Electronics > Projects, Designs, and Technical Stuff

Ferrite bead on op-amp output to drive long cables

(1/1)

ricko_uk:
Hi,
with reference to the attached schematic, I need to minimise noise picked up by the 3 meters long cable. Noise can be picked up by the cable and fed back into the driving (instrumentation) amplifier which in turns is amplified in its output. According to Fig 19 Page 8 of the attached application note, when driving long cables, you can use a resistor in series with a ferrite bead on the instrumentation amplifier output to reduce that issue.

Three questions:
1) is resistor required? why and how do you choose the value?
2) I am thinking of putting one also in the input (i.e. receiving side op-amp input). Is that correct? How do I choose the value for this one? Based on the op-amp input impedance maybe? How?
3) if the cable is both shielded and twisted pair, is the ferrite still required? In other words, is it worth putting it in or not for a 3 meters long cable?

Many thanks as always for all feedback! :)

awallin:
what is your signal frequency? DC??

R3 can probably be 50 or 100 ohms or so.
FB1 and FB2 can be chose so they have a 'high' resistance (200+ ohm?) outside your signal frequency range. (but small ferrites attenuate at RF such as 100 or 200 MHz, so if your problems are at 10 MHz you need big ferrites).
After FB2 you can put a new resistor to ground, 1k for example.

use a shielded cable, or coax?

David Hess:
Series filtering at the input will not work because the input impedance is so high.  Some type of shunt circuit like a capacitor to ground is required.  Feedthrough capacitors are ideal for this application but they make special three terminal RFI filters which are probably even better.  Note that RFI protection is often appropriate for outputs as well.


--- Quote from: ricko_uk on January 22, 2020, 04:36:31 am ---1) is resistor required? why and how do you choose the value?
--- End quote ---

The ferrite bead filters RF and may also prevent local oscillation of the output transistor when it drives the low impedance load of the cable.

The series resistor also prevents local oscillation (50 MHz and higher) of the output transistor however its purpose here is to isolate the load capacitance from the output to prevent loss of phase margin which could cause oscillation in the feedback loop.


--- Quote ---3) if the cable is both shielded and twisted pair, is the ferrite still required? In other words, is it worth putting it in or not for a 3 meters long cable?
--- End quote ---

It will depend on what is happening between the shield grounds.  Within the same chassis, it would be more common to use a common mode choke so the signal and ground of the twisted pair are wrapped through a small ferrite core.  With coax, nothing may be required.

https://www.analog.com/media/en/technical-documentation/application-notes/41727248AN_347.pdf

T3sl4co1l:
You probably want to do it like this,



although I guess that will be difficult with the in-amp as shown.  You may want to put a buffer after it, using the buffer in this way.

The ferrite bead alone, probably isn't enough impedance, at low enough frequency, to ensure stability.  It does help keep noise out, particularly when combined with a small capacitor (say 1nF) to ground, on the inside (i.e., between R and FB).

The other end, the FB is in series with a high impedance input pin, so its impedance counts for nothing.  Use a small capacitor to ground, to give it something to filter against.  This is also a good place for ESD protection, whether clamp diodes or a TVS.  A modest value series resistor between ESD diode and op-amp input further ensures safely limited current into the amp, protecting it.

Also since the input is high impedance, a series resistor can be used instead of FB, potentially giving better performance, but given tradeoffs.  The ESD diodes will leak some current (typically ~uA), causing a DC error in your signal.  If this is acceptable (an error of say 0.1% or more), then a few kohms can be used here, giving more impedance -- at all frequencies, not just around the FB's impedance peak -- than an FB can.  On the other hand, if you have high precision requirements, that leakage current has to be sunk into the source, with low voltage drop, preferring the FB.  Note: prefer PN diodes or zener/TVS over schottky -- they have lower leakage current.  Schottky are perfectly acceptable for lower accuracy signals (1%+ error?) and digital (~20% error?).

Tim

Navigation

[0] Message Index

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod