Are Opto-Isolators a DIY Solution for Comm Errors?

[gerryv]

Too much time on my hands I guess. I've been reading where a lot of the DIY type folks install low-cost opto-isolator boards in their controller boxes cum PC's to deal with comm errors. An added benefit is some of these boards also boost the signal output to a full 5 volts. Might these be a potential solution for those who suffer from comm errors often caused by interference and low voltage output from their PC? especially if they could be supplied in a small enclosure with something like USB in and out ports so it could just be popped inside the main controller box or PC as the case may be?

Randall, are you there? :-)

[dana_swift]

Gerald- I suggest you proceed with extreme caution. The comm between the PC and the bot is either RS-232 or USB depending on the age of your bot. Neither of these system uses "full 5 volts" from an (buffered) optoisolator.

[[Note unbuffered optoisolators provide only a 0.3v pull down, being the open collector output of an NPN phototransistor.]]

RS-232 needs plus and minus 5 to 15 volts, this could be achieved using commercial RS-232 isolators specifically built for the purpose. A simple off the shelf opto might work, if both sides of the RS-232 meet specification RS-232-C. (Go read about it http://en.wikipedia.org/wiki/RS-232 )

USB is much more complicated, it uses bidirectional LVDT signaling which 5v would possibly damage. I have not paid attention to the availability of dedicated isolation devices, I suspect such things have been built and can be purchased.

Once you have gone to all that trouble I doubt your comm problems will improve any, as a much more likely culprit is "services" running on your PC that you are unaware of. It is very unusual for either RS-232 or USB to malfunction in the presence of electrical noise, as each of them were designed for that purpose.

A much better direction is to use a free utility (download) from Microsoft "process explorer" and see what hidden things are using your pc you didn't know about. Typically there are 50-100 programs running simultaneously hidden from you to make your computer "friendly". These processes can take the CPU away from the SB3 program and keep it "too long" for the required (undocumented) communications handshakes. That is when I get comm errors.

I have a very old XP laptop I run my bot with, and having only 50% communications according to the SB link test, I don't get comm errors because I went into the advanced settings and nuked about half of these unnecessary CPU hogs. That is not for the faint of heart, as it is easy to leave yourself with a completely dead pc, hard to recover from shutting down the wrong things.

Easy solutions are faster computers so the hidden processes get done sooner, and the eternal USB 2.0 hubs documented on the forum.

Find the right problem and fix it. It COULD be that isolation is what you need, but I will bet against it.

D

[gerryv]

Hi Dana. Thanks much. Actually I'm not looking at doing anything on mine because I don't have these issues. I'd been on some of the DIY machine shop forums because I'm looking at converting my small metal lathe and mill and noticed this topic kept popping up. Because I knew that some of the fellas here had similar issues for various reasons I figured I'd simply post it for others benefit. Are you then saying that opto-isolation is not the answer or it's just not that easy. Sorry, I'm completely in the dark in this area. Actually, it was you I was thinking of when I posted this - Randall is probably wondering why I asked him. Call it a case of my confusing gurus (chuckle).

[michael_schwartz]

My feeling is that the potential risks would outweigh the benefits unless you have the hardware/engineering background to understand what is going on, and do this correctly.

Furthermore I would assume that there is a good chance that any com issues are due to another problem and that this might not solve the issue at hand.

I would only view this as a DIY solution if the project were more along the lines of an experiment, and a certain degree of risk has been accepted.

As far as my technical background I probably couldn't tell you what opto-isolation is.

[Brady Watson]

99% of all comm errors are a direct result of computer hardware. There are a few situations where grounding causes a problem, but more often than not, it is the computer to blame.

You have 2 choices: 1) Bolster your existing PC hardware with a PCI to USB expansion card + 2.0 USB hub. This should keep any other USB peripherals from interfering with the ShopBot's communication stream, such as a keyboard or mouse, and open up a 'Windows Enhanced Port'. 2) Build your own computer with good quality name-brand hardware that is robust enough for CNC duty, along with hardware mentioned in 1).

