Post by thread:[PICLIST] [EE] OpAmp magic

Hi all! Would anybody care to help a newbie out on OpAmp basics? I'm working with some pressure sensors, one with static pressure for altitude measurement and one differential for airspeed. I've searched the net for basic opamp designs and found some useful pages. I've set a span for the pressure ranges I'm interested in which is quite less than the full dynamic range of the sensor. So I'd like to amplify the signals to maximize ADC resolution over my range. The amplification for the altitude sensor is inverting, which is good since I get higher readings for higher altitudes with less pressure. I worked from the description at http://www.chipcenter.com/embapps/emba004.htm and now have a circuit which produces the right values when simulated in Protel. I do have trouble with the non-inverting one, though. There's a good example at http://www.tfs.net/~petek/rockets/altimeter/circuit.html, but I cannot figure out how to go from the first circuit, with two opamps, to the second with only one. I want to map a span of 0,2V - 0,439026V to the ADC's 0-5V, non-inverting. And there might be issues with noise etc since the interesting range is so narrow. Of course, I could do it inverted and handle that in software, but shouldn't this be a really basic thing?? I'd be happy to provide anyone with my calculations, circuits etc if anyone's interested. Regards, Thomas -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

> There's a good example at > http://www.tfs.net/~petek/rockets/altimeter/circuit.html, but I cannot > figure out how to go from the first circuit, with two opamps, to the > second with only one. I want to map a span of 0,2V - 0,439026V to the > ADC's 0-5V, non-inverting. And there might be issues with noise etc > since the interesting range is so narrow. > Of course, I could do it inverted and handle that in software, but > shouldn't this be a really basic thing?? Ok, frankly, you're going to need a pair of potentiometers and a DMM to do this and get the numbers that exact. Ok, the opamp needs to be biased at the average of the 2 extremes (0.319513V) and it needs to have a gain of 5V/(Vupper-Vlower) which is 20.9182264691. So, here comes the ASCII art: +5V | | | | |\ Vin O-- | ----------|+\ > | \ < | >---+----O Vout R1 > | / | < +----|-/ | > | |/ | < | | | | | +------+--/\/\/\/---+ | R3 > < > R2 < > < | ----- / / / (In case you havent seen ASCII art used before, copy this out of your email program if it looks like garbage, and paste it into something that uses a mono-spaced font like Notepad) Ok, R1 and R2 can be a single potentiometer with the wiper connected to the -ve input of the opamp, but if they are, then R3 should also be a potentiometer. So, to get the values of resistors required, here's what I did: Av = R3*(1/R1 + 1/R2) + 1 Vbias = 0.319513 = 5 * R2 / (R1 + R2) Solving the Vbias equation for R1, R1 = R2(5-0.319513)-0.319513 Subbing into the Av equation, Av = R3*(1 / (4.680487*R2 - 0.319513) + 1/R2) + 1 or Av-1 = R3*(1 / (4.680487*R2 - 0.319513) + 1/R2) Now, at this point, I used a graphing calculator to choose my values, as it makes the formula easily reusable. Since there is no third equation, we have to pick an arbitrary value I picked 180k for R3, because it gave 10k for R2, and that gives roughly 50k for R1. The high resistance is better for low power, to reduce the power loss in the divider, and since the opamp should have low output impedance, anyways, The actual values I calculated are the following: R3 = 180k R2 = 10.9678404701 kohms R1 = 51.3316396084 kohms If you use 2 pots, you'll need to do some careful calibrating. Configure the R1/R2 pot first, to get the correct bias voltage, then configure the R3 pot to get the right gain. Oh, and make sure you use a rail-to-rail opamp, or you'll get some clipping. Hope this helps. Regards, Brendan -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Brendan Moran wrote: > > > second with only one. I want to map a span of 0,2V - 0,439026V to the > > ADC's 0-5V, non-inverting. And there might be issues with noise etc > > since the interesting range is so narrow. > > Ok, frankly, you're going to need a pair of potentiometers and a DMM to do > this and get the numbers that exact. Uh, yeah and they need a ratio tempco of 1ppm/(temp range of operation) and the opamp needs a better than 1uV input offset and ... Wouldn't this be easier with a 40-bit ADC glued to a 40-bit DAC by a DSP chip to handle the offset and gain calculations ;-) Dave -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

David Minkler wrote: > Uh, yeah and they need a ratio tempco of 1ppm/(temp range of operation) > and the opamp needs a better than 1uV input offset and ... > > Wouldn't this be easier with a 40-bit ADC glued to a 40-bit DAC by a DSP > chip to handle the offset and gain calculations ;-) Well, I don't know about that... you'd have to write the software for the DSP If you ask me, which I know you didn't ;) then, since he's spending most of the gain of the opamp (usually has at least 250000 or 107.96 dB) then what you have left is pretty darn stable. Now, that said, the values he's asking for are unatainable. 0,2V - 0,439026V isn't going to happen. I'd aim for a range of 0.2V to 0.44V which is a much more likely target. As to opamp selection, my suggestion is an opamp with MOSFET inputs, and rail-to-rail outputs. Most devices simply don't have the precision to handle that number of decimal places, but then, I expect that that was already known. Good luck. --Brendan -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Hi Thomas, What is the average DC value of your Vsin input? Since you have a divider on the reference voltage (the 5V), it looks like you have an effective reference of 0.8197 V ( 10/(10+51) * 5). If your Vsin input isn't centered at this, then the amp will be amplifying the difference in DC offsets, too. Since your output amplitude is so small, I'd guess that the amp is saturated, that is, the difference between the input DC offset and 0.8197 V, when multiplied by the gain of the circuit, is outside the supply rails and so the op-amp has it's output at ground(with a very slight output voltage variation). One of the deceptive things about spice simulations is that they show results that are too clean. You might look at these results and say "hmmm, the output is right but the gain is too low", when in real life, compared to the noise levels, you wouldn't even see a 30 microvolt peak-to-peak output. Sean At 10:14 AM 7/1/2002 +0200, you wrote: {Quote hidden}>Thanks for all the help on the OpAmp circuit. I understand that the >levels in my question was a *little* too precise, but they where ideal >theoretical levels and I'm trying to get to a decent compromise between >ADC resolution and complexity of the circuit. > >I dont have any luch with the suggestion below though. I've tried >simulating the circuit below and don't end up with the nearly the right >values. I've attached (hope this doesn't violate any list policy) >pictures of my circuit and the resulting simulation. Can anybody figure >out what's wrong? > >Regards, > >Thomas > > > {Original Message removed}

Thomas <spam_OUTthomasTakeThisOuTBLUEICERESEARCH.COM> wrote: > Thanks for all the help on the OpAmp circuit. I understand that the > levels in my question was a *little* too precise, but they where ideal > theoretical levels and I'm trying to get to a decent compromise between > ADC resolution and complexity of the circuit. > I dont have any luch with the suggestion below though. I've tried > simulating the circuit below and don't end up with the nearly the right > values. I've attached (hope this doesn't violate any list policy) > pictures of my circuit and the resulting simulation. Can anybody figure > out what's wrong? I didn't try to work out Brendan's equations, but his results are not correct. You need a gain of about 20, and a Thevenin voltage for your bias circuit of about 0.21V. The 51K/10K voltage divider is providing a source voltage of 0.82V, which is way too much. The simulation is saturated at the negative (ground) rail, but the simulator is accurate enough to show the tiny amount of signal that makes it through the opamp anyway. Try R3=3.3K, R1=75K, and R2 = 62K (these are 5% values) If you have 1% resistors, try R3=1.02K, R1=23.2K and R2=19.2K. -- Dave Tweed -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads