![]() |
#1
|
|||
|
|||
Complex proxy suggestion
Hi
I see the suggestion to series Proxies in http://mechmate.com/forums/showthread.php?t=584 From reading http://www.automationdirect.com/stat.../proxterms.pdf I found the technical data for applying these proxies with the differences between steel and Aluminium in numbers for all you engineer types. My thoughts / will it work question is: Can we connect four proxies to eliminate the use of relays or logic by choosing three PNP NO sensors and one NPN NO sensor. All four sensors 0V connections joined together at the ground potential for BOB. The first PNP connected to +(12-24Vdc). The ouput of the first PNP connected to the second PNP +ve power lead. The second PNP output connected to the third PNP +ve power lead. The third PNP output connected to the NPN +ve power lead. The NPN output lead connected to the input lead for BOB. Discussion: I think this eliminates the use of logic and relays. Giving cost and complexity savings. It requires a complete path in normal operation, good for a stop system for rail jumps. It might be slightly slower to operate at electronic speeds if the sensors have any internal capacitance. This dissadvantage would be eliminated by determining how fast the machine can be driven towards home, I think this has to be done anyway. Finding which has detected a hole is a matter of following the path from the first PNP toward the NPN sensor rather than looking randomly at the LEDs. Usual digital Inputs and Outputs are used minimising noise issues. (Shielded cable no earth loops assumed.) No need for load resistors as each sensor is the load for the next or the pullup resistor inside BOB is the final load. Only three wires to BOB. (Assuming using 12V from BOB or BOB location) Further testing: I am not quite up to installing proxies, I just welded the table. I am using Al rails so I need a source for 12mm diameter proxies (Ideal) with a sensing distance greater than 4mm. Or I need to know if the al rails can easilly tollerate using 18mm diameter proxies. If no one, Mike ![]() Regards David |
#2
|
|||
|
|||
The PNP type sensor sources current and the NPN type sensor sinks current, so if three PNP proximity sensors (A, B, and C) were daisy chained and fed an NPN sensor (D), the circuit MIGHT work.
Assuming 12VDC and a voltage drop of 1V per sensor. 'A' would output 11VDC, 'B' would output 10VDC, and 'C' would output 9 VDC. 'D' would sink about 8VDC at the break-out-board, so you would still need to reduce that voltage to 5VDC via an opto-isolator or relay circuit. The voltage drop depends on the sensor. The minimum voltage of all my sensors is listed as 10VDC, but I have tested some of them at 9VDC. If the sensors did not turn on fully, because they were sensing aluminum instead of steel, the PNP sensors might not source enough voltage to run the next sensor in the daisy-chain. You would still be faced with the voltage difference between the 5VDC input on the break-out-board and the voltage output by the NPN sensor. (Yes, I know that a NPN device SINKS current, but when it is OFF, the output would float to a voltage higher than the voltage allowed by the BOB. Most chips are specified to have an input voltage no higher than VCC, which would be 5VDC on the BOB.) |
#3
|
|||
|
|||
Instead of mixing sensors types which I think is not a good idea why not just make a small "breakout board" for the sensors with some transistors? The PCB could be machined on the Mechmate.
