-
Welcome back Guest! Did you know you can mentor other members here at H-M? If not, please check out our Relaunch of Hobby Machinist Mentoring Program!
Needing more than a spark test?
- Thread starter graham-xrf
- Start date
Sort of an "oh pshaw" moment yesterday. I pulled the LMC662 off my pocketgeiger to replace it with a LTC6269 to lower the noise floor of the thing when I (belatedly) discovered that the replacement has a different footprint! Aargh.
I didn't purchase the LTC6269 for this project but for something else so didn't think about that when I started modding the pocketgeiger. Time to order the right one....
I didn't purchase the LTC6269 for this project but for something else so didn't think about that when I started modding the pocketgeiger. Time to order the right one....
Unfortunately, the LTC6269 isn't available in the SO-8 package. Making and installing an adapter board would be a pain and pretty messy so I did some browsing through Analog.com's list of low-noise amplifiers to find a reasonable substitute, and I think I found one. It's not near as fast as the LTC6269, but since the pocketgeiger circuit is designed to be slow as a turtle anyway I don't think that is a big issue. It's the AD8646, and, noise-wise on the low frequency end it looks better than the 6269 (and far better than the LMC662). It isn't in my LTSpice library but there is a spice macro-model available for it. Simulations looked good. Plus it actually is cheaper than the LTC6269, and DigiKey has it in stock!
I did realize that the pinouts were the same, but that the package sizes were different.
Just look on eBay for MSOP circuit boards. There are all manner of little adapter boards to convert to "bigger" formats, though I suspect the conversions are usually up to DIP size, but I haven't looked hard yet. I had not planned on using the PocketGeiger PCB unless I could conveniently hack it.
Points about using the LTC6269.
It is not just about the lower input noise figure. I chose the speed, in combination with the input noise characteristics, the input offset voltage, the input current and gain, used in conjunction with (some) bias in the PIN diode. The bandwidth was of the greatest importance!
The general design aim was to use a gain in the first stage sufficient to put put the signal/noise ratio beyond the point it could be assaulted by anything we would get up to downstream. 30dB would be more than enough. I went for 40dB to 50dB, trying to pile on as much gain as possile in that stage short of compromising the bandwidth. The gain in the next stage would then multiply that.
The LTC6269, it has sufficient gain-bandwidth product to deliver this while reliably amplifying a radiation pulse detection event. The multiplication gain required to deliver into A/D conversion, also provides for drivers, filters, etc. as needed. This can be provided in stage(s) without hurting the noise figure. It did seem to me that the LTC6269 could nearly do the whole job by itself, but I was prepared to use one more package (LT1807CS8)
as necessary, to interface with ADCs. Some had built-in differential drivers, others not. I wanted that the design did not have stored energy overshootsleading to all the games to re-establish q measurement base line
- - - - - - - - - - - - - -
I get it that one wants to be able to try and use the PocketGeiger board, and from there comes the motivation to find a op-amp in a package that fits. What I could not do is use a poorer op-amp, on the grounds it has the bigger package, and compromise the rest of the design. The rest of the comparators on the PocketGeiger board do not convert to op-amps layouts easily anyway, so I did not seek to use them in the longer term. To use the footprint, if at all you can, seek an op-amp with the high speed.
Indeed, if you can find a speedy op-amp in SO8, with reasonable low noise, I would try to hack the PocketGeiger.
Right now, I am viewing the A/D converter with similar concerns. I would love to have it that almost any enthusiast from HM would have a reasonable chance to DIY with enough instructions and a PCB source, and I reckon packages with 0.5mm pin pitch are not helpful!
I know I got side-tracked into other parts of the design, but I can report that the web page interface to smartphone worked. The (early trivial test) was seeing the Raspberry Pi4 program displaying a GPIO status changing as I was meddling with pins - but displayed on the phone laying about nearby, because the phone can see the WiFi, and hence the Raspberry Pi's XRF web page display. Of course, one will be able to interact with the web page, just like one can with any, but here, the Pi4 would be accepting information from the connected user.
