Charlieplexing and Ghosting

[threeneurons]

Well I'll go to the end, with the solution ... Clean your Boards !!!!

Back to the beginning. I'd been selling some multplexed nixie kits for close to a year and a half, now. It uses a unique form of multiplexing called "Charlieplexing", invented at Microchip. There it was used for driving LEDs. A clever member of this forum (when it was back at Yahoo), named Jason Harper saw that it could also be used to drive the base-emitter junctions of transistors, back ~2003. For our pleasure, base-emitter junctions, of HV transistors. In short, charlieplexing nixie tubes. That's what I incorporated in my kit.

One important issue with charlieplexing, is that it can only be done with uC tri-state I/O bits. It needs the I/O bit to not just output a one (1) or zero (0), but also present a high-impedance (Hi-Z). With charlieplexing, only one "node" (p-n, or base-emitter junction) of its matrix can be active at any time. One output issuing a "0", while another issuing a "1", with the rest "inactive" and presenting their Hi-Z state (DDR bit set to "input"). This becomes a key to the problem later.

Multiplexing, in general, and not just charlieplexing, can exhibit a problem called "ghosting". This is when more than one numeral appears to be ON at any time. Usually, the intended numeral is on brightly, while a second numeral is only slightly glowing in the background. Hence the term, ghosting. It can have multiple causes, that usually boil down to something being ON at the wrong time. A timing issue. This isn't an issue with direct drive displays, since they don't rely on timing.

Now to the problem at hand, and its inadvertent solution. A customer returns a board (for evaluation) on what's causing his ghosting problem. I turn it on to confirm the problem ... yep, ghosting. Pretty severe, too. I examine the board visually, but all the parts are in the right place, and the solder joints look okay. So, before probing around further, I decide to clean off the flux residue. When I'm done, I let it dry for a half hour, before applying power. Turn it On ... no ghosting. Apparently, the flux residue (and whatever contaminants it trapped) was conductive enough, to turn ON, inactive "nodes" (transistors). 

In conclusion, clean the flux residue off the board !

[Phill Scan]

Roger Dodger .. cleaning flux of do-hicky .. kimwipes at the ready !

[David Forbes]

I used to have an old Vox Continental Baroque organ from 1969, made in
Italy, with Ducati capacitors. It had a piano register and an organ
register. The thing had been through a hard life when ound it at the
thrift store, and I had to invent a new power supply and a top cover for
it to get it to work.

When I powered up the piano section, it made this multi-tone growling
noise. After a good bit of poking around, I eventually concluded that it
was due to the high-impedance, pressure-sensitive switches in the piano
keyboard. They had just enough leakage, when not pressed, to turn on
most of the notes just a little bit, all the time. It might have been
due to too many beers spilled on top of the thing in its former life.

On 2/11/16 6:22 PM, 'threeneurons' via neonixie-l wrote:
> Well I'll go to the end, with the solution ... Clean your Boards !!!!
>

> Now to the problem at hand, and its inadvertent solution. A customer
> returns a board (for evaluation) on what's causing his ghosting problem. I
> turn it on to confirm the problem ... yep, ghosting. Pretty severe, too. I
> examine the board visually, but all the parts are in the right place, and
> the solder joints look okay. So, before probing around further, I decide to
> clean off the flux residue. When I'm done, I let it dry for a half hour,
> before applying power. Turn it On ... no ghosting. Apparently, the flux
> residue (and whatever contaminants it trapped) was conductive enough, to
> turn ON, inactive "nodes" (transistors).
>
> In conclusion, clean the flux residue off the board !
>

--
David Forbes, Tucson AZ

[Quixotic Nixotic]

I experimented with charlieplexing nixies a few years back and while it worked well, it was very intolerant of faults. One open cathode circuit would make all the other cathodes in that group glow dimly, etc. 

I am amazed that flux was enough to cause a problem like that for you, but I will bear it in mind if I make any more.

  

 4x5 Charlieplex 2 tube pcb layout - board was single-sided home-etched:

John S

[John Rehwinkel]

I am amazed that flux was enough to cause a problem like that for you, but I will bear it in mind if I make any more.

I'm not surprised.  Just a few hundred microamps is enough for visible glow, and those transistors can have current gain in the hundreds.  Most flux is hygroscopic, so it pulls

moisture out of the air, and then the residual salts make it conductive.

- John

[threeneurons]

On Friday, February 12, 2016 at 7:48:07 AM UTC-8, jrehwin wrote:

I'm not surprised.  Just a few hundred microamps is enough for visible glow, and those transistors can have current gain in the hundreds.  Most flux is hygroscopic, so it pulls

moisture out of the air, and then the residual salts make it conductive.

- John

To make matters worse, it most likely was caused on the anode driver side. Each driver has two transistors. They are HV transistors (MPSA42 & MPSA92), so their current gain can be as low as 25, but usually twice to 4 times that. Even at the lowest values, to get 5mA into the nixie anode, it only takes 8uA to into the base of the low side transistor.

Usually, you don't think about this kind of leakage with digital circuits. Doing something, slick and fancy, never goes unpunished !

[Ross H]

Just wondering what you use to clean the flux off the board? I have one of your kits and have tried all other suggestions, resistor substitutions etc. for getting rid of ghosting with no real success. I love the clock and all it can do but the ghosting is a little annoying.

Thanks...

[threeneurons]

The solder I use is Kester 44. The rosin/flux it has can be cleaned with either denatured alcohol or acetone. I've read that some people use isopropyl alcohol. The key is how concentrated is the stuff. If its alcohol, make sure its at least 90% alcohol,The isopropyl from the drug store is only 70%, and only 50% from the dollar stores. I buy my alcohol and acetone from the Home Depot. Klean Strip brand "denatured alcohol" is ~ half ethanol (the stuff in gin and beer) and half methanol (wood alcohol).

On small boards, I just use acetone. The reason for having both, is that they evaporate at different rates. On larger boards, I start with alcohol, and let it soak a little. The alcohol takes a little longer to dry. I use a stiff tooth brush and a pair of shot glasses. One with alcohol, and one with acetone. I dip the brush in the alcohol then scrub the board with it. Once the really nasty crud is gone (or at least spread around), switch to the acetone. Dip and scrub until the dried board is no longer sticky. Brush in one direction, so that the tainted liquid sprays off the edge of the board.

If you still are having ghosting issues, after the board is really clean, with no sticky residue, then contact me thru the eBay messaging system. Also look at the clock's instruction page. Its updated regularly. Maybe, there's a note, that's been added since you viewed it last.

https://threeneurons.wordpress.com/nixie-power-supply/nixie-clock-kit-microcontroller/