Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: Davek0974 on September 21, 2016, 03:20:54 AM
-
If not, the idea was mine.... ;D
Having seen the carnage caused by the failure of a Z-probing routine for tool height a couple of times now, I was thinking if there was a probe plate on the market that was spring loaded - heavy springs will be needed, had the top surface as the normal probe plate and then as the springs compressed due to a failure of some sort, the extra movement triggered an e-stop?
Having heavy springs will make it appear as a solid to the probe routine, but still be able to compress and hit the e-stop.
Just wondering ;)
-
A few years back I designed a low profile tool setter which was intended as an open source project for all. There must have been something there that I had not realized because I was approached with a request to sell the design. Well I sold it (so cannot disclose any further details than have already been posted) but, to this day, I have never seen it on sale.
The related thread is here http://www.machsupport.com/forum/index.php/topic,21525.0.html and post #54 shows the overrun ring which triggered the EStop in case of emergency.
I use the prototype many times a day and it has never overrun so I consider the G31 to be 99.9999% reliable. Just why it behaves for some and not for others is probably one of those great mysteries that may never be solved.
If I built one, then there has got to be a fail-safe probe plate available on the market somewhere. ;D
Tweakie.
-
Hmm, seems they shelved it - i cant find it if it is out there :)
-
Some use a small cylinder with a couple of prox switches and a plate mounted to the end of the shaft. 1st switch seen as probe input, 2nd switch seen as estop. One end mounts to table/bed so that shaft is pushing in direction of spindle. Something like these. The right single acting cylinder (spring extend) may not even require air to work for this.
https://www.automationdirect.com/adc/Shopping/Catalog/Pneumatic_Components/Pneumatic_Air_Cylinders/Metric_Compact_Air_Cylinders_(H-Series)/H12M020MD-M
https://www.automationdirect.com/adc/Shopping/Catalog/Pneumatic_Components/Pneumatic_Cylinder_Position_Switches/4mm_Square_T-Slot_Cylinder_Switches
-
Thats quite a neat idea, i have no idea how accurate the trigger repeatability is for the proxy though - may not be good enough. I have a load of that stuff here so might be easy to test.
-
Decent switches will have extremely accurate repeatability. The rate the probe is moving at each time the switch is being made is likely the most important variable (slower = higher accuracy).
-
I have been doing a lot of probing on the lathe using G31. Have yet to have a problem and have tested it over and over for touching off tools, finding points etc.
BUT
In my probing for the lathe screen I manually jog to say about .1" of the surface and then probe at a SLOW federate. Repeatability is +- 0.0002" ( frankly I find it faster than the usual method of a double probe ......fast probe rate and then a slow rate ). At the slow probe rate. the amount of probe overtravel is less than most could even measure
( probe overtravel is calculable ).
To each their owne ..........
RICH
-
Solid advise Rich. :)
-
Yes a lot do use it and one of the fails i had was my fault although due to an unexpected issue - i was probing with a tool length offset applied and the routine did not like it so plunged the head into the bed at rapid rate :)
The other time was an engraving tool with a 0.2mm tip - might have been dirt or something but it drove that tool into the aluminium bed nicely :)
Both would have been prevented with a "safety" touch-plate so it does seem a good idea.
I have no idea how it works on a lathe though - do you have insulated tooling or still use a plate somehow or are you talking about an electronic probe itself???
-
what code are you using there is one out there that does crash if the home and work is under 25mm diffrences
-
This was the routine that comes with the 2010 screen set, it was most likely just crap under the tip or an intermittent ground connection - i was relying on the circuit through the spindle and motion bearings, slides etc.
-
Hi Davek0974,
I've had a look at Tweaky's probe and its really nice, his workmanship is superb.
I've not seen anything like it for sale and can only guess that whomever bought
his design went no further with it, a shame.
Except for a few operator bungles I've never had a problem with probing. Nonetheless
I have considered no 'overrun' to be less than desirable.
A good microswitch is not a bad idea. Both Ormron and Honeywell make a competitive
line of switches with good repeatability, hysteresis and overrun. I achieve .01mm repeatability
with 2.5mm overrun with one such switch, I also use the same series of switches for home switches.
More recently I became a bit more ambitious and started making a linear voltage differential transformer.
