Miniature GM detector, part I

First thing first. I tested the GM tubes from yesterday, and the results are more or less as expected. The huge tube, SI-22G, is really sensitive to the background, and it was nice to have it sitting on my desk and clicking. The STS-5, that one in the middle, was more or less on par with the SBM-20 I am using for my detector. The tiny one, SBM-21 does not seem to care about background much. I did not have any active source at hand, but I believe the results will be different when actually measuring something hot. The tiny tube is simply too small, compared to the monstrous SI-22G, and the chances that a particle will hit such a small target are simply... small.

I started to work on the miniature version of the GM detector, and I cloned the HV supply from my very first design, using the same parts, and duplicating the capacitors to achieve similar 

capacity as the original design (original caps are 18nF, these small blue ones are 10nF only). The results are positive - without load, the HV supply draws less than 0.5mA from 9V battery, and 0.25mA from 4.2LiPol 18650 accumulator. I am not sure what was the trick, though. The basic schematic comes from the Geiger Counter Circuits, the second circuit. Picture above is the component side, the picture on the left is the bottom side. I used SMD parts for pretty much all I could. Transistors are SMBT3904, 3906 and SMBTA42. I believe the transistor selection is vital, the SMBT39xx have relatively high hfe. Inductor uses cca 100-200 turns on ferrite core (hand wound). I don't remember what is exact inductivity, but it was pretty high when I measured it before, around or over 50mH. My impression is that bigger is better here. The ferrite core bears marks "SEM TM12". It's either coming from Pollin (part of the mixed inductor sale), or from local shop. Pollin is more likely.

The power supply seems to be relatively 'hard' - measuring the voltage with common 10MOhm multimeter shows 270V, while my previous constructions usually did not get over 150V. I still did not add the GM tube and clicker, that's left for tomorrow, or more likely after my vacation.

GitHub! (and fixing yet another old LCD)

Let me start with a picture today again. I swear I am not making it up!
This morning, my wife attempted to start her PC, and after a while, she said "Well, I don't want to sound stupid - but my monitor died." Let me remember that few days ago, I fixed two old FSC P19-2 LCD monitors. My wife uses the third one - and obviously, its time has come. I quickly swapped my wife's monitor for the one I fixed already, and went to work.
After arriving hope, I opened the monitor and discovered another failed Capxon, this time 100uF/25V. The top was bulged slightly, but as you can see on the picture, it was pretty much done. Luckily, I found some Samxon KM capacitors that seem to do the trick for the moment, and I already ordered some Rubycon and Panasonic ones.

The second picture shows the GM tubes that arrived from Bulgary recently. The middle one is STS-5, an equivalent of SBM-20. That lovely little tube is SBM-21 I want to use for pocket GM I could carry with me on vacation to Krkonose. The big one is SI-22G to be used for background measurement.
Last thing, the one that took me most of my evening and annoyed and frustrated me almost to death. As you might know, Google will shut down Google Code and suggests to migrate to GitHub. I am not going to rant here about Git - I just don't like it, I find it confusing and overly complicated for my purposes, and I am worried I'll hit the wall using it. But... well, that's the way it is. I had to manually import my Google Code repositories, as the default import didn't work the way I needed, and I already managed to mess the repository by a single click to such a degree I almost needed to import it from Google code again... I'll give it a try and see if I shall stay with it, or find another SVN service. Anyway, a hint for all who need to use Git - don't waste your time, and go and get TortoiseGit. It makes the whole process of using Git a bit less painful.

How much solar power II

I realized I did not post any pictures recently. In order to keep the blog more entertaining, this is my evening creation. I know, it's nothing, but at least a picture!
It's Arduino Uno with Ethernet shield (I bought two of them years ago and used them just a few times for testing, I usually use ENC28J60). I added two power 5.6Ohm resistors in series as a dummy load, and I hacked together a few demos to post the results of the solar panel performance on plot.ly. The graph shows voltage and wattage, the time is shown as number seconds after this evening. Not the best scaling, but it serves the purpose of giving me a better picture of what's going on, and I can monitor the results in realtime from work.
As for today's solar panel performance results - they're both promising and disappointing. After I moved the panel a little to face the morning sun, the wattage jumped to 3.73W. That was at 9:28 before I went to work. The power peaked at cca 4.5W, and changed with the clouds covering the Sun and disappearing again. The last usable reading arrived before 3PM, and then the power begun to decline. A rough guess is I was able to generate some 20Wh. It's not bad, although I'd expect the power to peak higher - those are three 5W panels! However, the position isn't optimal, and especially in winter, the building at the opposite side of the yard casts a shadow on our balcony in the afternoon, that's probably that sharp decline of power at 3PM.
I found some info on proper tilting the panels, and mine are obviously not optimally tilted. For bigger panels I plan, I'll need to come with better way to tilt them, and being able to adjust it over the year. Ideally, they should also trace the Sun, but that would be too complicated.
New goodies arrived - the GM tubes from Bulgary! Pictures tomorrow.

How much solar power?

The problem with my solar panels and experiments is that I have no clue how much power I will be able to generate. Theoretically, I should get 15W on ideal day, with panels properly set. I would expect much less later in the afternoon on a winter day, but so far, I am getting something like a few mW only. I can assume that part of the problem are less than ideal light conditions, not really optimal panels angle, and the fact I am at work during the day, thus all I can observe are results on late afternoon.
To overcome at least the last problem, I hacked together a quick measurement unit - just one of the RF24 nodes I have laying around. The only thing the modul does is measuring the voltage of the panel (bridged with 470 Ohm resistor), and report it to the RasPi that takes care about logging it into file. That way, I should see if the panels are worth anything.
That's pretty much all I've done today, I am afraid.
I received a nice box of goodies, I did not open it but I assume it's those old fans and heatsinks. I overdid it this time for sure, and I will end up with something like 20kg of Alu heatsinks. I am planning to use some of them for LED lights, and I wasn't careful when bidding on eBay - oh well, I can still sell them :)

No solar fun on this rainy day

Lazy Sunday here. I slept long, really long. My wife's bed is always so warm when she leaves it that I cannot resist to sneak under her blanket, and this time I fell asleep and woke up at 10AM or even later!
The Sunday wasn't only lazy, but also rainy and dark, not really the day you need when you're playing with solar panels. The ones installed on the balcony generated just a few milliwatts, not worth doing anything with it. I used the LTC3525 I received yesterday, and after a few experiments with coils, I made it work. The mistake was to use the small SMD coil I had at hand, but using bigger, 47uH one helped. The whole setup consists of the 2V solar cell, Schottky diode, 22F goldcap and the LTC chip. Unfortunately the goldcap was drained to 0.7V so it took some time for it to get charged. The power generated by the solar cell on this rainy afternoon was barely able to keep the LTC running without any load, but attaching an Arduino to it was definitely pushing it over the limit, and the voltage begun to drop steadily, mV by mV. To be honest, the Arduino I attached to it wasn't low power at all, it uses LP2981 LDO with something like 100uA of quiescent current, and red power LED with cca 1mA consumption, so I am not that surprised. Let's redo the test with fully charged goldcap, and one of the low power nodes I have.
I modded the meteo node a little. The DHT22 and BMP180 sensors are now powered from Arduino pin, thus I can control their power. The program was changed to not enable the sensors for battery (goldcap) voltages under 3.3V, but the program still needs some tweaking.
While tidying up my desk, I found the 24x2 LCD I bought recently. I had the idea of converting it to a Matrix Orbital compatible display that could be used with Winamp, thus I quickly hacked Arduino to it - just to find out that there's no working plugin! I found "LCDPlugin", but that one is pretty old, and reports that the COM port used for display cannot be accessed. It does work sometimes, but the chances of it working are around 10%. The other plugin is "LCD Smartie", but I don't like the idea of having that program running all the time. As Winamp seems to be dead (the official Winamp page didn't change for years, and just promises something new), I assume it's dead and chances of finding working LCD plugin are close to zero... well, I wonder if there are other players capable of all that Winamp could do, with LCD support. I tried foobar before, but I wasn't impressed at all.

