Still working through the TI Handbook and they sure need an editor. Was given this and something was obviously wrong with the LO cutoff freq being higher than the HI cutoff.

Breadboarded and spent a few days working on it. Tried a few different OPAs and ended up moving the grounding caps to the ±V pins instead of having them from the breadboard power rails to gnd rails. Did the calculations using actual component values.


With UA741 Odd little hump...


With TL071


Both show noise and then a LO corner ~700 - 1kHz and pretty flat up to where the sig gen cuts off @ 20MHz. So why am I not seeing any LO/HI corners anywhere near the calculated values?

 

What supply voltages to the 741?

 

±15V for all that I worked with. Took a bit of headscratching to get rid of the noise. Some was internal to the scope from leaving some inputs open that I was unaware of and some was the position of the grounding caps. But all tests were with ±15V.

 

hi Sam.
Have you tried the test using a more 'modern' type OPA. the 741 is a little buggy/
E
Update:
For reference, this is what LTS shows.

 

A wider range of bypass caps might help too. (A mix of ceramic AND electrolytic just to be safe.)

 

A Canadian I know would point out that the LM741 has a low slew rate, so when talking about 16 kHz you need to keep the signal amplitude very low.

The open loop gain isn't quite enough to get that 40 db at 16 kHz.

 

I just threw the 741 in for comparison but was a bit surprised at the strange "bump" in the Bode plot. The LTS plot sheds some light on that as the center frequency seems to match pretty well with the "bump". I did consider adding electrolytic grounding caps but the noise was pretty well filtered out and no oscillation problems. Without the Ri resistor the differentiator output is useless due to noise as stated in the handbook. Primarily working with the TL071 which is supposed to have some internal noise abatement. But even on the LTS plot I am not seeing any HI & LO freq cutoff corners? I dug into the operating manual a bit looking at the Bode Plot analysis function but all I find there is a basic description of what the various settable parameters are and nothing on how it actually functions. What I did learn by doing is that leaving the unused trace visibility parameters ON (which unfortunately is the default) when they are not actually used and connected to an input can induce noise on the Bode Plot.

 

Look at the massive 60Hz that is picked up by your "Mickey Mouse" breadboard.
Then calculate the frequencies caused by the 100k resistor and the capacitances between the rows of contacts and wires all over the breadboard.

Your circuit is a highpass filter that passes high frequencies.
The cutoff frequency or slew rate of the opamp with breadboard capacitances is a lowpass filter.

Your simulation has and shows a voltage gain of 100 and an input peak signal of 1V. Then the opamp is clipping like crazy.

 

Your simulation has and shows a voltage gain of 100 and an input peak signal of 1V. Then the opamp is clipping like crazy.

I agree it certainly should be, but it isn't... Something is a bit strange with this Bode Plot Analysis. I did add a 10MΩ load resistor to GND plus a 1nF cap across Ro to increase the quieting effect. Also showing Vout instead of Gain on these for comparison. For example. NOTE: using TL071CP

1Vpp and I forgot to turn the power supplies on!


Also @ 0.1Vpp


Then I turned on the ±15V power supplies. @1Vpp No 100X gain and very similar to the unpowered curve.


And @ 0.1Vpp powered. Also very similar to the unpowered curve?

So I'm still doing a lot of head scratching...

Here it is with Vi=1Vpp @ 5kHz and, Yes, I have double-checked the Ri=1k and Ro=100k, No Clipping?

 

You are calling for 20dB gain which is just not there above 100kHz.

 

What I am expecting to see is a LO cutoff corner ~600Hz and a HI cutoff corner ~1.6kHz according to the calculations. I'm not. I am seeing this ~600Hz... Which looks like noise.

and then ~80kHz (not 1.6kHz)... Although it is a "corner" apparently.

 

Look at the massive 60Hz that is picked up by your "Mickey Mouse" breadboard.

 

You wrongly said that Ci and Ro (feedback resistor) make a filter but they do not.

You said that you added a 1nF capacitor across the 100k negative feedback resistor.
Then it creates the only lowpass filter you have with a cutoff frequency of 1590Hz.
But your input 1k and 0.1uF highpass filter also has a cutoff frequency of 1590Hz.

 

I didn't say it, Burr Brown and then the editors at Texas Instruments said it. This is all from


This is where I started from in post #1 trying to replicate the HI and LO frequency cutoff corners. Which did not happen and thus my post. Note the errors in their given calculations that I appended the correct values for.

I achieved the high frequency noise reduction but there was no High Frequency Cutoff "corner" @ 1.59kHz. Instead, it was essential a flat line up to the point where the signal generator sweep ended @20MHz

And yes, the addition of Rf parallel to Cf gives Ci(Ri)=Cf(Rf) for low noise and drift compensation.

But my testing does not give me any "Double high frequency cutoff" results...

 

The Texas Instruments book makes no sense.
I simulated the double filters:

 

And that is NOT what I am getting... I'll go ahead and build a test module.

 

My apologies! this is a wonderful thread. Seriously!

 

Notice that my simulation has an input of 1V peak and an output of 50V peak and needs the opamp powered from 104V!
Since you use a +/- 15V power supply then the maximum output must be about 6.6V peak then the input must be 6.5V/100=0.065V peak.

The signal generator output impedance in my simulation is zero ohms which is much less than the 1k of R1. What output impedance is your generator?

 

50 or 1MΩ for the scope's internal AWG. Haven't worked on tying the scope to an external source as yet.

 

input must be 6.5V/100=0.065V peak.

Already planned on 0.1Vpp... Not sure how long this will take as doctor visit today and HVAC being stripped and replaced the next couple of days. Hopefully I can at least get a start on it later today...