1970's Nixie Tube panel meter to clock

[Jeff_Birt]

I scrounged up a couple of panel meters today that used Nixie tubes! They are 115VAC powered and have a +-0.1999VDC input range. I wired one up and used a programmable power supply as a voltage input source and it worked!! They use a vertically mounted neon lamp to the '1' and three nixie tubes for the other digits. There are also indicator lamps for '+', '-', and 'OL' (overload). They also have a hold feature and a BCD output (outputs current display). The decimal point is also programmable by grounding one of three input pins.

What to do with a 3.5 digit 200mv Nixie tube panel meter? The first thing I thought of was a clock. There are the correct number of digits for a 12 hour clock. It would not be hard to use a small micro to output a voltage corresponding to the time. Thinking then about mapping the D/A output to the correct panel meter reading (thinking it might not be perfectly linear) and it occurred to me that the micro could monitor the BCD output in case the reading drifts and adjust the output.

I need to find a couple of 1970's looking push buttons to act as time setting inputs and pop the whole works in suitable 1970's looking small box. Oh, then there is the small task of programming the micro and wiring things up

[bitseeker]

That sounds like a neat little project. I like your idea of monitoring the BCD output to compensate for any drift. Looking forward to watching your progress.

[mmagin]

I used to use a low-end 1970s (LED) frequency counter as a desk clock by driving it from the audio output of a sound card I never used

It was a linux box and I just had a cronjob run something like this every minute:

Code: [Select]

play -n  synth 60 sine `date +%H%M`

Which is basically take the time and make a string of "HHMM" out of it, then synthesize a sine wave of that frequency for 60 seconds.

[Jeff_Birt]

I used to use a low-end 1970s (LED) frequency counter as a desk clock by driving it from the audio output of a sound card I never used

It was a linux box and I just had a cronjob run something like this every minute:

Code: [Select]

play -n  synth 60 sine `date +%H%M`

Which is basically take the time and make a string of "HHMM" out of it, then synthesize a sine wave of that frequency for 60 seconds.

Very cool! I love putting common things to unusual uses. I did some quick calculations just now and found I need a minimum of 12bit D/A resolution to have 0.1mv resolution. Thinking of what you did with your sound card another option would be to use a frequency output form a micro, most of them have some sort of counter/timer type of thing these days, and a frequency to voltage converter. I guess that depends on the effective resolution of the VF converter though.

[bitseeker]

@Jeff
Depending on the micro you choose, PWM followed by some RC filtering may also be a useful way to get enough resolution for DC output.

@mmagin
How was the accuracy and stability on the least significant digit? I was thinking of doing a similar thing via microcontroller, which might be OK with an external crystal, but I haven't tried it yet.

[mmagin]

@mmagin
How was the accuracy and stability on the least significant digit? I was thinking of doing a similar thing via microcontroller, which might be OK with an external crystal, but I haven't tried it yet.

It was okay, not perfect, a little bit of instability in the 1Hz digit -- this was a HP 5381a, with such a simple counter there's only 1 Hz resolution at 1 second gate time.  Still better than having to turn around and look at an analog wall clock

[bitseeker]

It was okay, not perfect, a little bit of instability in the 1Hz digit -- this was a HP 5381a, with such a simple counter there's only 1 Hz resolution at 1 second gate time.  Still better than having to turn around and look at an analog wall clock

Yeah, no need to be somewhere with +/- 1 minute resolution, unless you're catching a train/plane/etc.

[ynfo]

I did something similar, see this thread: https://www.eevblog.com/forum/projects/a-digital-analog-digital-nixie-clock/



Still keeping good time!

[Jeff_Birt]

Nice!

[bitseeker]

Very nice. What happens when you switch the time to the 2000 range?

[ynfo]

Just tried it a moment ago:

Range    Display
20         09.18
200       009.2
2000     0010.

[Jeff_Birt]

Powered up the second panel meter yesterday and it did not work, the OL was on as well as a decimal point which is the mode it goes to when there is an overload. Took it apart and found someone had tried to repair it at some point and really butchered it. I sent an email to Newport (which is part of Omega) and got a replay back in 15 minutes. The replay was from a gentleman who had worked there for 27 years and said that the nixie tube model I have was obsolete 15 years before that. He said he has never come across any documentation in all of these years for the unit. The 200ASBS manual they have on line was the replacement model and hopefully the design will be close enough that the schematic is useful.

[Jeff_Birt]

More pics

Note to forum mods/admins if you are reading this. Instead of just having the forum SW complain that the image file size is too large why not just rescale it automatically to what is acceptable?

[bitseeker]

Woah! That's quite a mess, indeed. What's the deal with those scratch marks?

[Jeff_Birt]

What's the deal with those scratch marks?

I guess somebody thought they were repairing it. They really butchered it though.

I have been working on this as time allows. It is displaying the time on the panel meter, and I can read back the BCD output from the meter too. I've attached an image of the project. The three switches provide a way to set the time. The 10 turn pot lets me scale the 1.024V reference from the PSoC to 0.2V for the DAC.

You'll notice that all three decimal points are on. The documentation I have shows which of the three inputs to ground to set the decimal position. I 'assumed' these would be logic level since the other digital I/O seemed to be but they were pulled up to about 12V and seemed to go directly to the driver circuitry as even a small amount of current f low through a 500K resistor to ground would result in a dim decimal point. I dug up some 4N35 opto-isolators and that did the trick, now I can control all three. I was thinking that in time setting mode the decimal could blink beside the digit being set?

Using the PSoC for this has been fun. Most of what I needed was already built in as components in the IDE. The only downside I have found so far is that with the smaller PSoC 4 I have there are not enough digital blocks to have the SPI bus and a full UART going at the same time so I'm just using the SW TX only UART component to get some feedback on the PC while developing.

The IDE also has a RTC component which was simple to use but you set the time with BCD which is kind of quaint, in HH:MM:SS format. Luckily they have some helper functions so you can grab the BCD minutes, convert to decimal, and inc/dec and then convert back for time setting. Of course you can also do the unix type of time is that is your preference. I found that although I wanted to display the time in 12 hour format on the panel meter it was much simpler to set the RTC component to 24 hour format than deal with AM/PM when trying to inc/dec minutes and hours.

I need to close the loop with the BCD feedback from the panel meter now and finish up the firmware. Quite a lot of effort for a clock but quite a lot of fun too!

[bitseeker]

That's coming along nicely. The first thing that caught my eye were the three decimals. In addition to being used as time set indicators, the middle one can blink for each second.

Using 24-hour mode sounds fine. I got used to it many years ago, though it's more common elsewhere than in the US. If you were going to do 12-hour clock, you might also want an am/pm indicator (perhaps the right-most decimal; scratch that, there isn't one to the right of the right-most digit.).

[Jeff_Birt]

Blinking the decimals is a neat idea. The PSoC RTC API is set to run in 24 hour mode but I am limited to 12 hour mode on the panel meter as it is only 3.5 digits. The .5th digit is actually just a slender neon lamp! This panel meter is bipolar and has +/- indicator lamps (the + is on in the picture). I did not want to fuss with a bipolar power supply etc. so I think I will just use a small relay to swap the DAC output polarity so I can use the +/- as AM/PM indicators.

I was getting an occasional glitch on the BCD feedback lines and finally realized I forgot about the Data_Ready output form the panel meter; I was getting an occasional read of the BCD outputs before the outputs were stable. I added that wire and set up the input to trigger an interrupt and now have a nice stable minutes feedback.

[bitseeker]

Ah, yes. The old vertical neon as a '1' trick.

I was also thinking about the +/- indicator. That would work nicely. + for ante meridian and - for post meridian.