Ratemeter - should I buy one or make one?

[chrismb]

How easy is it to make a ratemeter?

I'm wondering because I put a bid on an ebay item that is a ratemeter with a variable tube voltage through to 1200V [good for scintillators];

http://cgi.ebay.co.uk/ws/eBayISAPI.dll? ... 0616173038

Seems to have attracted a number of bidders since, and I don't really like to get involved in bidding fights. (I like nice, clean, unfair ebay bidding where no-one else puts a bid in and I get it for next to nothing - as we all do, I am sure!!)

So I'm just wondering how complicated is it to make one, or what kinda price do folks here think this one worth to justify increasing my bid tomorrow before the sale ends? It looks like it'll go for more than GBP50, and that seems on the wrong price end of 'an ebay bargain'. Or maybe I am misjudging what could be a really valuable bit of kit.

What I was thinking was that I have various HV supplies so I could, perhaps, simply buy one of those cheap industrial rate meters that read up to 500Hz and I would put a decade divider between it and any future tube/PMT/whathaveyou I might acquire, so that I can switch the pulse rate to sub 500Hz so it can deal with the input signal.

Or do you have a better circuit I could build easily. [viz. less that 20 components!!]

[Doug Coulter]

Well, if what you're talking about is what I think -- just counting pulses vs time, just about any uP dev board will do that part easy, maybe with a 1 transistor or one opamp preamp/signal conditioner and pretty trivial software - they all have fast hardware counters built in. Of course my own favorite is the PIC stuff, as I've got that all tooled up. Using an 18lf2325 (28 pin dip), one bypass capacitor -- and whatever you want to use as display, well....there you are. I build mine with a max232 converter and rs232 cable to send data to a PC, and for standalone use a small LCD text display and a rotary/push encoder to have a bit of UI and some settings. Software for that is I think already up here (if not, I'll put it up), and you can add several counters (PICs have a few) and A/D channels to the data stream if desired. Our multigeiger code has all that stuff, but uses somewhat fancier hardware with a watch crystal for best timing accuracy. All or nearly all of that could be ported to the smaller PIC chips (easy until you run out of pins on the smaller ones), but I based that one on an 18f6520 because I had a bunch of them.

There are also TTL (and pretty sure CMOS) logic parts that are to-ten counters and 7 segment decoders for driving LED displays. Lots of parts, but a simple repeating circuit. If you want a true rate meter of course you have to gate the counter, and hold the last reading for display while the next one comes in. More useful in general in the lab *may* be a frequency counter, surely easy to find. Many have up to 10 second time gates, and you mainly need to clean up a signal to them so they count accurately - they tend to count each little wiggle/overshoot as a pulse. Of course, scaling to counts per minute is now something you have to do in your head.

[chrismb]

I tread a fine line between just buying what I think I need, and spending time building from basics. If I think I can buy 'value' then I tend to do so to avoid the hassles.

Yeah, digital-wise is easy - I was thinking just to run the output into a sound card, &c., but I just wanted something stand-alone, for ease of switch-on and watch. My experiments are tricky enough, without a whole other set of controls on something else to go pilot. You know the score!....

PICs and me don't really get on, you're an old hand at that I know, I'm a bit of a dunce and I've already learned far more electronics that I ever wanted to know (!) on my project.

There are some 10kHz ratemeter/counters from China for GBP14, which I could use, with a divider circuit helping out with pulse frequencies higher than that. I guess it is do-able, just a bit reluctant to get too involved in yet another side-project.

How much d'you think [how much would you pay, if you wanted it] that old ebay ratemeter/tube are worth?

[johnf]

Chris
for a nice linear type ratemeter a ADVFC32 chip from www.analog.com is linear over 6 decades of input and you can display on an analog meter just need a sig gen to calibrate
this is good to 500kHz
alternatively you could look for tachometer chips

[chrismb]

Thanks for the link, John.

[chrismb]

Hey, I just had a bit of a minor brainwave...

We all know that analogue dials are great for showing rates of change and rapid decadal variations that digital readouts just can't deal with. There again, a digital count is precise in a way that analogue isn't. For precisely this reason, 'plane instrument gauges give both to the pilot... I'm sure we all know this, 'cos we're engineers!...

