Re: [neonixie-l] Finished Nixie watch project

[Joe Croft]

Very pretty watch, it makes me drool like all of the other watches out there! As Homer Simpson put it 'Augh... Nixie Watch...' Now I just need to find a paper towel to clean the spit off of my desk.

-joe

On Feb 3, 2017 11:03 AM, "Craig Smith" <portrea...@hotmail.com> wrote:

I thought I would post a few pictures of my finished Nixie watch that I have been working on for a while and some pictures of the build.

The finished watch

The breadboard demo setup

After testing lots of different shift registers to control the tubes. I needed a shift register capable of doing the job and that would run at 3v. I went for Micrel MIC5841. Although these are not meant to run at 3v, after lots of testing they would work down to 2.1v very reliably, so operating them at 3v should be fine.

The microprocessor used is a PIC18F13K22
I ended up using an analogue accelerometer (ADXL335) over the i2c one as it drew a lot less current.
To get the 170v required to drive the tubes, I used a DC/DC converter IC (LT1308) and a small transformer 31105R (which I got from David Forbes at Cathode Corner)
To keep the time nice and accurately, I went for a real time clock IC. This is a DS3231M. It has built in temperature correction.

The total current drawn when sitting there not displaying is only 96uA


The watch displays the time in a similar way to most of the other Nice watches out there. You tip it to the viewing position and the hours flash, then the minutes. As I wanted the date to shown also, if you keep the watch in the view position, the date will also flash up, day then month.

When the battery gets to 3.5v, the minutes flash twice to show the battery is running low (I copied this idea from David Forbes). When the voltage gets to 3.3v the watch stops displaying.


When testing the circuit I did 45 time reads a day and the watch lasted for 18 days until it stopped displaying the time. It then ran for another week until the battery had dropped to 3v.

PCB's etched by OSHPark

PCB assembled.

--
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 neonixie-l+unsubscribe@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/722e8395-a6e6-4dbe-837d-a9f95986c03a%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Paolo Cravero]

Awesome Craig!

Thank you for sharing implementation details. Which tubes have you used? 

Paolo

[Craig Smith]

NL - 5870.

[Craig Smith]

A few more pics......

I was going to get the watch CNC machined, but it was going to cost a lot of money, so I ended up getting some aluminium laser cut to the rought size and machining it on my mate's lath and milling machine.

I also anodised the case myself in my garage.

I was originally going to use a stainless strap, but opted for a leather one instead.

[Jon D.]

Really nice job !!!

[gregebert]

Ahh, the purple PC boards from OSH Park. I just did my first PCB with them last month and was very happy with the results.

Nice job fitting everything together; It's hard to appreciate the challenge unless you've done it yourself. 

I got my idle power consumption below 2uA, and after 2 years my demo board is still running from the same initial charge on the battery. Definitely overkill on my part, but I really enjoyed the design challenge.

I wish I had the tools to machine my own case; yours is impeccable. Best I can do is 3D print, and it will look pretty raspy.

[Craig Smith]

2uA!!!

I couldn't get anywhere near that low, even if I was just powering the PIC. I spent weeks and weeks trying to get it to run as low power as possible. I'm running the PIC at 4MHz with all the hardware turned off (i2c etc.) until I need it. 2uA is very impressive.

The lathe and the milling machine belong to a friend of mine. They are pretty old, 1970s I think.

It took 5 evenings of 3 hours to machine them. The hardest part was setting the parts in the chuck and getting them centered, due to the very thin wall thickness. You can't go reaming the 4 jaw chuck up so tight you crush the 2mm thick walls of the watch case in, but then again you can't have it that loose that it will fly out of the chuck.

Anyway, I'm very happy with the finished result and am since Wednesday I have retired my Rip Curl analogue tide watch for my new Nixie watch. Although it is larger than my old watch, it only weighs 72 grams, my old one was 219 grams!

[gregebert]

The 2uA is approximate; I have a 50uA mechanical meter, and it barely moves from 0. My DMM has too much internal resistance.

Anyways, that's what the RTC uses plus and leakage in the DC-DC converters. Every 75 seconds, the RTC draws a 'spike' of current to do the temperature-compensation, and it pegs the meter.

Since my battery is Li-ion, it's voltage varies from 3.7 to 4.1V depending upon state-of-charge. The FPGA and TTL devices only run on 3.3V, so I had no choice but to use a DC-DC converter; it has an integrated low leakage power switch.

The HV DC-DC converter was a real pain, but I finally found a suitable NMOS device that had very low-leakage, low Vgs, and could tolerate the ~60V kickback from the transformer.

