Unstable HART Signal via AI Splitter


Thread Starter


We have several Endress+Hauser mass flow meters. Model is 83F40.
The AI signal is splitted to DCS and SIS.
Spliiter model: KFD2-STC4-1.2O-3 from Pepperl+Fuchs

All HART signals polled to the DCS appears unstable. The communication fluctuates between bad and good every few seconds.

A communication is easily established when connecting the HART communicator to anywhere at the right side of the AI splitter.

But connecting the HART communicator to anywhere at the left side of the AI splitter gives unstable connection. Eg. device not found.

I also tried putting at 250 ohm resistor at position * (see attached loop drawing), but no improvement.

<b>Moderator's Note:</b> Control.com does not support attachments. For people to see drawings, post it to a download site and put the url in your message.

Anyone out there know what the issue might be?

Thank you.
Obviously the signal conditioner is HART compatible. I suspect that you need to increase the loop resistance for the HART loops; a minimum resistance of 250 ohms is required for any HART circuit. You need to check that EACH 4-20ma loop has the minimum resistance.

Can you please let us know the brand of the DCS controller and also can you
please verify if the DCS I/O card is HART compliant?

Thx & rgrds
Yes, it is HART compliant.

We can get the HART signal, just that it is very unstable. I also did try adding a 250 ohm resistance to the flow transmitter side, but it didn't work.

DCS and SIS is Emerson DeltaV.
HART is normally pretty robust when it has a connection and sufficient resistance to develop a signal for the modem chip. Is the cabling shielded twisted pair?

Sounds like you need to go to the authority on this one to figure out why you don't have stable HART on the transmitter side.

The authority is Jonas Berge who has compiled years of experience with HART into one document, "New Tips and Tricks to get Value out of HART", for the benefit of the HART-user world. It includes the most comprehensive HART troubleshooting guide on the planet.

That document resides at


on LinkedIn's site. I am a LinkedIn member so I get access. I'm not sure if non-members get access or not. Try it and see. The document is valuable enough to register on Linkedin if only to get access to the document.

In my opinion, the document is valuable enough to copy and save, because one never knows when it will disappear from LinkedIn.
The manual at this link:

shows that the HART signal is always on terminals 26 & 27 (pages 12 & 13). There are other 4-20mA channels, but those are not HART enabled.

The spec portion mentions HART as an active output, but does not mention HART for a passive, loop powered connection.
Active: 0/4 to 20 mA, RL < 700 &#937; (for HART: RL&#8805; 250 &#937;)
Passive: 4 to 20 mA; supply voltage US 18 to 30 V DC; Ri&#8805; 150 &#937;
I suspect the absence of HART is merely an omission.

If loop powered, there are specs for a DC power supply used in HART circuit. That's worth looking into.

If I were there, I'd divide and conquer.

I'd disconnect the splitter and put a 250 ohm resister across the flowmeter AO terminals and "prove" the meter's HART functionality with something like the utility program that comes with the MacTek USB HART modem that polls the device's primary variable at one second intervals and displays the value, by putting the HART modem across the resistor and reading the device with the utility program.

If that works, reconnect the field cabling and move the resistor to the end of the cabling where the isolator/splitter is and make the same proof test (Still without the splitter). If the test passes, then the flowmeter and cabling is HART functional and the problem is in the isolator/splitter, its power or connections.
There's a note on the E&H splitter about adding resistance on the load side by using terminals 7-9 instead of 7-8, likewise 10-12 instead of 10-11. Are you doing that?

250 Ohms is not a magic number. It depends also on what type of a load it is.
There's a note on the E&H splitter about adding resistance on the load side by using terminals 7-9 instead of 7-8, likewise 10-12 instead of 10-11. Are you doing that?

250 Ohms is not a magic number. It depends also on what type of a load it is an inductive load requires less resistance because the HART can develop more Voltage.

I would think if you connected an oscilloscope you would be able to see the HART signal superimposed on the analog. You might be able to see where the signal is being lost and add something to boost it.

Does the HART communicator work from the test socket on the splitter?
Currently we are using terminal 26-27 at the flow transmitter, and it is passive powered. I was thinking if the passive mode is the root cause, but all other mass flow meters which do not have splitters are giving stable HART signals.