IN-18 direct drive question

1 year 4 months ago - 1 year 4 months ago #7497 by Ian
I'm thinking of designing and building an IN-18 clock and have a choice to make, and wanted to ask people here what they think about a really fundamental question: Direct drive or multiplexed?

There is a lot of folklore that it is better to drive IN-18 tubes directly (meaning not multiplexing them). The reasons given, as far as I can tell, is that if you multiplex large tubes (primarily IN-18 tubes I think) you can get a blue spot, , the life of the tube might be decreased, and the tubes might resonate with the multiplexing frequency.

Blue Dot

I have seen some real data/research on the causes of the blue dot here (thanks Paul for the link!) here: and it appears to be the fact that cathodes which are not driven float at some voltage and influence the driven cathode, causing an area of high current density. All open collector/open drain drivers have this effect, apparently, unless you bias the undriven cathodes.

So, here's the thing: clocks like the Spectrum 18 use the HV5622, which is a 32bit, latched, open-drain driver:

it direct drives the cathodes, but does no biassing. It should therefore have "blue dot" problems.

My understanding: This is a problem with the drive circuitry and not really anything to do with multiplexing or direct drive. Properly biassed tubes should not have a blue dot, and in fact in the multiplexing case, it is a lot easier to provide biassing (only 10 lines to bias instead of 60).

Tube Life

The folklore reason for direct drive being better is that the current is lower and therefore there's less wear and tear on the tubes, especially due to sputtering.

I can find NO data on this question at all. There is a discussion here!msg/neonixie-l/gqA6OpDbdXM/raYXUOMK3kIJ which does nothing at all to clear up the question.

My understanding of this issue: I have seen NO data (meaning I can have no understanding).

Tube singing

Larger tubes can resonate if the driving frequency matches the resonant frequency of some parts of the tube.

Solution: drive the tubes at a different (higher) frequency.

My understanding: Not really an issue, just a code implementation task.

So, here's the question:

So: I'm hoping some of you guys have some pointers to something I can use to decide whether the new IN-18 clock should use direct drive or multiplexed cathode driving circuitry. I have no reason to choose one or the other at the moment.

I can see some advantages to multiplexed driving at the moment:
  1. Dimming becomes easier, because you already have a division of the on time to do the multiplexing, and a sub-division of this time is trivial.
  2. Anti-Blue-Dot biassing becomes easier

I can also see some advantages to direct drive:
  1. Simpler board layout
  2. Less pins used for tube driving (=more pins for perhipherals)
  3. Easier to interface with something like an ESP32 (I want to transcend old-school controllers)

Any inputs?
Last Edit: 1 year 4 months ago by Ian.

Please Log in or Create an account to join the conversation.

1 year 4 months ago - 1 year 4 months ago #7503 by Pacman223
I like the research you have put into this.

I believe that Jeff Thomas was a strong supporter of direct drive clocks. He may have known something that we don't.

I think most tubes were built with direct drive in mind.

I think almost all nixie clock builders build multiplex clocks. PV electronics are mostly multiplexed and I have never had a problem with them. (I think their new IN-18 clock is direct drive)

For expensive IN-18 tubes, I would want them to look their best. That might be direct drive.

Also consider, direct drive Nixie clock kits are rather rare. This might be a good selling point for some.

If you make a really nice IN-18 clock, I probably will buy it (I need a good clock for my IN-18 tubes) I probably would rather it be direct drive, but it would not stop me from buying your clock if it was multiplexed.

Hope this helps
Last Edit: 1 year 4 months ago by Pacman223.

Please Log in or Create an account to join the conversation.

1 year 4 months ago - 1 year 4 months ago #7505 by Ty_Eeberfest
When this thread first went up I read it and thought I'd rely "later", then I forgot about it, so here's my input better late than never...

