purpose of TL081 in this circuit? - LowRippleHVsupply2.jpg (0/1)

Looks like a simulated inductor or "gyrator"

See description here

Simulated inductor,

. ,--/\/\--+---------, . | Rs | __ | . | '--|- \ | . | C | >--' . o---+--||--+----|+_/ L = C R Rs . | R . '--/\/\-- gnd

Used in a notch filter,

. ---/\/\--+------ . | . | ,--/\/\--+---------, . | | | __ | . | | '--|- \ | . | | | >--' . '--||--+--||--+----|+_/ . | . '--/\/\-- gnd

If you look carefully you'll see it's not an inductor or gyrator. No, it's a capacitance multiplier.

. 10k 10k . ---/\/\--+--/\/\--+---------- 1.5kV max HV out . | | 5nF . | '--||--- gnd . | . | C1 5nF R1 2k . +---||-----/\/\--------, . | C2 5nF R2 1M | . '---||--+---+---/\/\---+ . | | __ | . 16k '--|- \ | . | | >---' . +------|+_/ TL081 . | . 5k . | . Vcc/2

The opamp amplifiers whatever ripple it sees and presents this to one of the capacitors, C1, making it appear larger.

Charlie has two 10k resistors for the R part of a two-stage R-C post-filter for the output of his 1.5kV power supply. He's got three 2kV high-voltage 5nF cap to work with in this filter, but he's goosed-up two of the caps with a capacitance multiplier.

I analyzed Charlie's circuit and compared it with using the two 5nF caps tied directly to ground. The two 10ks and the passive caps are down -3dB at 800Hz and rolloff at 40dB/decade above 3kHz. Adding the opamp circuit, the attenuation starts earlier at 430Hz (after a +10dB peak at 320Hz), and rolls off at 40dB/octave above 800Hz. There's a little -82dB dip at 8kHz, afterwhich it bottoms out at -70dB above 12kHz, due to the 2k series resistor. In the critical 2 to 40kHz region the attenuation is 20 to 45dB better than simply using the two 5nF caps passively.

Charlie's circuit is an excellent "ripple stripper."

WOW, thanks a bunch Win. That's an awesome circuit!!!!!!!!

I had begun to wonder about using active amps to detect ripple and to feed the output of the amp to induce a 180 degree phase and then feeding back the phase shifted output into the ripple stream.....effectively making an active ripple reducer.

Is there any way to reduce ripple using a transformer to feed back phase shifted ripple content in order to achieve the same purpose?? Or, would a big 'ole inductor do just as well? I'm surprised no one has used miniature transformers in high voltage low current aps to reduce ripple...maybe it doesn't work?

For 20 kHz and up, it might be possible to reduce ripple nicely by using ferrite beads with 10 of 15 turns of fine wire, which would be easy to do and cheap. Of course, this might be a problem with magnetic flux in the proximity of a PM Tube, which causes it's own problems...PM Tubes are really sensitive to magnetic fields.

Thanks so much for the info Win.

You weren't reading a.b.s.e replies? I said capacitance multiplier more than two hours earlier ;-)

...Jim Thompson

Another good method is to float the HV supply, look at the hot end via a capacitor, and jiggle the cold end to keep the hot end still. I used this in a 3 kV piezo translation stage supply, and got ~ 100 dB of 120 Hz suppression. It used an RC tee network for DC feedback.

Cheers,

Phil Hobbs

Possibly used as a gyrator?

No, Jim...I am sorry, I looked in abse after I read your post here. I'm sorry, and did not mean to ruffle your feathers. Thank you also for the assistance.

Yes, I noticed that the high voltage was pretty close to the op amp and that a leaky cap or a sudden application or removal of the power might cause a problem.

Thanks for commenting.

Active noise cancellation techniques are limited to power levels and bandwidths that are relatively easily to handle.

Methods are as varied as cost and reliability implications permit.

Power supply applications represent only a small portion of potential or practical end use.

RL

Series vs parallel signal injection.

RL

I'm not ruffled... just funning with Win ;-)

BTW, Congratulations to Win and the Commonwealth of Massachusetts Legislature for over-riding the governor's veto of embryonic stem cell research. I knew there had to be SOME smart people in that state ;-)

...Jim Thompson

Clamp diodes from the relevant points to the op-amp's supply rails would be an advisable addition.

Graham

well, personally i would add a TVS (transient Voltage Suppresser) and maybe even a Zener to help protect that op-amp from an inrush of voltage that would destroy the op-amp.. putting that aside.

what the op-amp is doing is simply using the inverting input signal back to the source to cancel the effect there for, a high gain op-amp is generally selected for this application because afterwards the end results will be small. this has also been noted in some doc's as an electronic/active capacitor. personally, i wouldn't use it but that's me!

The circuit approximates an inductor.

By using the infinite open loop gain approximation, the voltage transfer function becomes ( S R2 C1 ) / ( 1 + S R2 C1). This enables calculation of the input impedance. The result is Z(S) = R1 ( 1 + S R2 C1 ) / ( 1 + S R1 C2 ). Expanding to second order in S gives Z(S) ~ R1 ( 1 + S ( R2 - R1 ) C1 - S^2 ( R2 - R1 ) C1^2 ).

Herbert

It's inductive below around 15KHz, but it's a capacitance multiplier above that frequency.

...Jim Thompson

Assuming I have derived the correct equation, the input reactance becomes capacitive when R1 > R2. Obviously when R1 = R2 the input impedance is purely resistive.

Bingo. Someone has just designed the simulated resistor.

Calculating the steady-state susceptance, B(jw), yields

Cin = C1 ( 1 - R2 / R1) / ( 1 + w^2 R1^2 C1^2).

divider.

Herbert

See....

Newsgroups: alt.binaries.schematics.electronic Subject: purpose of TL081 in this circuit - from S.E.D - PurposeOfTL081InThisCircuit.pdf Message-ID:

...Jim Thompson

Jim, you didn't include the opamp's falling gain and its rising virtual-node voltage. The critical frequency where the summing-junction-node voltage rises to equal the input "e" (destroying your current i = C s e) is surprisingly low, more or less, w_x = sqrt (w_T / R2 C), which is 11kHz for the circuit values in question, using fT = 4MHz. A spice plot shows this effect, plus a resonance notch at 10kHz.

This means the circuit looks like an increasing effective capacitance below the critical frequency, changing to an inductive nature above it. At any rate, its net Zin stays below that for 10nF of capacitance until above 160kHz, so it's still a pretty effective circuit in practice.

As far as the "capacitance multiplier" name, there seems to be no frequency region where it actually looks like a simple accurately-multiplied pure capacitance. Still, the practical effect of this circuit in a filter is that of a dramatically-multiplied capacitance value over a pretty nice usable frequency range.

I'm still thinking about R1.