how can people post things this stupid?

Naaaah! You have to admit, PALwHUB does attract flies better than anyone else here >:-}

Larkin can't handle such big words... he's just your ordinary everyday pompous asshole. ...Jim Thompson

On a sunny day (Sat, 01 Mar 2014 13:08:18 -0500) it happened snipped-for-privacy@attt.bizz wrote in :

Like I said, I connected the bases together, driven digital (I2C is digital remember).

They used a worm drive to turn a grating. Worm gears are unusual, because they rely on sliding instead of rolling contact. That in turn produces lubrication issues.

The total scan range is about 400 nm, out around 2 microns, which requires about an 8 degree tilt, total. That means that the worm keeps going back and forth across one tooth of the gear. With grease, that will gradually force the lube out of the interface.

The system noise comes from two main sources: real noise, mostly Johnson noise of the detector; and mechanical inaccuracy, which causes the center wavelength to be slightly uncertain. This X-axis noise gets turned into Y-axis noise by the slope of the spectrum, so that the noise shows up right in the interesting parts, i.e. the slopes of the absorption features.

If the lubricant film thickness changes, the wavelength calibration walks around, and since the changes we're looking for are delicate, this is A Bad Thing.

Keeping the system noise to 2E-4 AU on a typical slope of 1 AU per 100 nm means keeping the lube thickness changes down below +-70 nm.

Dry molybdenum disulphide is the only lube I know of that has any hope of keeping within that sort of tolerance.

Oh, one other thing: when I pinged them about having used grease, they said that they didn't have any dry lube on hand.... Unfortunately for them, I had already told them that the Fastenall location nearby had it, and had given them the SKU, a link, and everything. So they had to suck it up and do it right.

Cat herding as a way of life. :(

You betcha.

Cheers

Phil Hobbs

So, you liked that circuit?

I'd have to register, and they'd harvest my email address, and send me spam.

Hmmm... Isn't something like PECL essentially linear? It has no saturation , and only has cutoff when "over" driven. I admit that driving to cutoff i s the expected mode of operation. But with no drive or "under" drive, I th ink it is simply a class A power hog. No?

The design still sucks. You're not saturating the transistors, so dissipating power needlessly. Swap the transistors and it gets much easier.

MDs, business, finance, or just about any career other than engineering. One can make a decent living as an engineer but it's tough to make a lot of money.

I have offered, to a couple of similar publications, to check their stuff before they publish it. For free. They weren't interested. They want content and don't much care about the quality.

One of the mags had some guy who critiqued the Ideas for Design stuff, a little follow-up box after each article. He told the truth too often, so it didn't last.

er

oo.

nt driver it has the output in the collectors IIRC.

Krw seems to be thinking of the Darlington option - and a complementary Dar lington drops the the output swing by a Vce-sat which is less than a Vbe - when Jan was clearly thinking of the "synthetic diode" approach to replacin g the two biassing diodes which make the output class A/B.

This doesn't affect either the current gain or the power consumption or the output swing.

"Bill Sloman"

Use Darlington power transistors and bias stability is *very* poor as the driver undergoes the same temp rise as the output in the pair.

Reverse the devices and make the two collectors the output instead and problem gone.

See nice pic of early Motorola Darlington device in TO3 pack.

You can see two transistors but the C-E diode and two B-E resistors are invisible.

.... Phil

Isn't there a way of feeding back the position of the grating so the mechanical slop disappears?

You really are an idiot Slowman. Read the damned thread, for once.

Idiot.

...and extremely power hungry. I don't see it used much for TLDs. In fact, I don't see it used much at all anymore.

Lemme see...(1) write something/anything, (2) get to a computer with internet access, (3) post it. The basic implementation of the "how" is step #2.

(1) write something/anything, (2) get to a computer with internet access, (3) ? (4) Profit

Best regards, Spehro Pefhany

On a sunny day (Sat, 01 Mar 2014 20:13:04 -0500) it happened snipped-for-privacy@attt.bizz wrote in :

So, about 0.7 V at 11.3V output in 200 Ohm half the time makes about 0.7 * (11.3 / 200) = 0.039550 W (100 Ohm locally, 100 Ohm remote, 12V supply) The power in the local resistor being ((11.3)^2)/100 = = 1.276900 W

Your global warming fears should focus on the resistors. Always optimize that what sucks most first. Same with software. No use adding a 5 GHz 8 * multicore if you still run apps written in Java, For email that is....