Frontend OP amp test

[mickstv]

I have a few different OP amps that I've been testing in my latest detector design. I was using the normal NE5532 dual package, but have changed to a LM4562NA some cro pictures below showing the difference at the output with my test frontend.

The same circuit was used for both tests, I'll post the exact opamp config later once I get diptrace installed again.

NE5532






LM4562

[ODM]

The reduced settle time is also worthwhile, or is the ringing just an one off anomaly? With that ringing it seems the 5534 only settles after 12..14us, while the 4562 is nearly instantaneous with some flattening.

Is the 5534 close to oscillation with the same feedback RC parts?

[mickstv]

Sorry I made a mistake with the first picture it was actually a NE5532.


An extra capacitor in the feedback circuit gets rid of that ringing and becomes a nice curve, but your right in either case the decay to flat is around 14us. I know which one I'm using from now on.

[6666]

Hi Mick
thats interesting, they are cheap on fleabay to.

[mickstv]

I've just had a closer look at the nV/ Hz and it's not very good. It's actually as bad as the NE5532 at the same Hz. I have another op amp in my collection which I'll have a look at and post up the results.

[mickstv]

The top picture is the pick of them. The IC also has the lowest overall noise level, after looking at the datasheets.

[Aziz]

Hi Mick,

try the NJM2068 (NE5532 compatible). The Japanese dual op-amp is very cheap and is running very quiet too.
Aziz

[mikebg]

People, thermal noise is not important even at wideband (-: BBS, FBS, FBS2 :-) metal detectors. Thermal noise is
neglegible relative to AIR&GND signal in input of preamp, which is modulated by movement of search head. Synchronous demodulator suppresses thermal noise because operates with principle of lockin amplifier, but it is very sensitive to the modulation.
If you have a metal detector with low noise chip in input, make an experiment - change the chip with a more noisy to see if there is remarkable change in sensitivity.

[mickstv]

Hi Mick,

try the NJM2068 (NE5532 compatible). The Japanese dual op-amp is very cheap and is running very quiet too.
Aziz

Hi Aziz, I can't find much in the way of spec's for that IC. I have found the PDF but I'm not seeing any details about nV/Hz.


The OPA2822 has some good spec's and a reasonable price $3.70AU from Element14.

http://www.ti.com/lit/ds/symlink/opa2822.pdf

[mickstv]

People, thermal noise is not important even at wideband (-: BBS, FBS, FBS2 :-) metal detectors. Thermal noise is
neglegible relative to AIR&GND signal in input of preamp, which is modulated by movement of search head. Synchronous demodulator suppresses thermal noise because operates with principle of lockin amplifier, but it is very sensitive to the modulation.
If you have a metal detector with low noise chip in input, make an experiment - change the chip with a more noisy to see if there is remarkable change in sensitivity.

Why do the big players in the industry use low noise electronics IC's etc, if internally generated noise isn't an issue ?

[Davor]

Late samples?

I'd say every internally compensated op amp will have troubles with settling time thank to Mr. Miller. There are some better compensation options, but not widely available, and perhaps not a real solution. Perhaps the best way to tackle the PI input would be a discrete one. Using low Rbb BJTs would provide low noise too, and they'd perform deeply in class A. These would also have superior transient performance. Guess the classic paper by Otala "Transient Distortion in Feedback Amplifiers" is quite informative:
"The usual lag compensation techniques for obtaining stable feedback amplifiers are inadvisable. By reducing the upper cutoff frequency of the power amplifier this increases transient distortion."

Too much gain in a feedback amplifier is also bad for transient response. Simple differential pair can have both sufficient gain and high cutoff. And comes with superior noise too. It might require a bit tweaking and component selection hence it is not the best choice for mass production ... errr ... what mass production

[Tinkerer]

Sorry I made a mistake with the first picture it was actually a NE5532.


An extra capacitor in the feedback circuit gets rid of that ringing and becomes a nice curve, but your right in either case the decay to flat is around 14us. I know which one I'm using from now on.

The capacitor in the feedback also helps seeing targets that have a TC that would otherwise be too short to be seen at this delay.

Tinkerer

[Ferric Toes]

I find that it is the noise at low frequencies that is important. For example the 5534A rises sharply below 1kHz and is 10nV/rtHz at 10Hz. By comparison the LME49990 is only 1.5nV/rtHz at 10Hz and 3nV/rtHz at 1Hz. Always check the noise graphs rather than look at the manufacturer's best figure.

Often the manufacturer has as an A version which is lower noise. e.g. the 5534A is lower than a 5534.

With the 5532 tests above, is this being used as a two stage preamp? This is because a wider bandwidth and faster settling can be had by sharing the gain over two stages. I never use a single 5534 now as the gain does not recover as fast as the recovery curve as shown on the scope traces would lead you to believe. Sum in a 10kHz signal and you will see this.

Eric.

[Orbit]

LF 357 is often used What do you think about it ?

[Ferric Toes]

LF 357 is often used What do you think about it ?

Looks very noisy below 100Hz.

Eric.