From: Paul
Date: Sat, 30 Dec 2000 10:13:08 +0000
Subject: Re: [TSSP] NSVPI - Latter Results
Terry, I'm now beginning to process (again) the NSVPI results, both the new toroided data, and the existing bare coil data. I noticed something that you may find satisfying, concerning the 'jitter' in your measured gradient readings. When I printed out the 'predicted' curve on the original graphs, the gradient is calculated on a turn-by-turn basis, which gives a nice smooth curve. Against this smooth curve, your measurements were up and down, see http://www.abelian.demon.co.uk/tssp/pn2510/nsvpi2210b.gif for example. I decided to try comparing the data differently. Instead of printing out the predicted curve turn-by-turn, I calculated the voltage differences at 1" intervals, since that is what you were measuring. Unfortunately, 30 into 1000 doesn't go - in order to apply the probe you have to select the nearest turn. The result is a jitter of +/- 1 turn at each end, a possible variation of +/- 2 turns in an average 33.33 turns per inch. If I calculate the gradient in 1" steps, the calculation has to pick the nearest turn in the same way as your probe did, and guess what - the same jitter appears, and at the same amplitude. I've compared your original bare coil measurements with my 1" gradient predictions in http://www.abelian.demon.co.uk/tssp/pn2510/nsvpi-jitter.gif The conclusion is clear - your NSVPI measurements are probably much more accurate than previously supposed and the apparent jitter is caused by the necessity to 'quantise' to the nearest turn. Cheers, -- Paul Nicholson, Manchester, UK. --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.