One of the three fully analogue scopes I've retained out of respect for them is a Telequipment DM63 and it's *GREAT* - nice and simple to set up, unlike the DSO I spent over an hour with trying to get the settings right to display a simple 4Hz sine wave this afternoon.
Good to hear. You won't get a hernia tossing it in a dumpster a year or so from now.
--
This message may be freely reproduced without limit or charge only via
the Usenet protocol. Reproduction in whole or part through other
protocols, whether for profit or not, is conditional upon a charge of
GBP10.00 per reproduction. Publication in this manner via non-Usenet
protocols constitutes acceptance of this condition.
Youngsters often can't even operate a spectrum analyzer anymore so they'd likely would have never found out. A classic case:
Some electronics always became fritzed every few seconds. If the blinds were closed to keep out the afternoon sun the system ran much longer. "We think something must be light sensitive. But what?". Gazing out the window pondering the situation. "What's that reflection in the distance every few seconds? Looks like Indians doing mirror signaling but it's the 20th century and you guys don't even have Indians" ... "Ah, that's just an Air Force Radar, that's nothing ... oh ... *OH* ... OH WAIT!"
Another weird case: I found out that the noise was caused by reflections from turning fan blades. The folks at the client really thought I was off the rocker now. "It's got to be electrical, there IS no other explanation!" So I took a big rag, wrapped my hand and slowed down the propeller a bit with the palm. The modulation in the spectrum changed right along. There was some silence in the room for a while.
The most interesting consulting episodes are often those where the flight takes longer than the actual job.
You could have increased the voodoo factor: Solder a snippet of CAT5 pair to IN- and OUT, then twirl up some more until the response is flat.
This is why my spectrum analyzer has the full plethora of demodulators, NFM, WFM, AM, upper and lower sideband SSB. I wouldn't have bought without.
Then I have various receivers such as ICOM ICR-1500 (the "DC to light" receiver), AOR-3000, NRD-515 plus various others for baseband. The latter has a 300Hz wide a and very steep 8-crystal filter so you can listen to very faint noise right next to where Sutro Tower is blaring. That filter alone cost about $300 but was so worth it.
Sometimes when I noise-hunt with headphones on people look at me as if I was the snake oil dealer.
I live 35mi east of Sacramento and no matter how hard I try people won't let me really retire (getting into more volunteer work now so I kind of have to). However, most jobs are done over the distance and thanks to Zoom and GoToMeeting flights are mostly no longer required. To the detriment of airlines, rental car companies and hotels.
Also, I meant more like the rare species kind of engineering, in my case analog. FPGA or coding is different.
There's an app for (almost) andything and everything:
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
There was a generation of digital scopes that were awful. We demo'd one HP scope that literally had four buttons. Nobody could get it to do anything, including the HP rep. Most digital scopes now have a lot of obvious knobs.
Press DEFAULT SETUP and crank the time/div and volts/div knobs around, and you sine wave will be right there. The autoset may not work well for such a slow sine wave.
For 5 or so years now, I've been upgrading to successively fancier Rigol scopes. I give the old ones to junior engineers and interns. They all work fine.
We have the giant screen 1 GHz 4-channel Rigol too. It's fabulous.
formatting link
I think our last analog scope, a Tek 7104 on a cart, finally died. Only one guy here still used it. Not me.
Us older guys have to sometimes force ourselves to keep current... or give up serious electronics.
(But I won't give up my drafting table.)
--
John Larkin Highland Technology, Inc
lunatic fringe electronics
One can take thousands of data points, but only the curve next to the zero crossing is ever part of the result. You throw the rest away (and no, it does NOT help your signal/noise to have included them, unless the curve function is a very carefully tailored digital filter).
An FFT and selecting a small frequency range IS a carefully tailored filter. With all the data, you see a zero crossing, +/- .1 crossing, +/- 0.2 crossing, etc.
If you have a good approximation to a sine wave, the flat bits at the maxim um and minimum of the waveform don't contribute much to the phase informati on, but the slope of the waveform is tolerably high everywhere from 50% of full scale through the zero crossing.
If you go through the mechanics of calculating a best fit sine wave to a le ngth of amplitude versus time data this becomes fairly obvious.
Remember to reject DC (or any offset of a channel will contaminate the result).
FFT isn't harder than curve fitting, and it works fine with sines, triangles, etc. Why try to avoid it? By generating sine and cosine terms, it hands you the phases, on a (figurative) silver platter.
Right. May as well normalize both sines to zero offset and 1.000 RMS amplitude.
FFT is fine... my Python jocks do stuff like that all the time. We would, I think, need a long record so we can window the ends.
What happens if you FFT an un-windowed sample of a sine that's taken at some random time but an integral number of cycles? We can do that. Do the end artifacts cancel? Seems like they should.
I might not be able to guarantee exactly 2^n samples per cycle.
I understand curve fitting better.
--
John Larkin Highland Technology, Inc
lunatic fringe electronics
Actually, that's not a big issue; a measured sample in the middle of a big field of zeros would raise the FFT size, but be otherwise benign. It makes a fundamental frequency well-peaked, but with some width.
Yes, that amounts to circular boundary conditions, which is an antidote to end effects. The amplitude versus frequency peak gets very narrow, which is optimal for signal/noise.
The beauty of the discrete Fourier transform, is that it really IS a least-squares fit to the data curve, indeed is exact at every measured point.
The only tricky part, is that the peak-width of the fundamental is an artifact of the sampling (and sampling window), so the least-squares fit to 'the' phase shift number needs averaging over the peak's several channels. Best fit (i.e. minimum sum of ( squared errors divided by squared standard deviation) does, therefore, require that formula I gave earlier for a weighted average, because low-amplitude Fourier components have higher phase errors (standard deviation). Phase error is NOT the same at all points, but depends on amplitude. At 5% of peak amplitude, weight drops to 0.25% of what it was at the peak.
Phase (of the fundamental) will track any time delay. The second harmonic, if there is any, with the same time delay will show twice the phase difference. That means you want a frequency window that includes the fundamental frequency but little or none of any other (harmonics etc). The weight helps here, of course: it's very small away from the center of the peak.
and it works! We'll ship one to our customer so he can start talking to it, but it's not calibrated, and we do want to do the phase shift testing on every unit.
This wasn't bad, for a fairly complex function. There's an ECO to remove four caps, piggyback four caps on top some resistors, and change four other part values. No cuts/jumpers! Everything fits! There was no prototype, and this was built by production per released rev A docs.
We need a heap of Python now to test and cal.
--
John Larkin Highland Technology, Inc
picosecond timing precision measurement
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.