AD8310 - Possible Output Integrity Issue

Hi everyone,
Some time ago I started building my own scalar network analyzer. I already got some great help from Bruce_H on this forum so far for which I am very greatful! Do note that I am not a professional electrical engineer when answering/suggesting, but a chemist-turned-electronics-hobbyist, haha.

The main idea of my board is as follows:

AD9851 DDS chip outputs a sine wave with programmable frequency (5 - 60 MHz). This is passed through and LPF to a buffer amplifier which in turn sends it to a quartz crystal to assess its resonance frequency. I measure the envelope of the sine wave that passes through the quartz crystal using an AD8310 in single-ended configuration. The output is again buffered by an opamp to be sent to the microcontroller ADC to measure the voltage. I use python as my DAQ platform and to control the frequency programming over usb.

Thus far I can see that it works mostly as expected. However, I still have some issues related to noise. When I try to pinpoint the origin of the noise I can at least see that something strange is happening at the AD8310 output. During my frequency sweeps for some frequencies the AD8310 output voltage suddenly drops seemingly out of nowhere.. I know this is not much to go on, considering the full scope of the project but I was wondering if someone has had any similar issues? Most of the time it works just fine, but every once in a while it seems ot run into this instability. Could it be the AD8310 or do you suspect something else?

I posted an image of what it looks like in my DAQ software below. I would really appreciate some help/input.

If you require additional information or have ideas of what I could probe with my scope, please let me know!


  • 0
    •  Analog Employees 
    on Apr 2, 2021 8:03 PM

    Hi Rens,

    Glad to see your application is basically working out well, except for this glitch that, at times, seems to occur with some periodic repetition rate. The periodicity that we see above should be viewed as a clue to the problem. 

    AD8310 has been around a long time and I'm not aware of any problems with the design or output integrity. AD8310 is similar to AD8307, except for output buffer added, to simplify the output interface. 

    To help isolate the problem, you could test the AD8310 by itself: connect the detector directly to a SigGen. Run the level up and down, run the frequency up and down also. Observe detector output on an oscilloscope and make sure the output is clean, glitch free, and no excessive ripple or settling time. 

    If no detector problem observed, try programming for longer settling time before taking the measurement. If that fixes it, the problem might be the DDC programming commands are taking too long to reach the DDC. That would be a PC or computer issue. 

    Other readers, feel free to comment!  -Bruce H. 

  • Hi Bruce! Happy easter!

    I have done more tests over the weekend. It turns out I have made a 4G antenna on the input side. The periodic signal was a background app on wifi checkimg in to something. When I turned off wifi the interference was continuous as it switched to 4G internet. In airplane mode there is no interference at all.

    Good thing that there is no problem with the AD8310. Now I need to consider my options to mitigate this problem. I already ordered an aluminium project box, but the problem might be the cell in which I mount my quartz crystals as it has a long unshielded connection. Are there any other tricks I could use?

    Thanks in advance!

    Cheers, Rens

  • 0
    •  Analog Employees 
    on Apr 6, 2021 7:07 PM in reply to Xerrorable

    Hi Rens,

    Glad you figured that out! For the benefit of other readers, RF detectors in general are quite broadband, and sometimes very sensitive also. Cell phone RFI can sometimes be a big problem. RF filters are normally recommended for each custom application, to help prevent these problems. 

    in this case, several possible solutions exist:

    • keep leads short.
    • 5 to 60 MHz bandpass filters at the AD8310 input
    • slide a shield braid over the existing cable, and RF ground at both ends
    • or consider twisting the long leads to minimize the differential pickup. 
    • use AD8310 in differential input, not single-ended input config.
    • or just keep stray RF turned off or put away
    • a combination of the above.

    Hope some of these ideas help!   -Bruce H. 

  • Hey Bruce,

    Thanks for the reply! I did see that the AD8310 has a broadband input, but I did not consider that I could simply add a band pass filter. (Are there any BPF designs you would recommend?).

    Unfortunately, we cannot do anything to the wiring (twisting them would have been a nice idea) as it is embedded in a cell we bought.

    Using the AD8310 in differential input would mean that the output is basically a measure for the subtraction of the waveform passing the quartz from the waveform on the input of the quartz, correct? I guess care should then be taking in trace matching of both inputs to the AD8310?

    Right now, my solution is to keep RF turned off, but this is far from ideal. So, I will attempt some of your suggestions.

    Many thanks!



  • 0
    •  Analog Employees 
    on Apr 8, 2021 7:10 PM in reply to Xerrorable

    Hi Rens,

    AD8310 is inherently differential by design. Input is differential, pins 8 and 1, although good performance is obtained by simply using single-ended drive. The idea of differential drive would be that the desired signal is applied in normal-mode to both input pins 8 and 1, while the interference gets applied in common-mode to input pins 8 and 1. Thus the AD8310 measures the differential-mode signal, while rejecting the common-mode interference. Such a wiring configuration may or may not be possible, depending on the cable harness arrangement supplied by outside supplier. Note that this concept of 'differential' is a lot different from the differential described above. 

    In this case, it sounds like a conventional single-ended bandpass filter may be the best choice. If the frequency range is to be 5 to 60 MHz, best results are likely obtained by actually designing a low-pass filter at 60 MHz, cascaded with a high-pass filter at 5 MHz. This is because of the wide separation between 5 and 60 MHz. 

    Because AD8310 is sensitive to very low level signals, a passive filter design is recommended, as opposed to an active filter design which would produce too much noise.    -Bruce H.