Post Go back to editing

Questions: When ADA4930 driving AD9633

1. The input common-mode voltage range of ADA4930-2 should be 0.3~2.8V. But, i found this parameter has different range in Chinese version datasheet, where is 0.3~1.2V. Also, there is same issue about output voltage range.

2. There are two blocks between ADA4930 and AD9633, which need me to make choices: one is the mach resistor, the DEV kit used 24 ohm, our previous board (between AD8138 and AD9218) used 50 ohm, there is no match resistor in figure 60 of ADA4930 datasheet, however 33ohm in figure 59; aonther is the RLC filter, is this necessary for my application? And  will the inductance destroy the DC coupling fucntion?

 

3.The DEV kit schamatic of the ADA4930 is same as figure 55 of the datasheet, so the ADA4930 was configured as G=1 and input 2Vpp, output also 2Vpp. My system, the front signal is 1Vpp, so i should refer to figure 60, am i right? Which convert the input 1Vpp to differential 2Vpp.

 

4, In dev kit, use Rcm(1.82K) pull-up to +3.3V. There also have several pull-down resistors. Can i remove these pull-down resistor in my design?

 

5. Can you teach me how to calculate each node's voltage based on figure 60 of ADA4930 datasheet? For, datasheet said "For a common-mode voltage of 1 V, each ADA4930-2 output swings between 0.501 V and 1.498 V, providing a 1.994 V p-p differential output. " I can understand how to get 1994 V p-p, but i don't know how to get 0.501 V and 1.498 V.

  • Hi Coyoo,

    1. The ICMVR of ADA4930 varies with the supply voltage. I think you just misread the datasheet. To guide you, here is a table for ICMVR at different supply values.

    Supply Voltage

    3.3 V 5.0 V
    Input Common-Mode Voltage 0.3 to 1.2 V 0.3 to 2.8 V

    2. If you need to have a post-amp filter prior to the ADC, then use the RLC filter network. This filter if configured for 40 MHz corner frequency, if you ought to have a different cutoff, you can change the values to match the your desired corner frequency. I suggest you use 33 ohms just like in figure 59 for a better filter response. And NO, the inductance doesn't affect the Common signal on the outputs.

    3. Yes you can refer to figure 60. Although the best way to do it is by using the DiffAmpCalc tool, the benefit of this tool is that you can see the "real-time" signals and values are customizeable depending on the driver setting you like.

    4. If you notice on the dev kit, the 1.8k pull-down resistors are in DNI, this means that upon shipping those resistors are unconnected. It's up to the customer (you) if you like to center your inputs at Vs/2, which is 1.65V at 3.3V, by installing the 1.8k pull-down resistors. However, leaving those pull-down resistors as DNI, the effective DC offset will depend on the voltage divider network of 1.8k pull-up resistors and the Rin_se. On your current design, the DC offset (Vicm) is at 440 mV.

    5. Since the Vocm is set at 1V, therefore the outputs will have an DC offset of 1V away from ground. The differential gain is 2V/V (1.994V/V to be exact) but the single-ended gain (that means, each single-ended output, the +OUT and -OUT) is 1V/V. Given that the outputs has Vocm at 1V and the input is 1Vpp, therefore the outputs will be centered at 1V with a pk-pk amplitude of 1V. So effectively, the minimum voltage at each single-ended output is:

    Vocm - Vin(1V/V)/2 = 1V - 1V(1V/V)/2 = 1V - 0.5V = 0.5V.

    while the maximum voltage at each output is:

    Vocm + Vin(1V/V)/2 = 1V + 1V(1V/V)/2 = 1V + 0.5V = 1.5V.

    The best explanation for this voltage levels can be found in this application note. High Speed Differential ADC Driver Design Considerations.

    For more guidance with designing ADC drivers, I strongly suggest that you use the DiffAmpCalc tool in parallel with the application note for the computations.

