XPS analyzer control non-linearity problem

Print Friendly, PDF & Email

The 80-360 and 80-365/366 analyzer controls provide all the voltages to the 10-360 spherical capacitor analyzer that is used on many PHI (Physical Electronics) XPS systems.

80-365 boards Retard board is third from left

The retard voltage is used to slow down electrons and is essentially the sweep voltage.  In conjunction with the pass energy supply, the retard voltage controls the energy of the electrons that are being passed through the analyzer and into the electron multiplier and counting circuitry.

I recently had an interesting problem with a retard board on an 80-360 analyzer controller.  The issue was that the retard board output voltage was not linear.   Part of the calibration procedure for the retard board is to test the voltage at 11 specific voltages ranging from 253.6 volts to 1253.6 volts in increments of 100 volts.  This is a convenient way to confirm that the retard supply is linear.  The table below shows the hex commands that are used to set the voltages and the expected results.

In this particular case, the output voltages were close to correct at some points, but way off at other points as shown below. This non-linearity would present itself as inconsistent peak widths in the data as a function of kinetic energy.

These results at first glance look like a bit problem.  That is, the digital to analog convertor (DAC) voltages are likely off.  The DAC used in the retard board circuit is a 16 bit DAC and the output voltages should follow the voltages listed  below.

DAC Voltages

However, the DAC voltages were fine.  The retard voltage circuit comprises the DAC which drives a precision operational amplifier that in turn drives a high voltage switching supply.  Some precision high voltage resistors are  used to provide feedback.  The next most likely component that might be non-linear was the OP07 ultra-low offset operational amplifier (op amp).  The OP07 was replaced but did not solve the problem.

The next most likely cause of the non-linearity problem was the feedback resistors.  There is a total of 5 of these SX3730 5 watt wire wound high precision axial resistors in series.  To accurately measure those resistors, you need to lift one end off the circuit board.  Using my Fluke multimeter, I tested the resistance of each resistor and they all checked out as OK.  Ideally you will see very close to 1.0 megohms, but it might be off by .05 megohms. When a resistor is bad it will be open or be off by as much as .5 megohms. So, it seemed that the feedback resistors were OK as well.

That left not much in the circuit other than a few potentiometers. After spending some more time retesting all the components, I came to the conclusion that it had to be one of those feedback resistors. 

To test that theory I removed all 5 of the SX3730 1 megohm feedback resistors and replaced them with a single 5 meg ohm resistor.  And that worked!   So now I knew for sure that one of those 5 feedback resistors was the problem.  I measured the resistance of each resistor, and they looked OK.  But then I realized that the non-linearity is a function of the voltage applied to the resistor. At some voltages the resistance was OK, but at other voltages the resistance was off. 

I then decided to measure the resistance value again using a megohmmeter.  The model that I used was a Protek DI-2000M.  This megohmmeter (also called an insulation tester) puts 500 V across the resistor when measuring the resistance. I hoped that by putting 500V across the resistor that I would be able to see a greater difference in the resistance values. And that worked out as expected.  One of the resistors showed only 965 Kohms with the megohmmeter and 995 Kohms with the Fluke. I replaced that resistor and the calibration was perfect.  😊

In hindsight, duh. Since the gain of the circuit was changing as a function of the non-linearity of one resistor, the lesson is that when checking the resistance of suspected non-linear resistors, always use a megohmmeter since that will put much more voltage across the resistor than what a normal DVM will put out. Better yet, if possible, use the highest voltage that the resistor is rated for.

If your 80-360 or 80-365/366 analyzer control is not functioning properly and you need some help, please contact RBD Instruments for assistance as needed.

2 thoughts on “XPS analyzer control non-linearity problem

  1. Oh the hazards of “verifying” components under conditions far different than normal use conditions. I’ve been bitten by this myself once and another popular way to fool yourself when validating resistor performance is to check low noise resistors at the wrong level of current drive. Or at the wrong temperature. Or without actually analyzing the noise spectrum under bias.

Leave a Reply

Your email address will not be published.