Spectrum Analyzer... thoughts on this one ?

[Freq Band]

I bought this, it's being shipped to me now.

Spectral Dynamics SD345-2 Spectrascope III FFT Analyzer
is microprocessor based that analyzes frequencies up to 100 kHz with 400 lines of resolution (raster display). Any 25% of the display can be expanded (zoomed) for detailed examination. An alphanumeric readout on the CRT used with the marker functions provides additional information. Quantities can be displayed in engineering units such as mil/sec or g. Scaled X-Y recorder and display outputs are available on the rear panel. 10 bands are displayed for Octave bands and 30 bands are displayed for 1/3 Octave mode. Available center frequencies are 1.6, 2.0, 2.5, 3.15, 4, 5, 6.3, 8, 10, 12.5, 16, 20, 25, 31.5, 40, 50, 63, 80, 100, 125, 160, 200, 250, 315, 400, 500, 630, 800, 1k, 1.25k, 1.6k, 2k, 2.5k, 3.15k, 4k, 5k, 6.3k, 8k, 10k, 12.5k, 16k, 20k, 25k, 31.5k, 40k, 50k, 63k, and 80k Hz depending on the mode and frequency range chosen.

Some highlights are:

Sixteen frequency ranges in a 1,2,5 sequence from 1 Hz to 100kHz

Eleven full-scale voltage inputs from 10 mV to 20V in a 1,2,5 sequence

A X4 X-axis magnifier for closer examination if any 25% of frequencies displayed

Averaging from 2 to 1024 samples is available

Peak hold mode

Extensive cursor functions

X/Y plotter output w/pen lift

1/3 or 1/1 Octave Spectrum Analysis (option, is included)
-----------

Any thoughts ??
I've never used a hardware spectrum analyzer, but I have the pdf manual.
(I posted this over at DIYaudio first, but gained no responses.)

=FB=

 

[NewYorkDave]

Any thoughts?

Well, here's one:

I hope it works... Otherwise, you've got one hell of a paperweight!

:razz:

Seriously, I'm not sure what the question is.

 

[Freq Band]

noooooo paper weight !!!
...better not be.... :?
( fully usable rack handles though, and I can keep it in my passenger side seat, to balance the car when I'm driving alone.)

The question is, does it look useful for audio measurements ?

And another Q now that I think about it...do you recommend dedicated probes ? or is just a proper shielded cable good enough (with retractable clips) ??

=FB=

 

[Larrchild]

Extensive cursor functions

especially if it's full of dry caps.

what's the dynamic range of the display?
That's the big issue.

 

[gyraf]

Manual - http://www.cognitivevision.com/download/Legacy/SD345/Pt2-345.pdf - says 204dB on display, 65dB in any one-range

Good enough for rock music... Definitely a usable instrument IMO.

Jakob E.

 

[Freq Band]

SD345 User's Manual - Spectrascope III ....Gyraf linked #2 above.
....all three parts (3 pdf's):

http://www.cognitivevision.com/legacy.htm

 

[Larrchild]

Cool, if you connect it to a distortion analyzer that nulls out the fundamental, you can dig even deeper than 65dB.

Should work well.

 

[Freq Band]

[quote author="Larrchild"]Cool, if you connect it to a distortion analyzer that nulls out the fundamental, you can dig even deeper than 65dB.
[/quote]

(here I go again, risking a stubbed toe).......

Would a narrow filter do the same job as a "null" ?

i.e......A tracking filter I also bought on ebay ?
http://www.tucker.com/images/images_spec/00002582.pdf


(If my employer only knew what I was doing with that raise he gave me )



=FB=

 

[Larrchild]

I'm pretty sure it would, if it filtered the majority of the main spike (fundamental) out.

The idea is that if the, say, 1k spike is at the top of the display, you only have 65dB under it to see things.
But if you get the main spike to be reduced, every dB it is reduced, is a dB more of viewable dynamic range to see the rest of the stuff that lives down in the lower regions below -65 such as the 2nd and 3rd harmonics and all the rest..

 

[Guest]

How is it compared to software versions?
I'm using currently feedback analyzer for concerts, but for a record reproduction a spectrum analyzer is much better.

 

[Samuel Groner]

A tracking filter I also bought on ebay?

A tracking notch filter not bandpass! You surely can convert a bandpass to a notch with some additional circuitry but I doubt if the unit you bought has low enough distort to make it usuable for serious THD measurements.

Samuel

 

[mikep]

[quote author="Larrchild"]Cool, if you connect it to a distortion analyzer that nulls out the fundamental, you can dig even deeper than 65dB.
[/quote]

I had hoped to do this with my ST 1710A and a cheap soundcard + software based FFT, but the ST residual is not as pure as I had hoped. there is always 2nd harmonic about -82dB below the fundamental IIRC. my P.O.S. m-audio 2496 card does way better on its own. (subjectively, the A/D sounds like total ass). Ive got a "high resolution" realtime FFT setup now based around a PCM4222 eval board, but its actually not much more useful than the 2496. go figure. still need a super-duper LDO to make it worthwhile. I kind of lost interest.

the best distortion insights ive had are from *listening* to the residual.

 

[Larrchild]

Mike, it would be interesting fo find out where 2f was coming back in the ST 1710.
If it's an analog amplifier stage, maybe it can be hotrodded.

Or maybe it's some rail issue or notch filter issue and can be bettered.

Another distortion analyzer of better spec would allow troubleshooting.
Maybe even see the changes/improvements using the soundcard setup.

 

[JohnRoberts]

The typical source of THD in LDOs is from amplitude control circuitry. You can generally reduce distortion by tweaking loop time constants but the trade off is uncomfortably long settling times.

For general design the -82 dB noise/distortion floor is adequate. For the occasional spot measurements to see how good something really is consider a few fixed notch filters to knock down that second, There is probably a topology of notch that is also a LPF so it will knock down other higher harmonics also.

Perhaps filters dialed in for 1K and say 20Khz measurements will probably cover most specification needs.

Of course this filter needs to be quite good, but there are some modern opamps that are spectacular on paper. Use good resistors and film caps (perhaps polystyrene caps and MF res).

Depending on the circuitry and what you're looking for maybe make another spot filter for 20 Hz, but in general that is not heavy lifting for electronics.

JR

 

[Larrchild]

Also,
When you change frequencies on the oscillator, if the waveform level settles quickly on the "leveled" output, you can probably slow that loop time down as JR says and get lower specs.
Find the cap that determines the loop time, and hang something bigger across it for a quick test.