WDS preamplifier design

[John Donovan]

I'd just point to the peak shifts and window placement discussion in the "improving accuracy" thread to point to yet another issue that a solid state detectors would eliminate.  I do believe solid state detectors are the hardware improvement that will bring the next level of accuracy.  They will solve problems that microanalysts don't even realize exist.

I could not agree more!

I am reminded of the story I heard where, back in the day JEOL was contacted by the Seiko (watch) company, who asked JEOL to develop a FEG for their EPMA instrument. Apparently a few million dollars were provided to JEOL Japan for this!

Seems to me that a solid state detector and fast pulse counting electronics should be a much smaller and easier project, right?

Have you got a spare million bucks to be an angel investor on this?

[Les Moore]

There's an added factor that would favor SDD over FC, recovery time.  Try mapping something with SiC in it with a Si Ka at 200nA with bidirectional mapping and you will see a saw tooth effect on the boundaries. The massive change in countrate must be saturating the detector such that it won't reliably measure the Si Ka when it drops back down.
 

[sem-geologist]

There's an added factor that would favor SDD over FC, recovery time.  Try mapping something with SiC in it with a Si Ka at 200nA with bidirectional mapping and you will see a saw tooth effect on the boundaries. The massive change in countrate must be saturating the detector such that it won't reliably measure the Si Ka when it drops back down.
 

Could you share a picture?

Excellent observation about saturation... but completely wrong tree to bark. What you are saturating is rather the preamplifier and not the gas flow chamber itself. Also, did this observation happened with correct PHA settings to compensate the PHA shift? The PHA shift will happen gradually at transition of grain boundary, as discharging the feedback capacitor (which gets near-full charged during very high count rate) of preamplifier needs some time.
This thread is about preamplifier design which is the weakest part in currently used GFPC. Replace the outdated preamplifier + shapping oamps and discrete pulse counters, and give modern electronics to GFPC and it will leave SDD in the dust with the performance.

[Les Moore]

I think I'll just lower the current or choose another SPC rather than attempt to convince JEOL that their GFPC needs upgrading.

[sem-geologist]

I think I'll just lower the current or choose another SPC rather than attempt to convince JEOL that their GFPC needs upgrading.

As far Brian Joy had shared pictures some time ago (cant remember which of the threads), I think Jeol EPMA has very modular design. And theoretically whole module with pre-amplifier could be replicated as a copy with modified feedback loop (for better throughput).

Also, what you could try doing for free with no modifications is increase GAIN and reduce bias. That should charge up the previously mentioned (by me) capacitor much slower. SPC (single pulse counter) and MPC (multi-pulse counter) terms are a theoretical oxymoron confusing names. Pulses are counted sequentially in both EDS and WDS. "counter" needs to be able to count at least up to two to be eligible to be called a counter, else it is just flip-flop. Single Pulse Counter sounds like "clap with one hand" or "listen to the sound of vacuum".

The chain of pipeline:

X-ray photon -> electron cloud -> townsend's avalanche -> charge at anode of GFCP -> pre-amplifier -> (BNC cable to other unit on JEOL) -> Pulse Shapping Amplifier -> [pulse sensing and counting (integral) -> pulse height digitization -> filtering (differential counting)]

I believe pre-amplifier and Pulse Shapping amplifier are in separate boxes connected to GFPC through BNC like cables. Replacing counter (which you call a SPC) would not eliminate your observed saturation, which I suspect originates earlier in the pipe-lin. Well, unless Jeol's "SPC" is made much worse than I could imagine...

[Probeman]

I think I'll just lower the current or choose another SPC rather than attempt to convince JEOL that their GFPC needs upgrading.

No need for any of that when using the integral-baseline method for adjusting one's PHA settings:

https://smf.probesoftware.com/index.php?topic=1854.0

Once properly tuned, your PHA's will be able to handle count rate differences between your primary standard and unknown of 10x to 20x or more. 

In other words, even with these ancient counting electronics designs (and they are ancient!), once can now obtain a wonderfully linear response all the way from the pure element to zero, by using this integral-baseline PHA tuning method.

Try it for yourself!