A transistor, resistor and optocoupler per sensor and maybe something else should do the trick I think. |
#4
|
|||
|
|||
Hi Pablo
My aim is to eliminate the need for these extra components. Thanks for the comments Mike I think your 1 V per sensor will be about right (1.2 max on the spec sheet) so the minimum voltage to supply the circuit would be the minimum operational voltage plus three times the 1.2 volt per sensor drop. So some sensors may need slightly more than 12Vdc input to the chain. The specs for the digital sensors all show a repeatable threashold for the aluminium at a different distance than for steel so no risk of a different current with aluminium other than the position it switches at. (Thats what would happen with an analog proxy) I dont think that the floating output when the last NPN is off is going to be an issue. The very small leakage current (Link below shows this leakage current is less than 120 micro amps.) is only going to the input of the PMDX-122 which is a pullup resistor with series LED and a 10k ohm series resistor going into the buffer. This would be like using a single proxy to BOB. However if a 4.8 V zenner was added from the input pin to BOB and ground this small leakage current would not be able to drive the input above 4.8 V when the NPN was "off". http://www.automationdirect.com/stat...prox12mmam.pdf This would give our parts count to be Four proxies and one 4.8V zenner diode. Cheers David |
#5
|
|||
|
|||
David,
I was going to mention the Zener diode, but I decided against it because a zener diode requires a minimum current to flow before it regulates, usually slightly more than 1mA. I haven't tested proximity sensors as they switch on/off to know what happens as they change states. My untested feeling is that the state change is relatively long (as far was electronics go) and that during that transition state, a zener could fall out of regulation which would mean that the input could see full voltage. |
#6
|
|||
|
|||
Hi Mike
(I had to go back to the books for this one.) Yes the zenner will fall out of regulation at the low current however when it does the voltage must be under regulation voltage not over. I think I am at the point of testing. Unless anyone has a chinese source with the full data sheet for the cheaper sensors I will order from automation direct so I can get the data and the extended sense distance to suit my Al angle on the rails. (I have not yet got my head around the implication of a change to 18mm diameter) Does Jed or anyone from Western Australia want in on the order? (I have not purchased from automation direct nor have any links with the company) Regards David |
#7
|
|||
|
|||
David
I may, Rosses issues and some recent discussion has got me thinking I'll have a look into it and let you know. Jed |
#8
|
|||
|
|||
Ok Jed
It was the oscillation that Ross experienced that made me review the series connection as his battery charger must have been pulsing at 50 hz(or100) and I thought this is an acceptable detection time and if the prox had much capacitance this oscillation would not occur. I quickly ran this idea past another lecturer (who teaches Mechatronics) this morning and he said it should be fine. If you are going to order we can test a set up for steel rails (yours) and Al rails (mine) I am still thinking about where to put the NPN sensor. I am thinking of the X axis closest to the cable chain as this maximises the distance from the router / spindle as noise on the PNP sensors is damped (I think) by the following stages. (screened cable with one end of the screen earthed and single point zero volt connection for the sensors at BOB.) Cheers David |
#9
|
|||
|
|||
Hi David
Have been reading the posts and keeping an interest in your sensor proposition. regards Ross |
#10
|
|||
|
|||
Thanks for the note Ross
I suggest you continue with the sensors you have rather than buy three new ones. It will be a month by the time I get some from the US and test them Regards David |
#11
|
|||
|
|||
Hi all
I got my proxies in the post and tested on the bench: connected four proxies to eliminate the use of relays or logic by choosing three PNP NO sensors and one NPN NO sensor. (Due to my Al rails I chose three PNP NO sensors with 8mm sense distance.) I chose an 8mm diam NPN NO with a 3mm sense detecting steel on the z axis. All four sensors 0V (Blue) connections joined together at the ground potential for BOB. The first PNP brown + connected to +(12.7 Vdc). (Less than 10% ripple) The ouput of the first PNP (Black) connected to the second PNP +ve power lead(brown). The second PNP output (Black) connected to the third PNP +ve power lead(brown). The third PNP output (black) connected to the NPN +ve power lead (brown). The NPN output lead connected to the input connection pin 11 on BOB. PMDX 122 (Zenner diode discussed elsewhere was not necessary) I had to run from a minimum of 12.4 Volts to achieve the stated 10-30 Vdc range for the power on the third PNP proxy. The NPN has a voltage range of 6-30 V dc so the third was the crucial set up point. Only (slight) dissadvantage is that all lights after the one that is exposed to a hole go out. Saves on relays, decreases parts count. Cheers David |
#12
|
|||
|
|||
If all your sensors go out when you have a problem, you are not going to enjoy your trouble shooting. It is so much easier to locate the problem sensor when only the light on the problem one goes out.
|
#13
|
|||
|
|||
David
Thanks for the investigation as the price reduction here is significant if the relays are not used. Seems I still have a foot in both camps as one relay is still on the way from a wholesaler for testing and have also queried my China supplier for a supply of PNP proximity sensors. Either way you have given us with a further option for the sensor wiring and depending on our own preferences we can now select which option we prefer. Regards Ross |
#14
|
|||
|
|||
Thanks for the note Ross
Nils The sensors light in the sequence they are powered. So they dont all go out. Look at them in the order they are wired and the first that is out is the one that started the limit trip. Regards David |
#15
|
|||
|
|||
A good feature but I really question the need for it. My machine has yet to ride up...
|
#16
|
|||
|
|||
This is just added safety. Pluse it is cool to have folks collaborate on new ideas
![]() |
#17
|
|||
|
|||
Although I really admire new and different ways to handle problems, I tend to favor the tried and true methods that have been around for a long time.
