Nixie Clock - is it Ghosting? / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Nixie Clock - is it Ghosting?

Move Thread

LAN_403

P25416

Sun Jul 18 2010, 04:57AM

Registered Member #3009 Joined: Sun Jul 18 2010, 04:50AM
Location:
Posts: 4

Just completed a glowingtech clock (article in Nuts and Volts some time ago.)

I will describe the issue. All works well, but I can see the seconds digits in the hours tube. (Least significant digits on both.) The hours tube still has the correct digit lit, but dimmer, you can see the seconds changing.

Is this "ghosting"? Why just this one tube (especially when the 2 minute tubes in between them show no signs of problems.)

I have:

1) Cleaned boards of flux.
2) Tried different dimmer resistor values.
3) Added a large resistor from each anode to gnd.

The changing digits are still visible clearly.

Any ideas and I very much appreciate your help.

Nicko

Sun Jul 18 2010, 05:15AM

Registered Member #1334 Joined: Tue Feb 19 2008, 04:37PM
Location: Nr. London, UK
Posts: 615

It shouldn't be "ghosting" - that's an effect caused mainly by the capacitance of a single tube when the digit changes.

This sounds more like a leaky anode switch in the hours tube - I don't have the schematic for this design, but I suspect its multiplexed rather than direct drive as its based on an ATtiny2313 which hasn't enough pins for direct drive. That means you probably have an anode driver pair for each tube with all the cathodes for each digit commoned together with another NPN connecting to GND. If the anode driver for the hours tube is not switching off fully, and its digits are connected to the digits of the seconds tubes, then it will also show the seconds changing...

Each anode driver will consist of an NPN transistor (MPSA42?) whose base is driven from the cpu I/O pin, and a HV PNP transistor doing the actual switching (MPSA92?) - this is needed at the cpu is outputting 0-5V and the HV rail is at 180VDC.

You should be looking at this pair - it seems they are not switching off fully or fast enough for the hours tube - look at the resistors - the one from the base-emitter of the PNP, and the one between the collector of the NPN and the base of the PNP. Make sure the values are correct and they are soldered correctly - no dry joints.

If that isn't the issue, its possible the one/both of the transistors is/are leaky any needs replacing.

wrote ...
1) Cleaned boards of flux.
2) Tried different dimmer resistor values.
3) Added a large resistor from each anode to gnd.

They are not "dimmer" resistors - they are current limiting resistors - check they are the correct values and don't mess with them - they are chosen for a reason (which depends on the tube, the HV rail voltage and the multiplexing frequency). Adding a resistor from the anode to ground is also a waste of time - that's not the issue with ghosting anyway - it's caused by stray charge on unused cathodes - to correctly drive a nixie in a "rigorous" design, unused cathodes should be held at 1/2 the HV value, typically 90V, rather than being allowed to "float" when unused - most commercial kit designs don't bother with this because it costs more or the designer just doesn't know their stuff, but the better ones do...

As usual, I should point out that the best place for future nixie-related discussions is the Yahoo! NEONIXIE-L group.

EDIT: Found the manual - its what we call a 2x3 multiplexing scheme - in this case there are 2 sets of 3 tubes - tubes 1,3 & 5 are one group ("10s"), and 2,4 & 6 the other ("units"). This gives you 20 cathode drivers (2 x 10 digits) and 3 anode drivers (hours, minutes & seconds), so 23 MPSA42s and 3 MPSA92s.

The area to look in is R25,R28,R31,T23 & T26 (R31 & T26 in particular). The HV rail is at 170VDC, and as I suspected, there is no tie-off for floating cathodes, nor for that matter is the HV PSU (if you can call it that) regulated,

HTH,

Cheers

P25416

Sun Jul 18 2010, 03:21PM

Registered Member #3009 Joined: Sun Jul 18 2010, 04:50AM
Location:
Posts: 4

Nicko,

Incredibly good reply for not having seen the unit. Wow. Yes multiplexed. Yes again on MPSA42 and MPSA92.

The design seems ok, but with all the room on the board, I was surprised how close the pins were laid out to each other on the MPS92s. . Not necessary and tough to see/ troubleshoot.

I quoted an email from the company/ designer for the "dimmer" comment. Clearly this is a 15K 1/4w current limiting resistor with the effect of changing the overall brightness. Their suggestion was to change that value. In his defense, I explained this as ghosting and it appears that is not what the problem is. (Although it doesn't sound like a fix for that either from your description.)

