Electric pressure gauges

Earlier this year I made a load average meter out of an old ammeter. But I've decided I want to use some even older technology, so I've been refitting some pneumatic pressure gauges so I can display information with them.

I got a load of old gauges from Empire Exchange, the little Mancunian goldmine that supplied the ammeter. Opening these reveals an arc-shaped copper pipe which deforms when inflated, driving a lever and gear which moves the needle. Now, ideally, I would control these using a computer-controlled variable pressure pneumatic source, but I don't have pneumatic mains in my house (not at 140 bar, anyway), so I've had to go for an electro-mechanical solution. It's a shame to cut the proper workings out of them, but I don't think they would have gotten any better life in their present form.

A servo motor is the obvious choice to get a controlled rotational movement. However, most servos only have a range of 180°. I need a range of 270° to use this meter properly. You can get servos with a greater than 180° range but they are quite rare, especially in the small sizes I need to fit into the gauge's body.

In order to get the 270° range I've used a rudimentary belt drive. I made a pulley for the gauge's output shaft by clamping the shaft into a dremel and using it as a lathe to turn down a larger plastic pulley to the right diameter.

The servo's output shaft is about 5.6mm diameter, so I made a pulley as close to 4.2mm as I could get it for a 1:1.5 enlargement. The pulley I made is rather rough, but no-one will see it in the finished product.

I've attached the servo to the gauge body with Araldite, and used polyester sewing thread as a belt, with two turns around both pulleys. The belt is tied to the servo motor's horn and fixed with glue at that point. Ideally, it would also be tied to the output shaft, but the tiny scale of the components make this kind of engineering impossible for me.

A real engineer would use a toothed belt or a gear drive for this - or make their own servo motor. But for a few hours' engineering in my kitchen, I'm happy with this solution.

I've tested this with a remote control system and it works very well. The biggest concern is slippage of the belt, but the exercise I've given it so far shows that it's actually perfectly repeatable - it doesn't slip at all.

The next steps are to replace the scale backplate with something like a 'load average' or 'spare memory' scale, and to rig up a servo motor driver so I can control it from a PC. Servos are very easily to control, just needing a PWM signal - a little easier than pneumatic signals in this day.

If you want to know why I bother to do this, you are probably not my target audience. :)

Opening a WRT150N

Only a tiny number of people will be interested in this, but I recently had to open the case of a Linksys WRT150N router and found a few questions on the web asking how to do so. Doing this voids your warranty, of course.

First of all, find a T10 security torx driver. These are like standard Torx bits with a hole in the middle.

Start by flipping the router over and levering up the silver panel at the rear of the router. I'm calling the rear the side which the antennae are attached to. The rear half of this panel will pop out of the body, but the front is attached with hooks which will break if you try to lever up the front half first. Once the rear half is free, the front half will slide off easily.

This will expose four security torx screws. Remove these.


Now you can remove the top and side panels. Pull one of the side panels and its side of the top panel away from the router at the same time. The hooks which hold the side panels on should pop open one at a time without damaging anything.



Repeat this with the other side panel. You should now have a completely disassembled WRT150N, so you can get at the serial port connections you want to attach to.

Tea-making with gas and electricity

B3ta recently asked whether it was cheaper to make a cup of tea using an electric kettle or a gas one. My gut feeling on this was that gas was cheaper, because gas is a lot cheaper than electricity per unit energy. I've done an experiment to see which is actually better.


Starting with gas, I put a mugful of water in a steel saucepan and heated it over a low flame. I can measure the amount of gas used using my gas meter. I've completely shut off my central heating boiler, including the pilot light, so no other gas is used during the experiment.

This took 0.0092 cubic metres of gas to bring it to the boil. I'm calculating the energy and cost based on my last gas bill, which explains how to convert from cubic metres to kWh - ultimately by multiplying by 11.0786. So I've used 0.102 kWh. I pay 2.88p per kWh for gas. USwitch.com reckons I could get gas for 2.86p from another supplier, so I'll use that. That makes a total of 0.29p.

Next I set up an electric kettle with a plug-in power meter. I've got to use half a litre of water to cover the element and make up the 'minimum' mark on the kettle, which is slightly more than the mugful I boiled in the saucepan. To boil this took 0.06kWh. The cheapest electricity I could find on USwitch was 11.21p/kWh, so this cost me 0.67p. Even given the inaccurate measurements in this experiment, gas is the clear winner.

But what if I've got friends round? I did a test with a large teapot full of water as well. This time, the saucepan took .281kWh for 0.82p, and the kettle took .140kwH for 1.57p. Still nearly double the price of the gas option.

Boiling speed is another factor - the gas saucepan takes a lot longer to boil than the electric kettle. I did some more experiments, using the gas turned up higher than I normally would. I expected this would waste gas, because a lot of heat escape round the side of the saucepan. However, it was barely any extra. There's probably a 'sweet spot' for the gas flow - too high would waste gas, and too low would take so long to boil that heat gets a chance to escape from the saucepan. Still, I couldn't get it as quick as the kettle. The kettle isn't a fast boil one - it's about 2.2kW. Even so, in all the experiments I measured time for, the kettle is twice as quick. So for the extra third of a penny it costs to boil the electric kettle, and the convenience of having an automatic off switch, I may well stick with that for the time being.

You'll notice that the kettle uses less energy despite costing more. Does this mean the kettle is more efficient? Not necessarily. The kettle is more efficient in itself than the stove and saucepan, but a lot of the electricity comes from gas in the first place. About 36% of our electricity comes from gas-fired power stations, and the best stations are about 58% efficient. Another 37% comes from coal, but I haven't found efficiency figures for those yet. If our electricity came entirely from gas, the kettle would use about the same amount of energy as the stove. To do a fair comparison you would also need to know how efficient the national electricity grid is, and how much energy is used in pumping gas around the country.

http://www.berr.gov.uk/files/file39569.pdf
http://www.the-eic.com/News/Archive/2006/Nov/Article2799.htm

The data in full:

Fuel	Quantity	Speed	Time (s)	Volume of gas	kWh	cost/kWh (p)	Cost (p)
Gas	1 cup	slow		0.0092	0.102	2.86	0.29
Gas	1 cup	fast	192	0.0100	0.111	2.86	0.32
Gas	1 teapot	slow	1376	0.0254	0.281	2.86	0.8
Gas	1 teapot	fast	506	0.0258	0.286	2.86	0.82
Electric	1 cup	n/a			0.060	11.21	0.67
Electric	1 teapot	n/a	213		0.140	11.21	1.57

The by-products of potato juice

Today I have mostly been performing silly experiments in the kitchen. I like potato pancakes, but they are problematic to make because the moisture has to be removed from the potatoes before frying them. This usually involves dirtying a clean cloth, or trying to squeeze the moisture out by hand which doesn't work well. I've also tried using a ricer on raw grated potato which is quite effective.

I had a brainwave today - realising that I already had a device which does this job, namely a juicer. I pushed three potatoes through the juicer, and kept the pulp that normally gets discarded. Mixed with some chopped onion and seasoning, this makes really good, fine potato pancakes. I wasn't brave enough to try drinking the potato juice.

An earlier experiment to try separating freshly brewed coffee from its grounds using the juicer didn't work so well. It just tastes like soil.

12 volt mains

If you like gadgets as much as me you probably have a large number of AC adaptors with power bricks cluttering your house up. A lot of these have different output voltages, but I have quite a few that deliver 12 volts, or claim to. Is it possible to replace some of these with a single transformer to simplify the wiring and increase efficiency?

The experiments I've done so far suggest that the first part is quite possible, but surprisingly, I couldn't run my appliances as efficiently with a single power supply than with individual bricks.

The appliances I used for this were my Netgear ADSL Wifi router, my 20" LCD TV, and my mini-itx based server computer.

The first step was to give each a common connector. I split the cable on each and attached a PC drive power connector on each side. Drive power connectors are cheap, meant specifically for 12V and can be plugged straight into an ATX power supply.

Having done that I reconnected them to their original power supplies and switched them all on, checking the power consumption for each. Measured at the mains socket with a plug-in power meter, I got: PC on idle: 28W. TV: 50W. and the router: 7W. Total 85W. Incidentally, if you repeat this experiment, it might be better to plug all those into a power strip and plug that into the power meter, in case the meter has some non-linear error.

Using the first attempt - an "L&W" 250W power supply used 48W just to run the router and PC - obviously less efficient. The TV wouldn't work properly at all - I got sound but no picture. A quick check revealed that the 12V output of this power supply was actually only raising 11.05V unloaded - which presumably is not enough to power the TV properly. The PSU also used 12W without any load at all.

I tried again with an older Opus PSU which had the same problem powering the TV. It also used 8W unloaded.

Next was a Morex PSU, only a 200W model for a micro-ATX system but this was evidently of better quality and managed to get the TV running. However, just the TV and PC drew 92W from the mains - over the limit before the router was connected. It also drew 11W when idle.

The last test was my bench PSU. This is a bit of a beast, rated at 40A continuous, but given its higher cost I thought it might stand a chance of being more efficient. No such luck though - with all three appliances it drew 110W, consuming 18W unloaded.

So, while it might make the wiring neater if done properly, I've yet to find a power supply which will do the job more efficiently than using individual power blocks.

I suspect that computer PSUs are not the best choice for this job, as they are probably geared towards delivering a very stable smooth 5V and 12V rather than efficiency. The power bricks are not, so far as I can tell, very stable in their output voltage, probably relying on further conditioning inside the appliance.

As usual, I wouldn't recommend anyone else trying this for themselves - running things on too low a voltage could damage them, and not all PSUs output the voltage they claim - the router's power brick, for example, delivers 16V when unloaded.

Asus Eee 900

Hooray! Despite making me jump through a few hoops to verify my identity, Dabs came through and delivered my Eee 900 today. I'm posting from it right now - the keyboard will take a bit of getting used to but it's a lot easier to type on than my previous UMPC. It'd be hard to give a negative review to something I've just spent 330 quid on but it is very good so far. It's a lot better built than I'd been expecting - it may look like a toy but it's very solid. Next step - apt-get emacs and pygame.

Toaster video

I've been playing with my digital camera, trying to make a better video for the turbo toaster. The results aren't perfect, but still better than I expected for about an hour's work. The title and fade in was done with Kdenlive, which I found really easy to use. Pity that it doesn't seem to anti-alias the text properly when fading to the video clip. I brought up the colour a bit using mencoder with a few filter options. Anyway, there you go, flying toast in full motion video.

From the archives

I've uploaded a few of my older posts from my wordpress installation - the LED message board modifications, the event report from Robot Wars 2003, and the Turbo toaster.

Load Average Meter

I found this old milliameter in Empire Exchange in Manchester. It was built sometime in the 1960s. I then printed out a new scale for load average, turned the original scale plate around and glued my new scale onto the back, so as not to ruin the original one.

I've used a USB interface board from Maplin to drive it. The interface board has open collector PWM inputs, but the meter needs a bit more power than the board can supply - about 100mA on its smallest scale to go full scale. To get power, I've drawn off a drive power connector from the motherboard. The last component is a small perl script that does a bit of maths to convert the computer's current load average into a duty cycle.

Moving from Wordpress

While I'm very impressed with Wordpress, maintaining my own installation is taking up a bit more time than I'd like. I don't write posts very often and Wordpress seem to come out with a new version more often than I blog. I've decided to move over to Blogger so Google can take care of the service. I'll either write a script to keep them in sync or move srimech.com over to a republishing system.