XTR115: 4-20mA current loop transient spikes (resonant oscillations?) with capacitor across input/output per datasheet ref. design

Hi Matt,

You are right that the capacitor is there to help with RF noise. The capacitor in question is essentially acting as a supply bypass capacitor. This excerpt from the XTR106 datasheet talks about adding capacitors on the input pins (same concept as the XTR115).

If you could share your full schematic and layout, we can take a look to see if we notice any other factors that are contributing to the issue.

One clarification, when you are measuring the V+ with respect to ground, were you measuring anything else at the same time? Just want to make sure you weren't unintentionally shorting say the Io or Iret pins through the scope gnd. 

Hi Matt,

why are you probing with respect to earth ground?? This way you just see the noise generated by your medical grade PSU.

Kai

HI Katlynne, thanks for your response. I do have a PCB schematic handy of the signal conditioning circuit as shown below. 

I was not measuring anything else at the same time as V+. It can be inferred, but this same signature "switching" impulse noise also appeared across my shunt at the analog front end of the main board.

Its possible... yes. For what its worth, I obtain the same waveform when probing with respect to the main board's analog front end digital ground. It's not galvanically isolated--sure. But the odd part of the story is not where the noise is coming from, but rather why removing this bypass cap completely solves the issue(?) 

With this bypass cap removed, it doesn't matter where I probe there are no observable transients. My best guess thus far has been that whatever is in the output stage of the PSU has a harmonic with the cap at that frequency? But I have no proof of that. 

Hi Matt,

Unfortunately I'm no expert on switching power supplies. Is it possible to modify the value of the bypass capacitor to test your theory? 

Is the 220ohm shunt connected from Iout to ground of the power supply? This concerns me because the XTR115 has a compliance voltage of 7.5V across V+ and Io. Using a 220ohm shunt and a max output current of 20mA will put about 4.4V on the Io pin. It looks like above you mentioned the supply was 12V, but you schematic shows a V+ of 9V. 9V is too low to ensure the 7.5V compliance across the XTR115 with a max output current of 20mA. A supply voltage of 12V puts you close to the limit, but this is not factoring in any extra wiring resistance to the transmitter. I just wanted to make sure you were aware of this. 

Hi Katlynne,

I did test this theory and the results were somewhat linear. I reduced the value of the bypass cap by a factor of 10 (1nF) and the spikes decreased in magnitude by a decent bit (I did not happen to record the new value of the magnitude of the spikes) but it seemed to be about roughly half--though, still present. Removal of the cap altogether, as mentioned, completely removed the interference pattern--but again I did obviously notice more radio cross-talk now penetrating the system as expected. 

We did take note of that voltage comment you pointed above. We had changed the supply to +12V and the drawing has not been revved yet. But thank you for taking note! Thankfully we don't really require much at the high-end of the dynamic range either way.

Hi Matt,

Is there also a bypass capacitor present at the supply and ground on the supply side? Example below.

Thank you for your quick response, Katlynne.

I was unable to obtain a schematic for the Meanwell supply we use--I'll have to take apart one next time I'm in the lab. But I did also test using a LiFePo battery (w/internal protection but no bucking topology as I assume there's some bucking going on inside the Meanwell) and I did not observe these spikes, just pink noise. I think it could only be safe to assume there is a cap inside the supply in parallel with the bypass cap.

What were your thoughts that brought you to ask if there was a cap in parallel with the supply? I'm curious. 

Thank you

Matt

Hi Matt,

there's a certain chance of resonance, especially if the burden resistance (load) is small. The higher the shunt resistance, the higher the damping and the lower the risk of resonance or ringing.

As a remedy you might want to choose a somewhat higher decoupling capacitance like 100nF, for instance. Another option is to put a RC snubber in parallel to the decoupling cap.

But the main wrongdoer seems to be the medal grade PSU here. Correct me if I'm wrong, but medical grade PSUs have to limit the parasitic current into protective earth to such low levels that even the Y-cap between signal ground and protective earth must be omitted. This means that the common mode noise of PSU searches for an alternative path to protective earth and finds it in the burden resistance. Then, a cap in parallel to the burden resistance would be helpful to suppress the effect of common mode noise :-)

Kai

A++ Explanation, thanks, Kai. 

I was able to damp several dB of noise by adding a 0.1uF ceramic cap in parallel the 220ohm shunt. I cracked open one of the supplies to confirm my suspicions and there's certainly a 1500uF electrolytic directly across the output. 

I am no expert in medical supply design but my rudimentary understanding concurs with your analysis. I'll have to give that RC snubber a shot. When I get back into the office I'll set one up and report back with my findings. 

Thanks.

Katlynne,

Hopefully you are able to see this--a question I kept forgetting to ask in regards to this: is it not advisable to use a common mode choke on the 4-20mA lines?

Thanks,

Matt

Hi Matt,

Common mode chokes may be used. Most of the TI reference designs use ferrites, but we have seen customers use common mode chokes as well. We only have a behavioral model for the XTR1115, so it's not possible to simulate the effects as far as stability of the inductive load on the output. That's just something to pay attention to if you choose to incorporate it in your design. I believe incorporating noise reduction techniques is advisable, so the common mode choke would be an option. 

Hi Matt,

I wouldn't do that as it increases the risk of resonances. This must be developped very carefully, otherwise it does more harm than good. I would first try to find out where these strange spikes are coming from. Can you change the PSU for a test? Is there any need to use this medical grade PSU?

Kai

Thanks for the explanation Katlynne. W.R.T. the additional comment below I'll do some analysis before delving into using a choke.  

Hello Kai,

Thanks for the word of caution. It is certainly a compatibility issue it seems. Probing my supply I am unable to observe any obtuse ripple per the datasheet. Moreover, powering my loop test-setup via my benchtop SMPS doesn't seem to show any issues. 

Unfortunately this PSU component is part of our stock/supply chain for use in many products as picked by someone else years back. Not really in my scope to select a new part. As I said, testing using a "fancy" lab supply seems OK. Using a LiFePo battery I am getting about 150mV pink noise with the capacitor in place. Removing it reduces the pink noise to about 50mV. 

I have also tried:

Different XTR, different BJT, shielded cables, eliminating noise sources both radiated and conducted, different cables, different capacitor (not that it should matter, both low ESR (ceramic: X7R, and a Y5V). 

Hi Matt,

I would give this scheme a try:

Kai

And adding low pass filtering to the input of XTR115 is another option:

Kai