Non-sinusoidal periodic supply noise (rs232 caused)

Thread Starter

aeroguy

Joined Sep 16, 2009
40
Hi,

I'm using a Sipex SP3232 chip in a 5V circuit.

I discovered that as long as the sp3232 is powered, even when it's not actually connected to a serial line or being used actively, it creates a periodic disturbance in my supply voltage.

The waveform is at a frequency of 130 kHz, and consists of a distinct bump that is 13mV in amplitude. It almost looks like a 25% duty cycle PWM signal, to help you visualize. This behavior seems to correspond to one of the internal operations of the 232 chip, as it cycles through 4 internal processes.

The problem with this disturbance, is that it is causing troublesome noise for some instrumentation amplifiers sharing the same power source.

Anyway, it isn't obvious to me how to filter this sort of thing, since it's not sinusoid in nature. I already have numerous bypass capacitors of the common values used all over in my circuit (0.1, 0.01, 0.22 uF, etc.)

Has anyone here ever had to address this issue before? Any thoughts on filtering, etc?

thanks,
Eric
 

bertus

Joined Apr 5, 2008
22,270
Hello,

You could try to "isolate" the sp3232 from the rest of the ciruits using a 10 Ohms resistor in the Vcc line of the sp3232 and a 0.1 μF and a 1 μF capacitor from Vcc to ground on this sp3232.

Bertus
 

Thread Starter

aeroguy

Joined Sep 16, 2009
40
Hello,

You could try to "isolate" the sp3232 from the rest of the ciruits using a 10 Ohms resistor in the Vcc line of the sp3232 and a 0.1 μF and a 1 μF capacitor from Vcc to ground on this sp3232.

Bertus
Thanks, Bertus. I'll definitely give that a shot tonight.
 

JohnInTX

Joined Jun 26, 2012
4,787
Another thing to try if you can do it is to run separate physical power lines (Vcc and GND) to the 232 chip and to the IA/analog stuff all the way back to the power supply. Then heavily decouple them there. The object is not to share the V/I surges from the 232 with any of the sensitive analog circuitry by sharing current paths.
 

Thread Starter

aeroguy

Joined Sep 16, 2009
40
Another thing to try if you can do it is to run separate physical power lines (Vcc and GND) to the 232 chip and to the IA/analog stuff all the way back to the power supply. Then heavily decouple them there. The object is not to share the V/I surges from the 232 with any of the sensitive analog circuitry by sharing current paths.
Thank you, sir. I will try this as well.

All of the feedback is greatly appreciated!

Eric
 

THE_RB

Joined Feb 11, 2008
5,438
Serial 232 chips usually use a charge pump oscillator and some caps to generate the _/-12v voltages needed for RS232 serial.

It should have a pretty good decoupling cap, I would go a few times larger than the charge pump caps, so maybe 47uF for its decoupling cap. A tantalum is good.

And like Bertus said use a 10 ohm resistor before the cap, and a 0.1uF cap as well .
 

Thread Starter

aeroguy

Joined Sep 16, 2009
40
Hello,

You could try to "isolate" the sp3232 from the rest of the ciruits using a 10 Ohms resistor in the Vcc line of the sp3232 and a 0.1 μF and a 1 μF capacitor from Vcc to ground on this sp3232.

Bertus
Bertus! Thanks, man.

The solution was indeed a 1 uF cap, but I needed a 500 Ohm resistor to get the job done.

Thanks again, you saved me much time and madness.

Eric
 

Thread Starter

aeroguy

Joined Sep 16, 2009
40
Serial 232 chips usually use a charge pump oscillator and some caps to generate the _/-12v voltages needed for RS232 serial.

It should have a pretty good decoupling cap, I would go a few times larger than the charge pump caps, so maybe 47uF for its decoupling cap. A tantalum is good.

And like Bertus said use a 10 ohm resistor before the cap, and a 0.1uF cap as well .
Thank you. I may try this as well, because the larger (500 ohm) resistor I'm using does impact the rs232 transmission a bit. Will report back what happens with a larger capacitor.
 
Top