HelmutVonAutobahn wrote:DelSolid:
How can you quote the PLX at 36ms with all that noise? Do the widebands get credit for noise amplitude, when calculating the responses ?
I know that the 30-0300, Zeitronics, and MTX-L/LC-2 are all pretty quiet, noise-wise. The others ? Especially PLX ?
You bring up a really important issue. The noise levels of the different controllers were all over the place. Some were really good and some were so bad as to be unusable (IMO), despite herculean efforts to give the best possible power, ground and capacitance setups to the devices under test.
Ultimately the filter applied was decided upon by the independent testing firm. I personally do not like their choice because they decided to use a median filter on all the units, which is possible to do in a post processing action on an existing data set, and it shifts the data the least, but is not something you can do in real time which is how the analog output data is used for feedback. Our units would have fared better (actually, others would have fared worse, take your pick on wording) if we ran without any analog output filtering at all and enforced a strict t63 threshold rule (time only stops when you cross the t63 threshold and
stay across it). But some of the sensor outputs were so noisy that we literally could not have gotten a repeatable t63 time from them and people would be all angry. they would be accusing us of all sorts of shenanigans, So the filter was added along with the 8x test averages to give repeatable values and throw the one-time-wonders out.
But just remember, to keep things simple, this test is just measuring the response time of the sensors, not their accuracy, signal clarity or anything else. We wanted to keep the message simple.
