Isolated RS-485

Duane Degn

I think my boss wants to have an isolated RS-485 connection on a project with which I'm helping.

I initially thought an isolated RS-485 just meant optically isolated but I didn't see how something could be isolated if both sides of the optical isolation shared the same power supply.

Sure enough the power supply to the optical isolation also needs to be isolated (I'm not sure if this is a "need" or a "should" be isolated).

I find the following reference diagram in a TI app note.




Here's a 5v to 5v isolated supply I found on Mouser.

http://www.mouser.com/ProductDetail/RECOM-Power/RBE-0505S/?qs=sGAEpiMZZMvGsmoEFRKS8EFGozA47QQ4lf7QgFWqt78%3d

Am I correct to think it doesn't do much good to optically isolate a connection without also isolating the power supply?

We'll be use a Propeller which of course is a 3.3V device but we'll have both 3.3V and 5V on the board. Is there any advantage to using one input voltage than another? Should I be looking for a 3.3V to 5V isolated power supply or is a 5V to 5V supply fine?

Heater.

Duane,

Depends on what the requirement is. I have worked on equipment where parts of the circuitry are communicating over optically isolated connections and the "grounds" at each end are separated by 15000 volts?

I put "grounds" in quotes because basically there was digital logic at each end operating from what it thought was a ground and 5v rail. Just happened to be 15KV offset from what they were talking too.

In a situation like that I might suggest it is pointless isolating the communication signals if you are not also taking care of the power supply arrangements.

Duane Degn

Heater. wrote: »

Duane,

Depends on what the requirement is. I have worked on equipment where parts of the circuitry are communicating over optically isolated connections and the "grounds" at each end are separated by 15000 volts?

I put "grounds" in quotes because basically there was digital logic at each end operating from what it thought was a ground and 5v rail. Just happened to be 15KV offset from what they were talking too.

In a situation like that I might suggest it is pointless isolating the communication signals if you are not also taking care of the power supply arrangements.

Ouch! 15KV?

The examples I've looked at are good up to 2KV. I'll need to ask my boss what level of protection is needed.

I'm guessing the higher the voltage difference the more the parts are going to cost.

Thanks for the insight Heater.

Buck Rogers

Duane,
I typically use Maxim's family of RS485 communications parts, I believe they do include a set which is indeed isolated.

pjv

The ADMu chips provide isolated bidirectional digital communication as well as tens of milliamps of isolated power for exactly these requirements. High speed too, and a number of KV of stand off capability.

Cheers,

Peter (pjv)

jmg

Duane Degn wrote: »

I initially thought an isolated RS-485 just meant optically isolated but I didn't see how something could be isolated if both sides of the optical isolation shared the same power supply.

Sure enough the power supply to the optical isolation also needs to be isolated (I'm not sure if this is a "need" or a "should" be isolated).

Depends on what 'attacks' the isolation is to protect against.

Some designs do optically isolate, even though the grounds are essentially the same and that protects against ESD and faults propagating.
Adding local small, but isolated, power improves the degree of isolation as the pathways from the hostile outside world, to more sensitive electronics are reduced. - DC paths are broken, and AC pathways become the residual pF coupling.

When looking at isolated PSUs, check the pF coupling as well.

DigitalBob

Duane
I build isolation amplifiers and I use the Ti Iso124 this amp also has a compatible dc to dc converter in a dip package p/n dcp011515. So I would follow the ap note and use the recommended power supply. The iso124 data sheet shows a good 3 way isolation.

Duane Degn

Thanks for the information guys.

I'll try to figure out the pros and cons of these various parts so I can let my boss know what options are available.

I'll probably be back with a schematic for you guys to look over.

Thanks again for your help. I've learned a lot from this forum.

Publison

We have also learned a lot from you, Duane. Thanks for all your inputs!

Peter Jakacki

Here's a circuit I did that also includes multiple stages of lightning protection including gas tubes. It's a dual channel isolated RS485 and also includes selectable termination resistors. This particular one is made up as a plug-in module measuring just 20mm x 45mm.

Phil Pilgrim (PhiPi)

This is the way I would do it. Just jelly-bean parts, nothing exotic.



I didn't clutter the schematic with filter or bypass caps. You'll know what to do about those.

-Phil

Phil Pilgrim (PhiPi)

Duane,

It's not clear from your first post, but is this a multi-drop network app, or just point-to-point? If the latter, I would seriously consider going to RS422 instead of RS485. Full duplex is much simpler, both from a hardware and a software standpoint. In my circuit above, for example, you could switch to an RS422 transceiver and eliminate one of the optos. The only additional requirement is an added twisted pair.

