The (long) story of the relay clock

[Spemo]

Hello everyone,
in this thread I want to share the (long) story of the relay clock, how it came to be, the problems and what's going on today.

But first things first.

One day, I stumbled across a video made by look mum no computer titled "Retro Russian "NIXIE" VFD Clock ELEKTRONIKA 7" where he shows a russian clock and asked the question: But what does it sound like?
Well the answer was: like nothing. Silent. And that gave me an idea. Why don't I come up with a way to make a clock, which is "electronic", but sounds like it's doing something?
Since I liked electronics and a challenge, and had nothing better to do, it sounded like a good idea to keep my mind busy.

With that in mind, I started thinking about the idea. Which parts to use, how to make it, prices of everything, should the parts do the thinking oder use an IC and just let it pretend like it's doing something?

After a week or so I had my answers: relays (I like the noise they make), they should do the "thinking"(counting), they will be cheap shitty parts from china (somehow I ended up with "recycled" relays) and with actually making it - It'll work out somehow.

So I started to look for programs to design the circuit and the PCB, I ended up with Blackboard Circuit designer and a circuit simulator made by some guy named falstad I think.
It took me 4 weeks to figure everything out. I had some basic knowledge, but I have never made up something this big and this complex, all I've ever made never contained more than 5 active electronic parts.

so here is what was needed:

First off, I figured I'd make it with  decimal counters. So in total, it needed 6 counting modules to get the time format HH/MM/SS, and each one has to trigger the next one once it reaches 0(zero) again.
To make this work, I somehow came up with 2 relays per digit, with a delay circuit between them. So when the first one is activated, the other one activates a bit later.
I figured out that the delay was the key thing to make everything function, everything I tried before just triggered all the relays at once.

this is what the first working circuit looked like in Blackboard:



The IC sockets are connected to work with these relays:



So I tested this in a smaller way and it worked, kinda sort of.

Next thing: well ok, now you figured out how to make the counters, but you still need to get this to a readable format - the outputs of the counters had to be translated for a 7-segment display.
That alone took another week, and here is what I came up with:



In total around 350 diodes or something like that.


 So I went onto ebay and started looking for the cheapest parts I could get.
Here is the list of components that were needed:

I think 102 Relays (came from ebay, cheapest ones were around 3€ for 10 Pcs)
102 IC sockets for the relays
about 55 NPN Transistors (again, ebay)
55 or so Capacitors (I had enough overkill 2200µF 25V ones somewhere)
55 trimpots
500 diodes
500 LEDs
555 and support components for timing
loads of wires
and PCBs (perfboard). With these I ordered the largest ones I could get for reasonable money, so I ended up with 30x18cm ones.


After everything arrived, I put it all together. Very quickly I realized that I forgot to add a reset circuit, which resets the Hours when it reached 24 hours.
also, after noticing that the LEDs I ordered are very bright, I had to somehow get the brighness down...which I somehow managed with a PNP transistor.
After many errors and mistakes, and my basic soldering skills, I managed to get it to work and this is what it looked like:



it was huge. 30x18cm PCBs, and 4 boards in depth, it wasn't easy to find a place to put it. Also, I quickly made a housing for it, since it was very sensitive to movement (soldering skills.....)

after that, I noticed that with the best "tuning" it was never able to keep time properly - some googling later I found out that ne 555, resistors and capacitos shift in value with temperature 

So I looked for a simple 1 hz generating circuit, found one with a crystal, adapted it to the existing circuitry - and there came PCB number 5 with better controls for everything.