I have a very strong opinion on this subject but unfortunately very little in the way of facts or research to back it up. I take the position that Nixie tubes should always be direct driven. All my custom circuit designs use direct drive and yes, that includes my IN-18 Chiming Clock.

1) In the old days Nixies were not multiplexed. Thus multiplexing is not "historically accurate". Now I guess you could say the same about the HVxxxx chips I use in my direct drive circuits, since they obviously didn't exist in the old days, but somehow I can rationalize HVxxxx's to my self. B)

2) To the best of my knowledge Nixies were never intended to be multiplexed. The technique of multiplexing did exist in those days, so why do Nixie data sheets never provide any current (etc.) specs for multiplexed applications? Data sheets for the similar but different Panaplex type displays DO include info for both direct drive and multiplex applications. So from this I infer that Panaplex were designed to be muliplex-able and Nixies were not.

3) Direct drive with HVxxxx driver chips uses less I/O than any other drive scheme I'm aware of. I can run a display with theoretically an infinite number of tubes using 4 outputs: DATA, SERIAL_CLOCK, LATCH, and BLANK (aka ENABLE).

4) Dimming can be super easy - just do high-ish frequency (I like 4KHz for reasons and stuff) PWN on the BLANK line,

5) Multiplexed displays look like shit in videos (flicker) , direct driven do not. :dry:

That's all I can think of for now.

Look into it later when the dust is clearing off the crater.
Last Edit: 1 year 4 months ago by Ty_Eeberfest. Reason: Typo slop

Please Log in or Create an account to join the conversation.

1 year 4 months ago #7506 by Ty_Eeberfest
Also... Blue Dot. We have had that debate before on this forum but I'm too lazy to dig up the old thread. The discussion included a couple of the professional clock sellers, e.g. Mr. Nixie and somebody else well known, The consensus was that the blue dot issue arises when using the Russian K155ID1 knock-off decoder/driver but not when using a real 7441/74141.

I personally drive my IN-18s with HV5530s - two of them for a 6 digit clock. I've had no blue dots. IN-18 chime clock has run continuously for 5+ years now with no tube failures that can be blamed on electrical issues - I did lose one tube to gradual outgassing but that has nothing to do with drivers.

So far I've never had a direct driven tube "sing" no matter how much I dim it. I have several clocks running here with Russian IN-xx tubes, one with Valvo tubes, one with Burroughs tubes, and of course the monster with 16 Burroughs B-7971 segmented tubes. Also a couple Panaplex clocks. All direct driven, all using Supertex HVxxxx except the B-7971 monster which uses 240 MPSA42 transistors for drive.

Look into it later when the dust is clearing off the crater.

Please Log in or Create an account to join the conversation.

1 year 4 months ago - 1 year 1 month ago #7507 by Ian
Thanks for the input, it seems to be a subject that people find it really hard to stick to, but you guys have come up with some interesting points.

Weirdly, as someone who is usually pretty data-driven, the argument that I find most compelling is Ty's first point, that it is "not historically accurate", which is sort the the same thing that Pacman said with "most tubes were built with direct drive in mind".

The one thing that was holding me to the idea of a multiplexed clock was the blue dot story, which would be easier to solve with a multiplxed strategy (100k pull ups on the cathode lines, only 10 of them needed in multiplxed, 60 in direct drive). I can't count on getting 74141 in any number, and I sell bucket loads of these things, so a workaround is the only option.

I have a bunch of 5530's which I can start experimenting with.
Last Edit: 1 year 1 month ago by Ian. Reason: Removed possibly unfair statements about Spectrum

Please Log in or Create an account to join the conversation.

1 year 4 months ago #7509 by Ty_Eeberfest
Just to be clear... when I use 5530s I do not use any pull-ups at all and so far have had zero problems. The only resistors involved are the 6 anode resistors and just FYI I run my IN-18s a bit on the hot side so am using 5K anode resistors. Somebody is bound to cringe at 5K and insist on 10K minimum but the longevity and appearance of my tubes speaks for itself.

