silly circuits

Silly how? The first one has two RC poles and two complex ones--it trades off sharpness vs. lower op amp count and improved ultimate rejection.

Seems like the second could be improved a bit further by using an LM4041-adj and an extra resistor.

Cheers

Phil Hobbs

The three-pole single amp filter is common; there are lots of papers. I hadn't seen a 4-pole (google shows nothing) but it should work.

Yes. This one improves my regulation by about 3:1.

One goal of both circuits is to use only 1K and 10K resistor packs.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

On a sunny day (Tue, 24 Oct 2017 02:40:56 -0700) it happened John Larkin wrote in :

Now he tells us ...

I use a LM2596 3.3 V version, no resistors needed. It is more green. Yours need caps too, mine needs a diode and inductor.. My batteries last longer.

Why not a second opamp on the first filter circuit?

George H.

They are expensive, and the inductor and diode add more. But I want a quiet stable supply, and switchers are noisy.

I'm getting my +12 from a TPS54302 synchronous switcher, which is a great little part. It's tiny, quiet, and doesn't need a diode. Check it out.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Then you got one fewer opamp!

People keep too many resistor values in stock; I see these circuits sometimes with a dozen different values that would function materially indistinguishably if you replaced 2/3rds of them with 10k

Japanese manufacturers seemed guilty of this a lot; it seems like in the

1980s they took pride in making sure to use every resistor value in a set somewhere in a circuit

The opposite challenge is to do as much as possible with a small number of quad resistor packs. We do that a lot lately, to keep pick-and-place feeder counts down, and because it's a math puzzle.

Both single resistors and quad packs are so cheap that parts cost doesn't matter, both under one cent.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

The extra passive poles are all _real_. You need an OpAmp to maintain Butterworth or whatever. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| STV, Queen Creek, AZ 85142    Skype: skypeanalog |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 

                   Spice is like a sports car...  
     Performance only as good as the person behind the wheel.

On a sunny day (Tue, 24 Oct 2017 08:37:13 -0700) it happened John Larkin wrote in :

Yes, about 1 $ from gearbest for a whole module:

There are several fixed voltage versions and a variable version, I bought a bunch of 3.3 V those can be set to any higher voltage with a series resistor. Interesting is that that series resistor is not the same from one chip to the other... was probably laser-trimmed to 3.3 V. Almost got me.

Nice part, high switching frequency, spread-spectrum, about 2 $ for the chip for 10:

needs an extra boost-capacitor I see. Efficiency is very very good!

On a sunny day (Tue, 24 Oct 2017 11:43:52 -0400) it happened bitrex wrote in :

I do that deliberately, once I threaten to run out on the 10k SMD strip I start on the 8k2, 12k, 15k just keep the ratios OK. Else you get into trouble when you really need that 10k 1% for some project. This because I buy whole series SMDs, those come same number for each value.

What's interesting is that a small number of classic transfer functions are almost always used to design filters: Butterworth, Chebychev, Bessel, Cauer. I guess that's because the pole-zero locations are already worked out and blessed. But there is a world of other filters without pat mathematics, but no obvious path to synthesis. It's sort of fiddle (manually or with a computer) until you get something that looks good. That gets seriously hard with LC filters. I wonder if there's software to do that; Spice is clumsy.

The single-opamp 3-pole is often mentioned with a hint of distain, because it's not a perfect match to a classic form.

I've got a nice 4-pole lowpass with one opamp, two r-packs, and two values of capacitor.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

I'm not in disagreement about the filter's usefulness... I do that sort of thing regularly.

But I almost never use the Sallen-Key configuration on-chip because it's sensitive to sloppy passive tolerances. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| STV, Queen Creek, AZ 85142    Skype: skypeanalog |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 

                   Spice is like a sports car...  
     Performance only as good as the person behind the wheel.

Including the implicit increasing Zo as open loop gain falls off with frequency.

The topology just kind of sucks overall. For discrete op amp filters multiple feedback is the go-to, biquad if you have luxury. Sallen-Key if you only have tubes

The classic transfer functions are popular because you can mathematically verify that for any given filter cutoff freq and order generally at least one characteristic will be the best it can possibly be for that topology of filter, e.g. the Bessel filter has a maximally flat group delay at omega = 0. Of course to get that guarantee you often end up having to compromise other parameters more than you might like.

Yes. It can get by on a small inductor, too. But it gets pretty warm at 3 amps; I wouldn't use it past 2.2 maybe.

The spread-spectrum is radical, but it doesn't seem to add noise on the output. They must have shaped it right.

The soft-start is nice, and you can use the enable input as a low-voltage lockout, or to sequence supplies.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

I have a single supply, and I want a gain of +1.00. One virtue of S-K is that it has exactly a gain of 1.0 at low frequencies, irregardless of resistor tolerances or tempcos. That matters when you are making precision measurements.

We only use 1% or better resistors, even in packs. A modest-Q S-K works fine with 1% resistors and 2 to 5% caps.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

One problem with S-K is high frequency shoot-through, from input through the first cap into the opamp output. An odd-order filter can start with an R-C that whacks the fast stuff first. Two RCs are even better!

Or want DC accuracy.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com