I've built a Pease floating inductor circuit to use as a low pass filter instead of a large inductor. The circuit simulates fine, but when I input a 12V signal at Vin, I only get a tiny voltage on the other side of the input resistor (R20), maybe 200mV. I've combed the circuit for shorts and every node shows a high rising resistance. The only thing I can think of is perhaps the outputs of the opamps look like ground/low impedance and therefore we would effectively have a resistive divider of 3.32k / 100R.

The choice of the OPA2991 was because I first used NE5532s (all I had on hand) and fried them immediately with Bob's original values (1R, 100u, 1M). My guess is that's due to overcurrent between the inputs with the 5532's large offset voltage. The 2991 is something like 100uV Voffset, plus it's rail-to-rail, single supply and high voltage which all are required for this circuit. But the 2991 also has a weird input and output stage so I wonder if that might not be the issue.

Any ideas or things to pay attention to to get this circuit to work?

 

Link to the original article/circuit?

 

All the info I have found can be found by Googling "Pease Floating Inductor" but the very original source is this video link.

 

What are your power supply voltages?
Does your input signal have a DC path to ground?

 

What are your power supply voltages?
Does your input signal have a DC path to ground?

Supply is +30V to GND. Vin is an opamp buffer, so no it doesn't. Interestingly for some reason I thought the 1M resistors provided a DC path to ground but nope, they are both blocked by the caps. I'll try that and get back!

 

It will need a split supply to work: +/- 15V

 

It will need a split supply to work: +/- 15V

Bummer. Why's that?

 

Bummer. Why's that?

Because it works with AC signals.

 

Because it works with AC signals.

So even if we only want a fullwave pulsed AC input (from 0 to 20V or so), just like what we would see in a linear power supply before the filter cap? Is there any good alternative for a floating inductor that would work for single supplies?

I tried a 100k on the input and that did seem to pull a bit more AC information out but remained very low.

 

What is your application?

 

It's an audio sidechain, so it will receive audio that has been rectified positive about ground (0 to 10 or 20V, enough to overcome any losses on the way through). I want to see what the LC ringing does to the sound. A simple inductor would be ideal here but I don't know where to get a 1-10H inductor for a few milliamps of current.

 

Are you thinking of an LC circuit as below?
If so it is just a second order filter with f=1/(2PI * SQR(LC)) and Damping Factor =(1/2R)*SQR(L/C).
You can create the same response with a Sallen and Key filter simply by plugging the frequency and damping factor in to a filter designer.
http://sim.okawa-denshi.jp/en/OPseikiLowkeisan.htm

Mouser has inductors in the 1H to 10H region
https://www.mouser.co.uk/Passive-Co...tors/_/N-wpcz?Rl=wpczZer7uZ1z0wjb0Z1z0wgc8SGT

 

You can create the same response with a Sallen and Key filter simply by plugging the frequency and damping factor in to a filter designer.

Will it ring the same way though? When the LC or Pease circuit is sim'd, it will oscillate a bit on changes...that's the key thing I'm looking for.

Mouser has inductors in the 1H to 10H region
https://www.mouser.co.uk/Passive-Co...tors/_/N-wpcz?Rl=wpczZer7uZ1z0wjb0Z1z0wgc8SGT

Well, I know about those 1-10H inductors, but those suckers are all so big and pricey. I recall it was possible to get high-value inductors for small-signal stuff that were like a 3/4in or so in diameter a couple of decades ago...those would be perfect for this.

 

Will it ring the same way though? When the LC or Pease circuit is sim'd, it will oscillate a bit on changes...that's the key thing I'm looking for.

Yes.

 

What's my bar tab up to with you now, Ian!?

 

What's my bar tab up to with you now, Ian!?

That's a moot point with all the pubs being closed! You'll have to make do with a link to my local brewery:
https://www.bateman.co.uk/

To elaborate on my previous answer:
Generic RLC filter has a response of H(s) = 1/(s^2+ omega0/Q.s + omega0^2)
Sallen and Key circuit synthesises the same response, so if you select the same Q and omega0 then you will have the same response in both the frequency domain and the time domain as the original LC filter.

 

That's a moot point with all the pubs being closed! You'll have to make do with a link to my local brewery:
https://www.bateman.co.uk/

I totally got lost on that site. RUN A PUB!!! I am mildly-seriously considering doing that once things open up again. Plus, that would make it extremely easy to thank you and show my appreciation, wouldn't it? I have brother-in laws from Bath and Ipswitch so I'll definitely be making many, many more trips to the UK.

To elaborate on my previous answer:
Generic RLC filter has a response of H(s) = 1/(s^2+ omega0/Q.s + omega0^2)
Sallen and Key circuit synthesises the same response, so if you select the same Q and omega0 then you will have the same response in both the frequency domain and the time domain as the original LC filter.

Ya, between that design tool link you sent and LTSpice I've managed to get a near identical response after a bit of tweaking. Plus it saves an opamp! So much to learn yet. Thank you for saving me once again, Ian0, I really appreciate it!!

 

I've got interested in this and looked at the original circuit again.
If you step the input from 0V to 10V then I really don't see why it doesn't work on a single 0-30V supply, provided that any oscillations don't try to take the output below 0V.
The 100 ohm resistor shouldn't attenuate the signal as it is bootstrapped by the op-amp.

 

Hmmmm, I mean, it's a very interesting circuit and I'd be lying if I said I understood it other than the very basics. I still don't fully understand how much of anything signal-wise gets through the 1M resistors.

Now it's interesting you brought up oscillations because that is something I wondered about when deciding to try it. Normally we'd strap a low value cap across the feedback resistor to limit the bandwidth, but I didn't see how that would be possible with this circuit. We don't need a ton of bandwidth here so oscillation is a possibility.

I also wondered if it wasn't that this circuit may just sim nice, and the ~3V I got at C5 was never going to happen with real world components.

 

It was the 100 ohm resistors that fooled me when I first saw it. As it is drawn they appear to be the inverting INPUTs to the op-amp, but they are the OUTPUTs. The Op-amp is just a unity-gain follower.
So - I'm using my "in brain" circuit simulation (which I what I use if I suspect a dodgy SPICE model).
Apply a 10V step to the input - C1 will start to charge. The op-amp forces the voltage across C1 to be across the 100 ohm resistor, so the current increases, just like it would do if the voltage were across the inductor. But the current is sunk by the op-amp rather than passing though the inductor and out the other side. So the left-hand end works.
At the other end, the capacitor charges through the 1M resistor, the op-amp forces the capacitor voltage across the 100 ohm resistor, and outputs current as if it flowed from the "inductor", so my intuition says that it works. My intuition first said that Bob Pease isn't wrong!

To increase the damping, you need to reduce Q by reducing the load resistor (R21) as you would in any LCR filter.
The same applies to the Sallen and Key filter, except that Q is controlled by the ratios of the two resistors.
I just plug the Q and f values in here
http://sim.okawa-denshi.jp/en/OPseikiLowkeisan.htm
and let it do the calculations.
Q=0.5 = critically damped. Q=0.577 (1/SQR(3)) is the highest you can make it with no overshoot.
If it's an audio side chain, then wouldn't the oscillation cause a rather weird tremolo effect following a change in level?