when to isolate

Owen

I'm curently building a robotic device designed around the propellar chip. The device has accessory pushpull conectors on the case for future serial accessory use. Is there somthing I should/could do to protect these io pins? is it ok to have them conected to the outside of the case without opto isolation? when is it importient to isolate the pins on the propellar?

Thanks,

Owen

Graham Stabler

Generally you will use opto-isolation when connecting between your logic system and a higher voltage power system, for example in a CNC machine where you might have 70v on the motors and if something when pop it might otherwise get fed to the logic. They are also used to couple logic systems from the noise free logic circuitry to noiser power circuitry.

The choice of whether to use opto-isolation also depends on your confidence in your building. There are a lot of off the shelf modules with opto-isolation that strictly could live without them but as the vendor cannot guarentee that the user knows his stuff it can be a much safer belt and braces approach.

For what I'm interested in I'll probably use inverter chips as buffers to couple with my 5v systems and only use isolation in the case of noise problems. I'll also be interested to hear what others have to say on the subject though, good question.

Graham

Gavin

Owen,

There is a sticky note about I/O interfacing.
If you include a resistor in series with the pin the internal parasitic diodes will protect the pins to some extent.
Depending on the wiring length etc you could add external protection circuitry like transorb diodes etc then optos for longer wiring.
If it is a mobile robot and small then just the resistors should be enough, high voltage servos and fixed robot then add as much as you can. Welding robot then optos, ferrites, inductors, commonmode chokes and a bunch of stuff.

Gavin

LoopyByteloose

Isolation is often good protection in a habitual 'hot plugging' situation. In some cases, people impulsively want to reconfigure without first powering down the device.

For instance, in the Hydra the game card anticipates this problem and will tolerate 'hot plugging' with no problem at all.

But both keyboards and mice [noparse][[/noparse]which can use 250ma] cause problems with hot plugging ON ANY DEVICE and may damage themselves and other things. Documentation consistently warns against removing or installing these devices after a computer is turned on.

I have yet to see anyone actually propose opto-isolators on keyboard or mouse interfaces, but I am considering them as a more durible solution and they actually can allow one to jump from 3.3 to 5 volts at the same time.

Of course there is an added cost due to the added part count. So if it isn't a problem, isolation can be omitted.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"If you want more fiber, eat the package.· Not enough?· Eat the manual."········

···················· Tropical regards,····· G. Herzog [noparse][[/noparse]·黃鶴 ]·in Taiwan

Post Edited (Kramer) : 12/17/2006 8:36:42 AM GMT

Owen

In my aplication (which is a device that moves a camera in x y and later maybe in z axis taking still pictures at indexed positions) I have external accessories that will most likely be hot plugged many times through it's life. I'm not trying to go from 3.3v to 5v I just want to protect the propellar pins from harm. Would series resistors be enugh to protect the pins from hot pluging? also how woried about ground loops and noise should I be with accesories that plug in externaly? I will probably have 2ft to 10ft between the device and the accesory.

Owen

Lawson

another trick I've seen on Hot-Pulgging devices is to make the power pins on the connector longer than the signal pins. this allows the device to power up and get into a safe startup state before the signal connections are made. It would also prevent sourcing or sinking startup power through the signal pins. Another issue with opto-isolators is that they are one way devices. isolating say, an I2C buss, would be a bit of a challenge.

Also parallax is in the process of creating a data-sheet for the propellar. judging from some of the tests they've talked about on the forums, it'll be hard to fry a Prop pin hooked strait to a connector. Add a current limiting resistor and i'm nothing short of pre-meditated destruction is likely to kill a pin or the prop.

I think a larger problem with hot-plugging is on the software end. how do you recover from an unplugged periferial? how is a newly plugged periferial identified and configured? One of the key reasons PS2 stuff is not hot-pluggable is that PS2 stuff is only recognised during the boot process.

My 2 Cents
Marty

Joe "Bot" Red

Hey fellas... look at this solution from Analog Devices... http://www.analog.com/en/subCat/0,2879,767%5F827%5F0%5F%5F0%5F,00.html

From what I have seen.. its a chip-based micro-transformer IC solution combined with high speed CMOS.. interesting..

if you cannot use the link.. just go to www.analog.com .. then Interface ... then Isolators.... I haven't used this product... but it claims to be much more energy efficient than the opto-couplers alternative... and the gates are bidirectional.. which takes care of the one way direction nature of the opto-couplers (good point raised by Lawson)

Even better.... you can order samples of several versions of this product through their page.. (this product in particular comes in SOIC package only)


I hope this helps!!!

"Propeller ... Unleash the REVOLUTIONs"

LoopyByteloose

In your case, you have two dilemmas,
First is that isolation requires separation of power supplies into two different supplies.
And secondly, hot swaping may still damage the keyboard or the mouse or·other device·- reguardless of using opto-isolation of the Propeller.

Lawson mentions an idea which is basically a 'switching scheme' that is quite sound. And yes, the SXes and the Propellers are 'hard to fry' when compared to Atmel devices. Generally, they sink and source 150% or more than the competative devices.

Personally, I would prefer a case with a power interlock. If I wanted to reconfigure, the power is shut down whenever I open the case. It certainly is a less complex solution, but not one that is always possilble.

I really don't know if using curent limiting resistors will really do much in all situation.· After all, the assumption with them is that the inputs are roughly less than 5.5volts.· If you plug in 12 or 24 volts, there isn't a sacrificial chip inbetween and the resistor's value isn't enough.· Even using a simple inverter or driver chip can protect the Propeller better.· The chip gets destroyed, but not the Micro.· Lost at 74HC06 and 74HC07

Worry about ground loops and noise if you have a real problem.· These are hard to anticipate and often too much is added to a design.· Then it becomes are greater burden to analyize the true nature of the problem.· In other words, keep it simple.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"If you want more fiber, eat the package.· Not enough?· Eat the manual."········

···················· Tropical regards,····· G. Herzog [noparse][[/noparse]·黃鶴 ]·in Taiwan

Post Edited (Kramer) : 12/18/2006 10:01:16 AM GMT

bambino

http://www.maxim-ic.com/quick_view2.cfm/qv_pk/2073
I have used this one before from maxim. They have others as well.
basically you set the cutoff voltage and it fault protects your circuit.
plus you can switch the lines or just tie the lines that do the switching to vdd or vss if you don't need them.

they offer some with 75 Ohms and some with 100 ohms and thats all the effect it places on your signal.
And you can get them in dip's.

Lawson

lets see, assuming a 240ohm current limiting resistor and ignoring any resistance in the prop lets hook that resistor to 24 volts. About 86 mA would flow, it's a toss up weather a 1/4watt resistor or the Prop would blow first. With a 1kohm resistor that current drops to about 20mA. I think the Prop can take that from now till dooms day, not so lucky for that poor resistor. The poor resistor should burn in a minute or two. Another issue is ESD and inductive kick. both of those can easily build up to several kilovolts. The saving grace is that they are pulse events and luckily the silicon can take a good bit more punishment than normal for a millisecond or two. Another bonus of these events is the excessively high voltage, this allows a MOV and Ziener diode to be used to clamp the high voltage while minimally interfering with the signal pin.

anywho, have fun!

bambino

I personally don't know about continuous 24 volts , but I did try the max 4632 on a circuit that would deliver a 27volt spike on startup and then drop from 11 to 0 volts during it's amplifying stage. It never gave me any problems without any resistance on the lines, however the input came from a low pass filter, a high pass could be another story.
In cutoff the device only lets nanoamps through to the protected side of the chip.
I suspect the switching ability could be used to dicconnect the periphial's signal lines prior to removeing the device.