guard driver?

Thread Starter

mcike

Joined Apr 17, 2011
12
Hello,

i need to drive the inner shield conductor of a triaxial cable in order to avoid the effect of cable capacitance (phaseshift of signal). Signals will be 1to 5Vrms and frequency between 1Hz and 250kHz.

Since i'm quite a newbie i tried to build the guard with a OP27 in follower configuration. Unfortunately @67kHz and Signalamplitude of 1Vrms the circuit is unstable... amplifiers are ocsillating.

My question: Is there a pincompatible alternative for OP27, which can drive my shield properly. BTW: cable capacitance is about 150pF/m.

I thought about Buf634???

Many thanks for your help!
 

nsaspook

Joined Aug 27, 2009
13,086
I assume you're using it as a unity gain buffer amp. Did you try loading the amp output with a 2K resistor and using HF bypass chips near the chip power leads.
 

Thread Starter

mcike

Joined Apr 17, 2011
12
bypassing is done by 10µF tantalum in parallel with 100nF ceramic capacitors. i didn't put load to the ouitput yet.

do you think this could stabilise the circuit?
 

PaulEE

Joined Dec 23, 2011
474
Why does something oscillate in the first place?

1) the phase difference between input and output is zero or 360*
2) the gain between input and output is >= 1.

In a typical unity-gain amplifier circuit, there is NEARLY a 1:1 correlation between input and output phase. Since it's a UNITY gain amplifier, you've got a gain =1 v/v.

There must be an issue being introduced by the abnormally high capacitance at the output, the op-amp, or both.

Looking at an op-amp circuit whose output is driving a capacitive load, the phase between the input and output will change. Apparently, the phase difference is substantial; your circuit is oscillating. Also, in the datasheet, there is a phase difference from input to output in the op-amp itself equal to approximately 80* right off the bat.

Another way to look at this: there is a pole formed by the impedances in and around the op-amp.

To further complicate matters, the OP27 has input protection circuitry that matters in unity-gain op-amp configurations. See datasheet for details on the appropriate feedback resistor.

I would concentrate on using an op-amp with a lower input/output phase shift.

What is a phase-shift, anyway? It's just a shift in the time-domain of a periodic signal...what causes it? Stray inductances and capacitances; complex impedances.

I'd use a BUF634. It can supply more instantaneous current and is built for speed (minimum phase shift) from input to output.

Example:
Compare the phase charts at unity-gain for the OP27 and for the BUF634. The OP27 is about 80*, the BUF634 is about zero! (that gives you some headroom with cable capacitance).

Let us know how it goes!
 
Last edited:

PaulEE

Joined Dec 23, 2011
474
quote: bypassing is done by 10µF tantalum in parallel with 100nF ceramic capacitors. i didn't put load to the ouitput yet.

do you think this could stabilise the circuit? :end quote

No.
 

Thread Starter

mcike

Joined Apr 17, 2011
12
I'll givethe Buf634 a try... Unfortunately it's not available till 8th Of February. I will give a report on how things worked out asap.

Thanks for all your help!
 

Thread Starter

mcike

Joined Apr 17, 2011
12
finaly i tested the circuit with a buf634. now the circuit works quite fine. @200kHz the phase on output is about 3.5°, which seems to be a bit to much...
if i change voltage on inner conductor, the buf634 sometimes starts to oscillate and stops oscillation after few secs again. it also seems to be very sensitive to distortion of ground potential.

maybe i'll try to set the buf634 to wide bandwidth mode in order to minimize phase on output!?

could an ohmic load further stabilise the circuit?
 

Thread Starter

mcike

Joined Apr 17, 2011
12
Coax is about 2 meters long;
Signal source is Oscillator (Voltage source) of Lock In EG&G7260;
Load of the Buffer is capacitzance of cable only.
 

Ron H

Joined Apr 14, 2005
7,063
Are you concerned about phase shift between the input of the cable and its output?

Also, what is the capacitance between the two shields?
 
Last edited:

PaulEE

Joined Dec 23, 2011
474
EDIT:

A few posts back, I made an error. Here's what I said, and here's what I meant:

Things oscillate when two things are true:
-phase difference between input/output is zero or 180* and
-overall gain is >=1

What I meant:
-phase difference between input/output is zero or 360* (or multiples of that) and
-overall gain is >=1

This is referred to as the "Barkhausen Stabability Criterion". Sorry for the error! (I fixed it in the original post)

http://en.wikipedia.org/wiki/Barkhausen_stability_criterion

How'd that BUF634 work out?
 

Thread Starter

mcike

Joined Apr 17, 2011
12
Are you concerned about phase shift between the input of the cable and its output?

Also, what is the capacitance between the two shields?
Capacitance between the 2 shields is about 240pF/m.

the exact arrangement is: i have one conductor which brings current to a metallic sample sitting in a nitrogen flow cryostat (and therfore changing it's resistance value from about 2 Ohm to 300Ohm as a function of temperature), one conductor which brings the current back and feeds it to a reference resistor, and 2 wires which i use for measuring voltagedrop on my sample.
in order to minimize inductance of cabels the current wires (both separately shielded with guards) are the two conductors of a "twin-triaxial cable". i use a similar cable also for voltage measuring wires.

now my problem is, i need to compare the sample voltage (amplitude and phase) to the voltagedrop on my reference resistor. this i have to do as a function of frequency (1Hz-240kHz). with my setup, this is just possible, if the phasedifference is 1° or smaller.

So i thought, by minimizing cable capacitance, this should be possible.

Status: The Buf634 currently are a little bit to slow (9°phasedifference between conductor and shield@200kHz), and unfortunately also tend to oscillate.

1.)Oscillation
The strange thing is, if i turn on power for my circuit, the Buf634s are oscillating. As soon as they once saw a signal on input, oscillations are gone. in the moment, i use a 50Ohm resistor in series to the output of the Buf634. before doing so, the Bufs were completely unstable.

2.)Speed
To my understanding, the resistor which i use in series to the output is decreasing the speed of the Buffer!?? Is this true?
This would imply, that i can not use the wide bandwidth mode of the buf634??

Again thanks for all your help!
 

Ron H

Joined Apr 14, 2005
7,063
BUF634 is an open loop follower. Its gain is slightly less than 1,which is probably OK for bootstrapping the shield. Are you also using its output for making measurements? It may also have unacceptable distortion when used open loop. You might be better off using a wideband op amp as a voltage follower for the measurement, while leaving the BUF634 in place for the guard, or bootstrap, drive.
The datasheet shows a 50 ohm input resistor for all bandwidth plots. Why do you think this resistor precludes using the wideband mode?
As you have probably deduced, the buffer's primary capacitive load is between the shields. The capacitance from center conductor to inside shield, being bootstrapped, is minimal, as of course is your intent.
Are you digitizing voltage and current waveforms?
 

Thread Starter

mcike

Joined Apr 17, 2011
12
Fninally i understood, that the main reason for oscillation in my circuit was capacitive coupling from guard to inner conductor. therfore, a low pass filter at the input of the buf634 did the job. To my understanding the 50Ohm resistor at the input together with inputcapacitance already acts like a low pass...
now, i try to optimize the lowpass circuits at the input of the buf 634, in order to avoid current sinking to ground and at the same time to have acceptable phaseshifts of the guard voltage.

so far, my problem is solved, thanks again for all your help!
 
Top