approach control confusion

  • Anton Dubrau

    Anton Dubrau - 2007-05-15

    We in our lab are a little bit confused about auto approach. We feed the bias to the in to see what the sranger/gxsm setup is doing. Everything is behaving like it is suppossed to, retracting and approaching the z based on different signs of bias and CI (we set the setpoint to 0). with CI=0, the CP doesnt seem to do anything.
    Then we tried the auto tab in mover control. We notied a negative Bias (with setpoint 0) is apparently interpreted by gxsm as the tip being away from the sample, regardless of CI/CP. It seems that regardless of the sign of CI we always get the same approaching sawtooth, although the absolute value apparently controls the speed of the approach, is that right?
    The really confusing thing is that after stopping the approach sometimes the feedback control seems to be reversed. That is, with negative bias and CI the sranger retracts, but after stopping and then changing the bias to positive, it still tries to retract. Then changing the CI positive, the sranger tries to approach. When pressing stop approach again, the feed back control goes back to the usual operation. Is that some sort of glitch, a feature or is the approach control only suppossed to work with positive CI values?

    • Percy Zahl

      Percy Zahl - 2007-05-16

      What you are writing seams OK, in brief, it works like this:

      For tip retract and extension (checking) actually the exact same feedback is used by only temporary (on DSP) manipulation of CI and CP:

      (refer to FB_spm_autoapproach.c)

      There are the following Tip-Approach states:

      TIP_Ready: means the approach is done (tunneling, etc.) or was stopped manually,

      TIP_ZPiezoMax: check for Signal, i.e. the Tip is currently retracted with Z and the feedback will be enabled with CP and CI set by user, i.e. the current "error-signal" and CI will determine the approach speed as this will all stay constant until the tip gets close and the feedback will work normally -- if this occurs the approach is done and the state will be TIP_Ready, if the Piezo reaches Max-extension, the state will change to TIP_ZPiezoMin. (CP must be non zero for this to work)

      TIP_ZPiezoMin: Now, to "retract" the tip almost fully, we simply do a trick and set CP=0 and CI=CI * 20 (we remember the old CI for restauration if done) -- this will retract the tip (you notice, CI must be non zero for this to work at all!!) -- if done, we hold the feedback, restore the original CI and CP and switch to mode TIP_Delay_1.

      TIP_Delay_1: wait, then proceed with TIP_ZSteps,

      TIP_ZSteps: execute number of course steps (whatever is configured)

      TIP_Delay_2: wait, then proceed with TIP_ZPiezoMax.

      So, now you should understand it better and remember, in STM mode the signum of the input signal is (feedback in) in meaningless, as the abs value of it is used in STM mode.

      And one more thing, which actually right now is hard coded, the direction (the waveform is played) of the saw-tooth (or whatsoever curve you select) is independent of CI, however, you can download any curve, so if you need the other "direction" you can do it.

    • Percy Zahl

      Percy Zahl - 2007-05-16

      Correction (note in TIP_ZPIezoMax): not CP, but CI must be non zero for this to work.

    • Percy Zahl

      Percy Zahl - 2007-05-16

      Sorry I am writing a mess today.... one more correction:

      TIO_ZPiezoMin: .... CI = -CI * 20 .... (it must go the other direction for "retract with no current") so I set CI to the opposite signum and increase it by a factor of 20 to speed it up -- you get it now?

    • Anton Dubrau

      Anton Dubrau - 2007-05-16

      Thank you for the fast reply, everything is much clearer now.
      We are actually working in AFM mode.
      One thing i still don't understand is why after stopping an auto approach sometimes the behaviour is reversed, as if the CI is always multiplied with -1. stopping auto approach again corrects that behaviour.

    • Anton Dubrau

      Anton Dubrau - 2007-05-16

      Ok, we are measruing in tapping afm mode. That means that smaller input values means we are closer. In order to make measurements, we have thus used negative CI values to get the behaviour we want. Is that the right approach, or is there some setting that tells gxsm that small values are close as oppossed to big ones? In our setup the auto approach never starts the auto approach because the input is larger than the set point and gxsm seems to think the tip is already at the sample. Is there a way to accomodate this?
      Also we noticed that the CP in our AFM does not do anything, according to the manual its the proportional (i.e. z' = z+(z-set)*CP). As far as we understand we should be able to control the usual feedback loop just with the CP. Is there some setting that controls the CP other than the CP slider in the SR DSP control, or is it simply not used in AFM mode?

      • Percy Zahl

        Percy Zahl - 2007-05-16

        a) I think it is possible, even very unlikely as the retract state is very short, that if you cancel/stop the auto-approach the "-20*CI" is still set if you just hit it that moment. A little change of any of the Feedback parameters shoult correct that or as you wrote, stop it again... Usually I do not care, as nomally I will proceed the approach...

        But it is a good point and I will double check the worst case behavoir at "stop" and will see if it needs a fix.

        b) In AFM mode, sure, the SetPoint needs to be set correctly for teh auto approach to work. If it is beyond (to high force) the approach will certainly not start.

        Best is, to keep some offset adjustment (externally) to have the "Zero" Force at 0V. But not necessary.

        Negative CP AND CP are truely OK if you need it like this.

        CP is always working -- you can test it, set CI=0, but adjust Z to about zero before (by playing with CI). Then set CP to 1 and apply a voltage to the feedback-input and you will see the linear response, scaling with CP on Z.

    • Anton Dubrau

      Anton Dubrau - 2007-05-18

      We run our AFM in tapping mode and monitor the amplitude of the tuning fork as the input for the Signal Ranger straight out of the lock-in. To accomodate the reveresed polarity (smaller amplitudes closer to the surface) we're using negative values of CI. What we've noticed is that despite the correct behavior in feedback (extend when above the setpoint, retract when below the setpoint) the autoapproach does not start when the value is above the setpoint indicating that it is still away from the surface.

      Do you have any suggestion on how to do automatic approach in this case?

    • Nikolai

      Nikolai - 2007-06-19

      hi Anton,

      I had similar problem as I'm using gxsm for stm and afm. when I switch to afm I have to change z polarity. negative cp,ci would do the trick but autoapproch won't work. So, finally I'm using amplifier to change z signal polarity (invert).



Log in to post a comment.