Tuning fork/analog electronic clock idea

[spectre6000]

https://hackaday.com/2021/02/13/tuning-fork-keeps-this-throwback-digital-clock-ticking/

https://www.secretbatcave.co.uk/projects/electromechanical-clock/

I stumbled on these in one of my newsfeeds the other day. Macro scale Accutron clock project. A third clock project popped up at the same time, but it's not exactly a time based clock. I guess these are cool all of a sudden maybe? 

Meanwhile, back at the lab, I've been learning about analog electronics. Building a 100% analog phono preamp on the side of my desk not occupied by work related detritus (watchmaking, unfortunately, commands nearly entirety of my desk surface, so it's been playing 3rd fiddle at best lately). 

Anyway, digging into the first one got me thinking. He mentions at the end of the video that accuracy varies according to ambient temperature, and there's some conversation in the comments about ways of compensating for that. My knee jerk reaction to an image of the final project was that the 7 segment displays could be replaced with nixies. The schematic for the tuning fork regulator part of the clock could easily be done with a simple and readily available dual triode. I'm not sure about the rest of the circuit yet, but those are bridges that can be crossed.

The interesting thought involved the intersection of the analog electronics, the heat they produce, and the regulation of the tuning fork. The heating elements in the tubes put off a lot of heat. Especially if the entire thing were to be executed in analog circuitry (rectification, amplification, display, and I haven't figured out the totalizing/counting aspect of it yet). If all of that could be crammed under a double walled glass bell, there'd be some serious heat in there. Possibly enough to cause problems. BUT I think it might be self regulating.... As the heating coils get hotter, resistance drops. If the ambient temperature is higher, they consume less power, yadda, yadda, and I think you'd end up with a pretty stable, albeit high, temperature to keep the fork humming along at a pretty steady rate. That rate would NOT be the rate stamped into the side of the fork though, so there'd need to be some means of tuning that regulation... Not really sure about that yet.

Not sure I'm going to do this yet. It's a pretty tall leap to get from basic AC rectification and signal amplification to a totalizer circuit of some sort, but I haven't had time to do anything more than read that article/watch the video. I recall having seen another guy on youtube (handle has to do with rabbits in Japanese or something) that might have done something along the lines of a counter/timer. Might be worth looking into that. Videos require less eye strain and concentration than reading, and my toddler likes videos on daddy's computer, meaning I stand a chance at being able to do such a thing.

Anyway. Thought I'd drop this here to see what the horological hive mind thinks on the subject.

[LittleWatchShop]

4 hours ago, spectre6000 said:

analog

Are you equating analog electronics to tube electronics?

[spectre6000]

Yes. As in no microcontrollers/processors. I guess the required highs/lows could be considered digital-ish though. There really wouldn't be a lot of shades of gray in a project like this.

[LittleWatchShop]

2 hours ago, spectre6000 said:

Yes. As in no microcontrollers/processors. I guess the required highs/lows could be considered digital-ish though. There really wouldn't be a lot of shades of gray in a project like this.

But analog circuits can be made solely from semiconductors--no vacuum valves.

[HectorLooi]

7 hours ago, spectre6000 said:

BUT I think it might be self regulating.... As the heating coils get hotter, resistance drops. If the ambient temperature is higher, they consume less power, yadda, yadda, and I think you'd end up with a pretty stable, albeit high, temperature to keep the fork humming along at a pretty steady rate.

Huh?.... Is this correct I thought it was the opposite. 

[LittleWatchShop]

3 minutes ago, HectorLooi said:

Huh?.... Is this correct I thought it was the opposite. 

Hector is correct.

[spectre6000]

Re: analog/tube/valve circuitry. Yes, it can be made from semiconductors, but aren't glowing glass tubes cool looking? I've been learning how to design/build tube circuits lately, so that's where my head is. It's pretty much an art project I'm talking about here. It won't set any accuracy records or anything like that. I'll be happy if it's within a second or two a day. Vacuum tubes (at least the nixies) would provide that regulatory heat source [i]assuming[/i] it doesn't create thermal issues with the rest of the electronics.

Re: resistance/heat. Oops. Yes. Typo. Increases. The result for purpose is what I meant though. I mentioned the idea with my electrical engineer guru friend last night over drinks (the guy that made the prototype atomic clock resonator for NIST), and he said the same effect is used in (specific model/series number) voltage regulators. I have heat cables on my roof to prevent ice damming (100 year old house in the mountains), and they utilize the same regulatory effect. They heat up to a point (40°F or so), at which point the resistance increases to where they stop/decrease draw and then passively self regulate.

The idea is still in its infancy. I read the articles/watched the videos immediately prior to posting. The only time that's gone into it so far aside from posting here is bringing it up with someone smarter than I 3 stiff drinks in (yesterday was a bad day for anyone with kids under 5). 

