dynamic signal analyzer applications projects and accessories? (vibration and ?)

[coppercone2]

So after all these years I got a dynamic signal analyzer. Its main use was to look at <10 Hz noise on linear regulators and low noise rails. I tested it with the 100KHz band stop jim williams amplifier (pity this unit is only 40KHz).

But, given its portability, I was wondering.. what else can I do with it. Are there recommended transducers that can go with it? I tried it with a ultrasonic thingy (the one used for range sensors) and its not very interesting because it had such a narrow bandwidth.

Other then circuit measurements, I was wondering if there are some cool tools or sensors that can be used with one, say around a machine shop, cars, seismology...?

I have a general idea of what someone might have used them for but what I am asking is what are specific things I could get that I can hook up to it and explore something with that I am likely to encounter in my life, not something that came about in the depths of a engineering company trying to solve a very specific problem with a very specific setup.

I thought maybe some kind of microphone?

Does anyone have general suggestions for
1) some kinda useful microphone
2) some kinda useful siesmograph thingy
3) Other (i.e. I have a differential low noise nanovolt amplifier, I thought maybe if I characterize the CMRR it would be easy to hook up a E-field antenna to it, or maybe try some VLF antennas).
4) gasses, bubbles, welding?

I am more interested in I guess vibration/sound/the earth then Low frequency EM fields (I have a general Idea of what I can do with that.. its the mechanical/other I am more interested in)

Also, is there anything you can do with hydrology/lakes/water/etc? I have those near where I live and its quite portable and battery powered.

I even thought maybe something with pro audio, because I have been using generic dell speakers for the last decade.. dunno how I would apply it other then testing electronics.

Maybe something like.. general test system to find bad bearings based on noise with some kind of sensor?


Like, is it possible to create some kind of general purpose test bench of things or a kit with this to investigate things? The noise floor is like -90dB with alot of samples..

i.e. for RF work there are all sorts of wonderful things related to precompliance test stuff that you can make, so long you have a oscope/sa/amplifier I can get pretty specific.
-Various loop probes (with different shields, turns ratio), say square loop probe, shielded figure 8 coil, half moon ferrite directional thingy, other near field stuff
-clamps probes
-far field antennas
-field generators.
-all kinds of circuit interrupters

What is the stuff like that for a DSA with a little bit more specificity then 'microphone'. Maybe some kinda speaker to shake stuff with?

I am wondering if they are useful at all without "oh just get a piezo transducer and design a solid works model for a modified structural element with that transducer in it and then get it manufactured and catagorize it on a 30,000$ impedance analyzer then have fun. "

This thread died in a hurry in the TE forum

[trobbins]

I have this setup, that came from a site survey kit for electron microscope locating.
https://www.dalmura.com.au/static/HP%203560A%20dynamic%20signal%20analyser%20and%20sensors.pdf
https://www.dalmura.com.au/static/Site%20selection.pdf

The spectrum analyser is very old school, so won't get used much, as a good soundcard and software outperform it by miles.

Accelerometers can also come as very lightweight devices for low mass microphony assessment (eg. hifi related).

[Kleinstein]

The modern, relatively cheap electret micorpones are quite good.  The small ones may extend to the lwer frequency ultrasonic range (e.g. up to some 100 kHz).
The calibration is kind of tricky though - so one does not directly get absolute levels.

There are some plans for DIY seismometers. A simple, but not so bad version uses a speaker with weight in a pressure tight can to keep sound out.
There are other versions too (e.g. large pendulum and distance sensor of some kind (e.g. capacitive or optic).
Inside a building the vibration level can vary quite a bit, as a building often as resonaces and thus quite and noise spots.

Still much of the low frequency stuff today is recording data and than look at it at the PC.

[coppercone2]

can the sound card do <20Hz?

I thought the problem with sound cards is that they cut off at 20Hz and that the interesting seismic stuff happens at lower frequencies?

[Kleinstein]

There are a few sound cards that can do < 5 Hz. With a few one can remove AC coupling, but many have a digital high pass filter implemented as a feature that can not be turned off. Also the low frequency performance my not be that great.

If mainly interested in the low frequency part, one could modulate the sensor signal, or for a DIY seismic sensor directly build it to give a modulated signal. Some sensors (like optical ones) may be modulationed anyway.

[coppercone2]

is that what the industry is doing now adays?

What noise floor does a good sound card get at those low frequencies?

Do you have a sound card that is good in your opinion that I can put into a PCIE port?

My equipment is somewhere less then -90dbm, so I am wondering if that is useful or is modulated the way to go.

[coppercone2]

If I have 90 to 100dbm noise floor, does this exclude any type of device or category of device or type of general test ?

I am not exactly hounding for more decibels since when I have works and I have no idea what its capable of for non power supply circuit or voltage reference applications.