It appears that R25 (33K) feeds the base of T23, the collector R28 (470K) then goes to the base of T26 (MPSA92) which has R31 (100K) to the emitter for the hours. Is this what you are seeing? If I understand correctly, this transistor isn't shutting off fast enough such that I can see a dim tube 6 in tube 2 on top of what is supposed to be there. Have I captured your comments correctly?

(And yes, there is no tie off for floating cathodes based on what I have learned reading here.)

Based on the multiplexing scheme, shouldn't I see the same problem in tube 4 (least significant, minutes)? I don't have any "background" numbers showing there at all.

Again, thanks so much, , you really were able to understand the design quickly.

Nicko

Sun Jul 18 2010, 05:07PM

Registered Member #1334 Joined: Tue Feb 19 2008, 04:37PM
Location: Nr. London, UK
Posts: 615

Actually, I'm slightly puzzled - now I have the schematic I can see that each pair (h,m,s) has a single anode switch, so if one misbehaves, both should, i.e. both the 10s & units. You are seeing just the units act up. This could happen if tube 2's anode floats high, which is why they asked you to tie it to ground, so that when the anode switch turns off, residual charge is not left there, but is taken away to ground, so the tube cannot strike.

What resistor value did they suggest?

By the way, with the values they have chosen for R28 & R31, Veb will be about 30V when the anode is on - Vebo (Veb max) is stated as 5V in the datasheet, i.e. that 100K resistor is too high - it should be around 10K IMHO. With the values as supplied, I wouldn't be surprised if the MPS92s died or were at least pretty sick.

Cheers

P25416

Sun Jul 18 2010, 06:31PM

Registered Member #3009 Joined: Sun Jul 18 2010, 04:50AM
Location:
Posts: 4

Good point. There is some slight visibility of the most significant seconds in the most significant hours tube (1 displaying 5), it is not as apparent possibly because it is not changing as often. The value they indicated to use was to tie the the anode to a 470K resistor to gnd. It didn't appear to do anything. Then the suggestion was to tie both tubes' anode to gnd, again no visible change. I can find no cold solder joints, or shorted connections. The base signal on T21, T22, and T23 all look like clean digital signals on the scope.

Interesting analysis of T24/T25/ and T26. It sounds like you believe these are being overdriven....

(again, much appreciation for looking at this.) I am pretty stumped....

P25416

Sun Jul 18 2010, 06:50PM

Registered Member #3009 Joined: Sun Jul 18 2010, 04:50AM
Location:
Posts: 4

... On further review.

I just spent some real focus time in a dark room staring at the display (yes, my wife is getting concerned...)

Looking carefully it DOES appear that Tube 1 also has tube 5's digits displayed in the background, it is just harder to notice because it only changes once per minute. Interestingly, if I look very carefully at Tube 5 and Tube 6, THEY have Tube 1 and Tube 2's (respectively) numbers displayed in THEIR background. Tube 3 and 4 don't have anything really visible besides their correct display on them, or at least it is not very bright.....the plot thickens....

UltraMagnus

Mon Jul 19 2010, 11:37AM

Registered Member #2875 Joined: Mon May 24 2010, 08:28AM
Location: England
Posts: 42

were nixie tubes ever designed to be multiplexed like this? I know it is common practice with LEDs these days, but I thought nixie's were mostly directly driven.

Nicko

Mon Jul 19 2010, 01:03PM

Registered Member #1334 Joined: Tue Feb 19 2008, 04:37PM
Location: Nr. London, UK
Posts: 615

UltraMagnus wrote ...

were nixie tubes ever designed to be multiplexed like this? I know it is common practice with LEDs these days, but I thought nixie's were mostly directly driven.

Generally, most nixies are more than happy to be multiplexed - The very earliest were sometime a bit slow, but later tubes were even made with "keep alive" electrodes to ensure rapid firing in multiplexed designs - many Panaplex tubes (still made today) have these.

One thing to avoid is multiplexing them at the resonant frequency of the anode or any of the cathodes - that can lead to the tube "singing", which is very annoying!

Cheers

Conundrum

Wed Jul 28 2010, 05:50PM

Registered Member #96 Joined: Thu Feb 09 2006, 05:37PM
Location: CI, Earth
Posts: 4061

interesting...
wonder if this effect could be used to make a neon into an audio oscillator?

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob