By Peter Targett

A couple of years ago Model Engineer published information on a dynamometer car using the strain gauge from a cheap digital scale and an Arduino processor module. This was put on the ‘may do that someday’ project list. Supercheap Auto had recently advertised a digital scale for around $12, time to consider the project (Photo1).

These scales measure up to 50kg or 110lb the theoretical draw-bar pull for my 7.25” gauge 14xx 'Dart' is 76lbs so I thought that if other larger 7.25” gauge locos wanted to use it I would have to use 2 of these strain gauges in parallel and able to measure up to100Kgs or 220lbs pull. It may be the case that one of these strain gauges could well measure 100kgs with the correct scaling components in the measuring circuit, but I wasn't going to risk it.

The two strain gauges removed from the scales are connected to a pair of blocks, one end to each. The blocks are attached to each end of the draw-bar (Photo 2
). The load on the ends of the draw bar cause the strain gauge to distort and the resistance value of the strain gauge then varies proportionally to load. (The two bolts going through the blocks are there just in case the strain gauges fail as the strain gauges are only 12mm wide by 3mm thick.).

The basis for a dynamometer is to measure draw-bar pull, time and distance. With strain gauge based draw-bar, a micro-controller can measure time and a hall effect sensor with a magnet on a wheel can count the number of revolutions

of a wheel and distance then computed.

The ME article used an Arduino as the basis for their solution and the final circuitry ended up being a bit cumbersome. Arduino's are good for quick development, but they are overkill as there are other bits you just don't need and many of the useful processor functions can be hidden from the user. I prefer to use micro-controllers directly rather the  repackaged Arduino. In my view micro-controllers are the computing equivalent to Meccano and Arduino's Lego.

Unlike the unit in the ME article the intention is to mount the draw-bar and circuitry in the same box. This means keeping the component count down and the only external connections being the power supply and the wheel sensor. The added advantage being that any driving car can be used for testing, as long as it has a wheel sensor. If draw bar pull is only to be measured then the wheel sensor can be discarded.

To control the unit I'm using an Android based mobile phone, or tablet, connected to the unit via blue-tooth. It doesn't need an expensive phone and the software is just a terminal application available free from the Google Play app store. Basically a terminal emulator acting like a 1980s green screen computer terminal.

Like the ME article the signal from the strain gauge needs to be amplified to a level compatible with the analogue to digital converter (ADC) input of the micro-controller. I'm using two as I have two strain gauges. So the circuit has two amplifier chips, a micro controller and a bluetooth module (only about $10 on the internet) for the connection to the phone.

In Photo 3 is the main circuit with the three chips. The unit with the red light is the blue-tooth module and the unit with the green light is a pulse generator simulating the wheel sensor. In fact its a simulator for RCxel electronic ignition systems. With the system set up for a 4” wheel the simulator can provide pulses equivalent to speeds from 5mph to 160mph, more than enough for my testing!

The screen output is sent to the mobile phone in real time, every second, with the micro-controller calculating speed, average speed over the whole run, current draw-bar pull, current work, total work and current power. This can be displayed in imperial or SI units. (Photo 4 - click to enlarge).

Logging the data for offline analysis can be done with the phone app, terminal emulator log function with the log file stored on the mobile phone storage card for later transfer to a PC. It could even be emailed. For the data logging function there is a raw data mode in which every second the unit just sends the total number of wheel revolutions and current draw-bar pull, so providing the basic data for offline analysis. At the moment analysis and graphing will be done with an Excel spreadsheet. The free Gnuplot application could be used, or eventually, a stand alone application created for the phone or PC.

The table shows the first 30 seconds of a log file in CSV format to be read directly by Excel. The four values being Time hrs:mins:secs, total wheel revs, current draw-bar pull and the draw-bar units.

Next step is to get it all packaged up and some testing done at the track. It would be good to know what our locos can really do.

This article first appeared in Blast Pipe, the magazine of the Hutt Valley MES, New Zealand. A second article describes the technology used as a weighbridge.