Extrange behaviour on OPAMP

Sounds a bit like the input signal is pushing the OPAMP into a non linear part of its operation. The probes are changing the gain, probably by introducing some feedback, or compensating capacitance. A scope probe would be 30pf and 1M ohm X10 would be 30pf and 10M ohm.

--
Baron:

I think you are seeing the opamp bias current charging up the input node capacitance during a time when no multiplexer switches are connecting an input to it. Try adding a 1 meg resistor from the non inverting input to ground and see if the ramps disappear. If that works, you may have yo change your multiplexer drive algorithm so that there are no times when nothing is switched on. Think make before break, if you are driving more than one multiplexer switch.

Hi David,

"Saimazoom" schrieb im Newsbeitrag news: snipped-for-privacy@e3g2000cwe.googlegroups.com...

may be you forget the charge injection from your mux? the additional probe give this charge a path to flow, the OP not.

Marte

Your choice of an opamp and the circuit configuration is really one of the worst possible. PDs produce very little output current with low illumination and this opamp has already 160nA typ bias current(datasheet TPC4) and this changes at 4.2V to a positive value, when the complementary input transistor pair starts working. You will not measure your photocurrent unless it is very big. Take an OPA365, which has no x-over region and use it in a transimpedance configuration. The input capacitance will determine the noise floor and speed, so it is not good to multiplex there, but use 1 opamp per PD and switch the outputs.

--
ciao Ban
Apricale, Italy

Hi

Thanks to all for your help. Because of the tips I was wondering: There is some good reference (book,paper) that speaks clearly and good examples about all the characteristics of OPAMP ?.

I'll give all the ideas a try today, Thanks for your help =)

Ban ha escrito:

Hi

Is big enought to have masure, I've compared the results with the oscilloscope. The value of 160nA puts a floor on all my measures, but the signal is large enought so this is not really an issue.

This is not really a option for later use, because now I've just a few PD and is fine but later I'll need to work with a number greater than

100 PDs and submit all this info to the PC using a serial port. I'll order the OPAMP you suggest and give it a try, thanks =)

New opamps keep appearing on the market from so many companies that it is very hard to keep up a comparative list of all the important differences between them. But "The Art of Electronics" by Horowitz and Hill makes a valiant attempt to describe the various imperfections and limitations of opamps in a general discussion, and to catalog the best of class with respect to most of those imperfections. Unfortunately, that list is dated to the time the book was being written. A great boost to engineering in this area would be a web page that kept this list up to date. I spend a few hours a week surveying various manufacturers for the best opamp with respect to some limitation that I am trying to optimize in a particular circuit.

The imperfections that are most important to your application (I am guessing, here) might be input voltage range to the negative rail (it sounds like you never get anywhere near the positive rail, so a single supply design might be superior to a rail to rail compromise) offset voltage and bias current, more than voltage or current noise or bandwidth or output slew rate or output current limit or supply consumption.

Hello John,

But only if the table could be sorted by price. ONSemi is an excellent example on how to do that. Other than that it's Digikey. Unfortunately they don't allow listing by ascending price so one has to trudge through the pages. My mantra is, if Digikey doesn't have it then it might be a boutique part and I try my best to avoid a design-in. Keeps them purchasing folks happy.

--
Regards, Joerg

http://www.analogconsultants.com

You can try Googling these app notes

an001 an519 an9415 sboa067 an57 slyt174

: : I'll give all the ideas a try today, Thanks for your help =) : : : Ban ha escrito: : : > Saimazoom wrote: : > > Hi : > >

: > > I've a problem with a circuit that is puzzling me for a few days. The : > > circuit is a very basic one, I've a few photodetectors (BPW34), an : > > analog Multiplexer and a OPAMP (AD8519) conected as a Gain-Unity : > > buffer. : > >

: > > Everything seems to works fine: I can switch the photodetector I want : > > to meausure in the Multiplexer and the output of the OPAMP is the real : > > one. But ... very often I can measure a very big peak of voltage at : > > the output of the OPAMP, the range of voltage of the signals from the : > > detectors is around 100mV but this peaks are 4V (I'm using 5V as Vcc) : > > so it really mess everything when this happens. : > >

: > > I've discarded every other circuits just unplugging them, I've also : > > changed the OPAMP for a new one (but the same model) but this : > > behavious persist. With one probe at the non-inverting input I cannot : > > see any peak coming to the OPAMP, but when measuring with the probe : > > at the output this peaks appears. The shape of the peaks are a very : > > linear ramp and then it goes down very fast like the discharge of a : > > tiny capacitor, so is not really a 'peak'. : > >

: > > Something very funny happened when I used two probes: I attached one : > > probe at the output so I was watching this 'peaks' at the : > > oscilloscope, and just to be sure I attached the second probe of the : > > Oscilloscope to the non-inverting input of the OPAMP and ... the : > > peaks dissapeared and I had my signal as I wanted. If I unplug the : > > probe from the input the peaks appears again, and If I change the : > > probe to x10 the peaks also appears but with less amplitude. : > >

: > > So... What's happening ?, Do I have a pretty oscillator ?, Do I need : > > to compensate the OPAMP ?, if this is true how to do it properly ?, : > > I've tried different capacitors at the input and the output of the : > > OPAMP (220K) and it seems to lessen this behaviour but I'm afraid of : > > losing signals I can need later because of too much filtering. : > >

: > > Any help will be apprecitated. Thanks for your help =) : >

: > Your choice of an opamp and the circuit configuration is really one of the : > worst possible. PDs produce very little output current with low illumination : > and this opamp has already 160nA typ bias current(datasheet TPC4) and this : > changes at 4.2V to a positive value, when the complementary input transistor : > pair starts working. You will not measure your photocurrent unless it is : > very big. : > Take an OPA365, which has no x-over region and use it in a transimpedance : > configuration. The input capacitance will determine the noise floor and : > speed, so it is not good to multiplex there, but use 1 opamp per PD and : > switch the outputs. : >

: > -- : > ciao Ban : > Apricale, Italy :

Well, since the bias current varies and can also be 300nA, it will certainly contribute an important error term.

you really want to multiplex over 100 PD with a single opamp buffer? And you are measuring the current or voltage?

--
ciao Ban
Apricale, Italy

Hi

Yep, you are right. But is a more or less constant that I can substract later in the computer.

I'll use as much as three I think. So I can divide the work in different boards.

Voltage, I'm using a resistor in series with the PDs (1MOhm). Really the limit in the speed is great but I'm in the range of ms, so it seems to be working until now.

Thanks all for your help, really all the references and tips are very welcome. I've changed the way I accessed the Multiplexor and used also a 1MOhm from the non-inverting input of the OPAMP to ground and the peaks are still there, but they are not of so hight valuet.

Un saludo.