Post Go back to editing

RF SPST ADG902 switch time domain response

Hi there,

I intend to use the ADG902 switch for backscattering applications, with a control signal pulse of 50 MHz. The time domain response of the switch in the forward direction is shown below:

which matches the figure in the datasheet (ignore the blue RF signal, only the red envelope is meaningful in this figure). However, when I looked at the return signal in the same experiment, I got this result:

which shows clearly some heavy distortions. Such distortions happen for lower frequency of switching too. For instance, for a 10 MHz pulse the result is shown below in (d):

,but at lower frequency the distortions are less important as the period of the pulse becomes larger.

So in general, I'm wondering why such distortion is happening. Can we have a simplified model of the internal circuit?

I understand that usually the return signal is not important for many applications, but for my application this is especially important (the applications is a backscattering wireless node).

Also in general, similar switches (eg HMC194) also seems not to work at such high frequency of switching (50 MHz). What determines the highest frequency of switching for a switch? I don't think this data is available anywhere in the datasheet.

Any help would be much appreciated!

Update (Nov 22, 2018): To clarify, a schematic for my test setup is shown below:

  

I'm using a circulator to collect the return signal. In this setup, the switch is alternating between on and off every 20 ns, so exactly one half of the period of the RF signal should be in the through signal and one half should be in the return signal.



Added a schematic of the test setup
[edited by: Navid_rz at 3:25 PM (GMT -5) on 22 Nov 2018]
  • Hello,

     I had a look at the plots, the return path plot do look bit suspicious. Can you please clarify what do you mean by return signal. Is that return loss you are talking here? The return loss of the switch at 1GHz is 25dB (min) which means only less than 0.1% power will be reflected back so you will not see much reflections. One more thing can you please confirm does the RF traces on your board and also your set up has 50ohm impedance.

     Regarding the switching time, it can be calculated by 1/Ton which is around  277MHz. So ADG902 will definitely able to switch at 50MHz.

    Thanks

    Naveen

  • Hi Naveen,

    Thanks for your reply. Please see my update in the post which clarifies what I meant by the return signal. Basically, I meant the signal that returns from the switch while it's switching, not in a static mode. If you have a circulator and a fast time scope (I used a 40 GHz microwave transition analyzer), it's easy to run the experiment.

    Also, I have ruled out any problem with impedance matching as my measured s-parameters match those provided by analog devices pretty closely. See below:

    https://imgur.com/1aT9IWm

  • Hi Navid,

    I need to set it up in the lab to figure out this issue so need some extra time. I will get back to you soon.

    Thanks

    Naveen

  • Thanks Naveen, Looking forward to hearing about your results.

    Navid

  • Hi Navid,

    I did detailed investigation and figured out that the distortion which you are seeing on the return signal looks like is due to the extra board capacitance which is interfering with the return signal when you are switching at 50MHz. 

    I have replicated your set up using EVAL-ADG902EBZ. Below is the scope plot which I have measured at 50MHz. Yellow is through signal and green is return signal.I am not seeing those distortions on the return signal which you are observing.

    Scope plot at 10MHz, again yellow is through signal and green is return signal

    Can you please try to reduce the board capacitance or you can order a EVAL-ADG902EBZ evaluation board to replicate the results which I have plotted.

    Thanks

    Naveen

  • Hi Naveen,

    Thanks a lot for your detailed investigation. It's good news that the problem is with my board and not the switch! However, I still have a few questions that I'd appreciate if you could help:

    1. Could you clarify which capacitance on the board you are talking about?

    Here's a picture of my board: https://imgur.com/a/Ntv3pHi

    I basically tried to be as close as possible to the ADI eval board (please ignore that RF1 is shorted to GND, this was done for a later experiment). The only big difference is that I did not put any capacitance or resistance on the Control line, like it is done in ADI's eval board in Figure 4 in https://www.analog.com/en/design-center/evaluation-hardware-and-software/evaluation-boards-kits/eval-adg902.html#eb-documentation. I actually did not understand the point of having them, so if you could elaborate on that also, that would be great.

    Could this be the issue? 

    2. Could you also tell me what source you are using to drive the switch with? This would help me redo the results with the same type of source. I have tried both a pulse generator with a low output impedance and another CMOS clock generator, both gave me the same result. And I have also tried driving it with an FPGA, but got the same problem.

    Meanwhile I will order the ADI's eval board to redo the tests and will update you with my results with that board as well.

    Thanks again,

    Navid

  • Hi Navid,

    After looking at your board, I can't see anything wrong with your board. The cap on VDD and Ctrl line are the decoupling caps just to avoid any ripples coming from power supply.

    I am using Agilent 33250A 80MHz arbitrary waveform generator, HMC - T2220 signal generator, SFC2040A circulator and R&S scope terminated to 50 ohm

    Thanks

    Naveen