Looking for an isolation idea

Don M

I am working on a project to monitor a serial bus in a Master Slave setup. In most of the machines that·this setup is·used in the power supply ground and the serial communication common are at the same potential. However there does remain in the field some machines where there is a potential between the two. I have no idea how much but I am wanting to isolate the two if possible to avoid it.

The Master to Slave is not a problem as all the slaves utilize an opto isolator and all are tied together going to the input side of the opto. That I can monitor well. My problem is I also want to monitor the data going from the Slave to the Master. The Slave devices return channel are connected to the bus·using an opto-isolator open collector output.·The Slave to Master is held high (+5v) but if I hang an opto on it using the LED input it drags the·5 volts down too much even trying to use the highest value resistor to obtain the lowest amount of current through the opto's led and still work.

If I use a resistive voltage divider it works ok in those that have the same ground potential. Someone had·suggested using a "differential comparator" saying you could achieve a very high impedence between the ground and common but had no circuit to offer up. Is this possible?

Does anyone have any other ideas?

Thanks in advance.

·

Phil Pilgrim (PhiPi)

Don,

Which opto did you try? The reason I ask is that the H11L1 requires very little current to trigger and may not load the slave->master channel as heavily.

You could also use a battery-powered buffer ahead of the opto, which would barely load the line at all.

-Phil

Peter Jakacki

I'm not sure of the circuit connections you are talking about but you did say open-collector yet it sounds like you are connecting the led in a source configuration when it should be a sink configuration so that the led is tied to the +supply through a current limit resistor.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
*Peter*

Phil Pilgrim (PhiPi)

Peter,

You make a good point. But if the eavesdropping node is located remotely from the master node, that may be hard to pull off:



-Phil

_

kwinn

If the voltage powering the slave led is high enough it may be possible to put the slave and eavesdrop led in series. The current limiting resistor will also need to be changed.

Phil Pilgrim (PhiPi)

Don,

Something like this might work for you:



It uses the H11L1 I mentioned earlier, whose turn-on threshold current is at most 1.6mA. The key is the series resistor. It needs to have a much higher value than the pullup resistor at the master node. The master resistor needs to be fairly stiff for this to work, so that can supply enough current to turn the H11L1 on through its series resistor, without getting pulled down itself. Without the series resistor, the master node input will never see a voltage higher than the eavesdrop LED's V_fwd.

-Phil

_

Post Edited (Phil Pilgrim (PhiPi)) : 2/25/2010 3:47:39 AM GMT

Don M

Phil-

I think the H11L1 may work. You have the circuit right in your last diagram. Unfortunately I don't always know how stiff a pull up resistor is being utilized from the Master. The equivalent resistance for that load using the H11L1 is approx 4.2K. I don't have any H11L1's yet but I put a resistor of that value across the line and it only drew the voltage down from 4.85 to 4 volts and the slaves are still running ok.

I was using MOC206 opto's. I will be ordering a few H11L1's and test them out.

Thanks. If there are any other ideas please share.

stamptrol

The other thing to do is hang a '485 communication chip (75176 or the modern equivalent) where you want to eavesdrop. You can power it with an isolated power supply if required.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tom Sisk

http://www.siskconsult.com
·

Don M

This product looks promising from Analog Devices ADuM1200 / ADuM1201- http://www.analog.com/en/interface/digital-isolators/adum1201/products/product.html#ppa_print_table