robust Low Side Current sense

[ali_asadzadeh]

Hi
We are designing a PFC,and we need a good low side current sense amp, the problem is that we got all sort of noise,spikes and crap along the current using this standard low side circuit,  we have tested these op-amps, OP07 ,LMV358 and LM358, the OP07 shows better result regarding the noise and spike, but we got lot's of spike, the next stage to the Current sense is another LM358 in comparison mode so we could detect over current and switch of the mosfet,because of spikes and noise our MCU get lot's of over current interrupts, Do you have any idea what's wrong?

[ajb]

First step: add some filtering.  You'll need to decide what sort of bandwidth/response time you need and then select a filter to suit.  Don't use more bandwidth than you need.  An RC filter between the current sense amplifier and the comparator would work. 

First-and-a-half step: What are your op-amp supply voltages, what's the gain of the sense amp, what's your desired bandwidth, and what is the voltage drop across your sense resistor?  All of these will influence the suitability of various op-amps. 

Second step: Determine if the spikes and noise you're getting in the sense amp are a result of more serious problems in the rest of your circuit.  A certain amount of noise is to be expected, but excessive noise, and in particular large spikes, could be indicative of problems with your circuit design and/or PCB layout.  If you post those here

[ali_asadzadeh]

Thanks for the tips, I have already added the RC filter, here is the schematic and PCB

[bktemp]

Connect capacitors in parallel to R21 and R29. This will form a lowpass filter, limiting the gain at high frequencies.
R26 should also be much higher to form an effective lowpass filter.

Your OC voltage is set to 3.3V. That is very close to both the input common mode voltage range and output swing of a LM358 operating at 5V. Better use a rail to rail opamp or like a LMV358.

[ajb]

How is your current shunt connected?

Your RC filter has a cutoff frequency of 15.9MHz (f_cutoff = 1/(2*pi*R*C).  That's awfully high, and in fact exceeds the bandwidth of the current sense amplifier by a few orders of magnitude.  The LM358 only has a gain-bandwidth of 1MHz, so with a gain of about 40 you've only got about 25kHz of bandwidth to begin with.  How much bandwidth do you actually need? 

I suspect there will also be issues resulting from layout, but without seeing the schematic and layout for your whole power stage it will be hard to say.  As a band-aid you could add another filter at the point where the comparator output is used to prevent very short pulses from being acted upon. 

[ali_asadzadeh]

Thanks, I have changed the R26 to 1K ohm and C17 to 100nf, but there is still much noise and spike!