I know many have never had the opportunity to visit or interact with ShopBot's Engineering department. They have worked very hard to eliminate communication issues, and beef up the electronics so that they are up to par with what is suitable for a CNC tool. Even with my intimate knowledge of ShopBot's electronics, I was still very impressed by how much thoughtfulness, care and ingenuity goes into testing, evaluating and specifying hardware and software for our tools. All of this is done in-house - a ShopBot isn't another re-branded off the shelf controller. There is a lot to be said about that.

I am sure if a few opto couplers would cure all communication problems, then they would have been implemented long ago. I bet those that have persistent communication problems have never implemented the recommended USB/PCI card and/or have the 2.0 hub shipped with their ShopBot laying in the same box as their ShopBot mouse pad & hat. :rolleyes:

Got Com problems? Look at your computer first. Thousands of reliable ShopBots in operation can't be wrong...Sorry if I sound like I am on my soapbox, but - I have yet to see a Bot that still has com problems after properly grounding the machine/dust collection & getting the computer's USB interface up to par.

-B

[michael_schwartz]

99% of all comm errors are a direct result of computer hardware. There are a few situations where grounding causes a problem, but more often than not, it is the computer to blame.
-B

I would agree.

The com problems I had shortly after setting up my machine turned out to be a $30.00 fix. That involved replacing the generic USB keyboard and mouse I had been using with name brand equivalents. Problem solved... Junk hardware belongs in the trash.

Since then I have never had a problem while the machine is in operation.

The Dell I am using is going to go into the parts bin as soon as I find the time to build a replacement as I still have occasional com errors resulting from inserting/removing my USB thumb drive.

[bcondon]

The only time I have comm problems is typically when disconnecting a finger drive. This happens maybe 1 out of 10 times...

BobCondon

[gerryv]

Interesting and informative. The reason I'd asked the question was because I'd formed the impression from reading others posts that besides grounding, many of the problems seem to suggest the PC hardware and/or OS so it struck me that it would be nice if such a simple approach might be helpful but obviously it's not that simple. Ah well, I'm just very happy that my semi-old PC seems to be doing okay. Now I understand a bit more too :-)

[richards]

Most Opto-Isolators have slow response times (closer to millisecond range instead of microsecond range) - too slow for RS-232 and USB, even if the voltages were adjusted.

Putting opto-isolators in a proximity sensor circuit worked for me. Putting them in a push-button circuit also worked.

The proximity circuit modification worked because I was able to use 12VDC to power the sensors. The new sensors would not work on 5VDC, but they would work at 10VDC to 30VDC. Because the cost was in having a prototype board manufactured, I added a microprocessor control to the circuit to enable me to combine signals and to give a visual indication (flashing L.E.D.) that the circuit was working.

The push-button isolation worked because my wiring to some of the push-buttons was over thirty-feet long. That made a great antenna to pick up electrical noise. The slowness of the opto-isolator eliminated almost all of the noise.

Using opto-isolators correctly and for the right purposes will help isolate the controller from electrical noise, but they are not a substitute for proper grounding.

(I've seen industrial machines that took the concept of opto-isolation even further - they used fiber-optic cabling for all signal lines outside the controller.)

[gerryv]

That's interesting Mike. Out of interest, I'm sure there's a very good reason - maybe "perceived" extra cost - why a robust but simple PC isn't built right into the controller box? Maybe that would just worsen the interference challenge? Not that I'm planning on trying it :eek:

[steve_g]

Gerald:
My first experience with CNC was with a single axis feeder for sheet metal coil stock. The controller (the size of a refrigerator) was built into the machine. Everything was proprietary and required factory service reps to repair. I for one am glad that shopBot decided to use the customers PCs for the data crunching, I can update to my hearts content or not. I can also assure you that the “built in” controller was not without problems!

Steve