FPGA - Vibration Analysis

A ready-made ADC board such as the MCC172 that Jan and I reviewed, plus a Pi 4 and a USB SSD might be the lowest-cost way to start with. There's not a lot of MCC172 available but there were a few at distributors the last time I checked. You'd have to experiment a bit to enclose it all and it's not very rugged but maybe that's ok for the first phase, since it won't need to be the final setup. Does it need to be battery powered? The Pi plus ADC board will need quite a large battery (or charged and swapped out say daily). A low-cost large battery is the Lithium car chargers, I don't know if they are allowed on ships (I know nothing about ships : ( Some of those car chargers have quite high capacity, and 12V or 5V output. If you have access to an AC or DC supply, then that would be much easier.

For Pi models, definitely the Pi 4 is recommended since that's what is known working (both Jan and I used Pi 4 I believe), and other Pi models might need some further tweaking. We found no issues with using the Pi 4.

For the processing, Matlab home edition is reasonably low-cost (but you will likely need to buy toolboxes for it too, you can choose those as and when they are needed).

Thanks Shabaz !

The ships stay seven to ten days out at sea and engines run continually. Around 45% of the time at full power. I would like to see / analyze data for the entire period she is out there.

What size of USB SSD would be best ?

500 GB ?

Sensors ? IMI 603C01 ? 

I shall keep the testing / calibration part for the next time. Just a food for thought at this moment. Coffee Machines or small DC motors with eccentric mass.

I shall acquire the hardware the soonest.

Sincerely,

I think so, yes, although 1Tbyte might be more suitable, it's not much extra cost, and the spare capacity may make it perform better.

Regarding sensors I don't know unfortunately. That model sensor is reasonable cost, so if it doesn't work out, it could still be used for a testbed so I think it's worth a shot if there is no other recommendation (e.g. from the engine manufacturer).

Got it !

Shall keep you all posted. Honored !

I saw same sensor being recommended somewhere in their documents ie MCC172.

Sincerely,

Good luck on your journey. If you are tired of VHDL/Verlog coding for FFT algorithm on FPGA, you might give Matlab/Simulink tool chain a shot, which is a higher level than VHDL/Verilog.

Dear Flyingbean,

Thank you so much. Indeed, you are correct. But my aim is to have a standalone system in the future.

At the moment, cannot comprehend how to girdle MATLAB/Simulink along with its operating system AND other components in one basket.

Let me know please if any idea strikes you. 

Sincerely,

Dear Flyingbean,

Thank you so much. Indeed, you are correct. But my aim is to have a standalone system in the future.

At the moment, cannot comprehend how to girdle MATLAB/Simulink along with its operating system AND other components in one basket.

Let me know please if any idea strikes you. 

Sincerely,

Jawad_Malik said:

At the moment, cannot comprehend how to girdle MATLAB/Simulink along with its operating system AND other components in one basket.

Perhaps worth a closer look ?

"...With the push of a button, you can use MATLAB® and Simulink® to generate code and run it on hardware..."

https://www.mathworks.com/solutions/embedded-systems.html

Generate optimized C, C++, CUDA, Verilog, VHDL, and Structured Text

Prototype on popular boards from Arduino® to Zynq® using hardware support package add-ons

Dear Beacon_dave,

I stand in awe for such capabilities being offered by MATLAB® and Simulink® . Need to read more about it. 

Thank you for your guidance. Obliged for the help.

Sincerely,

Don't forget the approx £30k cost for commercial license for this !!!

(Not sure of exact pricing and Mathworks do offer a startup licensing scheme which may be cheaper.)

MATLAB is a great product but it isn't cheap.

Somehow I think Jawad is looking for  a lower cost solution.

MK

Hello Friends !

A bit of progress with my Project and some more issues cropping up.

I am glad that my senior management at work has now officially asked for a proposal whereas if we could assemble a prototype VA analysis system and test it on board. But the setting has changed from previously a slower moving Vessel 60-70m long, fitted with High Speed Diesel Engines

 To

 a high speed smaller Vessel, 13-15m in length outfitted with Mercury Verado 300 petrol engines (combination of 3-4 engines).

Added task is to route the OBD II data along with the MCC DAQ data and transfer it ashore after securing from the sea trip. The Data Recorder for OBD II chosen is; https://www.csselectronics.com/products/can-bus-logger-interface-cl2000. The same has a SD card for recording the data. The output is via a usb.

The latter i.e. “data transfer ashore” is under debate. Shall discuss thence.

Present issue is Vibration and High “g” accelerations / Heaving Motions when Vessel is running at speeds of 50-60 knots. The SD cards within the CL2000 from CSS Electronics and Raspberry Pi4 all shall be subject to forces. The usb connections too will rattle.

All Data Processing / Analysis is intended to be carried out ashore on a Desktop PC.

A high speed boat hitting waves can produce accelerations up to 6 g. A microSD card slot can probably handle that. Engine vibrations can be more severe. You will need to shock mount the electronics or hardwire everything and maybe even pot the electronics.

The manufacturer I was working with, was contemplating using 4G/5G when a ship was close to a port. That could be fine for even large volumes of data since the cost of the transfer is negligible compared to the data value. Another option could be to transfer less files, say, one file from (say) every hour or every ten minutes, so that there is a sample of history transferred quickly over 4G/5G, and then any specific period can be transferred say overnight.
Another option could be WiFi or Ethernet, e.g. someone carries on board equipment to pull the data, or even swap out the storage device.

The micro SD isn't a major issue for Pi 4 since it could always be manufactured with the Pi 4 compute card later on. In any case, even the normal Pi 4 can boot from other sources if desired (I have not tried this, but this was a feature that is supported).

Maybe there are other loggers that do not use SD cards (if this definitely is not acceptable for a prototype). I've no idea if this is any good, but there is code here to run a CAN bus logger on a Pi (if you do use this code, then it is probably a good idea to use a separate Pi for this initially). I don't know the quality of it since I've not used it, but it looks reasonable.

There will be a tradeoff between reliability and speed to develop (and associated costs) for a prototype however. There will already be slightly more rugged Pi devices although they too might only be suitable for prototypes. For example this one is a Pi 3, but more rugged (but it still uses a micro SD, but a slightly more reliable one): https://www.reichelt.com/de/en/netpi-core-3-iiot-control-netpi-core-3-p279926.html?CCOUNTRY=445&LANGUAGE=en&&r=1 

I've not checked to see that version of Pi has the 40-pin connector that you need however. And you might need to re-engineer it anyway, e.g. in a custom enclosure. There are firms that will do all that for you (but at $$$$$ cost). It's unfortunately a time and money decision in exchange for what level of reliability/level of willingness to lose some data during such a trial phase (or select a less harsh time period or type of vessel, I don't know how feasible that is though). Perhaps you can test the reliability of a quicker prototype (using a normal Pi 4 and nothing else) somehow (e.g. just write known data regularly, and leave it on the next vessel, and see if it fails when it gets back).