After all, this monster of a clock worked fine for about 3 Months or so, but it had its issues.
1. even with the crystal, timekeeping became more and more of an issue.
2. Somehow, it sometimes went mad and one counter module decided to erraticly stop somewhere inbetween and messed everything up.
3. the reset circuit had a mind of its own, sometimes it simply resets the way it was supposed to or startet "flickering", and finally managing the reset after 10 or so attempts.
4. The display. Either due to my poor soldering, or due to very badly made LEDs, a lot of them failed sadly. Also, it was still very bright and able to light up a room.
5. Power consumption and distribution. In the end this thing took a lot of power- about 15 Watts at full brightness - 10 at a more humane not sun-like experience setting. but still, a lot of power, which hat to go everywhere. it came to a point where I doubled every powerline in the circuit.
6. Size. Size. Size. It was just too big. The final dimensions were 35cm wide, 23cm high, 25 cm deep.


This is the story of the first relay clock that I made, and let me tell you, there were a lot of headaches with this one.
There is video of this on youtube, if you want you can check it out here:

That's all for the first one, I have made some more to try and improve the circuit and everything else. I will show you the others too.

enjoy your day,

Daniel

[Spemo]

So, the 3 Months it worked I got more and more fed up with it, it was sitting on my bench for reapair more than it was working and displaying the time.
So I got to work to figure something out.

My goal was to make it smaller, and a bit easier on the wallet for me (The original cost me around 75€ including the case).

after 3 "revisions" of the original circuit, lots of rearranging the parts and testing, some "revisions not working(I was trying to get rid of some relays), I finally I had something I was happy to try and build.

To keep everything in order, I called it CV4 (circuit version 4).

So this is what I came up with:
The idea was, to put everything onto small 5x7cm perfboards and glueing t hem together afterwards.
It was supposed to make throubleshooting and replacing non functional pars a lot easier (which it didn't).
This approach meant, that there would be a lot more wires, and due to spacing problems, I was stuck with the 555 again for timekeeping.

This is what it looked like :

The first finished and working parts:





Signal Converter



finished without display front



finished backside



Complete with case:





With this approach I managed to get one thing done - it wasn't that big anymore.
But it had loard of other problems this time.

With this many wires connecting everything together, failure was normal. It was as sensitive to movement as the original one - so no change there.

The glued together PCBs didn't look very nice.
and again, the 555....
Power consumption was down a bit, but with the same massive superbright sunlight feeling eye-burning display, there was no hope.
I added some features to this, like a better brighness control. but overall, the same story as before. Fragile and unreliable. So I had to make another one.......quickly.

[Spemo]

After some time and 2 more "revisions" of the circuit, CV6 was made. I went back to the 30x18cm PCBs, now packing everything onto a single board.

The idea with this one was, to have a small basic display on the main board, and replaceable otherones to plug in.
Also, the display got it's own "powersupply" with a transistor for every element.
Another addition was an auto dimming function, to reduce the brightness automaticly at night, with a setting for 20:00 or 22:00,, and automatic full brightness at 6 in the morning.
And it got some paint, but see for yourself:

The removable Display:




Frontside



Backside almost finished




With this one, it finally looked like I found a good compromise with everything.
It looked nice, worked as it should (mostly) and finally kept the time, it was off a minute after a week, I was happy with that.

It wasn't flawless though.

The removable display decided to break, I tried to remove it and all the solder pads came off, it was hopeless after that.

the new "powersupply" for the display was very sensitive to creeping voltages, getting all the contacts isolated from each other was impossible, so I gave up and bridged the transistors.

The built in display wasn't perfect either, again with those cheap white LEDs I had left over from the original clock....

the crystal circuit (let's call it clock generator) was still on a seperate board.

And then, with the attempt to make it more modular, and with my stubborn head refusing to finally use proper (and thicker) wires came the cable mess, which had it's own problems like the others before.

But it worked, and I knew, the next CV only needs a little bit of tweaking.....

at least it looked nice for once 

[Spemo]

So, after a while I decided to finally order another set of 110 relays for a second one of the clocks - and started to build CV7.

It finally ironed out all the problems I had with the previous versions, runs reliable and got some nice new features.