Solar panel installation

Not much done today, and no picture unfortunately for reasons explained below.
I was thinking about the best way to put the three small solar panels on my balcony. There were some options, and I decided to let the panels hang on the outer side of railings. I am not sure if that's 'kosher', as the local landlord seems to complain about things like satellite dishes - we'll see. After all, this is the land of alternative energy, solar panels are everywhere, so why not on my balcony! I had to rush to the shop to get some nuts and bolts, and I got cheese and anchovy as well while I was there.
I wired the three panels in parallel. I am still not sure if that's smart, but I'll see, and rewiring them shouldn't be that complicated if necessary. The problem is that those panels are 9V ones, which is sort of a strange voltage for common SLA accumulators. I have both 6V and 12V ones.
When I finished, and put the panels on the balcony, it was already dark, and raining, thus no pictures today.
The meteo node (powered from solar/goldcap combo) died at 4AM, more or less as I expected. The latest reported voltage was 2.71V, but the values from the DHT22 sensor were obviously wrong, and I think I should stop trusting the values much earlier. Let's declare the minimal acceptable voltage as 2.9V. At this level, the voltage begun to drop quickly, and the humidity reported by DHT22 went to the roof. I assume that the sensor simply went to weeds, and consumed all the power it could get before RF24 module gave up :)
I received the long expected LTC3525 chips today - I'll give them a try tomorrow, and also will try pairing it with that 22F goldcap, and see how long a node will run off it!

Fixing old LCD monitors - Fujitsu Siemens P19-2

I realized I'll need a secondary LCD monitor for my vacation. I had two FSC P19-2 LCD monitors in my basement, as I was planning on fixing them one day, and the day arrived :) Those monitors were really a high end ones when I bought them (okay, high end consumer ones), sporting PVA panel and pretty good speed and size. They're 1280x1024, 19 inches.
The first one did not start properly, and if it did start, it switched off itself soon again. It dies last year, so I still remembered what was wrong with it. The second one was sitting in the basement for a bit longer, and I just barely recall the backlight died, although I am not completely sure. Anyway, after watching Dave Jones' video on fixing the LCD monitors, I felt confident I'll be able to fix at least one.
Opening it isn't that difficult. A nice guide is on Diit, however after figuring out that the two important screws are hidden under the white rubber plugs, it did not take more than a few minutes. Just demount the monitor stand first (four screws), then the two under the white rubber plugs, and then gently pry the back cover apart. The electronic is hidden under the metal sheet held by three screws around, two screws near the power socket, and the four hexagonal nuts (or how they are called?) of the DVI and VGA connector. Then slide the metal cover off the holders, and that's it.
The source of the problem can be easily spotted - see that green capacitor on the left side of the picture above, close to the heatsink? It's bulged - not much, but it is. More detailed picture on the left. All I needed to do is replacing it with similar, but working capacitor. I didn't have any new one at hand, thus I used an older capacitor I recovered from an old Pentium mainboard - and, well, that was all. After putting the LCD together again, it works, at least for now. The original bulged cap showed no capacity at all!
By the way - I used the same method for fixing our late satellite Topfield receiver. Again, replacing the bulged cap with an old one recovered from old mainboard was sufficient. If you happen to have some old mainboard from the era before those capacitor problems, don't throw them away, keep the caps! And the coils, and maybe some other stuff - heatsinks, PS/2 connectors, resettable fuses...
The second LCD (again, P19-2) was a mystery. It worked when I tried it today, although I am sure it didn't work before. The cap in the power supply that I expected to be bulged was okay. I replaced it just to be sure, and also replaced another, smaller cap nearby that looked bulged, but I did not have adequate low ESR replacement, so I'll order some and will fix it once for all.
That's pretty much for today - oh wait, three small things:
 - the pre-WW2 vase with uranium-oxide paint arrived today. It looks nice, and is not as hot as I expected, showing something around 2000CPM (1mR/h). However, because of bigger active areas, there's a measurably increased radioactivity even some 20cm far from the vase.
 - we had partial Sun eclipse here. I watched it for a while, and although it doesn't compare to the total eclipse I saw in 1999, it was still nice.
 - the meteo node powered by a solar cell and 1F goldcap is still up and kicking. I am using RasPi to monitor it, and despite the fact the node isn't power optimized yet, and reports every 8 seconds, it's still at 3.47V (almost midnight here). The voltage drops by 10mV every 3-4 minutes and I expect it to die at around 3-4AM. With some optimizations, I am pretty sure it can make it through the night! (an afterthought - in emergency situations like when the voltage drops to dangerous levels, what about sending the meteo packet less often?)
Last but not least - I'll rework the sensor packet definition to give bigger range of voltages. Maybe one bit flag, specifying either 10mV precision as being used by now, or something like 50mV usable for higher voltages. Actually, that would be backward compatible!