When designing a circuit on the bench, sometimes things work far better than they do in real life. [Quinn Dunki] learned this lesson over the last few months as she struggled with one of her recent creations, the Dish-o-Tron 6000. We featured the Dish-o-Tron back in April, and at that point things seemed to be working out well for [Quinn]. As time passed however, she found the device to be an unreliable power hog. Aside from eating through a battery every few weeks, it kept spontaneously switching states from ‘Dirty’ to ‘Clean’ and back. It was time to take the Dish-o-Tron back to the bench for some debugging.
The random status flip from ‘Dirty’ to ‘Clean’ was a relatively easy fix, and required a small capacitor between the set pin and ground to eliminate the electrical noise that was tripping things up. She nailed down the spontaneous ‘Clean’ to ‘Dirty’ flip to a stuck tilt switch, which she swapped out for a mercury-based model, making things far more reliable. She solved her battery problems by wiring in a 12v wall wart, which might not be any more energy efficient, but it does save her from swapping out batteries all the time.
It’s always nice to see how projects evolve over time, and how the inevitable bugs are worked out of an initial design.
Thanks for the post, HaD!
The funnest part of this was ordering “illicit” mercury switches from China. I felt so naughty.
I can see that, re: the mercury switches. Then again, just about every home/apartment from the mid-90’s and earlier are all probably still using mercury switched thermostats unless they have been swapped out.
It really makes me wish I had held on to the old stat when I replaced it with a digital model at home. I feel worse for tossing one in the trash than I would for ordering one!
Mercury-based components should only be used as a last resort, IMO. Although you can’t get rid of mercury because it’s an element, improper disposal of mercury-containing components (and mercury switches contain a lot compared to, say, CFLs) is a real problem–as is, by the way, their production in places without effective environmental policies.
So yeah, it’s nice it works, but the mercury switch could have been replaced by far better alternatives. There are a ton of ways to do this, from the obligatory accelerometer to a magnet on a spring switching a hall-effect sensor to optical floor (i.e. door opened) detection. You just have to be creative.
tilt ball switches are available if people only bothered to look.
Digi-key is your friend.
I second this. The tilt ball switch is obvious : Mouser
Thanks for the suggestions, guys. I think my reply may have been lost in the noise below. There’s a section in the article about rolling ball tilt switches. I used one in the original design, and found they weren’t suitable. If you’re interested in the details, dig into the post a bit, where I lay out the tests I ran on them, and the shortcomings I found.
Depends on your application. The biggest reason to use a mercury switch is that they have the lovely property of not bouncing and being way less sensitive to vibration. Depending on the design of a ball tilt switch and the accompanying circuitry these can be real problems.
I also found that the little ball is remarkably prone to getting stuck in the open or closed position (details are in the article on how I determined this was happening).
kudos to you Quinn. Way to stand up to the overbearing nanny state. As long as you don’t break a mercury switch open they pose no threat to humans, pets, or the environment.
Mercury switches are dead simple and damn reliable. There is no replacement for them that can beat these qualities.
Suck it eco weenies!!!
I agree mercury switches are a last resort. I considered (and tested) many options before falling back to using one. Space is very tight in the case, and I did not want to redesign everything around that switch. There’s no microcontroller here, and no room for more passives, so it needed to be a drop-in replacement for a mechanical rolling-ball tilt switch the size of a couple grains of rice. After much scrutiny, I was forced to reject all my other ideas.
fartface, there’s a section in the article that talks about rolling ball tilt switches. They were used in the original design, and found to be poorly suited for this application.
My grandmother “invented” something similar for her dishwasher. She made a little magnetic holder that stuck to the door of the dishwasher, and a little sign which is inserted into it. It was decorated with cross-stitching, and very lovely.
Operation is simple: When you start the dishwasher, you turn the sign from “dirty” to “clean.” And after you completely unload the dishwasher, you turn the sign from “clean” to “dirty.”
There is a bug in her system wherein the sign says “clean” but the machine is still running, but it doesn’t take a rocket surgeon to XOR those conditions and determine the actual state of things…
This is all great, but what does the Dish-o-tron due? Does it wash Dishes? It would have been nice to breifly recape it’s function for us new comers.
Apologies if this wasn’t clear, Oscar. There’s a link in the opening blurb to the original device:
http://quinndunki.com/blondihacks/?p=92
I should make that clearer, I guess.
I like the “EXTREME” sticker.
+1
No wonder the batteries got chewed so fast.
#1: She used a 9V battery, which doesn’t really hold much of a charge.
#2: 555, a highly inefficient IC.
Also, according to the schematic diagram, this thing pulls the power supply low when CLEAN LED is lit. And she used potentiometers as resistors. What is the point of that?
This thing could have been done much simpler with a simple microcontroller (like ATTiny13) and 3 AA batteries.
Well, when lit, the Clean LED is connected to ground through a 5K resistor, drawing about 15mA, yes. I’m not sure I’m following your critique on that part.
The potentiometers are for adjusting the length of the door-open unloading timer, and the sensitivity of the thermistor voltage divider. Those are both real-world variable conditions that are difficult to predict at the design stage, so it’s to be able to adjust them later on.
A microcontroller would have been another way to go, but I chose to use 555s because it was a fun design exercise to do so.
Ah, I see what you’re referring to now- the collector-emitter path on the transistor there. That’s an error in the schematic- the circuit isn’t actually built that way (luckily, since that transistor would last about 15 seconds hooked up that way :) ). Thanks for catching it, I’ll make the correction.
My apologies. I am new to Blogging. I should have noticed the link, it’s very clear. I didn’t bother to think it was related to the article. Apparently some forums utilize linking in the same manner, but for advertising. Thanks for clearing that up for me.
Regarding mercury tilt switches, I still have some left over from playing with a bunch back in the early 90s, I made a couple of rudimentary tilt switch joysticks for gaming on my Amiga.
One had the switches packed into the handle of a busted joystick, another was a balsa-wood square steering-wheel affair. They were no match for true microswtich joysticks but a nice novelty.
However one bad aspect of them I discovered was that a tiny amount of mercury would get stuck on the metal pins inside the glass bead, meaning the switch became electronically ‘sticky’ as the mercury blob got stuck to the pins too easily and wouldn’t let go.
Some switches had this behaviour from the start and others developed it over time.
I also have a couple of 1978 games with a globule of the stuff in, Mercury Maze, I can remember someone dropped & broke one in school (in the 80s) and the corridor was blocked off whilst the offending metal was cleaned up – the game was subsequently banned from being brought into school from then on.
Version 3 should use an arduino and an accelerometer and have the ability to communicate over the net. The accelerometer will eliminate the tilt switch / mercury debate and the internet capability means being able to tell your kids to start the dishwasher.
There should be a separate input to monitor the saponifier (jet dry etc.) and report it when it is low and more needs to be ordered.
Or…you could just buy a good dishwasher that doesn’t require pre-rinsing everything before putting it in the dishwasher, then you could just open the door and see the dirty dishes…problem solved.
I like where your head is at, I won’t lie. ;-) A similar feature set may find its way into the next version (Dish-O-Tron Extreme Pro?)
I’d love a better dishwasher, but sadly I’m renting, so I gotta hack what I have to make it suck a little less.
When the dishwasher is empty, put soap in and close the lid.
If the lid is closed, the dishes in there are dirty.
When you run the dishwasher, the lid opens (releasing the soap).
After the dishwasher is empty (don’t put dirty dishes in, but taking clean ones out a few at a time is fine) reload the soap.
Works 100% of the time.
Great mod/hack! Its good to see some old technology still being used too (mercury switches). Can’t go wrong with KISS!
As a kid I use to break these open and play with the mercury. ;) I actually remember spilling a large vial of it in high school and the teacher was mad because of the price, not the contamination. Today if you break a thermometer, they shut the facility down and call the hazmat team.
Anyhow, thanks for sharing this project! Keep up the cool work!
The light should be green, not blue. “Light goes green, dishes clean”
Quinn,
Would there be room for a roll-your-own tilt switch? I’m thinking a small sheet-metal vane with a small hole drilled through it (acting as a pivot point) hung on a straight pin that is carefully epoxied to the top of an optical interrupter (scavenged from an old ball-type mouse or CD-ROM drive). When vertical, the vane would hang down between the legs of the interrupter, blocking the light. When tipped, the vane would hang to the side, permitting the light to pass through.
Or you could just be sure to take the mercury switch to a recycling center when you’re done with it.
I ask because I’ve been thinking about how to construct my own garage door open/close sensor. I’d been considering a commercial tilt ball sensor, but garage doors vibrate a lot and are in a harsh temperature, humidity, and dirty environment. It’s worse than your dishwasher, and you’re having problems, so I didn’t really want to replicate your entire journey. I just want something really simple that I can seal in a tiny dust-proof and spider-proof plastic container.
That’s definitely something I considered. Even a weighted contact hinged over a stationary one would be enough. There just wasn’t room, I’m afraid. The space is about the size of three grains of rice stacked up. The rolling ball switch was wedged in, to give you an idea.
Honestly, the rolling ball switch might work better in your case, because the problem was actually the smoothness of the dishwasher door. The ball would get wedged from sitting in one place, and the door’s motion is too smooth to dislodge it. A vibrating, shaking garage door may not have that problem. However, these switches have a lot of bounce also. The ball rattles around before settling into position. That may be an issue, depending on the circuit design.
For a garage door, though, I would think some sort of IR beam-break detector would work well? Like the kind garage door openers come with as safety stops?
Or one of those hall effect sensors that alarm systems use to detect open doors? Those are cheap.
One possible solution would be to attach a magnet to a spring and mount it in a tube so that when the door is closed, the magnet will be in position 1, pulled down by gravity. When the door is opened, the force gravity exerts on the magnet to counteract the spring will decrease, so the magnet is pushed (or pulled, depending on the spring) to position 2, where a hall-effect sensor or reed switch etc. is situated, triggering the “door open” signal.
“Quinn Dunki” is that your real name?