    Regards,

    Jino

  • Hi Coyoo,

    2. The filter will suppress noise, specifically those at higher frequencies. For 80MSPS ADCS, the maximum input you can process is 40 MHz, beyond this frequency the signal is useless and it may add noise and distortion effects to the signal of interest. If you have enough board space, I recommend you to implement a post-amp filter  for noise suppression as well as amp-ADC isolation.

    3. True. You can not implement an Rcm in the tool. But what you can do is play with DiffAmpCalc, find your operating points. Then if you have your target common mode, you can separately implement the Rcm to set the input common mode.

    4. Yes. It is set to 440 mV. To get this value, you need to know the single-ended input impedance. Given in figure 60, Rin_se = 1/[(1/50)-(1/88.5)] = 114.94 ohms. Therefore ADA4930's effective Vicm is 3.3V(114.94/(739 + 114.94)) = 444 mV (440 mV rounded). It's a very complicated process to come up with these resistor values, that's why I highly recommend the DiffAmpCalc tool to compute for those. Adding a pull-up resistor doesn't affect much so it's okay to disregard it in the tool. If you still unclear with the process on how this values are calculated, the application note above would be a great help.

    For the AD8001 question. just create a separate thread for it to be easily tracked down.

    Regards,

    Jino

  • Hello Jino,

    1. It's my fault. I confused between 3.3V and 5V version.

    2. I don't know If i need to have a post-amp filter prior to the ADC. Is this filter just suppress noise?

    3. For single power supply, there are two ways for input common-mode adjustment. I don't want to use DC biased source. So i want to use resistors, Rcm. It seems i can't implement Rcm in Diffamp-calc? I can only play it like this:

    4. Did you mean the ADA4930's input common-mode voltage is adjusted to 440mV? Sorry, i don't know how to get this 440mV. Vicm=(+Vin + -Vin)/2, -Vin=3300*(96.2/(739+96.2)), So is -Vin=380mV? But, how to get +Vin?

    BTW, we are debugging AD8001, and encounter some problem, can i get you help here? Or i have create another thread?

    Thank you very much!

  • Hi Coyoo,

    To enlighten you the major difference between figure 59 and 60, see the image below:

    Red Curve is figure 59 and Green is figure 60. To put it simply, the datasheet is for edit. Figure 60 should also include a 33 ohm for the filter network. If you will use AD9633 at 80 MSPS then use the same exact RLC values, otherwise calculate to match your desired filter response.

    Regards.

  • Hi Jino,

    2. Actually, our ADC is AD9633. I just want us discussing based on same part ^0^. For the filter prior to ADC, the inductors&capacitors, can i use the same values? Or i have to caulculate according some formula to get the values? Also, you said the impedence between ADA4930 and AD9633 using 33 ohms, why not 50 ohms? Is the filter will add impedence to the path? Figure 60 doesn't add any resistors between driver and ADC, how to understand the difference between figure 59 and figure 60 about this?

    3. Closed,^0^.

    4. There are some difficults for me to undertand this so quickly, i'm not a analog engineer, however, i want to process ADC's output data. So i am the board design ower, i have to take those analog knowledges which i have given back to my teacher. I'll read the App note carefully. So it's back to the primary question, because i'm not confident about the driver circuit, can i just design my driver circuit refer to figure 60. Just replace the AD9640 by AD9633, and keep the same resistors configurations of the ADA4930. Our input source also is 1V p-p single ended, and make good 50 ohms impedence match in front designs. The Vocm will connect to AD9633's Vcm pin (which output 900 mV). Also ADA4930-2 will use single 3.3V power supply.

    For AD8001 issues, i'll create another thread.

    PS: I create new thread here:https://ez.analog.com/message/284446 , if possible, please help to take a look.

    Best regard.

  • Hello Jino,

    Thanks, AD9633 will run at 100Msps.

  • You might as well consider in redesigning your filter at 50MHz, if you maximize it.

    Regards.