How to use FILT_LP and FILT_LP on CW308

Hi,

I am trying to understand how the onboard filter works in CW308. In default when Pin 1 and 2 of J14 are selected(?), it is said “Filter input set by FILT Pin from Victim Board (DEFAULT)”. If I understand correctly, this means the input is filtered by low pass and high pass filter together (bandpass filter). If I want to use lowpass filter only, is it correct to connect pin 1 of J14 to pin 5 of J13?

Thank you!

Hi,

I believe HP and LP there stand for high power and low power. Both pins are tied together and there’s only a low pass filter between FILTIN and FILT_HP/LP.

Hi,

what does low power and high power mean exactly? As for the low pass filter, what is its cutoff frequency?

I’m not sure the purpose of the LP/HP separation. They’re connected together on the PCB anyways.

I think the cutoff frequency of the LP filter is something like ~50kHz.

So the power source from FILT_LP and FILT_HP is combined together and pass through FILT pin for filtering. Correct? If I do not want to use low-pass filter, where should I connect the pin 2 of J14 to?

Besides, is there any reason (or reference) to choose 50kHz as cutoff frequency?

You can minimize the effect of the filter by connecting your voltage source directly to SHUNTH. My guess for the cutoff frequency is to filter out switch mode supplies, but this also just power supply filtering, so I’m not sure there’s a reason for the exact value.

Hi,
Do you mean connect J14’s pin 2 to pin 3 of J13? As for the power supply filtering, do you mean this does not affect the frequency of signal generated by MCU?

No, remove the jumper from J14, then connect pin 8 of J13 to pin 3 of J13. Some of the filtering will still be present. Is there any reason you’re trying to remove low pass filtering from the high side of the shunt?

Which MCU signal are you talking about? If you’re talking about power traces, you’ll get a noisier and less accurate result as the high side of the shunt will vary more with respect to ground.

The major issue is that we will need a good reasoning or reference support for selecting this filter in our research work. People might question about the 50k. That is why the raw one is better than the filtered one. :slight_smile:

I am confusing about the switching mode you mentioned. What are the switch modes?

Oh okay, that makes some sense. Honestly, you might be able to avoid most questions by calling it decoupling rather than filtering, which is more accurate anyways.

Most applications are going to have a DC/DC buck converter somewhere stepping down power, which will add switching noise in its switching frequency and harmonics. This is usually somewhere in a vague range of 100kHz-10MHz I’d estimate (the Lite’s DC/DC converter, the LTC3419EMS, for example, switches at 2.25MHz). Honestly, the 50kHz is just an estimate assuming the ferrite bead is acting like an inductor.

Hi sorry for the late response. What is “decoupling” about?

From wikipedia: In electronics, decoupling is the prevention of undesired electrical energy transfer (coupling) between subsystems.

Most commonly, this is just putting capacitors close to the power pins of chips to limit high frequency noise on the power supply. Unless you’ve got something that’s very sensitive to noise (like an FPGA), it’s pretty common just to just throw some 0.1uF or 1uF ceramic caps on the power rails as close as possible to the chip.

Also, a correction for above: this is actually an inductor and not a ferrite bead. I had mixed up what was on the CW313 and the CW308. Also the cutoff frequency is ~800Hz, not 50kHz.

Thanks for the reply. Is there any similar low pass filter design on ChipWhisperer Lite (CW1137)? Besides, I found very interesting that I still find many signal has higher frequency than 800Hz after fourier transformed the power trace. Now I am confused again why I can still observe these high frequency signal again. :sweat_smile:

Yup, the Lite has a similar design. The goal is to have a stable high side of the shunt resistor. We’re measuring the AC component of Vshunt_l, which is equal to Vshunt_h - Vshunt_drop and we want Vshunt_drop, so we therefore want to minimize the AC component of Vshunt_h so that it’s as close to 0 as possible.