162mV pkpk ripple on 5V is not going to interfer with microcontroller?

[opampsmoker]

Hi,
Please confirm that the attached waveforms, providing 5V to a micro, which is doing ADC readings and comms via I2C, is not going to result in microcontroller malfunction?

You've probably guessed that these are outputs of an ON/OFF controlled (sepic) converter.

 

[KlausST]

Hi,

There is a good chance that all digital part will work properly. Not nice..but maybe not off specifications.
Don't expect any precision from your analog part, especially the ADC

Why not add a capacitor to the power node? Or an LC?

Consider to use a clean voltage reference for the ADC.

Klaus

 

[opampsmoker]

Thnaks, its the output of a 5W SEPIC, with cout = 100uF (tant) and LC post filter of 470n and 22uF ceramic.
Its only a sim of course...and i actually think that skin effect will carve away some of the sharpness in the real thing.

Wherever i go to work, i always here engineers saying how micros are badly affected by smps....and i just dont see why......i mean i would say a micro would be fine even with say 0.5Vpkpk ripple at 100khz as long as it is relatively smooth....would you agree?

It would be good to know, what is the function of a micro that is most affected by supply voltage ripple?

(starter is as you say when its getting a ref voltage from the rail.....and of course in this case i woudl agree using a ref regulator is the way forward here)

 

[betwixt]

It would be good to know, what is the function of a micro that is most affected by supply voltage ripple?

That is difficult to answer because each micro behaves differently. As a general answer:
1. if it uses an internally generated clock (NOT crystal controlled), its frequency might shift in sympathy with the instantaneous supply voltage.
1a. any peripherals using the clock may also lose timing accuracy.
2. If the ADC uses an internal reference, or uses VDD as the reference, it will scale each step from the reference and that might cause inconsistent measurement.

The logical functions should still work fine, they are quite immune from supply noise as long as the minimum voltage condition is met.

I would caution you about how you measure ripple. In practical terms, the MCU is completely unaware of its surroundings, it only sees what is on its pins. That means when you measure ripple it has to be directly between the VSS and VDD pins, not referenced to a ground somewhere else. Similarly, any analog voltage is measured across the pins, usually an input and VSS so you need to consider where the voltage drop causing the ripple actually occurs. A voltage drop generally means current is flowing through a resistance, make sure that drop isn't influencing the MCU operation.

Brian.

 

[KlausST]

Hi,

The first two pictures show the same signal? Then I guess there is a swtiched load that draws pulses of about 1.5A.
Here the power supply has no influence.
--> add capacitors.

The third picture is a "no load" picture. Burst mode or reduced swtiching time.
Improvements:
--> Higher value inductance. RC or LC post filter, add some dummy load.

Klaus

 

[danadakk]

Not all caps, for same Capacitance, as effective in terms of ESR -

Regards, Dana.