How far can be located an SPI digital accelerometer from the processor without loosing signal propieties???
By itself, SPI isn't intended to be "cable ready," so SPI ports on processors, sensors, etc are not typically going to provide "cable length ratings." From an experimental point of view, we have been able to communicate with the ADIS16223, ADIS16227 and ADIS16228 over distances of ~12 inches, when using 1mm ribbon cables on our evaluation tools. With proper trace design on the PCBs, a cable driver near the ADIS162xx device and the use of twisted pair cables, we have been able to extend operation out to ~60 inches. Of course, cable design for industrial instruments is a separate topic on its own, as the performance will be somewhat dependent on protection from EMI during communication. I hope that helps.
And digital mems can handle longer distance?, why Analog device develop SPI accels with that higher frequency response, 22 KHz, but with a pour comunication system? (for my aplication...jejejeje)
Since the Spanish "J" sounds like the English, "H," I am going to assume that I can interpret "jejejeje" as "hehehehe." Am I correct in interpreting your "poor communication system" as a joke? For those who are thinking this way and are not joking, I would encourage consideration of the product's claims and its intended application space, before committing to such a thought process. Including every communication standard into one product would severely limit its value because this would significantly grow its size, cost and power. The SPI enables us to preserve the size, cost and power advantages of these devices, while providing a standard interface that most embedded processors can manage. Using a localize processor gives the customers the opportunity to manage their own I/O preferences, more efficiently than we could in a "one size fits all" product. I hope that helps with this perspective. Your communication structure will need to be a key point of consideration for this project and it would be a key point for any industrial engineer who is trying to develop a similar system.
Mark, thanks again for answering, but again, I´m an acoustician with many new electronics things to learn. And of course it was a joke (it seems that traying to be funny in other language its not so easy), well, for my project I was ideally thinking in four monitor points (for phase analysis too) each one with a triaxial accel, all of them conected to the Arduino, and then sending those signals to an analysis aplication, like matlab, with a wireless system. So from my searching came out the ADIS16223 as the best alternative. But now it seems to be a really big problem, because I can´t find another accel with a frequency response up to 10 KHz and triaxial...and analog or just digital...and with the price of a mems. Probably one new idea is to send the signals in a wireless way directly from the ADIS. Fortunatly I have some time for thinking. Well, thanks again and I hope we still are friends.
I knew that you were joking but since this is an online forum, where others will review the content in the future, I had to establish a clear "ADI position" for this. I sincerely hope that you did not take offense to this, because it sounds like you have a great sense of humor (something I always appreciate). Out of curiosity, what is driving your 10kHz bandwidth? That is pretty normal for predictive machine health/maintenance applications but I was curious if you had any machine-specific insights that were guiding you. How soon do you need to be up and running?
The 10 kHz bandwidth is for picking up the harmonics from belts or rolling elements on any machine, there´s no especial machine. The history is this, I´m an acoustic engeneer from Chile and I some time ago I was looking for something to get more especific, and then I found predictive maintenance based on vibrations, and I said thats what Im looking for!!, I know pretty good vibrations but I dont know machines, and it became my new professional challenge, I made an online course on Mobius Institute from Australia and I became Category II ISO 10836 certified Vibration Analist, I bougth some equipment, and looked for work. Then in my contry was really difficult to find job been acoustician with a vibration especiality, they are looking for an mechanical engeneer, not an acoustician, and then I decided to continue studing an I found a master on vibration and acoustics for mechanical engeneers in a university on Brazil, it was for free, two years, an island on the atlantic ocean...thats pretty good, I came to brazil last year and I´m studing for master degree. When I was here I found a foundation that suport engeneering projects for industry, meet then and they said, we want to make a device for remote continous monitoring for predictive maintenance, and we need the vibration signatures and a desing of the prototipe considering mems accels, we have contacts with Shulz the can provide a new machine and part with failures for the research, we don´t know nothing but we want to do, can you? obviusly I answered yes I can, and I´m here answering this to you. I have time until feb 2014 for finishing my project, and I been driven it in the way that will conect with my further estudies for getting a doctor degree.
Wow! You have a nice story and experience set. Let's keep in touch, because we are continuing to invest in developing products for this market. Perhaps a future offering may help you in the future.
Mucho gusto, mi amigo nuevo!
Would you care to share what driver and possible receiver you used to extend SPI?
Yes 10-11kHz aliasing filtered range is the lowest limit that can be accepted for CM use to be
able to do the required magic for bearingcondition measurement, our current range of instruments that we
now are working on to replace have that range and we used that for 6-7 years now with good results.
There are people out there fighting the datasheet war that go beyound 40kHz but the real use for that can be argued. We also happily include the sensor resonance in the measurement range using the excitation of the sensor resonance by the spikes from the bearing damages to enhance the sensitivity and early detection of such faults. We then also get a amplitude modulated signal that may be demodulated to extract the impacts from the bearing faults. Anyone that got a good easy digital demodulation algorithm? It´s all got so digital when using these things. Olov
With respect to sharing our buffers, there are a variety of ways to extend the SPI. We do not have an official reference design for this, but our test systems use a buffer with a Schmidt-triggered input, like the one shown in the figure below: NC7WZ17P6X.
Some customers have found that they can extend these signals up to 5 meters, when translating them into differential signals for the transmission lines. It all depends on the environment, speed and distance requirements. I hope that helps.
For the demodulation algorithm, does Fourier analysis allow you to discriminate the bearing fault signature in terms of magnitude and frequency, or are you referring to another approach?
Yes FFT is one thing but you get all kind of faults so low level bearing signals are not so easy to find. Classic analog transducer procedure include a HP filter to get rid of unbalance, alignment and other mech problems then a full wave rectification and a demod filtering by the anti aliasing filter before the A/D and the you get to do the FFT giving a noise reduced signal mostly containing bearing faults, if it´s empty all is ok, the more it looks like a garden fence the closer to voluntary or involuntary bearing change you get. Olov
Thank´s good help I will do some testing. Olov
Retrieving data ...