ASIO and Hypermiling THD

Hypermiling is the act of driving a car with extreme efficiency so as to maximize the miles per gallon achieved. It requires a careful understanding of the engineering of the vehicle: Where is the car most efficient, how much energy can be captured via regeneration for example. But in the end, it allows drives to achieve range figures far in excess of published figures for MPG. 

Is it useful, or just a curiosity? It's hard to say. Probably a little of both. Hold that thought for a moment...

The next release (release 1.52) of the QA401 software will start laying the groundwork for an experimental option to shift in ASIO mode. The way this will work is that you can use the QA401 as you normally do, but when you need ASIO operation, you select that option from the Experimental menu. The FX2 USB controller is then re-programmed and the QA401 drops off and re-enumerates itself as a new device. It's no longer visible to the QA401 application, but once you have installed the ASIO drives (which will be another release down the road) you can then use the QA401 as an audio in/out box for any application that supports ASIO. If you want to switch back, just re-plug the QA401 and it's back to normal.

ASIO in Development and Validation 

For development, ASIO is a handy way to cross-check the maths used in the QA401 application. For example, below is a shot of the QA401 hardware being used with ARTA software in loopback mode. 

This shows a THD achieved of 0.00015% (-116.5 dB) and a THD+N of 0.0030% (-90.5 dB). 

The same experiment is repeated with the QA401 software with the following result:

This shows a THD of -116.7 dB and a THD+N of -90.9 dB. Putting the results in a table:

THD -116.5 -116.7
THD+N -90.5 -90.9


Thus, for this particular measurement, there is very good agreement between the two applications. 

How does this tie into hypermiling? And why is the measurement above being made on the right channel? And why aren't the window sizes the same? Glad you asked.

As a side study to related to the comparison above, we went down the rabbit-hole of trying to understand the limits of THD on the QA401 hardware, and what the various settings might mean to measurements. 

First, there is a measurable difference between right and left channels--several dB can be observed in fact. And there's an equal amount in play from the output channels. 

The table is above is interesting because it makes it clear that while output R- looks to be exceptionally free of distortion when going into R+, it looks unremarkable when going into L+. Notice the right channel has a worse average THD, but the combination of R- out to R+ in was a great starting point. 

And this is how we got onto the topic of hypermiling. What if our goal was to squeeze everything we could from the platform? What kind of THD figures could be had?

The above shows a staggering -120.1 dB THD. An a particular unit, it can be easily replicated over and over with some tweaking as the measurement is sensitive to the position on the desk. Note that rectangular windowing was used, as Hanning and other windowing degraded the measurement by a dB or so. The 128K FFT window wasn't as important, but when aiming for bedrock, why not? And of course, the right channel on this particular unit excels. 

Wrapping Up

Like hypemiling, the utility of measurements at this level is marginal. You generally want to be a long ways away from the limits of your equipment when making measurements. For example, If you want to measure millivolts, your equipment needs to be able to resolve hundreds of microvolts. 

Humans have always been intrigued with limits, and this is no different. Take some time to learn the limits of your equipment. The quality of your measurements will improve, as will the confidence.