Agreed - but also, this thread is, during the crucial discovery phase, mostly a open design discussion. I will try and post more pics, but hopefully of meaningful stuff. I am not a fan of "unboxing" videos, but for some stills, I think it's OK, and agree they can mark progress. The last picture I posted on Saturday. if zoomed up, does cause us to swallow hard! Mark is right. Maybe we have to look for something of realistic size.
Before I start trading off on the performance, I did want to try and push the limits of resolution and accuracy, but I am well aware that actual experience of X-ray detection events might well involve a re-spin of the front end. I am hopeful that we won't encounter some of the stuff that drove the development of the PocketGeiger to places I didn't like. I do have, in the back, my PMT tube and scintillator resource, which is fun, but I think not for the average HM member hoping to put together a little kit, turn up a shield, adapt and construct some sort of casing, download some software, and reach for his phone.
With a different front-end circuit, the PMT tube can use the same samplers and software.
You can bet that if this gadget starts delivering XRF identification spectra, it will feed a new thread, presenting the outcome.
Whoops! Can one get a LTC6269 in a bigger package?
By any chance, have you come across these guys - 5 PCBs for $2.00
--> JLCPCB
They also supply boards with the components fitted.
Last edited:
Yup, I'm using JLCPCB. FYI, in my experience if you request some colors other than the defaults your PCB may take a couple more days to be fabbed. I haven't used their board-stuffing service yet because the components I've used so far can be hand soldered. A friend of mine is using laser-cut stencils in a 3D printed holder to put down solder paste to solder down his parts and reports good results. If you get Gerbers from your PCB design S/W you can send the solder mask layers to folks who will make you a stencil for not a lot of money.
Unfortunately the LTC6269 doesn't appear to be available in the SOIC-8 package so I couldn't use it. However, the AD8646 is pretty close in terms of its noise performance, and actually is noticeably better when it comes to flicker noise -- in the .1-10Hz band the LTC6269 is spec'd to have in the neighborhood of 13uVp-p noise while the AD8646 outputs 2.3uV. The LTC6269 clearly is meant for high speed applications because the voltage noise specification is for 100KHz. Their noise current specs are about the same, in the fA region.
Amplifiers with BJT front ends can get down to a few nV/sqrt(hz) or less, and tens of nV noise in the .1-10Hz region but their noise currents can't compete with JFETs. However, if you're building a TIA with Rf's in the 1Mohm region that shouldn't be an issue. Amplifier GBW may be the main issue when it comes to choices there.
I got some AD8646 units and installed one on the pocketgeiger (they were delivered yesterday). My initial evaluation suggests that the noise floor isn't much better than the LMC662 so most of the noise probably is coming from the detector itself. The bias voltage to the detector is filtered by a 100K/1uF low-pass filter, which would appear to be pretty good (with a 10Hz cutoff frequency) but it's just a 1 pole filter. Circuit gain is very high so it's possible that some switching noise is present in the output. The bias voltage is brought out to a pad on the back of the board so it would be fairly easy to add more C if need be. I guess I need to try that.
I had removed the copper shield around the detector and amplifier so I could swap out the amplifier and voltage regulator, so I reestablished the shielding by using a combination of copper foil and aluminum guilding. Installing the copper foil was easy, the aluminum guilding was anything but. It took several tries to discover the various things that could (and did) go wrong -- all the way from cutting the guilding to size without turning it into a wad of useless foil, to figuring out how to tape it down in place. It's incredibly easy to tear, and is attracted to even the faintest electrically charged surface or object. I didn't want to use double-sticky tape or zip-grip because those are nonconductive, and it DID need to have good electrical connection to circuit ground, so for temporary use I was going with tape. Inside a lead box this may be a non-issue -- we shall see!
A lot of the noise coming out the the modified pocketgeiger is higher-frequency so another option would be to add a low-pass filter block to my signal conditioning board. The Teensy Audio library has a 4-pole state-variable filter so that's yet another possibility, plus it would be a "piece of pie" to change the filter parameters (quote taken from the movie Space Odyssey 2010). I'd want to implement a Butterworth filter characteristic to get the least amount of phase distortion. Come to think of it, the audio library also has an FFT function so I could check for systematic noise, too....or to help tweak the filter characteristics.