TCXO (shopping for replacement)

no they don't have to be high Q to injection lock

There may well be an effect, but the explanation doesn't seem very convincing.

Jeroen Belleman

The HP 5370 in TIA mode is pretty good for measuring rms jitter. Unfortunately, it doesn't tell you where it is coming from. You need a phase noise plot to see that.

A true time domain measurement, such as with a sampling scope, is needed when you have multiple distributions.

So phase noise, TIA, and sampling scope are all useful and needed.

Power supply sensitivity is the bane of the industry. Better oscillator design and improved power supply filtering is needed. This is trivial if you know what your are doing.

System layout, bypassing, filtering, grounding and shielding are all part of a low noise design. Then you fight coax cable leakage and multiple ground loops.

Transfer of timing signals using fast rise pulses is convenient. However reflections from poor terminations can cause errors. In these cases, pure sine waves can be useful.

Much of this has been discussed in the tiniest of details in the time-nuts archives going back to 2004:

If you can get John Miles to respond to a question, he is clearly the heavyweight expert in this area. Listen carefully to what he has to say.

Yes. Production managers may have been clutching at straws to explain the poor yields. But it worked. The data is there.

OTOH, the epoxy binder formula was never understood very well. There were constant improvements, such as using acicular particles, then aligning them on a horizontal plane to improve phase noise.

The end came when the industry abandoned rust and converted to a sputtered plating. This was done in a vacuum and immune to outside influences.

The main theme of the disk industry is the phenomenal improvement in cost per byte and areal density over the years. Nothing else can come close.

We owe the disk industry a great debt of gratitude. The internet would not exist without huge banks of disk drives spinning in the dark.

Fun story (I may have told it before.) We sell this torsional harmonic oscillator. Vertical wire, big mass, some permanent magnets so you can drive it, or measure the angular velocity.

The magnets are the interesting part for the story. There were a whole bunch of these (say 8) lined up next to each other. All oscillating at about 0.5 Hz. I'm half watching them and notice that the amplitude of any one is growing and then shrinking... they are all coupled together! (by the permanent magnets.) A colleague hooked up the SRS spectrum analyzer to one, took data for ~1 hour. (undamped Q >500) And measured all 8 normal modes of the system.

(there was no locking to one frequency in this case.)

George H.

Well, you're in that small region of the lunatic fringe where stuff actually works. ;)

Cheers

Phil Hobbs

Nothing wrong with measuring delay jitter in the frequency domain. Any repetitive signal has a spectrum. Just get the delay to retrigger itself, or retrigger it from a stable source.

Any time you make a measurement that you haven't made before, there is a good chance of learning something new. I highly recommend it.

Quantitatively, it's the energy loss per cycle as a proportion of the energy stored in the system.

Measuring the energy dissipated in inelastic distortions of the earth and the moon due to tidal forces is a non-trivial exercise, but it has been done.

Well, we rarely try to sell the failures.

Sure. But today I'm blowing up resistors.

Didn't you post images of a thin film resistor on a heat sink that could absorb huge amounts of punishment?

Why are you messing with wirewound resistors that cannot hope to compete?

Why are you wasting your time?

For a given coupling strength, the higher the Q the stronger the locking effect. For instance, coupling between two LC circuits goes like Q times the mutual inductance.

Cheers

Phil Hobbs

If you had any real interest in electronics, or asked civil questions, I'd answer you.

You're just here to be nasty. Envision a much less polite way to say "go away."

John Larkin's idea of a "civil question" is one that includes the flattery to which he thinks he is entitled. "Interest in electronics" has to be nuanced in the same sort of way.

Which is to say you haven't made him feel flattered.

Why bother. A better translation would be "flatter John or he'll flounce off".

Legitimate question. Deserves a civil answer.

Why are you so sensitive?

Here's the resistor you posted:

Here's another similar one:

If you had any interest in electronics, you'd explain the benefit of wirewound resistors to the masses.

Wire wound resistors are usually used for power stuff. There's a few page summary of resistor types in the back of AoE3 if you're interested.

(I wish wire wound could be made smaller, I use Caddock thick film but have toasted some. I never tested for pulsed power... which sounds kinda fun.)

George H.

In say a millisecond, there isn't time for a pulse of heat to diffuse out into, say, an epoxy or thermoplastic body. The resistance element heats itself as if it were alone in a vacuum, but maybe without even radiation cooling. So we need a resistance material with high specific heat and a high melting point... like nichrome. And it needs a surrounding medium that won't melt or decompose when in contact with the high temperature element. Spreading things out helps, if you have the volume. I don't.

I blew the top off a Vishay power-pad 5-watt, at about 10 joules.

I'm pulsing a TE SMW5 surface-mount wirewound now at 20 joules/shot. If it survives the night, I'll crank the energy up.

A ceramic-body uncoated MELF wirewound might be ideal, but seem hard to get.

I'm thinking that a classic axial, leaded wirewound might gobble a lot of joules, especially the virteous enameled kind, where the hot nichrome won't smoke anything organic. We'd just have to hand-solder

So your original claim was false.

Good to know, but not that I care much.

Get a hundred 1W carbon composition resistors and connect them in series or parallel as appropriate. Spread the heat around. Betcha it will take more abuse than your single wirewounds.

I did this once with a bunch of carbon resistors soldered between two copper plates as a RF load. I could not generate enough RF to kill it.

Lasted forever.

piMZ

or

Sure you couldn't. Carbon has a negative temperature coefficient of resista nce, so it forms "hot channels" when you put enough volts across it. The ho t channels have very little resistance - I've see roughly 1R in a 10k 0.6W carbon film resistor - so there's mot a lot of power being dissipated.

Once the channel cools down the resistor looks - and seems to be - undamage d.

You are not allowed to use them as current limiters in intrinsically safe d evices for exactly this reason.