simulate this circuit – Schematic created using CircuitLab

Just bumped into a problem. When I am trying to make a voltage subtractor based on AD8055 op Amplifier, I get rail-to-rail oscillations/noise one the output.

The signal is 0.5sin + 1.5V at 1MHz. I wish to shift down the signal to 0.5sin + 0.5. Any ideas why I do not get a desired output?

Edit: enter image description here

  • \$\begingroup\$ The art is great, but you can also add schematics using the little schematic button above the text editing area, if you want. :) \$\endgroup\$
    – JYelton
    Mar 4 '14 at 21:32
  • \$\begingroup\$ There you go. Hope it will help to solve the problem \$\endgroup\$
    – Nazar
    Mar 4 '14 at 21:43
  • \$\begingroup\$ There was nothing wrong with the original cubist (or Bob Pease -esque) drawing. \$\endgroup\$ Mar 4 '14 at 21:59

This is a pretty fast op-amp. It's almost certainly due to layout or decoupling.

Do you have ceramic decoupling caps right at the supply inputs to a ground plane? Is this on a PCB with short traces and a tight layout or is it on a breadboard?

If it's the latter you might want to try a lower bandwidth op-amp. Since your differential gain is 1 but your noise gain is 2, you need an amplifer with a gain bandwidth of say 10MHz or so. (Assuming the 1MHz signal is the highest frequency you need to work with.)

This is a 300MHz amplifier, which is probably making your life more difficult than it needs to be.

  • \$\begingroup\$ 1. It's a breadboard. Trying to design it first. \$\endgroup\$
    – Nazar
    Mar 4 '14 at 21:45
  • \$\begingroup\$ 2. I did not use caps. Was hoping to get an approximate expected result before adding more components. \$\endgroup\$
    – Nazar
    Mar 4 '14 at 21:46
  • \$\begingroup\$ 3. The signal will be a 40MHz square wave, so I need a high bandwidth amp. Want to stick to this one. 4. I used the same op amp in the inverting gain -1 configurations with +terminal to ground and Rin=Rf=390Ohm. Did not have any problems. Did not use caps either (ofcourse I will use the later). \$\endgroup\$
    – Nazar
    Mar 4 '14 at 21:50
  • 1
    \$\begingroup\$ John! Holly cow. You were right. Decoupling caps fixed the problem. Thanks. \$\endgroup\$
    – Nazar
    Mar 4 '14 at 22:00

In all the application circuits for this device I never saw a value of feedback resistor or input resistor that was bigger than 1kohm. They even have a section on the data sheet dedicated to giving this information and again, they do not recommend a resistance greater than 1kohm - take a look it's on page 14 of the data sheet.

There are plenty of other things that can make this high-spec 300MHz op-amp sing and ring - decoupling capacitors - these are specified as being 10uF and 100nF on each power pin to gnd. It goes without saying that these need to be very close to the pins on the device.

It also goes without saying that this needs to be mounted on a carefully designed PCB for best results.

enter image description here

  • \$\begingroup\$ Andy, I used figure 35 page 11 inverting configuration test circuit. They have Rin=Rf=402 Ohms. Decoupling caps fixed the problem. I just did not think it would be such a BIG deal. \$\endgroup\$
    – Nazar
    Mar 4 '14 at 22:03
  • \$\begingroup\$ The signal is from CCD array which requires high input impedance buffer. I thought that configuration for negative gain -1 would work for me except I did not use 50 Ohm resistor at the input (see new figure I posted). I needed G=-1 to pre-condition my signal. Then, I wanted to further alter my signal by using subtractor to decrease the offset. \$\endgroup\$
    – Nazar
    Mar 4 '14 at 22:16
  • \$\begingroup\$ 10k feedback is likely to have other problems like poor bandwidth and maybe resonance effects. Try it with a signal generator or at least use a simulator that has the AD8055 model. \$\endgroup\$
    – Andy aka
    Mar 4 '14 at 22:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.