Needing more than a spark test?

Even the high-end x-ray spectrometer we had at work would make crazy guesses at what it was looking at. One time I had a sample analyzed and it claimed it had Dysprosium in it!
 
Just in experimentally messing with a single op-amp package LTC6268IS810#PBF, not on a PCB, but super-glued upside down onto a scrap of PCB, as in "dead bug". It proves to be a tiny handful!

It needs very low leakage around the transimpedance amplifier input!
It's looks like it is going to work OK, but comes with a caution on stopping it becoming a powerful little oscillator. It takes around 1pF or 2pF across the feedback resistor Rf to get it going sensible, and that is easy enough, except for I think the capacitor has to be of extraordinary quality! The leakage current of most capacitors I could find handy is way more than the input bias femto-amps. An old-school axial polystyrene cap with the thin wires worked, but is almost the size of the whole chip. Two short wires stood close to each other with some stripped off PTFE sleeve pushed over one also worked, but this is only crude breadboard-style messing with it.

I read that ceramic capacitors have leakage that is orders of magnitude more than the input bias current of this amplifier.
It makes me wonder if a casual bit of FR4 printed circuit board is going to be up to it. It almost seems to be getting into the class of electrometers which used glass. Low leakage should be easy with caps having values of only a few pF, and I think even pH meters have used FR4. I think this may be a non-problem, only an artifact of my sloppy breadboard, and sweaty hands.

Now looking into surface mount low leakage caps.
 
graham-xrf said:
Just in experimentally messing with a single op-amp package LTC6268IS810#PBF, not on a PCB, but super-glued upside down onto a scrap of PCB, as in "dead bug". It proves to be a tiny handful!

It needs very low leakage around the transimpedance amplifier input!
It's looks like it is going to work OK, but comes with a caution on stopping it becoming a powerful little oscillator. It takes around 1pF or 2pF across the feedback resistor Rf to get it going sensible, and that is easy enough, except for I think the capacitor has to be of extraordinary quality! The leakage current of most capacitors I could find handy is way more than the input bias femto-amps. An old-school axial polystyrene cap with the thin wires worked, but is almost the size of the whole chip. Two short wires stood close to each other with some stripped off PTFE sleeve pushed over one also worked, but this is only crude breadboard-style messing with it.

I read that ceramic capacitors have leakage that is orders of magnitude more than the input bias current of this amplifier.
It makes me wonder if a casual bit of FR4 printed circuit board is going to be up to it. It almost seems to be getting into the class of electrometers which used glass. Low leakage should be easy with caps having values of only a few pF, and I think even pH meters have used FR4. I think this may be a non-problem, only an artifact of my sloppy breadboard, and sweaty hands.

Now looking into surface mount low leakage caps.
Click to expand...
It's always interesting to see what happens when actually building up some circuits!

Thin film capacitors are available in SMT format and aren't too expensive for picofarad-level values. Their dielectric is a silicon dioxide-silicon nitride composite so they should be pretty good when operated below their maximum-rated voltage.

You're correct about the relatively high leakage associated with multi-layer ceramic capacitors, and the high-K varieties are the worst. Ceramic caps with a low temperature coefficient are the best you can get for low leakage, because you've got a better dielectric. Another clue: don't use capacitors with a significant variation of capacitance vs voltage that's a big clue regarding the type of dielectric.

Dead bug should be pretty good w/regard to leakage -- after all, even with a good guard ring PCB layout you're still dealing with leakage around the package pins. So that's about as good as you can get, given the limitations of the package. I _still_ don't understand why vendors sell opamps with FA-current-level inputs when they're immediately adjacent to VCC/VEE/output pins, what a stupid way to shoot themselves in the foot. Gotta wonder how many designs have been burnt by that.

An alternative to full-on dead-bug is to lift the critical package pin(s) and directly solder to them. I've done some designs using that approach. For a TIA that would be just one pin. My rant still applies :rolleyes:.

FYI, the LTC6268/9 series have about the lowest input capacitance I've seen in a wideband FA-input-level amplifier...at least, for devices where the input capacitance is spelled out ( I suppose one could get an idea with SPICE simulations of the target amplifier that are designed to elicit that parameter...or maybe examine the SPICE model to see if it's spelled out). They're pretty good for low input-C, but if it's achieved using some type of bootstrap scheme they could be abnormally sensitive to the impedance presented to their input terminals.
 