But how about a ratemeter that is kinda a bit analog-y and a bit digital, and above all else dead cheap and easy to make.

My solution - that I will almost certainly try out if I get around to building my own scintillator: Put your pulsed output into a regular binary counter. Prob best to link two together. Then take each output and feed an led. .... that's it!!

Y'see, what you will get is a row of flashing lights, but there will be one which is nodding back and forth every second or so. It's quite easy for the eye to pick that kind of rate out from a bunch. Well, the one that is flashing around that rate will indicate a rough value for the count rate. Of course, reading it will take a little practice, but sounds like it'd take a lot less time to practice the skill of reading that scale than building something too complicated.

What we end up with is not only nether analogue or digital, but it is also a counter as well as an 'analogue ratemeter'. Just hit the 'freeze' button, and you have a binary readout of total count. To assist the eye, a timed led could also be set to flash nearby and so it might be easier to see and compare that rate with whichever one in the led chain is flashing at that rate.

Do you guys like the idea, or is it a duffer?

[Doug Coulter]

Well, it's certainly interesting, and in fact I've even seen it done in the old decimal counters in similar fashion, in this case, one neon bulb per number. Yes, you could get a feel for rates just glancing at it. Trying to do this with leds is going to give one a lot of noise problems -- the current spikes tend to add up, and that was a real bear with the early led-driving bargraph chips.
The lack of a way to have accurate gate timing is a flaw indeed, however. For counting say, an activation you need a succession of correctly timed counts so you can curve fit and "pre-traploate" back to what the count would have been when you powered off the neutron source -- you know you've missed some decay time turning off the gear and moving the active sample to the counter. A timestamping thing tells you how much time you lost doing that, if for no other reason than that it was counting away when the power was on too from radiation the device was making -- you can see the count drop to zero, then resume at the activated sample rate when you do the move from the oven to the counter. But since you've been taking readings right along, you also can know precisely how long that took.

It's a heck of a lot of parts and complexity -- order or two of magnitude more than a PIC, one bypass cap, and rs-232 converter (and its caps), which gives gate accuracy, time stamping, and possibly a bunch of other features -- the designer can put in whatever is wanted, including a pwm output to drive an analog meter (I love those things..paint the needles with day-glo paint and put some UV on them to read them in the dark too) -- and it could even be a log function, for wider dynamic range shown on the meter than usual. You could do log->analog meter all analog as well, a couple people make the chips for that function, and that's a lot (heck of a lot) less total parts. You just have to threshold and pulse width normalize your detector input (eg comparator and one shot) then low pass filter the result to get volts proportional to count rate. This would work for all but fast phototubes, unless you had a range switch to shorten/lengthen the one-shot pulse width depending on count rate. This has been done, of course, and you can see a copy of one in Carl W's nifty demo of 3He tubes, in that video he made. His unit had a special meter as well, that had a lot more degrees of pointer deflection available than normal -- a lot of "cool" factor there.

For my uses, I need the timestamps as I need to know how long from power down to first count to really get any kind of accuracy on how activated a short life (silver) sample was right at power down, as well as *accurate* gate times I can't get with my hand on a pushbutton.

I am also a huge fan of simply hooking detector signals to an audio amp, perhaps with a bit of pulse stretching first so there's enough audio power to hear. The ear seems awfully good at this, and you can have more than one going at a time. This helps to notice correlation (such as when the thing under measurement is bursting, or you're counting EMI instead of real counts) that the eye can't do so easily. A multichannel scope is also very good for relative measurements, and in fact I'm using mine for that too -- a couple detectors, an EMI antenna, and a faraday probe in the tank all at once and time-aligned on the screen is pretty nice. PC software can capture those scope screens for later examination as well (free with the scope, or free on linux anyway, which supports most DSO's directly). Further, most of my DSO's will count pulses in a known time base for you if you like (frequency counter built in).

But the audio amp is both super simple and has one huge advantage -- you don't have to be looking at it so your eyes are free for other things. And it's as real-time as it gets. Another thing it can tell you (if you don't over stretch pulses) is when a burst of them all come in at once -- more bass in the sound! I use a little microphone transformer in my lashup to get rid of ground loop noise and stretch the pulses "just enough" due to the transformer self resonance. Works quite nicely. The kids were impressed to see (hear?) that in our recent demo for them.