7q6gUZkY5peRvRW47_Ah4oXoaG7TPT%3DPJr9_NwA%40mail.gmail.com > > <https://groups.google.com/d/msgid/neonixie-l/CAOQ6x0GwffP7q6gUZkY5peRvRW47_Ah4oXoaG7TPT%3DPJr9_NwA%40mail.gmail.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, visit https://groups.google.com/d/msgid/neonixie-l/0312b92c-6f8d-46b5-a399-87d74d86aaean%40googlegroups.com.
