From: Paul
Date: Wed, 27 Mar 2002 11:41:27 +0000
Subject: Re: Ready :-)) - Re: [TSSP] short H/D and stuff
Hi Terry, Sorry I haven't sent any code over yet. I'm just not happy with the program yet. Had to unwind some of the code and redo it, but it's getting there. Still have to think of a name: did consider 'Coil Resonance Analysis Program' but decided against. Maybe tcma: Tesla Coil Mode Analyser? Don't worry about the trace captures, they appear to be absolutely fine. I have quite a suite of test data files generated by software to contain certain mixes of signals and noise. The program isn't quite working correctly on these yet. I'm concentrating on getting 'tcma' working properly first, and then I'll go on to do a more accurate analysis of the current profile results. I'm tempted to mention the progress so far to the pupman list, because the results already achieved demonstrate the elevated current maximum very nicely - almost identical to fig 6.1 in pn2511, and this is very strong evidence that we have quantified the internal capacitance correctly, and applied it correctly in the model. This feature of solenoid resonance is the main departure from the kind of transmission line behaviour described by the familiar telegraphist's equation, so it's nice to see it so unequivocally demonstrated. > I will also change the timing to help get higher frequency data. Well I'm not seeing any problems attributable to the input file in this respect. So long as there are 2 or 3 steps per cycle, we're ok. An important thing about the traces is that they contain a small amount of pre-trigger. Doesn't need much, but it's important that the program sees the start of the event. In this respect the existing traces are fine. > and let the scope average over say 16 waveforms. Ought to work fine with single shot. Averaging over 16 could give up to 4 times improvement in sensitivity, but at the moment we're not at all limited by the noise floor. At some point the traces will be limited by the Y-axis quantisation noise, so averaging beyond a certain number will not help any. At my end, one optional technique is to inject artificial noise in order to reduce the sharpness of the correlation functions - makes for better initial convergence. There's definately a tiny little resonance circa 190kHz in these traces, Q of around 14...hard for the program to lock onto... 3-26/TEK00001.CSV PK FREQ kHz (Error +/-) Q FACTOR (Error +/-) LEVEL 1 229.940 (0.01%,20Hz) 290.32 (5.23%,15.2) 98.1% -0.2dB 2 578.003 (0.03%,187Hz) 162.62 (5.23%, 8.5) 16.9% -15.5dB 3 903.931 (0.07%,635Hz) 103.88 (3.92%, 4.1) 7.4% -22.6dB *4 201.616 (0.37%,736Hz) 13.91 (25.48%, 3.5) 14.7% -16.7dB Accounted for 98.0% of input signal and from 3-26/TEK00004.CSV PK FREQ kHz (Error +/-) Q FACTOR (Error +/-) LEVEL 1 229.958 (0.01%,26Hz) 221.01 (6.04%,13.4) 97.7% -0.2dB 2 578.098 (0.03%,193Hz) 152.73 (1.07%, 1.6) 18.1% -14.9dB 3 903.931 (0.07%,632Hz) 101.37 (6.04%, 6.1) 8.9% -21.0dB *4 185.355 (0.21%,381Hz) 16.12 (25.48%, 4.1) 8.7% -21.2dB Accounted for 98.2% of input signal Have you another coil in the room that's responsible, I wonder? Wanders around in frequency quite a bit - maybe it's the pinger tank capacitance resonating through the gap circuit inductance? > The bucket also has a bottom and a heavy support near the top... > The coil sits on a Styrofoam box. It could be suspended by > string too. I should leave things as they are for the moment. No indications that these things are having adverse effect, and the stability of the test setup is desirable at the moment, while refining tcma. > I will try reducing the size of the CSV files too. Maybe put > them in one file with only one set of timing numbers and ZIP > them. No need. If you like, just zip the directory containing the 12 (or so!) csv files. > I got the RS-232 adapter for the scope... The RS-232 thing came > with a big programming book that explains the like > 300 commands... Yeah, they're usually very easy to program. We use loads of automated waveform capture stuff at work - I wrote most of it. Is the RS232 command set documented on the web? If so I can maybe help, but I won't be able to supply code for windows. You can probably use windows 'terminal' to manually command the scope, then when you've got a sequence of commands that work, just paste them into a 'basic' program. As regards the Delrin, etc, it's probably premature to trust the accuracy of the Q extraction enough to recognise the extra loss. -- Paul Nicholson --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.