-Phil

Duane Degn

@Peter, Thanks for your schematic. I confess that I don't understand it all but I'll try to figure out the parts I don't understand.

@Phil, I a had originally thought something like you suggested would be used but then I found out there were chips which include opto isolation with the balanced signal producing circuit.

I think the sensor will most often be used in a point to point configuration but I'm pretty sure it needs to support multi-drop. We're using a Modbus protocol to send and receive data and I think single duplex is all that is planned. The sensor will be a slave device and only reply when data is requested.

I believe we'll use a 24VDC supply with both 5V regulator and 3.3V regulator (discussed in an earlier thread) on the sensor board (with a Propeller). I don't think we'll have a second (external) power supply to power the circuit on the outside portion of the opto isolation. Which is why I'm wondering if we need an isolated DC to DC supply.

Would a beefy diode and a second 5V regulator from the 24V supply be a viable option? I haven't researched all the options mentioned earlier so maybe one of these parts would do the trick. I kind of like the "jelly-bean parts" concept since I think I can understand what each part is doing. (I hadn't heard the phrase "jelly-bean parts" but I think I understand what it means from context and analogy.

BTW, I helped with an earlier version of this sensor but my boss is planning an upgrade and wants to include isolated outputs. I'll probably be asking about isolated 4-20mA circuits once I know enough to hopefully ask an intelligent question on the topic.

Phil Pilgrim (PhiPi)

Duane Degn wrote:

...Which is why I'm wondering if we need an isolated DC to DC supply.

If you're powering the isolated side of your circuit from the non-isolated side, yes, you do need an isolated DC/DC converter.

-Phil

Phil Pilgrim (PhiPi)

Duane,

Once you get to the 4-20mA stuff, take a look here for another "jelly-bean" approach:

http://forums.parallax.com/discussion/comment/631965/#Comment_631965

-Phil

Peter Jakacki

The schematic is straightforward enough, the two terminal components are either resistors (240R, 1K etc) or a ferrite bead FB, or a polyfuse PTC. The gas tubes are the 3-terminal components on the right side and they need an "earth" (not signal ground) for proper lightning protection. In conjunction with the 5V transzorbs SA5.0A the line side is not only isolated but also protected to a very high degree from stuff that tends to find it's way down the line and turn chips to char. Bear in mind that this is industrial grade stuff if you are running lines out in the open or between buildings or up poles etc.

Duane Degn

Peter Jakacki wrote: »

The schematic is straightforward enough, the two terminal components are either resistors (240R, 1K etc) or a ferrite bead FB, or a polyfuse PTC. The gas tubes are the 3-terminal components on the right side and they need an "earth" (not signal ground) for proper lightning protection. In conjunction with the 5V transzorbs SA5.0A the line side is not only isolated but also protected to a very high degree from stuff that tends to find it's way down the line and turn chips to char. Bear in mind that this is industrial grade stuff if you are running lines out in the open or between buildings or up poles etc.

Thanks Peter.

I think the meter I'm working on is intended for industrial stuff but I don't think it will be connected to things in other buildings. I doubt we need lightning protection but I'll check to make sure.

@Phil, Thanks for link. I haven't seen a current driver done like that before. Not that I know a lot about this sort of stuff.

Tracy Allen

Attached is photo of the pcb for isolated RS485 using the Linear Tech LTC1535. The chip includes a 420kHz driver for the transformer seen along with a dual schottky rectifier and filter capacitor for the isolated supply. Pretty simple, but 5V Vcc, not 3.3. It's been around for a long time. (The extra circuit at the top is something I patched in for power control.)

I've also used the LTC2862, which is a 3.3V or 5V part, but not isolated. I mention it because it has high ESD ratings (15kV HBM), as well as extended common mode range and fault tolerance. Other manufacturers make such things too. If you are mainly concerned with ESD events and not with chronic ground loops and widely different industrial ground potentials, it might be okay to skip the extra hassle, power consumption, and expense of full isolation. Or should it be, better safe than sorry whatever the expense?

Buck Rogers

Phil Pilgrim (PhiPi) wrote: »

Duane,

Once you get to the 4-20mA stuff, take a look here for another "jelly-bean" approach:

http://forums.parallax.com/discussion/comment/631965/#Comment_631965

-Phil

That's an interesting idea. I've always used the Burr-Brown now TI XTR101/RCV420 parts. But that's just me.

As for the RS485 stuff, I use the parts from Maxim for the stuff. (Sometimes also from TI.)