homebrewed said:
Dead bug should be pretty good w/regard to leakage -- after all, even with a good guard ring PCB layout you're still dealing with leakage around the package pins. So that's about as good as you can get, given the limitations of the package. I _still_ don't understand why vendors sell opamps with FA-current-level inputs when they're immediately adjacent to VCC/VEE/output pins, what a stupid way to shoot themselves in the foot. Gotta wonder how many designs have been burnt by that.
Click to expand...
This is the whole reason I went for the 6268 single op-amp in the package in the SO8 size package. It gives (some) opportunity to sneak a tiny thin guard ring track between the pins.
homebrewed said:
An alternative to full-on dead-bug is to lift the critical package pin(s) and directly solder to them. I've done some designs using that approach. For a TIA that would be just one pin. My rant still applies :rolleyes:.
Click to expand...
In view of what you say, I think a sure-fire approach would be to set the chip down with one pin deliberately lifted up, and tack it to a very thin wire that drops into a guard ring. In my circuit, there is a 10pF capacitor between the diode and the input, which is part of my AC-coupled scheme to leave the bias current all in the diode. That could help in connecting to the pin. My circuit is not like the DC-coupled TIA circuits I see, but it seems to work in simulation.
homebrewed said:
FYI, the LTC6268/9 series have about the lowest input capacitance I've seen in a wideband FA-input-level amplifier...at least, for devices where the input capacitance is spelled out ( I suppose one could get an idea with SPICE simulations of the target amplifier that are designed to elicit that parameter...or maybe examine the SPICE model to see if it's spelled out). They're pretty good for low input-C, but if it's achieved using some type of bootstrap scheme they could be abnormally sensitive to the impedance presented to their input terminals.
Click to expand...
Would that be the same as "Miller" capacitance between drain and gate in FET amps? I don't know what is really in the model. The device could be like a microwave op-amp. I suspect the input capacitance is genuine.
 
Looking at the simplified schematic in the data sheet, it appears that indeed some of the input capacitance is bootstrapped away. If you look at the "cmos input buffer" you will see two signals referred to as "input replica" on the other side of some ESD protection diodes. The input replica signal forces the voltage on the other side of the diodes to follow the input, so the diodes' capacitance disappears. At least, up to some maximum frequency beyond which the buffer's phase shift causes the capacitance cancellation to become less and less effective.
 
RJSakowski said:
There is some danger associated with amassing quantities of Americium 241 so caution should be exercised.
Click to expand...
The NRC would take a real dim view, that's for sure. I don't think they want that to happen again.
 
K30 said:
Nuclear Boyscout
Click to expand...
Hi @K30.
This is a epic huge thread, and I never expect anyone to read it all through, but from memory, I know that we fully calculated what emanates from the very tiny amount of Am241 in a smoke detector, and what it can do to one. @RJSakowski has always been there for our reality backstop, and he helped at the time. The major radioactive energy from Am241 drives alpha particles, which are the things needed for smoke detectors. They are charged helium atoms nuclei, You would see them pinging away if you put them over dry ice, all under a bowl. (Wilson cloud chamber high school experiment).

Our XRF uses the remaining other by-product energy, this being a 60KeV gamma ray photon. The amount of Am241 in there is so small, it's quite hard to get enough of the photons to hit a material, get it to "glow" out some lower energy x-rays, and end up with a diode detector current above 40,000 electrons. We can safely take apart the ionization chambers, down to the little metal button with the smidge of oxide in it, and handle it normally, so long as we don't swallow them, or crush them up. The alpha radiation would not even get through a piece of paper!

The 1995 case of the David Hahn is well enough known. He was not using a few pellets of Am241 smoke sources for XRF. Instead, he bulk-bought possibly 200 for a dollar, and added those he managed to steal over a long time. He also added a startling quantity of radium from a bottle of paint used to make clock hands glow in the dark. He used nitric acid chemistry, and various procedures to concentrate his stash of precursors, including uranium (ore or cake), and lithium from batteries. He was intent on making an effective neutron source to use as a "gun" to make a small breeder reactor, hence letting it irradiate beryllium, and various other stuff. The Am241+ radium was a starting step to breed a better cocktail. Leaving his "ball" of reactants all wrapped up in aluminum foil and Duck tape in the trunk of his car for days on end, with the count rising every day, was a mistake! It was detectable at the end of the street! Even when he thought to separate the stuff, he was incredibly inept.

There is not the remotest prospect that our smoke-detector Am241 traces can do anything like that, even if you stack them together. I plan to use a circular array of 6 or 8, or as many as I can pack into a couple of inches circle.

Of course, the public and media misconception confuses, or lumps together what is the difference between exposure to radiation, and what is exposure to radioactive materials from tiny amounts getting onto or into our body tissues. We have evolved to live in and among quite high levels of the former.

I know its a fun graphic, so I include the PDF comparing "doses". I have no idea how anyone figured that you have to sleep with 10 people to rack up 0.5uSv of radiation dose. :) Somehow, it neglects what they would have received from you.
(It's a big image, you may need some zoom)

radiation.png
 
Last edited:
If you took that as anything more than tounge in cheek I apologize! I'm well aware of the low levels you're dealing with, and how little clue the general public actually has. I hold an NRC license to operate a reactor.
My first thought in this thread was that you'd get most your components just taking an old detector that was being surplused out of a power plant, but it appears theres not much luck to be had with the internal components after sitting on the shelf for 40 years. General Atomics will gladly make you new, for dumptrucks of cash it appears :rolleyes:

Hey! That's neat! I've never seen that graphic in Seivert before, only REM.
 
You must log in or register to reply here.
Back
Top