Assessing electrical noise with a 70MHz scope with "Bandwidth limit" deslected.

[zenerbjt]

Dear Engineers,
We are doing some test measurements on a Control PCB which will get used in a noisy environment, where several 1kW BLDCs and their drives are being operated. We wish to have tests which can see how noisy the control PCB is.

The control board contains microcontrollers, as well as Radio Transceivers and multiple PWM channels sending speed control PWM signals to the electric drives. Also, there are comms busses carrying data. Also, there are 4 DCDC power modules. The DCDC modules output 5V and there are 4 of these, partly to give redundancy. Vin to the DCDC's is a 48V battery. Total control board power useage is some 10W.

We want to test all the rail voltages on this control PCB for noisiness, and also see how noisy are the PWM speed control signals on this board.

As such, to test for electrical noisiness, we will simply scope the voltage rails (and PWM signals) with a 70MHz scope in AC coupled mode, and without “Bandwidth Limit” being selected. We will use a roll-your-own “coaxial cable scope probe” which is 1:1 and does not have a “dangling ground clip”. We will observe the high frequency noise, and record the peak-to-peak of it, to assess noise.
We will observe the PWM speed control signals in the scope's DC coupled mode, and see how noisy are the mark and space of these pulse trains. Also, we will look out for sharp undershoot and overshoot spikes at the high-to low ( and vice versa) of the PWM signal transitions.

Do you think this is an effective ploy to test for noisiness in the PCB?


Scope is a Keysight DSOX1204A
https://www.keysight.com/en/pdx-296...z-4-analog-channels&cmpid=72858?&cc=GB&lc=eng

 

[unclejed613]

for a 5V rail, about 50mV of noise is usually the maximum amount of noise you want to see, but it depends on how sensitive various systems on the board are to noise on the rails. good filtering of power, and adequate bypassing of rails (usually a 0.1uf or 1uf cap across the power pins of each TTL chip as an example) will keep the noise level low. proper power and ground layout is also important. for mixed signal systems, there are usually separate analog and digital rails and ground systems. also you didn't mention what frequency range the radios were operating on, and where in respect to the board the antennas are. pc board traces that are near 1/4, 1/2, or full wavelengths of the operating frequency of the radios may couple noise onto the board if the antennas are too close to the board (this is also affected by the amount of power the radios operate at... if it's very low power it's probably ok to have the antennas on or near the board (such as 50mW). if it's more than 100mW you want the antennas elsewhere. you will also need a spectrum analyzer if this is a production board, as most countries require some kind of testing and certification for radiated noise.

 

[rjenkinsgb]

Without a local ground, how do you tell what the "noise" is relative to - there could be high levels between the scope power ground and a machine ground?