AD8138

Hi Rabeeh,

The AD8138, with matched feedback networks, amplifies the difference between its inputs whether they are balanced or not as long as everything stays linear.  A common case-in-point is when one input is DC and the other is varying.  Can you post a schematic of your amplifier circuit?

Best regards.

--Jonathan

Hello Jonathan,

Thanks for helping me on this problem. I use the ADC8138 in two different places on my board. one is for making differential input to the ADC and the other place is for turning on and off a differential MOS devices inside the ADC. I have attached the two schematics in here.

The input to the ADc should be a signal that goes from 0.2V to 0.5V with a common mode voltage of 0.35V. Since the AD8138 needs a common mode of at least 1V, I use a voltage divider at the output of AD8138 to make the common mode 0.35 and in fact I am using POT to have this ability to change the common mode.

The other signal has to be a signal going from 0 to 1.8V. how ever if the common mode of signal be a 1V instead of 0.9V, it does not make nay problem for my design. I actually use a jumper on my PCB board so I have this ability to supply the AD8138 with a -5V instead of ground for better performance in the case of 0-1.8V . However if I use 5/-5 supplies instead of +5/GND as a supply the IC gets too hot.

Thanks,

Rabeeh Majidi

PHD Candidate of Electrical Engineering

Worcester Polytechnic Institute

Worcester, MA  01609    (USA)

Tel:  Work:508-831-5644

          Cell: 408-888-8474

Hi Rabeeh,

Thanks for the additional information and schematics.  The problems you are seeing with output swing are due to exceeding the lower input common-mode and output swing limits on a single-supply.  Please refer to the single +5V Specification Table on pages 5 and 6 of the AD8138 data sheet for these limits.

It's important to note that with DC-coupled feedback networks the output common-mode level is divided down at the amplifier inputs, placing the input common-mode level lower than the output common-mode level.  Additionally, in single-ended-to-differential applications part of the output appears on the input common-mode as ripple.  This ripple causes the input common-mode level to get even closer to its lower limit.  If your 50 ohm input is AC-coupled, you will need to AC-couple the lower loop (put the same value C in series with Rg as is in the upper loop).  The two feedback networks must be identical in order to avoid unwanted differential offsets in the output.

The simplest solution is to use a negative supply.  You indicated that the part gets too hot with +/-5V supplies, but this should not be the case -- going from a quiescent power dissipation of about 100 mW to 200 mW @ RT should not present a thermal problem.  There may be something else causing the part to get too hot.

I've recorded a couple of webinars on the topic of driving ADCs with differential amplifiers that cover this situation in more detail.  If you're interested, they are available at:

http://www.analog.com/en/content/WC_DRIVING_ADCS1/webcast.html

http://www.analog.com/en/content/WC_DRIVEADCS_2/webcast.html

I wish you well in your Ph.D pursuits.  I'm a WPI alum (even enrolled in the Ph.D program, but didn't continue due to job commitments).  Please give my regards to Kaveh Pahlavan when you see him.

Best regards.

--Jonathan

Hello Jonathan,

Thanks for your email. I would try the negative supply then.

Thanks for your good wishes. I would let Prof Pahlavan knows about our email conversation.

Best regards.

Rabeeh Majidi