A small overview:
- Mainboard has no display built in - small modular display units, easy to replace, easy to remake
- fading LEDs - looks a lot smoother
- night mode now turns off blinking leds and the module for displaying seconds
- removed the option of setting the night mode time
- Crystal clock generator now on mainboard
- simplified wireing on the backside
- lower power consumption (5 watts at full brightness)
- simplified the way of setting the time (only 2 buttons)
- rearranged componet layout for easier wireing
- new brightness dimming circuit


take a look:

front side



back side



in case (in night mode)




This one has proven to be the most reliable yet - no failed leds, no hiccups, nothing. And I really like the fading display, I will make a short clip of this soon.


The only thing it still has is that it's still pretty large, like a big picture you hang on the wall.

The next step is to make it more compact, to fit into smaller spaces.
Also, an alarmclock would be nice aswell.... all I can say is, CV8 is built in blackboard, I'm just waiting for parts.....we'll see.

that's it for now.

have a nice day

Daniel

[digsys]

I applaud your effort .. but you're definitely crazy :-)  Do you intend to make a proper PCB at any stage? Have you considered adding other mechanical noises / circuits for a more "varied" tune? Wondering if the same repeated relay "click" would get boring?

[Spemo]

I applaud your effort .. but you're definitely crazy :-)  Do you intend to make a proper PCB at any stage? Have you considered adding other mechanical noises / circuits for a more "varied" tune? Wondering if the same repeated relay "click" would get boring?

Thank you 
I have considered the proper pcb - but since the relays are so tightly packed, there would be little to no benefit to this.

The noise changes a bit with temperature, and since the components are not matched to each other, each digit sounds different.

Oh, anf after a while, you get used to the noise....

[digsys]

It will be interesting to hear it when you post the clip.

[james_s]

The benefit would be reliability and ease of servicing, also it would make it far easier for someone else to duplicate the project. It's definitely cool but you have to be a masochist to wire so much stuff up by hand. A modular design with PCBs that plug together and no rats nest of wires would solve most of your problems. Also using a 555 as a timebase is hopeless, you'd never be able to get an accurate enough oscillator, that's not what that IC is designed for.

[digsys]

The benefit would be reliability and ease of servicing, also it would make it far easier for someone else to duplicate the project. It's definitely cool but you have to be a masochist to wire so much stuff up by hand. A modular design with PCBs that plug together and no rats nest of wires would solve most of your problems. Also using a 555 as a timebase is hopeless, you'd never be able to get an accurate enough oscillator, that's not what that IC is designed for.

Those were my thoughts ... for the clock, it could be a plug-in, and let the user decide the accuracy, if he wanted to make a more accurate one

[rs20]

I have considered the proper pcb - but since the relays are so tightly packed, there would be little to no benefit to this.

Clearly you don't value free time very highly?   

But seriously though, very impressive project.

[Spemo]

sure, ease of servicing is the other thing.

Honestly, I don't know how to design a proper PCB, and also, I don't know if it would work out - I'm sure I'd still have to solder in a huge amount of jumpers ....look at how bad the wires cross each other



I'm don't know how to route this onto a PCB.

I have a huge schematic for this - maybe someone can walk me through this?


And the thing with the free time - well sure it takes up a lot of time to wire this all up - but that was the goal, to keep me busy 

Thank you for the nice comments, I'm happy that you guys like it

[digsys]

Can you post a link to the schematic, and give part# details of the relays / components used? Maybe someone can help

[rs20]

sure, ease of servicing is the other thing.

Honestly, I don't know how to design a proper PCB, and also, I don't know if it would work out - I'm sure I'd still have to solder in a huge amount of jumpers ....look at how bad the wires cross each other

...

I'm don't know how to route this onto a PCB.

I have a huge schematic for this - maybe someone can walk me through this?

So don't let me tell you how to live your life, I'm just foolin' around. But if I wanted to do something like this, this is one of those situations where autorouters* would be great. The clock speeds are low so there's really no such thing as a "poor PCB layout".

So I'd capture the design in KiCad (or, if I already had the schematic like you, just use the corresponding netlist I suppose), lay out the components in the PCB editor, and then figure out how to use an autorouter (either the one in KiCad or some third party situation.)

