|Les Jones 1||10/05/2020 14:24:03|
|2134 forum posts|
7 BCD Scales
Static state of clock line between frames Low
Clock in the data on the negative going edge.
28 bits in a frame of data
CPI 2540 (Metric)
LSB sent first
Static state of clock line between frames Low
21 bits in a frame of data
|5921 forum posts|
Timebase should be good because the scope software uses the sound-card which has an accurate clock.
I've had a half-hearted go at measuring lathe vibration and a slightly more serious attempt at detecting track vibration from Dynamometer Car data. Duncan Webster's been experimenting with a dynamometer car for measuring steam loco performance. He measures and logs rpm and drawbar pull from which engine power and other data can be derived in the time domain. I was interested in analysing the same data in the frequency domain to see if spring in the load cell coupling was a problem because Duncan suspects this might cause errors, and - my mad idea - to see if vibration caused by track defects could be detected as a maintenance aid. Unfortunately load data is measured once per revolution, which doesn't translate straightforwardly into frequency because loco speed varies.
Underlying problem with both lathe and dynamometer - my maths! I got as far as understanding the dynamometer data needs to be munged into regular frequency form before applying an FFT, but don't quite understand how to do it. Nor do I know how to apply FFT appropriately either! I got as far as deciding to go back to lathe vibration because the data comes off in frequency form and then stalled.
Got interested in vibration ages ago after reading it was the first test applied to a newly installed industrial lathe. Said to be vital to finish and done before checking for run-out and parallelism etc. The tool used before electronics doesn't look difficult to make: About 12 pendula are suspended on piano wire from a gantry mounted on a stiff plate about 18" long. The pendula are each tuned to a different period. The gantry is plonked on the ways and the machine taken through it's speed range: when the machine vibrates, one or more pendula swing in sympathy and indicate the period. The hard part is working out cause from the observed frequency and the rpm, and how to dampen out any resonances.
Don't know how common vibration is in home workshops. My WM280 lathe had a couple of resonances, much reduced by putting roofing felt under the stand as a damper. Before installing the damper I just avoided the guilty speeds, about 800 and 1250rpm. Like as not root cause was failing to bolt the stand down because I didn't want to puncture my garage floor's DPM.
1656 forum posts
Just saw this fwiw.
|Andrew Johnston||10/05/2020 16:11:02|
5553 forum posts
I'm not surprised; it's hardly a tutorial. It's incorrect to state that the Fourier transform of a real function is even. The real part is even, but the imaginary part is odd. Of course the magnitude will be even.
Strange that they don't mention the number of samples per transform. While the number of samples can be any highly composite number, the original FFT works best when N is a power of 2. To the extent that if the data set isn't a power of 2 long, zero value samples are added to pad out the data sequence.
|Robert Atkinson 2||10/05/2020 16:25:52|
698 forum posts
The Picoscope software has an FFT function built in so they also act as spectrum analysers. Their automotive software has specialised vibration analysis and balancing functionality and they sell accerometer kits for it. https://www.picoauto.com/products/noise-vibration-and-balancing/nvh-overview
5290 forum posts
Some interesting responses there, thanks. It seems there is always someone on here who has some experience in just about every topic one can come up with.
|Robin Graham||13/05/2020 00:02:05|
|725 forum posts|
Thanks for replies - too many to respond to individually, but all appreciated and I've followed up on suggestions..
I was asking about 'kits' (in the sense of DIY soldering etc) because I thought they were the only alternatives to 'proper' bench scopes. Much as I'd like a proper scope, I can't justify the cost for what I do.
From the discussion it seems that there is a middle ground - pico etc. But for the mo it seems that the KKmoon offering that Neil recommended will do what I want for not much money.
Duncan - many thanks for your kind offer. Your old pico may well be good enough for what I need to do. I'm typing on an ancient computer with a parallel port! All my ancient computers run Linux though - maybe that would be a problem. I'll PM you.
Les - thanks for your post giving data on the various protocols used by different manufacturers - it should be made a sticky.
|Brian Oldford||01/06/2020 16:27:06|
673 forum posts
If you're not in a rush. **LINK**
Or if you want it quicker **LINK**
|Leslie Rix||01/06/2020 17:31:52|
|10 forum posts||
|Clive Foster||01/06/2020 18:29:30|
|2245 forum posts|
For pulse timing applications the more channels the better, within reason.
Chasing timing errors down with a two channel scope can be a miserable experience because you pretty much have to devote one channel to a reference so with several potential pulse string to pulse string error possibilities you are forever swopping around. Not ideal for the inexperienced.
8 channels are much nicer in my experience but then connections can be difficult. Once all hooked up youhave a clear view tho'. I often found 16 too many.
Bigger screen helps.
Ideal for pulse timing and similar would be a very limited box able to display pulse trains linked to a tablet for the display and control. At modern electronics prices a one job tool is viable. Needs to be buy, not build, with robust software and a manual.
|Leslie Rix||01/06/2020 19:04:58|
|10 forum posts|
I have an Instrustar IS205A usb, datalogger and spectrum analyser which does a fair job on frequency analysis but no way can I get a printout. Current cost is about £65 from China via ebay.
Also have a Hantek 6022BE similar usb scope but not sure it does FFT; cost around £40. Best to search both on ebay for full specs.
Sorry I didn't get back to you on PM re. vibration lathe analysis but I just bought a Raspberry PI and I think I have fried my brain. What the learning curve is I daren't guess. Over the years I opted away from an electronics career to be a noise and vibration consultant - my workshop is testament to the deep allure of swarf and oil.
Basically for a manual lathe f(0) is mandrel rpm dictated by balance quality riding on background slush (noise) generated by lubicant swirl and sundry resonances plus motor speed and electrical noise. All other frequency peaks relate to gear meshes and bearing noise don't forget beat frequencies including relative speeds and rolling element speeds. Hewlett Packard used to do a handy booklet on this.
Hope this helps, Les
|5921 forum posts|
Thanks Les, very interesting! I didn't know it was possible to buy a logic analyser for that kind of money. I'm going to order one!
I sympathise with your RaspberryPi shock - that little card has 50 years of software complexity in it! Hours of 'fun' ahead!
Your last paragraph opened up some interesting ideas as soon as I followed the clue. For example, motor vibration caused by 'soft foot' , which is when the motor doesn't sit evenly on it's mounting. Detected by putting a DTI on the motor and slackening eacjh bolt in turn; if the DTI registers more than 1 thou of movement the foot should be shimmed. I'd never have thought of that!
|Andrew Johnston||01/06/2020 20:48:14|
5553 forum posts
They're available and called logic analysers They concentrate on displaying time relationships between channels and are less concerned with voltage levels. They usually display a channel as 0 and 1 based on the levels of a selected logic family. Relatively cheap ones are available, although I can't remember the details. On the other hand fancy ones (32 channels and upwards with bandwidths into the GHz) can approach £100k. The better ones can morph a number of channels into one line with a binary or hex equivalent of the channels. When looking at address and data buses that's often more useful than displaying individual lines.
28 forum posts
Go with the moderator mate, save yourself money and KMoon is incredibly clever - I have used several of their innovative devices - I have Hameg scope and all sorts of radio analysis equipment, but Kmoon takes value for money - I got one of these little scopes - way acurate enough by it's very nature, and a little signal generator (same people) - totally impressed - scope even comes with a probe!
Please login to post a reply.
Want the latest issue of Model Engineer or Model Engineers' Workshop? Use our magazine locator links to find your nearest stockist!
You can contact us by phone, mail or email about the magazines including becoming a contributor, submitting reader's letters or making queries about articles. You can also get in touch about this website, advertising or other general issues.
Click THIS LINK for full contact details.
For subscription issues please see THIS LINK.