I am trying to re-design the TI Precision Design - Low Level (5uA) V-to-I Converter. My design goals are:

  • Increase bandwidth from around 1kHz to 100kHz (ideally 1MHz in the future)
  • Increase output range from 0-5uA to 0-50uA

To increase the output range I simply decreased Rset by a factor of 10 down to 10kOhm. To increase the bandwidth I needed to select a new op amp and instrumentation amp; while selecting these amplifiers I prioritized GBWP > input offset voltage > offset drift. My reason for this is the TI app note says that the key design specifications for the amplifiers are offset voltage and drift, but with my goal of increasing bandwidth I moved GBWP to the top.

I settled on replacing the OPA333 with the OPA277 and the INA326 with the INA118. Note, I switched from singled ended to +/- 15V power, this was necessary to find amplifiers with the bandwidth that I wanted. My schematic of the entire redesigned setup is at the very end of this post.

I was having troubles simulating with various SPICE tools (I couldn't for the life of me get the TI models to work in OrCad PSPICE and TI's TINA kept freezing on me) so I did a quick turn PCB with the circuit after getting some vaguely similar simulations running. What I found in the lab is that my circuit oscillates at around 25kHz and is just flat out unusable. I am positive that it is my compensation network, as I didn't change any of the values from the TI design so it makes perfect sense that my bandwidth isn't my desired 100kHz. I am struggling to understand how the compensation network values are picked and how I can design something that is stable. Can anyone point me in the right direction or offer some advice?

I should note I'll be attempting to drive this current in to a capacitive load (approximately 10uF after several feet of cabling). For the first level bench top testing I was attempting to drive the current in to a 50kOhm load (no large capacitance yet).

Precision V-to-I Converter

  • \$\begingroup\$ Go to the design document ti.com/lit/ug/slau507/slau507.pdf and read paragraph 2.5, then do some research. It is entirely possible that your components simply aren't fast enough and you'll have unacceptable loop stability with your desired bandwidth. \$\endgroup\$ – WhatRoughBeast Jun 14 '16 at 23:43
  • \$\begingroup\$ Do you need to have a grounded load for the current source? Do you need it to be bipolar? (Current in both directions.) You only see the oscillations when driving the big capacitance? Or always? I think when doing a composite amplifier you want the second amp to be faster than the first. \$\endgroup\$ – George Herold Jun 15 '16 at 14:57

To meet the criteria for an oscillator, you need to have positive feedback at a gain of one or greater. Your op amp as configured will act as an integrator at higher frequencies, so you have a 90-degree phase shift (in addition to the 180 degrees of the negative feedback) and with the instrument amp's higher gain at the high end, you have met the criterion for oscillation. The original TI circuit relies on the roll-off of the response of the instrumentation amp (providing lower gain at higher frequency) to stay away from this condition. If you went back to the original instrumentation amp with its R2C2 roll-off network, your circuit would likely be stable but with the lower, 1 Khz frequency response.

You may also want to consider that, because I = c dv/dt, your 50 uA into a 10 uF load will only provide an output voltage slew rate of 5 volts per second. At 1 Mhz or even 100 Khz, any change in current will be primarily charging the large capacitor at this very slow rate, so the voltage response at 1 Mhz will be a few microvolts.


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.