DC coupled input path on ADS62PXXEVM - THS4509

Anthony,

It sounds like you are using a THS4509EVM to drive the ADS62P24EVM? Are you intending to drive each channel of the ADS62P24 with a THS4509 (one channel using the on-board THS4509 then using a THS4509EVM to drive the other)? How have you configured the on-board THS4509 on the ADS62P24EVM - supplies, interfacing components?

With regards to connecting the 1.5V CM output of the ADC to the CM input of the THS4509, the answer is, it depends. If you are powering the THS4509 with single +5V supply, you can tie the 1.5V to the CM pin of the THS4509 to set the output common mode of the THS4509 to 1.5V (the output common mode will default to midsupply, or 2.5V in this case).

However, I would recommend powering the EVM with +4V/-1V supplies if possible, as this will automatically center the output common-mode of the THS4509 to its midsupply voltage level of 1.5V, which is the required common mode input of the ADC. On a single 5V supply, you will need to shift the output common mode 1V down from the 2.5V midsupply to 1.5V, which limits your differential output swing before you start clipping. If you look at Figures 11 and 12 of the THS4509 datasheet, you will see that the distortion performance degrades as the THS4509 output common mode is shifted from its midsupply for this very reason.

A schematic with all supply voltages labeled would help me understand the situation better. What is the input signal to the THS4509, in particular, the signal amplitude, single-ended/differential, DC common-mode voltage?

Kristopffer,

Thanks for the reply. I copied my post in the high speed converter forum below as I was directed by Kang Hsia that it would all be answered here. I already answered some of your questions here. I saw the recommendation of powering with a +4/-1 supply, but was thinking the CM pin driven by the adc would give me the offset that I need. Your comment above state that connecting the output of the ADC to the 4509 CM pin will set the common mode of the 4509 to 1.5v, but then you say that it will default to 2.5v - I thought it would default to 2.5v or mid supply if the CM was unconnected. I do see the shift and clipping that you describe. I will try to find another supply. However, I still think I have issue of attenuation that may have to do with impedence matching the to the ADC input. I will try to include my Tina schematic.

Thanks for the help.

Anthony,

I apologize if I wasn't clear before. Yes, if you leave the THS4509 CM pin unconnected, the output common mode will default to mid-supply, which for the single 5V supply case will be 2.5V. Tying the 1.5V from the ADC to the THS4509 CM pin will move the THS4509 output common mode to 1.5V, however, as mentioned, it could cause clipping if your output signal swings beyond the the output swing range. For example, on a single 5V supply, each single-ended output can only swing from 1.1V to 3.9V. Since the output is centered at 1.5V, each single-ended output can only swing down 400mV before clipping.

But this is separate from the issue you are seeing. You are seeing a steady 1.5V output common mode because the input of the THS4509 is being saturated. On a single 5V supply, the input common mode range is going to be from 1.1V to 3.9V. With a single-ended input and output common mode set to 1.5V and circuit configured for G=10dB, the voltage at the IN+ and IN- pins of the THS4509 will be at about 400mV, which is below this range. By using 4V/-1V supplies, the input common mode range will be shifted down and there will not be an issue.

Since your signal input is only 20-50mVpp, which  will result in an output swing of only 63-158mVpp with a gain of 10dB, there is another option. You can use a THS4508 on a single +5V supply to avoid the need for the awkward 4V/-1V supplies. The THS4508 input common mode range goes down to the negative supply. The THS4508 is pin-compatible to the THS4509 so you can order samples and just swap out the parts on your EVM and on the ADC board.

Please see attached 2 TINA circuits (v9), one which uses the THS4509 with 4V/-1V supplies and the other which uses the THS4508 with a single 5V supply.

5282.THS4508 SE-Diff 7-1-2011.TSC

2450.THS4509 SE-Diff 7-1-2011.TSC

Kristoffer,

Thanks, that was a big help. I am now seeing the output levels and frequency response that I expect. I am seeing a noisy input signal, but I don't have any filtering on the ADC input, so I will need to add that back in. I am also seing a DC shift in the output that I don't see in simulating your design. I included the waveform from the FPGA. Let me know if you have any thoughts on correcting for that.

Anthony

Anthony,
Have you probed the ADC inputs with an oscilloscope and do you see the same ripples as shown in your ADC output data?

When you say you see a DC shift, is this something you measured on your board? How big a DC offset? Note that the original schematic you posted and the schematic in my TINA simulations differ in how the 50ohm source is terminated. The extra 50ohm resistor to ground on the non-driven input side in my schematic is used to balance the fully differential amplifier feedback networks since the 50ohm input is coming in only one side. In your schematic, you had a 50ohm resistor to ground on the same side as the signal source. Imbalance in the fully differential amplifier (FDA) feedback networks will manifest itself as a differential DC offset.

Kristoffer,

It turned out to be a loose cable. The circuit looks good now. I may need to increase the gain at some point. I if I adjust the gain that will affect the input common mode voltage correct? In the spec sheet for the 4509 the discuss using Rpu to set the input cm voltage and there is a table for each gain step. Can I use the Rpu instead of shifting the supply rails to -1 and +4. If I do change the gain and don't use Rpu, I assume I will have to adjust the supply rails for that gain?

Thanks,

Anthony