On my own workbench, I've been able to design different ways to solve old problems, but many times, my "new and improved" methods were not nearly as reliable as the old and proven methods. Eliminating relays or a circuit board is a justifiable reason to experiment, but assuming that all proximity sensors will function the same as the sensors in the test circuit may lead to problems. Most electronic parts have data sheets that list specifications. Ignoring those specifications because the "test circuit" worked is inviting future problems. I'm not saying that this new method is faulty or that anyone will have a bad experience by doing something different. What I am saying is that many builders don't have the electronic background to experiment. Just because something works today does not mean that it will always work. The "standards" in electronics takes into consideration all of the possibilities. Saving $100 on a $6,000 machine may cost much more than expected. |
#18
|
|||
|
|||
Hi Mike
I think if a builder wants to walk the "tried and tested methods" the best suggestion I have seen on this site regarding proxies is yours with the opto isolators. I dislike the use of relays in this instance as digital signals are being converted to mechanical signals and back to digital again. So I have chosen a simpler, cheaper route. Relay contacts need a minimum current called a wetting current to operate. see http://en.wikipedia.org/wiki/Wetting_current or http://www.control.com/thread/1026229568 This is the "specification" that is not being met with the relays. If there is a specification I am ignoring I would like to know what it is. regards David |
#19
|
|||
|
|||
David,
The specifications for my proxy sensors give a range of voltages over which the sensor will give a reliable result. The specifications also list the current required for reliable operation for those sensors. The kind of problem that worries me is that running a sensor near its lower voltage reading can eventually lead to errors or daisy-chainning sensors can starve individual sensors. Heat causes power supplies to put out less than the anticipated voltage. It also causes devices to draw more than the anticipated current. In my final designs, I like to use a voltage for sensors that is well above the minimum and well below the maximum. I like to verify that I have at least 2X the required current available. I like to also verify that the sensor can trigger at 2X the 'gap' and that the distance traveled during the transition is not measurable. There is nothing about your design that causes alarm, but I would put your design at the 'alpha' stage, meaning that it is a good starting point but that time and testing might cause you to modify your design. It is not uncommon for me to have more than a dozen designs before I finally decide that any further changes would only be a waste of time. Not all of those dozen designs ever get built, but most of the time readings with an oscilloscope show that what I had assumed to be a good design needed 'adjustment' before it was 100% reliable. Forums like this let us work together to find that 'ideal' design without the need for each of us to go to unnecessary expense or to waste too much time to find that perfect design. Varying the design allows a whole slew of variables to enter the picture. Each of those variables must be addressed and tested before the design can be certified as being reliable. I'll admit that I'm guilty of over-designing equipment, but doing that eliminates the nit-picky errors that drive me nuts. To me, there is nothing so frustrating as facing a dead machine and then finding that the problem was caused because the designer trusted untested specs. In most of my work, I deal with customers who can't afford down time. It's not unusual for down time to cost more than $2,000 per hour. They can't afford a minimal design and I can't afford to lose a customer just to save a few chips. |
![]() |
|
|
![]() |
||||
Thread | Thread Starter | Forum | Replies | Last Post |
A nice tool to draw complex patterns : marabesk.com | grandpi | Utility software | 2 | Mon 04 January 2010 16:59 |
gauge bar suggestion | lumberjack_jeff | 10. Base Table | 2 | Wed 17 June 2009 05:34 |
Complex 3D volumes to 2D parts ready to cut. (Flatten twisted surfaces to 2D sheets) | grandpi | Graphics & CAD | 4 | Fri 18 July 2008 09:04 |
Limit / Home switches - Mechanical or Proxy? | fabrica | 70. Control Systems | 44 | Thu 21 June 2007 01:07 |