Op Amps… G=1 stable & decompensated

RE: Op Amps… G=1 stable & decompensated

Bonnie Baker - WEBENCH Design Center — Thu, 25 Apr 2013 20:51:02 GMT

Mengchi Chen ,

About using uncompenstated op amps with current out DACs, I took some time to write an article about that a few years ago (EDN, 2007, July, "Transimpedance strikes again", www.edn.com/.../Transimpedance-strikes-again-Current-to-voltage-conversion-with-MDACs). In that article I summarized that actually your noise gain, that deterimined stablity, is dependant on your feedback capacitance and the output capacitance of your DAC. This is very hard to balance, as the output capacitance and resistance of a DAC changes, dependant on the DAC input code (or output analog current source). Please read the article.

RE: Op Amps… G=1 stable & decompensated

Mengchi Chen — Thu, 18 Apr 2013 05:54:36 GMT

Hi Bruce,

Thanks again, now I figure out that :)

RE: Op Amps… G=1 stable & decompensated

Bruce Trump — Thu, 18 Apr 2013 05:41:10 GMT

Mengchi-- Sorry for misspelling your name. I have old eyes.

Imagine a transimpedance amplifier. R2 is the feedback resistor. R1 is the impedance of the current source driving the summing junction. The impedance of a current source is infinite. The gain if you were driving the non-inverting input is 1+ R2/R1 where R1 is infinite. -- Bruce

RE: Op Amps… G=1 stable & decompensated

Mengchi Chen — Thu, 18 Apr 2013 05:25:31 GMT

Hi Bruce,

Thanks a lot for your clear explanation.

But I want to ask why is the noise gain of a current source driven transimpedance amplifier unity?

Is that because the op amp internal noise source will see the current source's impedance as the input impedance of this closed loop configuration?

RE: Op Amps… G=1 stable & decompensated

Bruce Trump — Wed, 17 Apr 2013 22:27:34 GMT

Mangchi-- Just to clarify terminology, the OPA637 is a decompensated amplifier. An uncompensated amplifier would require external components for stable operation in virtually any common configuration.

The noise gain of a transimpedance amp when driven from a pure current source is unity (beta=1). Feedback makes the impedance at the summing junction low, but this does not mean that beta is zero. In fact, it is the strong feedback that causes the impedance to be low.

A typical DAC does not have an infinite output impedance. It's not really a current source. In fact, its output impedance is code-dependant and this creates a code-dependent noise gain. Furthermore there may be a significant output capacitance which affects the high-frequency noise gain and therefore the stability. All to say, it's a more complicated situation and depends on the specifics of the DAC. I suggest that if you have further questions you should post on our E2E forums as this will be highly product-dependent. -- Bruce

RE: Op Amps… G=1 stable & decompensated

Mengchi Chen — Wed, 17 Apr 2013 20:12:21 GMT

Hi Bruce,

I have a question: How to examine the stability of uncompensated op amps employed in DAC-buffering I/V converters? Take the OPA637 for example, it will be stable when its noise gain is greater than 5, but when it comes to noise gain, it represents a "voltage gain", and transimpedance amplifiers do not have voltage gains, so what's the condition makes these kind of configurations stable?

I have some ideas: G>5 means β<1/5. From the voltage point of view, the input impedance of this inverting configuration is 0, so β=0 thus the op amp will be always stable. From the transimpedance point of view, the β of a transimpedance amplifier is 1/(feedback impedance) so to make op amps stable we should use feedback resistance greater than 5 Ohm. Which is correct?

RE: Op Amps… G=1 stable & decompensated

Alex Striganov — Mon, 15 Apr 2013 23:10:34 GMT

Thanks for your observations, Bruce, they are quite useful. I can confirm them for OPA637 / OPA627.