ADL5519 output noise

Hello,

We've been using multiple ADL5513 for VSWR measurement, but I am considering using the ADL5519 instead. Low output noise is critical for this application, especially at low frequencies <1kHz. I had asked previously here about noise performance of the ADL5513 below 1kHz, and got a very helpful response. I was hoping to get similar data for the ADL5519, both for the single OUTA/B and difference OUTP/N pins.

Another question: I was very surprised that the datasheet shows near identical noise plots for the single channel outputs vs the difference outputs. I would assume the difference outputs to be 3dB higher since they will sum the noise from each single channel. The additional opamps used for the difference outputs should add even more noise. So how can this be real?

Regards,

Mike

  • +1
    •  Analog Employees 
    on Apr 28, 2021 11:24 PM

    Greetings Mike,

    Sorry about the delayed response to your post. 

    It was very fortunate that we were able to find the ADL5513 output NSD plots that extended to lower frequency. I've looked but have not been able to locate such data for ADL5519. 

    About the NSD graphs, Figures 43 and 44: Looking close, we do indeed see the diff. outputs OUTP, OUTN 3dB higher for noise at 1MHz and higher. For 100kHz and lower, NSD is about the same, as stated. Its quite possible that at low frequencies, the IC internal bias supplies are dominating the output NSD as measured at both OUTA/OUTB and diff. OUTP/OUTN. 

    Be aware that ADL5519 channel-to-channel isolation degrades up at the higher RF input frequencies. Datasheet pages 5 and 6 show 46dB isolation at 2.2GHz, quickly decreasing to only 20dB isolation at 3.6GHz. Below 2.2GHz, isolation is 46dB or better, typically. Other dual RF detectors to consider are AD8364, HMC1030, and LTC5583.   -Bruce H. 

  • Hi Bruce, thanks for getting to my questions

    I now have the ADL5519 dev board so I'll attempt to measure its NSD with a good LNA when I get the chance.

    As for comparing the noise of the difference/single outputs, I'm not sure I follow you. Noise on the biasing of the opamps should only appear as common mode output noise, right? I was assuming that fig 43-44 are depicting the differential noise, but is it actually showing the noise at each pin in a single-ended measurement?

    Just a crazy thought: might it be possible to measure the detector outputs at the CLPA/CLPB pins, since they connect to the input of each driver for the "real" output pins? Any chance it might have higher SNR (along with a different slope/intercept)?

    I'm working at frequencies between 50-500MHz, thanks for the tip though.

    Regards,

    Mike

  • 0
    •  Analog Employees 
    on May 7, 2021 2:54 AM in reply to mtwieg

    Greetings Mike, 

    Again, sorry for the delayed response. 

    1. DS figure 43 should be a differential measurement, but right now I cannot confirm that. 

    2. Using the CLPx pins as output for lowest noise should be OK, provided a suitable external high input impedance buffer is utilized. You would need to do your own testing to validate if there is any benefit. 

    -Bruce H. 

  • Hi Bruce,

    I spent a lot of time doing my own measurements of the noise. My setup looks like below:

    I take several measurements with different connections between the ADL5519 and the LNA. All the outputs (OUTA/B/P/N) have 14kHz RC lowpass filters on them (200ohm/56nF). I acquire 200ms of data (20000 points) on the scope for each measurement. Here's the spectra of the data:

    Overall seeing much higher noise levels than suggested by the datasheet (around 10x for the OUTA/B outputs between 1-10kHz). I refined the setup and repeated things a few times trying to close the gap, but I'm confident this is the real output noise (you can see the noise floor when my LNA inputs are shorted is far below the other traces). I tried using different synthesizers as well but didn't see any significant effects. Powering things from batteries didn't help either.

    Absolute noise levels aside, the relative magnitudes are as I expected them. OutN and OutP are each about 3dB above OutA and OutB (but OutN and OutP have the same noise as (OutA-OutB), also expected). Furthermore, (OutP-OutN) is 6dB above either OutP or OutN, again no surprises.

    Overall I think the performance is not bad, it's very similar to the ADL5513 I had been using, and I'm fine with that. But I think the datasheet really needs some corrections here...

    I will also check the CLP pins with my LNA, just out of curiosity.

  • +1
    •  Analog Employees 
    on May 19, 2021 7:21 PM in reply to mtwieg

    Greetings Mike,

    Thanks for sharing your results. Perhaps others can benefit. There does seem to be a gap between datasheet specification and actual NSD measurements. 

    This test requires a super low AM noise on the SigGen. To help confirm, you might try some other lab-grade Signal Generators to see if there is a difference.   -Bruce H.