One of the few things I've rubbed any brain cells at is the frequency of the tuning fork. Higher frequency would be better accuracy, and I could conceivably find something around 25K Hz and it would be inaudible, but then it would also be so small as to lose some of the visual effect. Below that threshhold of inaudibility, higher pitches don't travel as far, but are more grating and perceived as louder when you can hear them. Lower frequencies would be more pleasing, but less accurate, travel further and easier, etc. I keep stumbling on some concept of "most pleasing frequency", but we're talking clean sine waves here, and I don't think any such thing exists. Just one of those stumbling points my brain likes to trip on in that constant system optimization mindset.

I'm pretty confident a basic digital clock circuit, and probably even one utilizing nixie tubes as the display will be extremely easy to track down. What I haven't figured out yet, and suspect might be a challenge, is a means of regulation. The tube heater based thermal regulation idea will decrease the frequency of the tuning fork by an effectively arbitrary amount, and I'll need some way to compensate for that.

Finding a large bell jar will be another road block. It'll need to be sourced very early, because everything will need to fit inside. That said... Mindgrapes squeezing... I might be able to better insulate thermally/acoustically if I make something from wood/glass... Glass is cheap, double panes are easy to do, wood looks nice (I have a literal ton of exotic hardwoods in my stash)...

[LittleWatchShop]

Analog and digital are characteristics of the signal being processes and not the technology in which the processor is implemented. That was where I was going.

Yes, tunes are nice. Especially in the winter. Snuggle up to the clock and stay warm.

[spectre6000]

Yeah. I got you with your initial question. It'll be a digital signal. Old school electronics tech is what I meant, because I'm in that mode lately, and because I think the thermal regulation element of it is a pretty neat way to solve a problem while making the entire thing even cooler over all.

I think this will be the next project on my docket, at least to figure out if it's something I can figure out. I think all of the problems are solvable, I just need to figure out a solution for regulating to the unknown frequency of a hot tuning fork.

[spectre6000]

Current thoughts:

Power supply will be 170VDC tube rectified. The heater supply will be split off and rectified (solid state) for 5V and/or 12V depending on how things shake out.

Tuning fork pickup/stimulator will be amplified via a dual triode. Preferably something readily available and in current production (though with current events...). Likely a 12AX7 of similar if I can make that work with the 170VDC (preferred by most nixie tubes).

I don't know that I'll be able to handle the counter logic work and be tunable without a microcontroller... I have a handful of Arduinos, and another similar microcontroller board or two, so that should be easy.

I had a neat idea to control the nixies from the microcontroller: an array of reed relays! They'll click quietly, but for different reasons than your typical clock. Additionally, there will be more information in the ticking, as there will be extra ticks every 10 seconds, minute, ten minutes, etc. It would take a total of 40-something relays (one for each number that is used, plus a few for some neon lamps) if I want to keep it simple. I might be able to figure out a way to multiplex them, but I'm not sure it really gives me any benefits.

Enclosing the whole thing will bump up the temp and keep it stable to keep the tuning fork thermally stable. The point of the microcontroller counter is to be able to reduce the frequency being counted to account for this AND any manufacturing variances. Both of these factors will be unknown until it's done and in service, so the ability to control it on the fly is necessary for any accuracy.

I'm hoping to be able to figure out how to suspend the tuning fork, pickup, and stimulator magnetically. I know it can be done actively, but I'm not sure it can't be done passively. If it can't be done passively, I'll use a series of springs or something. Total mechanical isolation, so the only way the hum can get out is through the air. Similarly/convesely, if I want the ticking relays to be audible, I'll need to mount them/that assembly (however it works out) either to a soundboard of sorts, or I might be able to get away with just mounting them to the side of the case somehow.

So... The big question is accuracy... I'm pretty sure I'll be able to get the accuracy to within half a Hz of whatever frequency the tuning fork oscillates at. 440Hz is sort of the standard tuning fork frequency, but higher frequencies obviously decrease the potential error rate. It needs to be big enough to fit the look I'm going for (I don't know exactly how it's going to work proportionally), and it needs to be magnetic (apparently, a lot of tuning forks are made of aluminum these days). Assuming 400Hz (decreasing the 440Hz for temperature and to make a round number) and spherical chickens in a vacuum, I should be able to just barely edge out an Accutron at +/-1 minute per month. I'll have a higher frequency, more stable temperature, and no motion to account for. If it is close enough that it's not meaningfully out (~+/-5 minutes) over the course of 6 months, I'll be thrilled.

I'm 99% sure this is my next big project. I've got another similar project currently in the works that I need to finish to make room though, and there will be some learning curve to this. I intend to prototype a subsystem at a time, probably starting with the tuning fork pickup/stimulator. I'll copy an existing transistor based circuit to start. It's a semi-known quantity, and I'll use it as a baseline.