[Craig Smith]

Impressive. Have you got any pictures of your creation?

[gregebert]

I posted about 2 years ago on this forum. To be honest, I'm ashamed of it's appearance compared to your machined case.

It's handmade acrylic (plexiglass), and was sealed-shut. Something failed during endurance testing, and although I've opened it up I have yet to debug it.

The main difference is mine uses a 3.5 digit 7-segment display that was intended as a timekeeping device. It even has a colon and AM/PM. My father gave the display to me back in 1975, and it took me almost 40 years to find additional units and make a wristwatch out of it.

Once I get my backlog of nixie projects under control, I'll fix it, 3D print a case for it, and finally wear it. It's supposed to last 3 months on a charge when used 10 times/day.

[Craig Smith]

I just took a look. I like the 7 segment Nixie.

I have to correct an earlier post. I run my processor at 32KHz, not 4MHz. Still at 32KHz, I still can't get the power consumption as low as I would like.

I suppose I am using a linear regulator to drop the lithium battery to 3v, so some wasted power there.

[gregebert]

I suppose I am using a linear regulator to drop the lithium battery to 3v, so some wasted power there.

I dont know what your battery capacity is, but if it's around 1000mA-hr, your watch will run in standby mode at 95uA for a year. So even though it's wasting some energy with the regulator, it's probably of little consequence.

I had some challenges getting my LT3561 to work, notably soldering that tiny thing, and noise/lead-inductance from my bench supply. I've often wondered if I should have stuck with a simple LDO.

My original intention was to make a sealed watch so it would be waterproof, and I would turn it on by touching 2 exposed metal bumps (skin resistance). To charge it, I would apply about 5V on the same bumps. And to set the time, it would be a serial protocol, again on the same bumps. I ran out of FPGA gates to do the serial protocol, and my skin resistance was too high to overcome the ESD-protection on the 2 bumps. If I had more time, I would replace the FPGA with a microcontroller and solve the 'skin switch'.

[Instrument Resources of America]

Next time that you're trying to machine something thin like that in a four, or three jaw chuck, plug the end that fits in the chuck with something solid that fits fairly tight, and when done just knock it out. I've had to that a time or two to keep from caving something in.   Ira.

--
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 neonixie-l+...@googlegroups.com.

To post to this group, send email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/dbf5c253-e834-4f8c-bbc8-29fc11c42597%40googlegroups.com.

[Craig Smith]

My battery capacity is only 120mAh which is about the biggest I could get to fit my design, and it's re-chargeable. Circular enclosures are a bit of a pain to fit stuff into.

I guess I could get more life out of the battery if I set the under voltage limit lower in my code, but to be safe, I have it stop displaying the time when the battery gets to 3.3v.

I haven't tested how waterproof my watch is yet. I designed it so it should be pretty water tight. The bottom fits to the top with a small lip, which was a nice snug fit when I finished machining, but after anodising, it is an extremely tight fit. I was originally going to have a tiny o ring in a groove, but I decided it was going to be too difficult to do in such a thin wall. When I attached the front part of the watch to the back, I smeared a small amount of Vasoline around the lip, just to help keep out moisture. The glass is glued in, so should be water tight there and the micro USB port on the back for charging has an o ring around it to keep out the water.

I don't plan on going in the water with the watch, (especially not surfing) but I'm pretty confident it will be fine in the rain.

I don't have any external buttons on the watch so there ain't a chance of water leaks there.


Just out of interest, to set the time you do...


Tip watch fully to the left for 3 seconds, then to right (within 1 second) and hold it there for 3 seconds (but no longer than 6), then tip it fully to the left again for 1 second.

The watches flashes the battery voltage (for example, 3.8v will be displayed as 38)

The watch is then in setup mode. The watch is tipped left and right to move through the basic menu and is tipped forward to accept.

If the watch is in the setup mode for more than 30 seconds without the 'accept' being used, it exits back to normal operation. This is in case you somehow get into the setup mode by accident. I've tried to make it pretty tricky to actually enter setup mode.

In the setup menu, you can only set the time and the date. I didn't bother with a setting for viewing angle, as it's a one off watch for me and I have hard coded where I want it to display the time.


Anyway, I've been wearing it now for 3 days with no problems. It's going to be interesting to see how long the battery will last in real life operation. I would like to think at least 2 weeks before it needs charging, but as it only takes 1 hour to charge the battery, it wouldn't really matter if it only lasted a week on a charge really.