Vibration measuring with Spectrum Analyzer

[CompNut]

Hi everyone, I want to be able to perform noise and vibration analysis on high end audio turntables.  Does anyone have any ideas on a cheap piezo transducer and the interface required?

[0culus]

Wouldn't an oscilloscope be a better tool for this? You're going to need a DC coupled spectrum analyzer to do this and you will have to be extra extra careful to not stick a DC bias in the front end (expensive mistake).

[CompNut]

I am looking to do measurements like this:  https://youtu.be/xIrYWc6zyBo

[richnormand]

In similar measurements I ended up using my HP 3561 dynamic signal analyser with an electromechanical L15B 3-axis seismic detector.
Both are available on ebay at affordable prices these days (compared to new!)
That way you can measure to very high SN ratio, see real time waveforms and get a FFT display of frequencies from below a few hertz  to well in the audible range.

With a modern scope with FFT capabilities and some good piezo or home made voice-coil transducers you might be able to get some good data.
All depends how DIY or pro you need it.

[CompNut]

I do have an HP3585A Spectrum Analyzer and also an HP3562A Dynamic Signal Analyzer on the bench.  The only digital scope I have is a 1054Z as I mainly service audio gear and use the Tek 2465B most of the time.  I also use ARTA for computer based FFT measurements.

[richnormand]

The reason I mentioned the 3 axis detector is watching the video you mentioned.
I would assume the z direction (R+L) is important for vibration but the way stereo work on a record the x/y perpendicular to the groove is the the L-R signal and is also susceptible.

With your equipment it looks you could do the FFT and amplitude trace. You may need to jury-rigged a good sensor that is direction sensitive and do multiple measurements.
The L15B was about $40 if I remember correctly or do the inverse of a loudspeaker essentially. I don't know how good pzt are in that range.

Another idea would be to use an old phono pickup cartridge mounted on a cantilever to touch the table in different directions to sense the vibrations on that axis.

Good luck with it.

[DaJMasta]

Any decent sound card will get you more than enough sample rate and resolution to do full analyses of a turntable.  Your nominal dynamic range of turntable playback is going to be about 70dB, and this is limited by noise picked up from the surface texture of the material itself, even with the best mechanical hardware used for the arm, turntable itself, vibration damping, etc.  That means a decent soundcard is going to be able to give you 20-30dB more dynamic range, as a baseline, and could get 40dB on the high end if other amplifiers or cables aren't the limiting factor.

Otherwise, if you want a dedicated piece of gear which isn't a sound card or audio analyzer, something like those dynamic signal analyzers would be appropriate.  It is unusual to find a spectrum analyzer down all the way to DC, and it's extra cost to get it below 9kHz on the frontend in most situations because they are optimized for high frequency work, so while it's a similar readout and basic measurement, a spectrum analyzer's architecture is going to have trouble giving you good dynamic range on the low frequency end when compared to dedicated devices - very often their converter is only 16 or 14 bits, partly because its bandwidth is at least a few MHz - if your converter being used doesn't need nearly that bandwidth, it's much simpler to make a lower noise, higher resolution design, even at a lower price.

[CompNut]

The HP3585A has a bandwidth of 20Hz to 40MHz and and my Dynamic Signal analyzer is DC - 100KHz.

I am looking for a small transducer like they used in the video I posted, they have the transducer sitting on the chassis right behind the motor mounting. I would like to be able to measure in various places on the chassis and main spindle bearing.

[0culus]

The HP3585A has a bandwidth of 20Hz to 40MHz and and my Dynamic Signal analyzer is DC - 100KHz.

I am looking for a small transducer like they used in the video I posted, they have the transducer sitting on the chassis right behind the motor mounting. I would like to be able to measure in various places on the chassis and main spindle bearing.

A quick google for "vibration transducer" will turn up a lot of results. Why not just get some cheap ones and try them? I have no idea what the fellow in the video is using.

[CompNut]

 :I found a CEC transducer on ebay that is 15-2000Hz and outputs an analog signal for $150. https://www.ebay.com/itm/294117896724?hash=item447acaf214:g:w6UAAOSwE85gdJX1

[DaJMasta]

Since you're in the US, there should be a number of accelerometers which may be inexpensive if you find the right one used - they are export restricted so they can be harder to get abroad.  I'd look at PCB Piezotronics and Endevco/Meggitt as the most common availability of these quartz based accelerometers, and you're either going to get a charge mode accelerometer (easier to find, cheaper, requires an external charge amplifier), or you can get an ICP/IEPE accelerometer which has the charge amp integrated and which is powered through a current source on the coax into it.

There are a bunch of potentially expensive pitfalls, you're going to be dealing with mostly single access stuff, and mounting method is going to be critically important if you want any sort of high frequency measurement (>1000Hz), but these industrial accelerometers offer great sensitivity and linearity.  If you can live with somewhat lower maximum frequency, you can get decent MEMS integrated accelerometers that can readout digitally or analog, but which probably aren't going to get to nearly the same frequency response.

Because the volume of people looking for them is low, you can sometimes snag a deal, but normal prices can be higher, especially if you need a charge amp or signal conditioner or similar in between your transducer and your analyzer.

[Conrad Hoffman]

Will a MEMS device work for this? They're inexpensive and quite sensitive. IMO, you don't need a spectrum analyzer other than for bragging rights. A scope and a brain will usually do the job.

