rectifier waveform

"Helmut Sennewald" schrieb im Newsbeitrag news:e5d9o0$kba$00$ snipped-for-privacy@news.t-online.com...

Hello,

I have done today a few real measurements on rectifiers with an analog oscilloscope.

I tried with different rectifiers, two bridges and one with 4 diodes. One bridge rectifier showed no spike. The other one produced a medium height spike(10V) and the 4-single diodes produced a spike nearly as high as the peak voltage in everey cycle. (20V) The medium high spike has had a width of about 1us and the big spike has had

I used a 1.5A load current. The transformer has a secondary output with

Compared to my simulation the measured spike doesn't ring many times. It was only one strongly damped spike. Maybe the reason is the more soft switching of real diodes compared to the SPICE equations.

Best regards, Helmut

Helmut Sennewald wrote: (snip)

.. Or the high eddy current losses at higher frequencies with a laminated silicon steel transformer core.

You're the one having trouble with your ABC's of elementary circuit behavior, and you're a few days late realizing the timing error caused by line fluctuations.

Time-of-day clocks don't care about phase noise, they only care about frequency errors. If phase error is bounded, as it will be with any reasonable filter/discriminator, the longterm time error is zero. At any instant the clock may be a millisecond early, or a millisecond late, but that will average to zero longterm.

If you disagree with that, we have nothing more to say here. I'll leave it to others to give their opinions, but you of course may not respect them, either.

Incidentally, the US Naval Observatory uses my logic fanouts in their atomic clock farm, because they drive cables hard and add only a few picoseconds of jitter. And we designed and sold the master timing system and the beamline modulators for the NIF laser, timing about

John

No it will not average to zero, and "clocks" are event accumulators, they are not frequency counters.

This has nothing to do with respect, it is all about simple observation.

That is a completely different noise model and has no bearing on this application.

But the digital electronics is clocked off a crystal oscillator. The oscillator generates a sine wave, and the sine is squared up by a comparator to clock the timing. The sinewave varies in amplitude, and the oscillator/comparator combo has phase noise; all this is the same as the AC line-Schmitt thing, just faster and with smaller relative phase noise. But the NIF delay generators stay aligned... I'd sure hear about it if they didn't.

In both cases, the phase error is bounded, so longterm time errors don't accumulate. Phase error *is* time error.

What's the difference?

John

It's called phase error because it has no mean offset from center frequency. In the line-operated clock trigger, the waveform+comparator is nonlinear, so what goes up does not necessarily come down, *and* because the inter-cycle perturbations only occur during the intervals of mains *transitions*- there is absolutely no reason in the world to conclude the perturbations average to zero.

Fred, How can you be so ignorant? We're COUNTING ZERO CROSSINGS. We don't give a damn about some flutter.

...Jim Thompson

It doesn't matter what you call it.

You've just proven that all digital logic makes errors. If a 100 MHz logic signal has 100 ps of jitter, your logic suggests it will accumulate enough phase error to make a mistake on the order of once every hundred operations.

John

Hey, Fred,

see Fig 4 for the proper way to recover a clock from the AC line.

John

You can pretty much eliminate false zero crossings due to noise spikes by using a low pass filter and hysteresis. Then the main source of error would be line outages caused by lightning strikes, distribution transformer failure, transmission line switching, current surges due to large motor startup, and so on.

These events are called power failures. They can cause clocks to lose count of the zero crossings. It is not a failure of the circuit to detect the zero crossings properly when they do not exist, or when random line transients exceed the normal range. This is the price we pay for such a simple and inexpensive method of telling the time.

Most of the time it doesn't matter. The hundreds of millions of people who use clock radios to tell the time usually have to reset the clock every six months for daylight savings.

For applications when greater accuracy is needed, we have many alternatives. Simple programs can read atomic time from the internet and reset the computer clock. Many people have gps receivers that display the correct time. You can buy clocks that receive wwvb at 60KHz and automatically reset themselves.

Perhaps the single biggest obstacle to reading the correct time is the clock setting routines on vcr's. These are rapidly becoming extinct, but for decades they simply flashed 12:00:00 since nobody could figure out how to set them properly:) Regards,

Mike Monett

lol, the original version of TTL

martin

Beautiful. A 4-pole Sallen-Key lowpass filter made of triodes.

I wouldn't want to trust that clock to wake me up for an airplane, though...it's got as many firebottles as Eniac.

Cheers,

Phil Hobbs

I liked the dedicated transformer winding and full-wave bridge, too. I assume he derived 100 Hz, and not 50, because it would take more tubes to process that way.

John

And 60Hz would require too long an extension cord.

Best regards, Spehro Pefhany

Europeans are weird. They use fewer cycles but more hours than we do.

John

Okay- take a 60HZ symmetrically bipolar triangle at +/-1V and give it say (1+r) amplitude modulation at 10Hz with r=r(t) a 1Hz sawtooth from

"martin griffith" a écrit dans le message de news: snipped-for-privacy@4ax.com...

/Digital-Roehrenuhr.htm

Toomuch-Toobs-Logic?