PRT-Alpha prox. sensors

[richards]

Last Friday evening, after Shopbot was closed, I started having intermittent problems with my X-axis proximity sensor. I had some BALLUFF BES M12Mi-NSC40B-SO4C proximity sensors on hand, but they required 24VDC. The PRS-Alpha uses 24V sensors and feeds directly to input #2 and input #3, but without schematics and without being able to ask someone whether my PRT-Alpha had the same input circuitry, I decided to replace both sensors and to build a small interface board using opto-couplers to isolate the 24V from the 5V logic. At the same time, I added a 74LS08 AND chip so that input #3 would have only one device connected to it.

The BALLUFF sensors are rated for a distance of 4mm. With feeler gauges, I set the distance to 2.5mm. The higher voltage sensors are dead-on repeatable. Using a digital dial indicator, there is NO variation in the sensing location.

I'm attaching a schematic of the interface board (hand soldered on a 2" X 3" Radio-Shack proto board).


292

[jnbish]

What?

[beacon14]

I think what you just heard was the sound of Mike's electrical engineering knowledge going right over my head....

[Gary Campbell]

Mike...
As you state the PRSa uses a 24V prox sensor. Do you think that the higher Sn spec is responsible for increased accuracy,the higher voltage sensor, or is it due to the opto/electro isolation of the sensors?

Did you keep the OEM PRT NPN NO configuration?

Reason I ask is I have added 2 NPN NO Sn=1 sensors to my Z & A upper travel (in parallel) and sometimes get inconsistant results.

Thanks for your time.
Gary

[richards]

Gary,

My PRT-Alpha used 12V proximity sensors on a 5V circuit (at least the data sheet showed that the sensors were rated 12V). To make them work reliably, I had to grind the heads of the target bolts flat (a tip that Brady W. posted). The two sensors were also connected directly to the same input port (#3).

Because the original configuration worked and was repeatable within 0.003 inch, I never closely examined the circuit. However, when my X-axis sensor started to act up, I decided to change things.

A basic rule of thumb is that any input should have one and only one connection (fan-in). However, a single output can drive more than one input (fan-out). I decided to add the 7408 AND chip so that I only had one output going to the #3 input. (Many process control computers use the same design as Shopbot, so the design is accepted, just not optimum. With transistor logic, it us usually reliable.)

I stayed with NPN (sinking) logic and I used sensors that are NO (normally open). Shopbot uses NC (normally closed) sensors on the PRS machines. NC is better, but they are not available off-the-shelf from my normal supplier.

The opto-isolation's function is to allow a 24VDC device to connect to a 5VDC circuit. The opto-isolators actually slow down the signal slightly (propagation delay). However, compared to the original sensors, the BALLUF sensors have a high SN (signal to noise) ratio. With the 2K resistor in the circuit, the BALLUF draws about 10mA through the internal LED of the opto-isolator. That means that the opto-isolator's photo-transistor is getting sufficient signal to operate properly.

Finally, each opto-isolator feeds a separate input on the 7408 chip. That keeps the output transistors on the opto-isolators from fighting each other.

The 7408 chip has eight inputs, so you could still use that one chip to handle at least four sensors.

I just noticed that I didn't give the values of the resistors on the schematic. R1 and R2 are 2K, 1/2-watt, R3, R4, R5, and R6 are 4.7K, 1/4-watt. Resistors in the range of 1.5K to 3K would work in place of the 2K resistors and resistors in the range of 2.2K to 10K should work properly in place of the 4.7K resistors.

One thing that I always try to do with proximity sensors is to use extended range sensors (4mm) instead of standard range sensors (2mm). By doing that, I can set the distance between the sensor and the target in the 2mm to 3mm range and still get and excellent signal without the worry of crashing a sensor into the target.

[Brady Watson]

I think what Mike is saying essentially is...the 12v sensors, if used at their rated 12v, are more reliable than using the same sensors @ 5v. Opto-isolation helps to buffer the signals coming in & cuts down on false triggers caused by voltage spikes (usually static electricity) - I'm pretty sure his electronic 'black box' filters out a lot of the noise issues, and improves signal strength and operating range of the sensor itself, since a 12v sensor used @ 5v has less usable range than if it is supplied it's as-designed 12 volts.

