Author Topic: 2nd order sallen key LPF : stability issue  (Read 1636 times)

0 Members and 1 Guest are viewing this topic.

Offline sdoubleTopic starter

  • Frequent Contributor
  • **
  • Posts: 280
  • Country: fr
2nd order sallen key LPF : stability issue
« on: July 17, 2019, 05:51:01 am »
Hi all,
I am trying to design a 2nd order sallen-key Low pass filter. Bessel type. The 1st challenge s that I am requiring a cut off frequency of.. 150 Meg.
Second issue, I do need a  DC gain of 3. I have big issues with the design : the LPF is basically a very nice oscillator.
here is the design.

Any help would be highly appreciated.
 

Offline awallin

  • Frequent Contributor
  • **
  • Posts: 694
Re: 2nd order sallen key LPF : stability issue
« Reply #1 on: July 17, 2019, 09:55:19 am »
depending on the op-amp and your board-layout etc. you will have 10pF stray/input capacitances all over the place.

better to dimension it so that your caps are >1-10 nF so that stray capacitances don't matter so much?
 

Offline blackdog

  • Frequent Contributor
  • **
  • Posts: 765
  • Country: nl
  • Please stop pushing bullshit...
Re: 2nd order sallen key LPF : stability issue
« Reply #2 on: July 17, 2019, 10:52:42 am »
Hi sdouble,


What you want to do is very difficult with opamp because you need a lot of bandwidth, think of opamps with a few GHz bandwidth.
awallin already indicates that your parasitic capacity of your construction will cause major problems.

And then there's the next problem, what's the signal level you're working at.
Be aware that e.g. 10pF at 150MHz gives a load of 106 Ohm.

I think it's better to build a separate amplifier stage that drives a passive filter and then terminates the filter with the right impedance.
You can also take two amplifier stages where the filter is in between, so you have more freedom to choose the impedance on which the filter works.

Remember this too, a normal 1:10 probe of a scope will cause a high load when you measure for example at the opamp output because off the probetip capacity.

I hope this helps a little.

Kind regarts,
Bram
« Last Edit: July 17, 2019, 02:00:17 pm by blackdog »
Necessity is not an established fact, but an interpretation.
 

Offline MaxSimmonds

  • Contributor
  • Posts: 29
  • Country: gb
Re: 2nd order sallen key LPF : stability issue
« Reply #3 on: July 17, 2019, 11:24:50 am »
Regarding the DC gain, I looked into a sallen key filter before, low pass, and I can't exactly remember why, but if you look at the transfer function, if you introduce a DC gain greater than 2 or 3, you get a right hand plane pole, and it becomes unstable. I wouldn't go any more than 1.5, or you'll get a bit overshoot and possible instabilities.

I needed a gain of 15, so I just had a unity gain sallen key filter, and then a gain stage at the end - this was sufficient.

Hope it helps!
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: 2nd order sallen key LPF : stability issue
« Reply #4 on: July 17, 2019, 01:24:43 pm »
Be aware that e.g. 10pF at 150MHz gives a load of 106 Ohm.

Right. That's why passive LC filter with separate gain stage is much better choice at those frequencies.

 

Offline sdoubleTopic starter

  • Frequent Contributor
  • **
  • Posts: 280
  • Country: fr
Re: 2nd order sallen key LPF : stability issue
« Reply #5 on: July 17, 2019, 03:21:02 pm »
Hi guys,
thanks for the help.
Just a comment about the parasitics. I'm used to deal with very low parasitics. Basically, the routing is such that the parasitic cap is sub pF every where. very small packages (201 and 0402 only) thin and sub mm traces...
The AD8045 that I use is supposed to be usable for high gain sallen key LPF. this is clearly stated in the doc.
 

Offline blackdog

  • Frequent Contributor
  • **
  • Posts: 765
  • Country: nl
  • Please stop pushing bullshit...
Re: 2nd order sallen key LPF : stability issue
« Reply #6 on: July 17, 2019, 03:46:25 pm »
Hi sdouble,

The AD8045 that I use is supposed to be usable for high gain sallen key LPF. this is clearly stated in the doc

The above does not mean that you can build an active filter with a crossover frequency of 120 to 150MHZ.
Let's first look at Figures 48 and 49 in the datasheet.
These are the input and output impedance of the opamp from 100 to say 500MHz, this means no ideal opamp values!

And now for some horror values, look at figure 46 this is the suppression of interference signals on the power lines.

And finally the gain at capacitive load of the opamp output, see figure 10, 18pF load at a gain of 2 gives +5dB gain!  :-DD

I do not mean to say that this is not a good opamp and also that it would not be suitable for filter applications.
But the designer will have to take into account at least the properties that I have described above.

Kind regarts,
Bram

Necessity is not an established fact, but an interpretation.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf