Trying to understand W-Gun physics on SX100

[sem-geologist]

Our SX100 has nearly 25 years. We experience lately severe instabilities of beam if heat is in canonical regime (above 180), the "stable heat" value was systematically dropping during the last 3 years, I could not understand what's the matter, until I found in wardrobe used photo resistor tube (LDR-light depending resistor) in package from 2009. Also In our paper copy of HV tank schematics I see some modifications (with pencil) which marks that where only one LDR is left. Anyway, after extensive testing the HV control circuit, I suspect that LDR is failing and has its resistance severely dropped. The outcome I think is very low Bias voltage between Wehnelt aperture and filament. I measured the resistance between the filament socket and Wehnelt metal contact and it is only 120kohm. Lighting up the bias control diode internally could only reduce the resistance, thus the maximum bias Voltage for measured emission current of 140uA is about 16V. (Or actually it is reverse, because resistance can't go above 120kohm, the biasing of filament can achieve only 16V - such extremely small electric field cant suppress sideway emission from the tip and can't focus correctly the beam at crossover near anode. Normally bias voltage of tip to Wehnelt ring should be in hundreds of V range. Am I wrong?

It seem to me that some work with HV tank had to be done previously (that used up photoresistor from 2009). Had any of Your SX100 had HV tank repairs, and how does it looked? I suspect it will cost a lot as it will require shipping the tank to the France and back. Considering its weight and content I fear for the worst (logistic and price problem). Any thoughts or insights?

[sem-geologist]

The possible things to do I thought about:

1. Ask Cameca for inquire for HV tank repair
2. Buy a new Cameca HV tank and utilize old one locally
3. Adapt system to dry HV supply of other vendor (this looks most future prone to me, especially considering my successful upgrade of gauge)

The trickiest part is as far I looked through different options of other vendor HV supplies, that bias on those is controlled by setting desired bias voltage, where on Cameca tank it is controlled by  desired emission current. Unless I miss something? Cameca HV supply seem to have 3 options: a switch - reg I <-> reg V, a switch - W <-> LaB6, jumpers 30kV <-> 50kV. While I intuitively understand what LaB6 and W switch do, or jumper for 30kV-50kV (partly), regulation by I or V slips away my understanding. It looks it is applied for filament heating regulation, but I will ask here:

So, do anyone know why regulation is set at particular regulation mode, and what other mode gives? Some suppliers allows to buy either I- regulated or V- regulated power supply as additional option.

But maybe I am overlooking something, what kind of oil could be in the tank? It was produced after PCBs worlwide got banned, thus maybe it is not of toxic type?

[zorch]

In one instance, Edgar replaced the IR LED that drives the LDR. That got things going again, but I'm not sure if this was a bias issue he was addressing. Just going by the discussion here, I would think the I or V regulation choices are just another way of stating W or LaB6. If your filament is a W type, you want V regulation, and conversely, if the filament is LaB6, you want I regulation. W needs V regulation since I regulation causes premature filament burn-out. The reason for this is that when the filament thins as it ages, its resistance goes up, driving current down a bit. If I regulation is used, the same target current is forced through that weakening filament, causing runaway destruction of the filament. V regulation does not care that the filament is becoming more resistive as it ages, it just keeps applying a nice steady V across the filament.

LaB6 prefers I regulation since it is very sensitive, emission wise, to changes in heating current. Since the block of LaB6 is in the current path, it acts as a current limiting resistor, in fact, a linearizing element in the circuit, preventing runaway power dissipation that would occur if W were run with I regulation.
Things to look for in older systems: That IR LED mentioned above, and, dried up heat sink compound for the power transistors (more of a problem with air cooled heat sinks). Remove old heat sink compound and apply fresh compound.

[sem-geologist]

Just going by the discussion here, I would think the I or V regulation choices are just another way of stating W or LaB6. If your filament is a W type, you want V regulation, and conversely, if the filament is LaB6, you want I regulation. W needs V regulation since I regulation causes premature filament burn-out. The reason for this is that when the filament thins as it ages, its resistance goes up, driving current down a bit. If I regulation is used, the same target current is forced through that weakening filament, causing runaway destruction of the filament. V regulation does not care that the filament is becoming more resistive as it ages, it just keeps applying a nice steady V across the filament.

LaB6 prefers I regulation since it is very sensitive, emission wise, to changes in heating current. Since the block of LaB6 is in the current path, it acts as a current limiting resistor, in fact, a linearizing element in the circuit, preventing runaway power dissipation that would occur if W were run with I regulation.

zorch,
thanks for this explanation - It was missing piece for me to understand fully how this system works. Technically the switch (I<->V)  directs the feedback loop to heat-regulating OPAMP either directly (at I) or through the capacitor (at V). Interestingly I see no such switches on other vendor HV PSU's, which declare compatibility with W and LaB6. Would that imply that Cameca regulation have some specific W-tip lifetime saving feature? LaB6 and W switch changes to shine different pair of LED and they have different light apertures between LED and LDR for LaB6 and W configuration. Short-cutting the reasoning behind and inner workings, it seems that LaB6 needs higher bias voltage than W, Or because bias "is pumped" cyclically (OPAMP used for switching the LEDs has integrating feedback configuration (capacitor), so smaller light aperture between LED and LDR for LaB6  result in smaller bias voltage "pumping" steps.
So in case if I would look for an HV PSU alternative, the Tip-Wehelnet ring bias range does matters if we would want to use LaB6 somewhere in the future, although I am not sure if that Axial/in-axis ion pump is good enough for that. Our situation with HV tank can be more complicated as it probably was modified decade ago leaving single LDR and diode. My intuition tells me that it would be more beneficial to revert it to original form (two of LDR and two LED).

In one instance, Edgar replaced the IR LED that drives the LDR. That got things going again, but I'm not sure if this was a bias issue he was addressing.

If he did that at site, then these are details I would like to know. My particular concern is the handling of transformator oil:
1. any health hazards?
2. draining out all oil (for recycle) and filling new after the procedure?
3. if oil is flammable < 250C, then it requires mandatory whole PCB cleaning before soldering new IR LED so it would not lit up in fire during soldering (cleaning procedure, agents used) (in my case, different to changing of LED, changing LDR needs no soldering)?
4. refilling tank with same type of oil, or new type (green, expensive, high ignition temperature) can be used?
5. the final tank closing procedure to keep bubbles and air away?
6. In case of reverting  to the original design (Filter board inside tank), should not HMP solder be used (resistors are rated 2W) so it would not de-solder.
7. what kind of IR LED should be used (model?)?

This is why I am considering alternatives, i.e. going to dry HV supply as in long term it looks to be less problematic.

In one instance, Edgar replaced the IR LED that drives the LDR. That got things going again, but I'm not sure if this was a bias issue he was addressing.

I believe in your case it could be LED. No light from LED translates to huge resistance on LDR (if LDR is functioning correctly) -> which translates into very high bias voltage -> which leads to very low emission current. Burned LED will result in low emission current and wont react to setting the current higher. In our case it is opposite, we get much too high emission current, currently with heat 175 we get 140uA, and if we set heat 200 we get 440uA! and it does not react to setting the emission current lower. My testing on regulation board shows that measured emission and set emission is correct, and OPAMP which switch the LED is not draining its negative rail (correct behavior). Additionally direct measurement between filament pins and wehnelt's ring connection shows only 120kOhm (it should be at least few megaohms, as else parallel connected few 10Mohm resistors in the schematics makes absolutely no sense). Our used old LDR (inside 2009 marked bag, used spare) shows 240kOhm then wrapped in light-blocking material. Albeit, probably it would be wise to change LEDs if HV Tank if it would be opened, just in case, as LED's age too. The two LED, LDR configuration (original) is more prone to LED failure (however failing one of LDR would still let the system work correctly), where single LED-LDR configuration (modificated) is more prone to LDR failure, failed LED would leave it at somehow usable state. Thus original two LED-LDR configuration is more resilent, as in case of double - it needs repair only if LED fails, and in single - it needs repair if LED or LDR fails.

Things to look for in older systems: That IR LED mentioned above, and, dried up heat sink compound for the power transistors (more of a problem with air cooled heat sinks). Remove old heat sink compound and apply fresh compound.

Oh, these are known to me, I did that to many power transistors (while hunting the source of problem looked for clues in other systems, also due to other problems which I had fixed) and some had burned and were replaced.

BTW, on HV regulation board, did your R14 (filament heat regulation) and R42 (used for V regulation) 0.5W resistors had burned or have signs of burning the dye? Mine were with clear burning of dye signs, thus three years ago I had replaced those with 1W type (actually original carbon R42 was completely cracked). While it did the job for R14, R42 (as I saw at last inspection) is again with burned dye. I think placing 0.5W in these positions is a design mistake (for some unintelligent reasons 53V output from toroidal transformer (at standard wall input of 230VAC) is noted with +28V in schematics, maybe that mislead the circuit designer), quick simulation of that circuit portion in circuitjs shows me that R42 will disipate 1.6W , thus this time I had replaced it with 2W type. But even if supposing just +40V (as denoted in schematics, ignoring the real +48V at load or +53V no-load conditions) 0.5W rated resistor there is overhelmingly too little.

[sem-geologist]

After short research of electronic market, I see that thanks to that european RoHS (again! sic!) nonsense, the availability of photoresistors are extincting. Thus in decade they could be unavailable completely thus Cameca also would be not able making these LDR tubes anymore. Thus it is an additional reason to develop the plan for moving with HV supply of different vendor. I am considering Spellman PSU's as their supplies are used in ThermoFisher SEM's.

* addendum:

...replaced the IR LED that drives the LDR...

Oh, so it is not visible light LED-LDR system, but IR. What I had found when I wrote the above were Visible light LDR based on CdS; IR LDR is based on PbS chemistry, and I could not find even a single place where to buy any IR LDR - looks they are not manufactured anymore. Thus price of these are going to be space-grade (if not already is).

The RoHS nonsence is hitting not only us, but also audio/sound industry. Photo Resistors have this overwhelming advantage (which is marketed nowadays as big constrain/disadvantage compared to photodiode), they react to light with very huge delay measured in seconds. It allows to have extremely simplistic system for regulation, basically integrating OPAMP which enables temporary the LED which slowly changes resistance of LDR. Replacing LDR with photodiode will need lots of stabilisation as photodiode reacts to LED very fast, and OPAMP loop will need be much more complicated. It will need to go to high frequency switching for bias regulation. Probably other vendor new HV supplies does exactly that...