Interesting, thanks for the heads up. Our test area is enclosed, so I might try some low speed tests, but I will have to order water jet cut metal parts then.

We posted two videos on our (open source) GUI. I think some will find the second video especially interesting due to the scripting abilities. We hope in the future to integrate dynamic test with the scripts.

https://www.youtube.com/watch?v=eA0wLIQhi70
https://www.youtube.com/watch?v=2CYWFWXhhQs

I implemented an optical RPM sensor. It allows very precise measurements even during acceleration. My CD inertia wheel shipped yesterday. We have our dyno enclosed, so it can explode without risks... I am planning to install multiple reflective tapes per turn to obtain high accuracy and precision on the rotation.

While doing this implementation, I improved the precision of the electrical RPM sensor quite a bit. It was accurate before, but noisy. Now, the sensor is super precise, with pretty much no delay. At 20kRPM, the interrupt triggers every 0.0015 seconds, and the sensor measures the time between interrupts with 0.000004s accuracy. I am using the interrupt with timer 0 on the Atmega 328. The signal is very smooth. The following image is the graph of the two sensors. There is no low pass filters.


http://i.imgur.com/Gj1amND.png

Anyhow, I think all of this will be important for dynamic tests. You really need a precise and accurate RPM measurement.

We also have a sales for cyber Monday, 150$ off.

Out of curiosity, can you explain what you had to do to improve the smoothness? Hardware changes, or software changes?

Sounds like a separate optical RPM sensor isn't necessary, if it's matching up so nice!

Software. We were counting the number of interrupts in a fixed amount of time. Consequently, if the real value was 20.5 interrupts per unit of time, we were counting 20 half the time and 21 the other half. This process was quick, but a bit noisy. Now, each time the interrupt is triggered, the triggered time goes into a rolling buffer. Each time the board can send the speed information, the time to complete the last three interrupt is used to compute the frequency.

An optical sensor might not be necessary, but:

It will allow greater angular resolution, as there can be multiple optical tapes per turns. That could be useful during acceleration.
It can be used on brushed motors.
We can be certain that we don't miss a rotation just after startup.

Isn't it possible to use the sensor of the motor?

I am using it for measuring the FDR:

http://www.euronet.nl/users/tooms/fdr/sensorsignal.jpg

https://www.youtube.com/watch?v=aw1Gb_VJo-Y

Yeah, I was going to say, there's even three of them! But I think they want to support sensorless flyer motors. ;)

Yes! However, those sensors could also be used. They output a square wave at 5V right? The board detects the square wave from the the optical sensor.

We are trying to gather information for what would make the ideal dyno for racing. I think an optical sensor would be best for those racing with brushed motors. The electrical rating of the dyno would have to be the same as for the most performant ESCs.

I think our scripting system would be great to do more advanced tests. For example, it would be possible to simulate a race with a flywheel, and objectively compare batteries for that extra "kick" at the end of the race.

Is anyone still racing with brushed motors? I don't think so, and I'd suspect that if I'm wrong, the racers still running brushed motors probably wouldn't really bother with a dyno. :)

It's digital, but not necessarily square. You'll need a pull-up resistor.

For sensored motors, you'll be fine if you use one of the Hall sensor outputs. But if you want to have more than one pulse per motor revolution to get more data points during acceleration, stick with the optical sensor. While it's true the Hall sensors can be processed to give six pulses per revolution, those pulses will not be evenly spaced due to inaccuracies in sensor placement and rotor magnetization. It's pretty common to see a 3 degree error, which is 5% of the 60 degree steps between commutation. That's a very large error!

There are also plenty of racing classes using sensorless motors, and they will need the optical sensor.

Thanks for the info, I thought there were still classes for brushed motors...

There are but most of them are stock classes where motors are sealed. Just mount and run them.