This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

PGA113: causing distorted signal on pre-amp circuit output

Part Number: PGA113
Other Parts Discussed in Thread: LM358, OPA364

Hi all, 

I am trying to design a programmable amplifier board. The intent was to be something small with the PGA113 at its heart, being controlled by a PSoC5LP. However, I have noticed that when the gain is set to anything other than unity on the PGA113, there is noticeable distortion on the INPUT signal to the PGA113. 

Here is an image of the schematic being used. It is designed with am LM358 preamp with LPF and adding a DC bias at half voltage: 

the signal Sig1_Preamp is the signal in question here. I have noticed that when channel 1 on the PGA113 is enabled and the gain is set to 5 or above, what was once a clean signal from the pre-amp circuit becomes noticeably distorted. Those distortions get amplified and are seen on the output of the PGA113. When I disable the channel or switch to the other channel on the PGA113, the signal returns to its normal, clean state. I have more or less followed all of the design recommendations for the PGA113, so I am unsure what exactly could becausing the problem here! It's like the PGA113 is loading the pre-amp circuit despite it having relatively high impedance inputs with very low input bias current! 

Any assistance is appreciated. 

Some extra info: 

Signal input is capacitvely-coupled (ac-coupled) through a 3.3uF cap into the SIG1 input on the schematic. I am using a 100mVpp @ 5kHz sine wave test signal. If I shrink the input to be vanishingly small (so small that it's mostly noise), I can make the lower gain values (2x and 5x) lose the weird noise signal. If it is desired I can provide scope shots with specific input signals. 

  • Hi Xavier,

    Can you provide oscilloscope shots of the "SIG1" node (next to R2, on U1A) and  "SIG1PPREAMP" (at U1A output) first when the PGA113 is set to G=1, and another set of oscilloscope shots when the PGA113 gain is set to G=5?  Please set the oscilloscope to DC coupling.

    Is the issue occurring on one programmable amplifier PCB board or on various PCB boards?  If occuring on (1) pre-amplifier board, have you attempted replacing the PGA113 with a fresh device?

    Thank you and Regards,

    Luis 

  • Hi Xavier,

    have you noticed that the datasheet recommends a more complex midsupply driver for the VREF pin of PGA113 ? See figure 72. There must be a reason for it.

    Kai
  • Hi Luis,

    Here are the scope shots you asked for. I used a slightly larger input signal to make the scope shots a bit cleaner. In this case, it is 500mVpp @ 5kHz. 

    Channel Off:

    Channel On, Gain = 1: 

    Channel On Gain = 5:

    I ordered a set of boards with this configuration and they all present the same issues with the same noise on the waveform. 

    These scope shots were taken with a second person simply sending commands that should change the channel and gain selection while the scope probes were held stable. Do you recognize any of this weirdness? 

  • Hi Kai,

    I did notice that it's different, but the functional difference is the LPF in the feedback path and a bit more stability on the signal thanks to C_L2. The noise on my VREF line is very low and it's at a pretty stable 2.48V, so I am happy with it so far, but I could be convinced to change it if we can't identify the problem elsewhere.
  • Hi Xavier
    The LM358 requires the common-mode input signal range to be from 0V to (V+)-1.5V; which should be okay if VREF=2.5V and V+=5V; and SIG1/SIG1_PREAMP are centered at the 2.5V DC level. As you have mentioned, the PGA113 CH1 terminal is expected to be very high impedance, since you are looking at a multiplexer and the non-inverting input of an operational amplifier.

    If possible, please confirm the voltage level of the +5V supply with an external meter when the issue occurs.

    Thank you,

    Best Regards,

    Luis
  • Hi Luis,

    The +5V rail is clean. As an experiment, I powered the LM358 from +/-9V to make sure the distortion wasn't induced by being too close to the CM input range of the LM358. The distortion was still present unfortunately.

    I doubt it, but could this be a layout problem? I can't really think of how other than a parasitic feedback path, but the layout is so simple...

  • Hi Xavier,

    the OPA364 will make a huge difference as it is much faster than the LM358. The LM358 is a dino and shows a much poorer performance than the OPA364 and this in many respects.

    CL2 is no simple low pass filter. It's a storage cap and minimizes the source impedance of VREF driver at high frequencies, exactly in the frequency range where you observe the distortion...

    CF is no low pass filter either but is a phase lead capacitance which helps to recover the phase margin of OPA364.

    I don't think that the LM358 is a good choice here, neither at the VREF pin nor in the signal path at the input of PGA113.

    Kai
  • Hi Xavier,

    can you use an external power supply to provide the 2.5V reference voltage, just for test? And don't forget the decoupling caps at the VREF pin of PGA113 and the +inputs of OPAmps U1A and U1B. Also test the performance with a better OPAmp for U1A and U1B.

    Kai
  • Yea, that was one of the first notes I made. I didn't realize it when i selected it but the LM358 is actually pretty bad. I will consider the OPA364 in the next rev. Thanks for the suggestion.

    CL2... Makes sense that it minimizes the source impedance at higher frequencies, but given that I largely expect VRef to be DC, I don't know that I understand why we would sniff that out as a problem? 

    CF... again makes sense that it would give you back a bit of your phase margin, but same as above, I don't understand why in this case we are worried about recovering phase margin? Am I thinking about this too simply? The VRef signal goes into the feedback block of the opamp internal to the PGA113 so is it like... the back coupling of the internal opamp back onto the VRef signal? The VRef signal, when probed, is very clean, so I am a bit lost here. 

    I'm going to see if I can get my hands on some other drop-in OpAmps for the LM358 today and maybe that will help? 

    Edit:

    2.5VRef signal DOES have a blip on it! I shall make efforts to clean it. I suspect this is the problem. Will report back. 

  • Also, to all reading, when I lower the frequency of the input signal, the noise is still there all the way down to 100Hz, HOWEVER, it starts to occur closer and closer to the VRef signal crossover point. So in this case it looks more and more like crossover noise? weird.
  • Hi Xavier,

    the PGA113 makes internal switchings. Think of the MUX and the digital interface. All this could inject noise into the VREF line. That's why CL2 might be needed. Again, there must be a reason for the complex scheme shown in figure 72.

    This noise might also appear on the input lines of PGA113 and a good OPAmp with a low output impedance might be able to suppress it.

    Sorry, that I cannot tell more. I haven't designed the PGA113. What I can tell here is only intuition drawn from experience...

    Kai

  • Hi Xavier,

    the LM358 is known for crossover issues...

    Kai
  • KAI WINS.

    Somehow I didn't notice the noise on the 2.5VRef signal. Adding a capacitor between it and ground cleaned everything up immediately. I didn't even have to switch opamps from the LM358. It's pretty obvious to me now that it needs it. Thanks Kai!
  • Hi Xavier,

    these are good news... :-)

    Good luck!

    Kai