Unusual functions of cheap parts

Am Mon, 28 Nov 2005 22:14:17 GMT schrieb Rich the Newsgroup Wacko :

One time I used an old Iron as a dummy-load for a 230V/1kW TRIAC power control circuit (we had it in the lab for improvised BGA soldering). To "satisfy" the thermostat I used a 30cm room fan.

I once owned a Knight transistorized amp that used incandescent bulbs in the output stage to limit current. When you had some brighteness, you had a problem.

I remember that it was the very worst-sounding amplifier that I ever owned.

jon

Heh heh. I have a Knight kit-built amp that glows too, but that's a bias problem in the tube output...

...No, I don't use it regularly...

Tim

-- Deep Fryer: a very philosophical monk. Website:

Maybe 2N4261 ? Have not looked into the datasheet, but as i remember, i have seen them sometimes in complementary with the 2N2369

Jorgen

I've always been led to believe that this wasn't a bug, but a feature. Really! (E.g., you can often get away with one diode when you'd otherwise need two if the things actually recovered quickly...)

Hi Jim -

And on what delay timescale it works?

regards - Henry

"RST Engineering" schrieb im Newsbeitrag news: snipped-for-privacy@corp.supernews.com...

BA682.

A step-recovery ("snap") diode works on the principle of stored charge in the diode. During the forward biased half of the AC waveform, the diode is a very low impedance and it stores excess charge; during the reverse biased half of the waveform, the diode remains a low impedance until the stored charge is depleted, at which time the diode "snaps" into high impedance. This snap acts much like a spark-gap transmitter, in that a tremendous number of higher order harmonics are generated. In general (and there are ways to enhance this), the power available from any harmonic is around 1/n * Pin, where n is the order of the harmonic and Pin is the RF power input to the diode.

Biasing the diode simply varies the point on the reverse cycle of the AC waveform where the diode snaps. For maximum power, you try to get the diode to snap at the peak of the waveform. However, by varying the diode bias, you can get it to snap before or after the peak of the waveform. Generally you can get it to snap plus or minus about 30 degrees about the peak before the snap action degrades.

Jim

the problem is, that it is an rf-transistor and can't be driven at

If I simulate with the 2N3906, the frequency response is not worse than with the 2N2894A.

mfg. Winfried

National Semi's (now Fairchild) 2n5771 was a gold-doped PNP. ft>=850MHz. For avalanche mode one might try the lower-Vce-rated PN3640 (12v), or PN3639 (6v).

I might even have notes on this. I tested/compared various BJTs in avalanche mode some years ago, trying to find the "best." ISTR picking the 2n2369, both because it was fast, and because it avalanched reliably where other types wouldn't.

James Arthur

Hi, James,

Interestingly, the best avalanchers aren't usually super-fast transistors, but old klunky things. The Zetex avalanche transistors have lowish Ft's and are made in Russia, maybe on an old process.

John

Howdy John, I was unclear: by "...it was fast..." I meant the 2n2369 was one of the devices with the fastest avalanche edges.

Digging through some of my notes, I don't see the BJT comparison, but a 2n2222 biased to +100Vce, banged / triggered by a 74HC-series gate, gave synchronous 750pS risetime pulses. Not very impressive, really, though good for higher-power stuff than I needed.

Interestingly, I found a 74AC00 driving an MPS2369 was faster & less trouble: 360pS fall (turn on) time, & 570pS rise (turn off) time, and no nasty high voltage supplies. It was possibly even a little faster than measured--at 360pS I was pushing my poor little 7S14 1-GHz sampling plug-in pretty hard.

Best, James

That was a high-frequency part for the time, spec'd at 1200MHz...

An old Raytheon datasheet says the 2N2894 was doped with platinum.

BTW -- in AoE, we list the 2n5771 as a PNP complement to the NPN 2n5769, both 15V plastic versions of older metal-can parts.

Thank you Jim for your longly explanations. I already knew the charge storage process, but the phasing aspect was new and interesting. My question about phase delay was in another direction. To be concrete: How to delay (=phase shift) a 145MegHz signal (mostly sinus waveform) with a snap diode? After reading your explanation I cannot see how to achieve a non-snapping action here. Maybe that would work with the diode if you modulate it with dc current getting delay in the ps timescale. Another question would be if it possible with the snap diode to make a power amp in some form of ringing oscillator. Of course, it should be modulable at least with FM.

- Henry

"RST Engineering" schrieb im Newsbeitrag news: snipped-for-privacy@corp.supernews.com...

is

biased

diode

Generally

before

Thanks for tracking that down, Win! Gold in a PNP was certainly troubling my ancient remembrance of semiconductor chemistry.

...Jim Thompson

Very interesting. The 1N4007 seem to be very versatile devices. They are available with a SOD-57 glass envelope, too (1N4007G?). These are fairly well photoconductive. When illuminated by a high efficiency IR LED (HSDL-4230 or so) current transfer ratios of 0.001 can be achieved. Not too much, but with two LEDs 100uA of photocurrent is obtainable. This is OK for a pass element in an "electrostatic" power supply for e.g. electron or ion lens systems.

A high-voltage optocoupler; cool.

I've posted a schematic for a hv opamp (400 v p-p) that uses two optoisolators as the output push-pull stage... it's very cheap and simple. A higher-voltage photodetector, like a glass power diode, sounds useful, too.

I worked once with a company in Southern California that had a neat gadget: it was a truncated cone of silicon with gold contacts on the base and the flattened apex. It would stand off something like 5KV until you whacked it from above with a laser, illuminating all the sides of the cone, whence it would conduct hard. I think they went out of business, though; it was pretty obscure.

John

While many companies* are making 1n4007G glass-passivated diodes, it appears they all cover the glass with plastic. I wonder... where one can get a 1n4007 with an all-glass package these days?

• Including unusual semiconductor manufacturers, like: Won-Top, Bytesonic, Leshan Radio, Formosa Microsemi, Gulf Semiconductor, Dachang Electronic, Goodwork Semiconductor, etc.

I've got a few 1N4003 and 1N4004 diodes with all-glass packages but I'm not sure how recent they are. They have the (older?) Fairchild logo of an italicized "F" with the middle stroke extending on both sides of the vertical. I'm about 90% sure these came in one of those "20 rectifiers for $2" packages from Rat Shock, so who knows how old they really are. The bodies are about 2.5 mm diameter by 4 mm long. Inside the glass, the ends near the leads are orange, with a clear strip less than 0.5 mm wide near the middle.

Testing them with the "diode check" on a $40 multimeter and either a 40 watt clear globe lamp or a TV remote control doesn't show much photoconductivity, but I suspect I would need to look a little harder than this to see it.

Matt Roberds

How many do you need, and can you use ones pulled from PC boards? I may have some left that I pulled from damaged boards.