You'd be surprised at the complex connectivity you can achieve with a 4- or even 2-layer PCB, especially when the relays you're using have such an incredibly huge pin pitch compared to modern components.

Having said all that though, this is probably not the best "Learning how to make PCBs 101" project. At all. You'd want to gain experience doing a simple, 2-relay, manually routed board first just so that you're not completely confounded when dealing with hundreds of parts.

* A feature of PCB tools which lets you place all the components wherever you like, and it figures out how to lay out all the traces for you.  Also, a feature which is widely maligned as cheap way out, although I think there are still niche situations where they are applicable.

[Henrik_V]

A trick I was told by the son of Konrad Zuse (Build the first relay computer Z3) that his father used to extend relay contact livetime (at that time) and bouncing contact problems:
He tried to switch as much relays as possible without power at the contacts.

[schmitt trigger]

When I see such a complex project completed and operational, all I can say is: "I am not worthy of tying your shoe laces!" 

Seriously, my sincerest congratulations and respects. 

One comment about your video: it seems that there are sudden brightness jumps on the LEDs. Having attempted to video record LED displays, I know that sometimes the camera's auto-exposure function gets confused by the extreme contrast and will cause this.
Are the brightness jumps real, or is it a camera exposure artifact?

[edavid]

Hey OP, how about posting schematics of your counter circuits?

I think I would have been tempted to start with a regular MCU based clock and add some software controlled relays as noisemakers - but I'm lazy.

[JVR]

Damn. I wish I had the time to do this kind of stuff. Well done on getting that ratsnest to operate, even if it was a little flaky in the first revisions.

[SiliconWizard]

Given there would be so many ways to get the same result and appearance, or pretty similar, with much simpler appraoches, I'm baffled how the OP got the courage and patience to do this.

[Spemo]

Hey,
Schematics will be posted - once I'm done remaking then - to maybe make them more understandable. I hope I have it done this night.

When I see such a complex project completed and operational, all I can say is: "I am not worthy of tying your shoe laces!" 

Seriously, my sincerest congratulations and respects. 

One comment about your video: it seems that there are sudden brightness jumps on the LEDs. Having attempted to video record LED displays, I know that sometimes the camera's auto-exposure function gets confused by the extreme contrast and will cause this.
Are the brightness jumps real, or is it a camera exposure artifact?

Thank you!
The brightness jumps are intentional - they are different brightness settings. This got removed in later versions.

I don't have problems with exposure there - the bench is extremely well lit, one day I found some LED panels with 512 high power SMD LEDs... at the moment I have 6 of them lighting up everything, 2 additional smaller lights (40 watts each) in the background. powerconsumption for all of this: 420 Watts  But it's the only way to get my phone to shoot some usable footage.

[Spemo]

As promised, here is the "Schematic" I use to build this monster - maybe one of you can find some use with this.

Have a nice day

Daniel

     

[HB9EVI]

remembers me a bit my Z80 board with all that wild wiring, but it's still small compared to your clock.
I guess I can imagine the thrill letting the idea for it grow in your mind and make it real, step by step, although I wouldn't do it myself - at least not a rather trivial thing like a clock - like already said by someone else: too lazy for it; I appreciate a RTC in an STM32 with a small TFT - but I admit - there's no challenge down that road.

[iMo]

Fantastic project!!
PS: do you plan a wrist version of the relay watch? 

[Spemo]

Fantastic project!!
PS: do you plan a wrist version of the relay watch? 

sure, why not!
3 Kilograms in weight, 50000mAh Batterypack to get through the day, and a massive 35x22cm case on your arm, does that sound ok for you?
Price would be 1599,99$...you know, because of the wrist strap.

[rs20]

What program did you do the schematic in?

[German_EE]

You're crazy, but it's a nice kind of crazy that happens to engineers when they get an itch that needs scratching. One question though, where did you get those large sheets of perfboard?