Rather than listing parts and schematics to woodworkers, perhaps an assembled kit would be more appealing to those that want better accuracy & repeatability with their proxy switches? Just a thought...

-B

[gabepari]

I had some 7408 chips with my PB&J today for lunch


Mike, very good info. Cached and filed into my "do when I get time" folder.

Thanks,

Gabe

[Gary Campbell]

Mike...
Thanks for the info. I checked today and actually had the prox about .080 gap from the trigger bolt. Thats a little fat for a 1mm sensor.

I used the NO config so that I could parallel 2 on 1 input, as I am getting low. (Still jonesing for the new board!)

We have virtually eliminated all eroneous stop hits with the shielded cables to the gantry. I was looking to possibly get rid of the last few with your isolator board. Thanks again.
Gary

[bcammack]

Gosh, I get all nostalgic hearing all this electron-pusher lingo. Reminds me of my wasted youth in the business back in California.

[richards]

Brady,

I just sent an order off to have a few prototype printed circuit boards made for the PRT-Alpha machines. The boards will have separate connections for up to five proximity sensors. After isolation and gating, the interface board will output one signal that can be connected to Input #3 (or other input [1 - 8], if desired).

If there's sufficient interest (at least twelve units), I'll assemble some kits.

[gabepari]

Mike, put me on the list.

Thanks,

Gabe

[richards]

As they say, "A funny thing happened while waiting for the prototype boards ... ". I designed a microprocessor controlled board (3.8" X 2.5") that has five opto-isolated inputs and six TTL level outputs. Because it is computer controlled, it can be programmed to use Normally Open inputs or Normally Closed inputs.

The way I see it, that board would make a nice interface to those of us who use the PRT-Alpha machines. Two or three inputs could be used for proximity sensors, one input could be used for the Z-Zero plate, and one or two inputs could be used for the E-Stop switches (as long as the total number of inputs was five or less).

Cycle time to poll the inputs is 26uSecs, so the controller can check each input 38,000 times per second.

Because it is controlled by a micro-controller, proximity sensors can use a common output (reducing the number of I/O pins dedicated to prox. switch sensing on the Shopbot controller).

An L.E.D. is on board that can be programmed to flash at various rates to show the status of the various input signals. (Sorry, but there is not enough room on that board to run a 7-segment status display, which would also require a 2nd micro-controller chip - but I do have those items on my prototype tester that I built to test the idea).

The board is designed to be used with NPN type sensors, which means that it would also work with switches that are active when they SINK current (like the Z-zero switch and the E-stop switches). PNP type proximity sensors would not work with this board without inverting the opto-isolator circuitry

Anyway, you get the idea. Cold weather, gout, too much time on my hands and I hide in my process computer design corner and play with micro-controllers.

[richards]

I've had a month to play with various proximity sensors and finally found one that was basically a bolt-in replacement for the original sensors supplied by Shopbot with my PRT-Alpha. (The sensor to the extreme left in the photos.)

That sensor has a sensing distance of 4mm, although I have my sensors set 3mm from the targets. The sensor is about 50% wider, so it is not nearly as finicky about the target. In the photos, you'll also see that I've also connected a quick-disconnect adapter onto the sensor so that replacement is quick and easy without the need to splice into the existing cable.

The new sensor only required an additional washer on the X-axis so that it was able to clear the target. The Y-axis worked without modification. Of course, I had to change the offsets slightly to adjust for the minor differences between the original sensors and the replacements. The offset adjustment was about 3/16-inch.

For reference, there is also a photo of some of the other sensors that I've tested, both 12mm and 18mm. Some have quick-disconnects and others have a permanent cable attached to them. All of the sensors worked perfectly. I particularly liked the 18mm sensors, because they have a default sensing distance of 5mm.

The circuit board is one of the prototype boards that I use to connect the 24VDC sensors to the Shopbot controller's inputs. So far, in about one month's usage, everything has worked perfectly.


293
294

[jastod]

Mike: If you still have any unclaimed prototype circuit boards, we'd be interested. Thanks.

[richards]

Alison, please send an e-mail (miker@xmission.com (mailto:miker@xmission.com)) and I'll see if I can help.

[lukejenson]

Have you looked at TURCK prox sensors.

http://www.clrwtr.com/TURCK.htm