[DH7DN]

Hi everyone, I want to be able to perform noise and vibration analysis on high end audio turntables.  Does anyone have any ideas on a cheap piezo transducer and the interface required?

In the video, he's probably using a Brüel&Kjaer 4370 piezoelectric transducer (sensitiviy of 10 pC/(m/s²), very sensitive), along with a charge amplifier (probably B&K type 2525 or similar). He's also measuring vibrations in one direction (one axis), I hope he's not rendering wrong conclusions from this experiment ;-)

B&K 4370 goes for like 100-150 EUR on eBay, the charge amplifier goes for 100-250 EUR. If you're lucky, you will get the special low noise cable with your transducer, otherwise the microdot UNF 10-32 cables go for 60ish EUR per piece. I wouldn't suggest microdot cables from China.

MEMS sensors can be used as a cheap option, although they are very limited in the use. Their cutoff frequency is in the range of few hundred Hertz so you won't be able to see higher harmonics. The sampling frequency is in the order of 1 kHz so you won't be able to measure frequencies reliably past 400-500 Hz. You also have to adjust the measuring range of the MEMS sensor in order to measure small vibrations...

For signal analysis, you don't need a spectrum analyzer. Use either your sound card (for analogue signals) or in case of a MEMS sensor any free availible scientific software (GNU Octave, Python/Numpy, ...) to do a FFT analysis.

I'd like to mention that a usage of different springs or rubber bands does not eliminate the mechanical problems (bad construction or design), it's just helps mitigating the interference or to move the disturbance in a different frequency band where it's not critical.

Good piezoelectric transducer manufacturers are Brüel&Kjaer, Endevco/Meggitt, PCB Piezotronics and Kistler, just to name a few. If you can borrow a laser interferometer (e. g. Polytec), you will be even happier

[JeffK]

I do have an HP3585A Spectrum Analyzer and also an HP3562A Dynamic Signal Analyzer on the bench.  The only digital scope I have is a 1054Z as I mainly service audio gear and use the Tek 2465B most of the time.  I also use ARTA for computer based FFT measurements.

Since you already have a HP3562A Dynamic Signal Analyzer, that would be good tool to start your investigation. It already has everything you need to replicate the FFT plot as demonstrated in the Hanze Hifi YouTube video. You'll need a good, low frequency optimized accelerometer and a signal converter/amplifier depending on the accelerometer you choose.

As mentioned by DaJMasta, Piezoelectric Accelerometers generally are available with two possible electrical interfaces - "Charge" or "ICP(or IEPE)". Either one will work but each type will require some type of signal interface box to provide your HP3562A with a suitable voltage signal. Here's an overview on Charge Mode accelerometers https://www.pcb.com/resources/technical-information/charge-mode. And an overview on ICP/IEPE accelerometers https://www.pcb.com/resources/technical-information/introduction-to-accelerometers

Unfortunately selecting the optimal accelerometer is a process of compromise. A balancing act is required to trade off the desirable low frequency response and desirable high sensitivity with the undesirable physical size, mass and noise floor of the accelerometer. One of the few white papers I have found that directly compare accelerometers in practical use is titled "MEASUREMENT SYSTEM NOISE" and was written by Gary C. Foss. He distills a massive amount of hard earned knowledge and practical use of accelerometers into 6 pages. Table 3 contains the key information. You can get a copy of the PDF document here: https://docplayer.net/137622267-Measurement-system-noise.html      

If you are interested in going down the rabbit hole on modal vibration, here are some good resources:
HP created several application notes that really do a great job on the subject. Look for App notes 243, 243-1, 243-2 and 243-3. https://www.hpmemoryproject.org/ressources/resrc_an_03.htm
Of course, not complete without credit to Bruel & Kjaer: https://www.bksv.com/en/knowledge 
Sound and Vibration online articles. Good content by experts from the field. Ancient website, navigate to "back issues" for content: http://www.sandv.com/home.html
U Mass Lowell -  Dr. Peter Avitable, lots of great information in app note format: https://www.uml.edu/research/sdasl/education/modal-space.aspx 

[tggzzz]

Hi everyone, I want to be able to perform noise and vibration analysis on high end audio turntables.  Does anyone have any ideas on a cheap piezo transducer and the interface required?

You already have one at your disposal, on the end of the turntable arm!

If you have any other specification in mind, let us know

[_Wim_]

:I found a CEC transducer on ebay that is 15-2000Hz and outputs an analog signal for $150. https://www.ebay.com/itm/294117896724?hash=item447acaf214:g:w6UAAOSwE85gdJX1

I would look for "ACCELEROMETER MMF" (https://www.ebay.com/sch/i.html?_from=R40&_trksid=p2380057.m570.l1313&_nkw=ACCELEROMETER+MMF&_sacat=0)

They have very nice sets with cables and mounting hardware, and specs for older models are still available here:
https://www.mmf.de/obsolete_accelerometers.htm

I have a few different models of these, and quite satisfied by them.

As others have said, you will need a charge amplifier to use these. I use a robotron 00 042 with them (https://www.ostron.de/Mechanik/Mechanik-allgemein/Schwingungsmesser-Robotron-00042.html) , but these are not so common on Ebay (but cheap when they do appear). Bruel & Kjaer is much more common and off course excellent quality.

[_Wim_]

Warning: for your purpose the robotron 00042 is not useful as high frequency response is max 4kHz.

[_Wim_]

A DIY charge amplifier is also very easy to build:
https://www.analogictips.com/understanding-and-using-charge-amplifiers/