I see pictures online ranging from -140 (dubious) to -120db with pro audio sound cards. Not that I am really interested in using a computer for this since this is a nice unit, but what would that unlock for me?

My best oscillator has a harmonic of ~-90dbm down in the range of the instrument. (0.002 % distortion or so), if it works (4 equipment repairs in the last month, its like restarting the engineering room in a derelict space ship).

When would I come across the need for the computer system ?


I thought the first thing to do might be to put a seismograph in the basement or in the well. I think this is the most interesting project since it seems interesting to build, I am not sure what use I have with microphones, maybe next year in the winter because right now its birds, and in the winter its usually the high way, but at least its not birds and crickets.

[Kleinstein]

For something like a seimogrph it is more about data logging - one usually want's all the data, even though 99.x% would later be just ignored. What time is interesting can depend on other sources (e.g. you look at your data for the time just after a distant earthquake). Those data are like more interesting in the time domain, not so much in the frequency domain.

Modern low cost ADCs can get some 100-120 dB of dynamic range, depending on the frequency band.  Especially with lower frequencies and thus less BW for the noise the dynamik gets better - so for low sampling rate 140 dB dynamics is possible, though the costs may be a little higher and one may have to exclude mains hum.
 
Just for data-looging it would not need a full PC to run. It could be a small µC based system or maybe some Raspberry.
Especiall low frequency data (e.g. 60 SPS to get rid of mains hum) could be written directly to memory with little compression:  32 Bit per point at 60 SPS would be just some 1 MB per hour, would would allow nearly 1 year for a 8 GB SD card. Depending on the system there may be a few more data to log (e.g. temperature, time sync ), but this schould not be that many data.

Modulation is more like an idea to use cheap soundcard hardware. This would than need some real time calculations to demodulate and reduce the data rate to the real data contend. It very much depends on the sensor. With some systems one would use modulation anyway and could do the demodulation the digital way instead of classical analog. The limiting part for the dynamics is likely the sensor anyway, not the ADC. The harmincs are likely even more set by the sensor.

[trobbins]

Perhaps 'Just do it' is your best way to initially progress.  Dive right in and start the learning curve of acquiring and analysing data - and keep searching for tools and sensors and interface equipment that refine your outcome.   Report back on what is interesting for you, and what you can practically achieve, and what limitations you just can't seem to get around.

Open-ended discussion ends up being just open-ended until you dive in to something.

Just my 2c.

[Berni]

I built my own sound card for measurement use. It used a XMOS MCU to run the async USB audio, so no resampling is needed. Then stuck on the best spec DAC and ADC that TI makes on the other side of a galvanic isolation gap to remove ground loop issues.

It ran at 192KHz 24bit and was DC coupled. Running a FFT on it the noise floor sat at about -155dB, so it could see some very tiny signals. It could clearly see harmonic spikes down in the -130dB, but im not sure how much of that may have been the sound card itself since i would need a perfect sine source to find out, but it was showing THD numbers in the ballpark of 3 zeroes after the decimal place in %.

[Conrad Hoffman]

Balancing man, balancing rotating objects! Fans, motors, power tools or whatever. Start with the second line here- http://www.conradhoffman.com/chsw.htm
I can also find you the original Nicolet app note if you want to delve into the math. What you need is some sort of vibration sensor, maybe a PZT accelerometer, a small loudspeaker with a mass attached or a MEMs sensor.

[coppercone2]

this one has a serial uplink to a computer.

Yea the sound card curve does NOT look nice, this only had some spike on like <3% of the screen when the full scale is 3.12Hz, so I think its like 0.1Hz where the noise is, it is flat from around 0.1Hz (or less) onwards if i connect it to a turned off function generator, floating it looks crappy, but I think for the 3Hz full scale you need like 512 samples. I just got a 110db variable attenuator, when the low noise oscillator is repaired and tested I will see what that does.

But yea if you get 140dbm thats 50db better. quite alot of dynamic range for audio


The balancing thing seems useful, perhaps I can do a small shaft with my proxxon mill and a grinding stone attachment on a 1/8 inch shaft and the round object 3 jaw vise ? Grind a little and check? unfortunately no lathe

[trobbins]

To augment your dynamic signal analyser, it can be quite easy to set up a soundcard for continuous and some impulse inputs courtesy of the hifi and speaker fraternity - I use REW with a USB soundcard and a laptop to avoid mains AC connection and ground loops, and loopback calibration allows flat signal bandwidth to extend down to 2Hz.  In addition, REW allows any test frequency, in continuous or burst mode, down to 1Hz to be generated in a way that suppresses the harmonic distortion presented to the DUT to below 0.0001%, by nulling out harmonics generated in soundcard DAC and any subsequent interface amplifier.