The only downside I know of with 5530s is that they want 12 volt Vcc AND logic (!) so there's one more voltage on the board and some sort of level shifting is absolutely necessary on the 4 signal lines from the MCU to the 5530s. I have read of people cheating by feeding the 5530s with low Vcc (8 or 9 volts) and hooking up 5V signals straight off their MCU. The idea being that the chip sees Logic 1 as anything within a few volts of Vcc, so reducing Vcc to the ragged edge lowers the logic level requirements. Supposedly this works. I've never tried it and don't mind spending a couple extra dollars to include level shifting.

Look into it later when the dust is clearing off the crater.

Please Log in or Create an account to join the conversation.

1 year 4 months ago #7513 by judge
First of all I want to say that I have no particular axe to grind either way here - Inown both types of clock and am happy with them all. I think that the idea that direct drive is better than multiplexing is largely one of technical aesthetics. So throwing that consideration aside here's my two cents about the issue:

I think the blue dot question is pretty much answered by the link in the first post and is basically orthogonal to the whole question of multiplexing.

I don't find arguments citing the original intention of the tube designers very persuasive, unless you can actually cite that intention. In fact I am pretty sure that I have seen patents and data sheets for nixie multiplexing circuits - I will try and dig some up.

If you are using any form of dimming, you are turning cathodes on and off quickly. Multiplexing is just turning the anode on and off quickly as well. You could argue that this is quantitatively different, but I can't think of any actual physics that would back that up. Conversely, the history of development of these tubes clearly shows a progression based on experimental data, so me not being able to think of physics that would support the 'multiplexing is bad' position doesn't really mean much. Experiment would be better.

Experimentally, the only evidence I have seen actually supports multiplexing. This is Ian's own evidence of apparently increased lifetime from multiplexing IN-1s. Obviously not definitive, but also I haven't seen anything citing decreased lifetimes for multiplexed tubes.

Finally here is a video of one of Ian's multiplexed clocks. No flicker in sight :)

Please Log in or Create an account to join the conversation.

1 year 4 months ago #7514 by Ty_Eeberfest
Actually there is blatant flicker visible in that video. Do you seriously not see it?

Multiplexing is not the same thing as dimming. Yes they are both "turning the tube on and off rapidly" but I'm not so stupid as to have been suggesting that doing it to the anode is any different than doing it to the cathode. Of course it doesn't matter which you are switching. However, a multiplexed tube in a 6 digit clock is only lit 1/6 of the time so for a given perceived brightness the multiplexed tube must be fired with approximately 6 times the current of a non-multiplexed tube.

You sound like you DO have an axe to grind... with me rather than with the topic. Or maybe it's just your writing style.......... :S Can't say I was trying to persuade anybody actually.

Look into it later when the dust is clearing off the crater.

Please Log in or Create an account to join the conversation.

1 year 4 months ago #7515 by judge

You can put it down to writing style, or I guess debating style. A conversation would be better, but we are limited to what can be conveyed in text. Apparently not very well!

Actually I can't see the flicker. I looked carefully at it before I posted it - partly to see if there was any difference between the YouTube version and the version on my phone! I have certainly seen videos on Facebook that show an obvious beat between the clock's refresh rate and the video camera's sample speed. Now I'm going to have to post a video of one of my direct drive clocks to see if I can se the difference in comparison!

Good point about the 1/6th time thing. Though I still haven't noticed any difference between multiplexed and direct drive clocks in real life.

Please Log in or Create an account to join the conversation.

1 year 4 months ago #7516 by judge
Hmm. Now I have watched it play directly in this forum, the beat is obvious. So I went back and looked at the YouTube version and the version on my phone. Now I can't unsee it!

Please Log in or Create an account to join the conversation.

Moderators: AccutronTy_EeberfestIan
Time to create page: 0.171 seconds


Tube Suppliers

Go to top
JSN Boot template designed by