Crystal Types

AT-26 is a cylinderical package designation for small AT cut crystals. Note the minimum frequency for AT-26 is 6MHz.

However, something is wrong here. I'm seeing 60KHz crystals for sale on eBay and Aliexpress in AT26 packages as "quarts column resonators". My guess(tm) is that a "quartz column resonator" is half of a tuning fork, crammed into an AT-26 package. Other than the for sale items, I can't find any specs on such a device.

I've seen quartz crystals in the audio range, flexure mode of a slab.

Useta be, in the post-WWII military surplus shops, you could buy crystals down in the low kHz, and package styles even for ham and CB radios followe d military designations. I remember pulling and regrinding FT243 rocks to g et frequencies that weren't commercially available. You had to make them sl ightly convex, but not too much.

These days it seems each mfgr and vendor has their own idea about package nomenclature.

Mark L. Fergerson

Yup. And there's a lot of packages out there that have different names, but will go onto the same PCB pads. It's frustrating.

From the brief web search I just did I'm guessing it's a package style -- the page I hit

Why is that frustrating? Being able to use the same footprint for multiple devices is a good thing!

Yeah, I found this page that shows several similar packages where the name seems to be related to the dimensions.

Yes, but will it fit inside an AT-26 case, which is 2mm dia by 6mm long? I don't think so.

Back to AT-26. I think this is the relevent patent which shows the rod like column resonator: "Mounting for rod-like crystal oscillators"

This is a great clip on the subject. These old military films are so clear and precise (in the days before the dumbing down of the language, obviously).

And I do like a man in a uniform. ;-)

The one I saw was inside an octal tube jug, all visible. I'm told you could see them visibly blur when oscillating.

This is pretty:

like a little piece of jewelry. It's glass, vacuum inside, AT-cut, turning-point temperature noted on the bottom.

Having to plow through data sheets to see if a device from manufacturer A is compatible with pads designed for manufacturer B's device is frustrating. Particularly in a world where a SOT-23 is a SOT-23, so you know it can be done right.

That's not what you said. You complained about multiple names for the same package. An AT26 is an AT26 in the crystal world. What you said would be like the ?SOP, MSOP and ?MAX all having the same footprint, or the TQFP and the VQFP having the same footprint. Sometimes it seems like every FPGA maker uses different names for BGA packages including their "generic" name and their proprietary name, both different from other manufacturers. There are many other examples in the transistor domain as well. So this issue is not limited to the crystal makers.

I must be old- I think I watched that film in Tech School in 1971.

That's what I wore, but with fewer stripes.

Mark L. Fergerson

Except when they're not.

Put in a RT9818 recently. (Or maybe that wasn't the one, but it was a something.) "SOT-23-3" on the datasheet and on Digikey.

Flagged by DFM. Whaaa?

Turns out they use a slightly larger package. After a quick correction to the footprint, sent off for proto. Perfect solder joints first time.

A lot of Japanese parts get around it, I think, by calling it something meaningless like "SOP-3". Of which, the Panasonic ones I think are usually SOT-23 compliant after all, but YMMV.

Tim

ly

I was given some old handheld barcode scanners to play with. They have so me weird 4-pin and 8-pin SMD chips that don't match any outlines I've ever seen, with some leads fatter than others (not the middle pars either like o ld-fangled power-handlers). I suspect they're proprietary as their markings don't show up e. g. on the SMD Codebook web page:

By the way; the 3-pin laser diode packages are unmarked and the driver ci rcuits are... weird. I know that LEDs can be used as photodiodes and was wo ndering if laser diodes can too. My idea is to shine light into the package aperture and read induced voltage on the pins, then drive them oppositely to see which way lights up the laser. If I cook the built-in photodiode I w on't care...

(I mention this here because alt.lasers is moribund and there don't seem to be any other active laser-specific groups.)

Most of the rest of the bits (SMD components I *can* read, connectors etc .) are salvageable, but I'll have to pitch the chips as unusable.

Mark L. Fergerson

Welllll... what do they connect to? Trace the circuit a bit? If they're doing power stuff, that should be pretty obvious (caps, inductors, fat traces around), or signal processing (weird op-amps?), or driving (motors, galvos, lasers), or etc. There are limits to how much you can glean from a cursory glance (I wouldn't expect a mixer to be very obvious), but maybe it's something.

If it's a multilayer board, the "cursory" information is even more limited, unfortunately...

Hmm, certainly oughta...

It's not the wrong ways 'round, though. Same ways. Think of it this way: a diode generates voltage up to a level limited by its own current draw. Photocurrent is balanced by forward current. If it were backwards, what would the voltage be set by? It could build up to nearly breakdown voltage, or, well, limited by nu*h/e anyway. But that's not the observation: solar cells don't depend on the light energy (above a minimum threshold, the bandgap), the excess is absorbed, and the practical maximum output voltage is determined by Vf.

Also means the V(I) curve of a solar cell is more-or-less the V(I) curve of the diode, which is why peak power point is a modest fraction of Vf, and not simply half, as a resistor would give.

And you reverse-bias a photodiode to suck all the photocurrent charge carriers out of the junction, and reduce junction capacitance. You're putting work into it, this time (the light is dissipated as heat, plus the external power supply delivers power to the leakage current through the junction): a fair sacrifice for high bandwidth.

And you... nevermind, I've repeated enough photofacts for one night!...

Tim

Perhaps it was an SC-59(TO-236) package? I can't tell the difference between SOT-23 and SC-59 without my calipers:

er

One brand of scanner has one pin of the diode connected to a proprietary Metrologic chip (MLPN26319) that doesn't seem to have ever had published da tasheets and the other to the OUT pin of a plain old TL555C.

The other brand's diode connects to one of the unidentifiable 8-pin SMD c hips I mentioned earlier and to an AMI(!) microcontroller, also with no ava ilable datasheet.

Fairly clearly the first unit drives the diode with the 555 and the other with the SMD chip, now that I'm not too tired to think about it. At least n ow I can figure the driving polarity, voltage, and current of one of the di odes from what the 555 can deliver.

d,

Nah, only two-sided, easy peasy. These things are practically antiques am ong scanners.

was

Whew!

: a

D'oh! I plead posting while tired- don't know what I was thinking.

ge,

of

not

I don't mind at all- apparently I could use the refresher.

Mark L. Fergerson