Project not finished yet but testing shows .002mm absolute accuracy with +-2mm stroke. The example
I'm working on has an overall height of 110mm so not compact. Subsequent designs will be about 50mm
height 4mm stroke and 1um accuracy.
Given that probing works as well as it does this is probably a waste of time but I'm having fun!
Craig
-
Its the fun that counts :)
Sounds good, no idea what a linear voltage differential transformer is though, but i'm sure i need one ;)
-
Hi Davek0974,
some good vid's on youtube about them. Reamarkably simple tho, three coils side by side
with the centre one energised by some AC voltage, few kHz is common, mine is 160kHz.
A magnetic core slips down the middle of the coils, if nearer one end than tother the outer coil
gets energised by transformer action, shift the core to the other end and that coil becomes energised.
Little bit of electronics and you have an infinite resolution linear gage with core in the centre
equal zero V.
They are available comercically as rugged contact free transducers. One with 1mm stroke but centre
accuracy of 100nm right through to 200mm and more stroke with centre accuracy of 20um.
I can't afford to buy one but I can make one! My current one offers 38mV/mm sensitivity and a linear
range of plus/minus 2mm. Not great but getting better.
Craig
-
Learn something every day, nice:)
-
I LIKE LVDTs, but the commercial ones tend to be $$$ (and these days it's my wallet, not someone else's).
What are you using as a core? Round lump of ferrite?
Any more details?
Cheers
Roger
-
Hi rcaffin,
yes that is exactly what I'm using.
I considered using a soft nickel-iron but eddy loss limits frequency which requires
multi thousand turn coils in VERY fine wire which I deemed to be beyond reasonable
means at home.
Ferrites on the other hand while being lower permabilty behave well loss-wise in the
100's of kHz and I can wind multi hundred turn coils in 0.2mm wire without undue
difficulty.
The current example offers 38mV/mm/Vexcitation and best measurement to date with
passive de-mod a noise floor of 200uV at 4.5V excitation for resolution of 1.2um.
If I can shrink future examples to 50mm overall height I might end up with something
useful at a fraction of the cost of existing commercial designs.
Craig
-
All understood.
How about going to active demod? Could seriously improve the resolution. There's an AD chip, but pricey, or you could improvise.
Cheers
Roger
-
Hi rcaffin,
when I say passive demod what I actually mean is a double balanced modualtor
based on two transformers with centre tapped secondaries and four diodes
in a ring. Such devices are used a lot at RF frequencies and some manufacturers
have models with input bandwidth down to tens of kHz. They are rugged/reliable/
simple and broadband. They all suffer some loss of RF to IF conversion in the
region of 5-7dB. They are commonly quoted with noise figures of 7-8dB which is
dominated by the loss. They are in fact quiet devices excluding the loss.
The actual voltage loss is about 2.5. I can readily amplifiy the output to whatever
level I desire and provided I use decent opamps used properly insignificant signal to
noise degradation. Active demodulation achieves the same result but all I end up
with is more signal not increased sensitivity.
The Analog Devices chip uses a different demod technique based on a ratio method
so avoids the requirement for a synchrous local oscillator. Great idea. The chip is
limited to 20kHz and so way lower than my device. Additionally see that they sell
for around $50US The handful of components I use cost less than $20 and in all
other respects the match for the AD device.
Sensitivity of an lvdt is related to it magnetic properties rather than signal processing
electronics. The higher the effective permability of the core at the frequency of interest
the greater the change in output voltage for a given distance of core shift. At low
frequencies nickel-iron or soft silicon irons outperform ferrites but become too lossy
beyond a few kHz.
If you're interested I could post a few pics and diagrams. I would be (already am!) guilty of dragging
this thread way off topic so not sure where to post.
Craig
-
Sure I would be interested.
Fwiiw, I am a retired research scientist (physics, metrology) and used to supervise several electronics engineers at the full technical level. So blaze away.
Cheers
Roger
PS: yeah, I reckon the AD chip is too $$ as well.
-
Hi All,
rather than further hijack this thread I've posted in 'show'n'tell' board as Linear Voltage Differential Transformer.
Craig
-
It's not really off topic so no worries :)
Damn interesting stuff.