I posted recently a great idea to measure noise levels from our power supplies. Yesterday I managed to put together this small interface circuit. I used a remainder piece of double layer PCB big enough to fit the bulky capacitor, the transformer and the output BNC connector. The input is a just a simple set of copper turrets. Special care is taken in laying out the ground planes to avoid ground loops. Also the transformer is grounded at one side only of the case. A finished interface looks like this:
The performance of this noise level measurement interface is really good. I only get 1dB loss at 50Hz. Sound Card interface adds an extra 0.44dB loss at 50Hz so the correction factor for measuring the 50Hz component is +1.44dB. The response is flat and -3dB points are around 35Hz and 15kHz:
This device needs to be shielded in a metal box. I haven’t found any suitable metal box at hand yet, so for the time being it will be used inside a big metal screen box temporarily. Looking at the noise that the interface picks up without any connection:
This measurement was calibrated at 100mVrms/0dB for 100hz.The 50Hz component is actually -85.1db+1.44dB=-83.66dB or 65.6uV!. My interface also picks up some noise at 450Hz and 25-28kHz which I presume is the LCD monitor cable being close enough as Video card is on the way unfortunately.
Now, looking at my bench HT power supply with feedback regulator:
Noise levels are:
- 50 Hz: -14.66dB (184mV)
- 100Hz: -66.4dB (478uV)
- 150Hz: -50.9dB (2.85mV)
Which clearly shows that the mains transformer is inducing noise in the supply somehow as the 100Hz level is really low which shows that the filtering stage and feedback regulator are doing their job pretty good.
This noise level interface circuit proved to be really good. I can measure now HT noise levels greater than 100uV and up to 1kV thanks to the massive input capacitor added. One word of caution: if you use this interface you have to ensure that the input capacitor discharges after use. Best way is to keep the test load connected so there is a path to discharge the HT charge in the 10uF/1kV capacitor.
