What would be the best FDA for AD9690-500

We are designing a high speed pulse digitizer that needs strictly DC coupled signal path through out (pulses are max +/-500mV and not necessarily DC centered). For ADC the AD9690-500 will do just fine but I have a hard time finding a good enough FDA to go with it. The main problem being the high Vcm that is required for the input buffer of the ADC (2V). All the high speed FDAs are restricted to 5V total supply range and will not work if wasting 2V on Vcm on their output I suppose. Any suggestion?

Regards/Ramin

Parents
  • 0
    •  Analog Employees 
    •  Super User 
    on Sep 5, 2019 1:52 AM

    Hi Ramfar,

    LTC6404-4 is one FDA capable of driving AD9690-500 (~2Vpp diff out, 2.06V CM, 14 bit specs) running off of 5V or even 3.3V supplies. You can see the operation using LTspice as shown below:

    LT6404-4 AD9690 FDA driver EZ 9_4_19.zip

    However, the LTC6404-4 harmonic distortion (HD) limits its usefulness (<-95dBc) to 10MHz or less!

    I think if you use ADL5565 with a 5V power supply, you'd get low distortion (<-80dBc) to 250MHz with Vocm = 2.06V and 2Vpp diff output, per AD9690-500 requirements:

    Hope this helps.

    Regards,

    Hooman

Reply
  • 0
    •  Analog Employees 
    •  Super User 
    on Sep 5, 2019 1:52 AM

    Hi Ramfar,

    LTC6404-4 is one FDA capable of driving AD9690-500 (~2Vpp diff out, 2.06V CM, 14 bit specs) running off of 5V or even 3.3V supplies. You can see the operation using LTspice as shown below:

    LT6404-4 AD9690 FDA driver EZ 9_4_19.zip

    However, the LTC6404-4 harmonic distortion (HD) limits its usefulness (<-95dBc) to 10MHz or less!

    I think if you use ADL5565 with a 5V power supply, you'd get low distortion (<-80dBc) to 250MHz with Vocm = 2.06V and 2Vpp diff output, per AD9690-500 requirements:

    Hope this helps.

    Regards,

    Hooman

Children
  • Hi Hooman,

    I thought you might suggest ADL5565. But, and here is the rub, because of its limited rail I don’t think I can use it in my particular application. Here is my reasoning and correct me if I’m wrong: My signal consist of irregular pulses. The base line, amplitudes (and lengths) which are all highly variable could be randomly anywhere between +/-500mV and the pulses are not DC centered. So I have to have DC coupled signal paths and +/- rails which in itself is a hassle with ADL5565 not having a VEE pin (but I’m sure it works with GND connected to VEE). Say we chose VCC to be +3.5V and VEE to be -1.5V and configure for a gain of 1. Then considering Vcom of 2V which I have to have because of the ADC input, I get as close as 1V to both rails on inputs and outputs. And I think ADL5565 would object to this resolutely by producing excessive HD2 and HD3.

    Question #1 do you agree with me?

    Question #2 If yes, is there a way round this? I know there are no better parts than ADL5565 when it comes to linearity and BW but can we do any biasing outside the device when DC coupled to remedy this?

     

    Regards/Ramin

  • 0
    •  Analog Employees 
    •  Super User 
    on Sep 5, 2019 5:32 PM in reply to Ramfar

    Hi Ramin,

    If you run the ADL5565 with 5V single supply (VEE = 0V or ground), you would not have any issues getting 2Vpp diff output with Vocm = 2V, per plots shown below (f = 100MHz):

    10dBm = 10mW --> Vout_diff (RL = 200ohm assumed) = 4Vpp_diff (2Vpp at each output pin)

    As you can see, the input / output headroom issues you've noted should not be an issue according to these plots, albeit these plots are with AC coupling condition. From these plots, 1V headroom to each supply seems adequate.

    I think your question about ability to run the device with asymmetrical supplies (V+ = 3.5V, V- = -1.5V) is possible with ADL5565. However, I don't think you need to do that with the conditions you're facing per above. I would try single 5V supply as it seems the headroom to V+ needs to be larger than the headroom to V- based on HD plots.

    BTW, I've not tested the ADL5565 on the bench. These are my best guesses based on the plots. The only caveat being the datasheet is heavily tilted towards AC coupling vs. your DC coupling condition. I hope I'm not overlooking any issues with DC coupling (but I cannot be 100%).

    Regards,

    Hooman

  • 0
    •  Analog Employees 
    •  Super User 
    on Sep 5, 2019 5:34 PM in reply to Hooman

    I forgot to mention the ADL5565 minimum gain setting is 6dB.

    Regards,

    Hooman

  • Hi Hooman,

    What many of us who are used to AC coupled signals tend to forget is that there is a world of difference between an AC coupled signal that has consequently zero DC component and a DC coupled signal that has zero DC component. Both look the same if you plot them on a graph but the former is floating and hence could be connected to the inputs of a single supply diff amp whereas the latter is not floating and if connected to the same diff amp results in clipping and my signal is not even zero DC it can have any DC plus and minus that should be conserved through the diff amp. Only if the diff amp is supplied with + and – rails it will work with DC coupled signal. Believe me I have a ADA4927-1 board on my bench that I experiment with a lot.

    So no I cannot do as you suggest with single supply which brings us back to my original question: is there a way to bias the output CM level without pushing the diff amp too close to its rails?

     

    Regards/Ramin

  • 0
    •  Analog Employees 
    •  Super User 
    on Sep 6, 2019 6:38 PM in reply to Ramfar

    Hi Ramin,

    Yes, you can run the ADL5565 with skewed supplies you've noted (V+= 3.5V, V-= -1.5V).

    I think I see your point about DC coupling having hidden issues which I had overlooked.

    One approach would be to make sure the device input / output CM voltage ranges are not violated in your particular DC coupled application: 

    5V operation:

    Input CM voltage range: 1.2V to 3.8V

    Output CM voltage range: 1.4V to 3.0V 

    Assuming 6dB operation (RF = 200ohm, RG= 100ohm, SE to diff operation, Vsource_DC = 0V), here are the input / output voltages you'll encounter when DC coupled (VCOM = 2V, 2Vpp diff output, 5V single supply):

    • Input_hi = Vo_hi * 100ohm / (100ohm + 200ohm) = 2.5V * 0.333 = 0.833V (violates the input voltage condition!)
    • Input_low = Vo_lo * 100ohm / (100ohm + 200ohm) = 1.5V * 0.333 = 0.50V (violates the input voltage condition!)

    Skewed Supplies (V+ = 3.5V, V- = -1.5V) operation:

    Input CM voltage range: -0.3V to 2.3V (shifted by V- and V+ shifts respectively)

    Output CM voltage range: -0.1V to 1.5V (shifted by V- and V+ shifts respectively)

    Now the 0.5V to 0.833V input voltage excursions fall in the Input CM voltage range. But now your output voltage range no longer supports what you need (VCOM= 2V)!

    Here is an idea: Use ADL5565 with single 5V supply but add pull-up resistors to bring the input voltage into compliance. This would not impact gain. I'm talking about R5, and R6 in the zip file below. I'm using the LTC6404-4 LTspice model to predict the input / output voltages only:

    LTC6404-4 ADL5565 LTspice IO pullup 9_6_19.zip

    I've included the LTspice file if you'd like to experiment with it.

     

    Regards,

    Hooman