From: "Terrell W. Fritz"
Date: Fri, 26 Jan 2001 20:31:18 -0700
Subject: Re: [TSSP] Top voltage testing
Hi Paul, I got a 1 watt 10 ohm surface mount resistor and four wired it for a current shunt with Kelvin connections. The resistor measures 9.915 ohms. The load resistor and wiring measures 501.97 ohms. Taking into account the shunt resistor's value I get the following frequency response table now with the HP. F(kHz) Vrms Irms (mA) Rcalc 1 0.66497 1.3311 499.56 2 0.66097 1.3235 499.41 4 0.66193 1.3256 499.34 10 0.66134 1.3246 499.28 20 0.66262 1.3269 499.37 40 0.66419 1.3295 499.58 100 0.66426 1.3276 500.35 200 0.66640 1.3285 501.62 400 0.66855 1.3258 504.26 1Meg 0.70392 1.3800 510.09 So the HP just needs a bit more signal for it's rated accuracy. I bet if I read the manual really hard it would say +- percent of full scale somewhere... There is skin depth and dissimilar metals involved as the reading will drift over time but it is still pretty darn close :-) The signal generator only puts out a fixed 2 volts peak to peak so I can't turn it up. So the Zft is now: Fo = 105087 Hz Vin(HP) = 0.66586V Iin(HP) = 1.1742mA Vtop(Tek) = 53.1V Zin = 567.08 ohms Gain = 79.75 Zft = 45.222K Note that this is with a 10-11 ohm source resistance. I note the Qv is a bit higher tonight too. I don't think I can get much closer... If I get the new signal generator I can calibrate the levels much better still which will help. Cheers, Terry At 03:58 PM 1/26/2001 -0700, you wrote: >Hi Paul, > >I measured the current for the 501.88 ohm resistor at many frequencies with >the following results: > >F(kHz) I(mARMS) VRMS(in) (meter) Rcalc >1 1.361 0.67798 497.148 >2 1.353 0.67343 496.731 >4 1.354 0.67495 497.486 >10 1.352 0.67435 497.780 >20 1.351 0.67568 499.133 >40 1.341 0.67721 504.004 >100 1.309 0.67710 516.265 >105* 1.290 0.68130 527.140 >200 1.157 0.67928 586.105 >400 0.814 0.68102 835.634 >1Meg 0.0848 0.71268 8403.23 >*I took this measurement later and the level drifted somewhere. Need that >HP 33120 :-)) > >So the higher voltage is good but tiny voltages start to loose it above >20kHz. Fortunately, the loss is predictable (this is when the extra $$ for >the HP meter pays off :-)) Probably the only multimeter in the world that >can do this stuff). > >> Meas Model Error >> Q: 68.3 68.3 Adjusted to fit Qv >> Vgain: 71.6 74.5 +4.1%, >> Zft: 42.6K 44.7K +4.9% >> > >So at 105kHz, 527.14 / 501.88 = 1.05033 >Zft = V/R or corrected = V / (R / 1.05033) = 42.6 x 1.05 = 44.73K > >I would submit the measurement error is found and corrected and all is well >with Zft :-)) We are working at the limits of the equipment (wish the Tek >scope was closer to the HP**) but the errors all seem to be converging to >the model's predictions eventually. If I win that HP signal generator on >Ebay, it will help a lot to determine error since it can act as a very >accurate calibration source for this stuff. If I get out bid, I will just >buy one... > >I will try to get the metering to simply work right in the first place. >This correction table can be applied to my last post... > >**The HP is AC coupled true RMS while the scope is AC+DC True RMS. That >may explain the error a little... But when I switch the scope to AC >coupling nothing changes... Perhaps the far higher scope bandwidth is >picking up high frequency noise or it is sampling error in it's rather >course vertical digital sampling... From the HP comparison it appears to >be high, but it has always seemed to give the right answers... > >Cheers, > > Terry > > >
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.