TI Bulk Acoustic Wave (BAW) clock oscillator

So why did you suggest that you'd need to use it?

In other words you don't known much about them and never used them.

Few people do. I did suggest "Presumably there are programmable logic parts around now that can also run fast enough".

That's what matters when driving transmission lines. Coaxial ceramic resonators aren't going to need a specific DC voltage level, and it's not that difficult to set up a transmission line transformer to deliver a specific DC voltage

If they didn't move they wouldn't resonate. To work out precisely what moves - and how far - you would need an expert, and neither of us is that.

Digging deeper tells me that even the coaxial parts rely on piezoelectric ceramics, so the ceramic is distorting under the applied electric field. As Galileo said, it moves.

I was doing a thought experiment, considering an alternate to the way we have been building instant-start oscillators.

Is there something wrong with exploring new ideas?

We might find an FPGA that could work with a 600 MHz clock, but it sure wouldn't be data-sheet legal. A CCRO oscillator, with a 5 or 10 ohm resonator, would make a low amplitide signal, probably not enough to go into an FPGA LVDS input.

I've used 10K, 10KH, 10EL, 10EP, GiGacomm, and the Analog Devices very fast ECL/PECL/CML comparators. Fun but rarely economic.

How about you? What modern ECL parts have you worked with?

There are some low frequency piezo parts that are called ceramic resonators. Common frequencies are 455KHz and 10.7 MHz.

CCRS are not piezos and do not use motion. They are real coaxial transmission lines, and TDR as such.

Galileo didn't have a good TDR. CCRs aren't mechanical devices.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Would a pair of them in series, each a quarter wavelength but different characteristic impedance, give some intermediate ratios? Even better, if they were not a quarter wavelength but paired so that their reactive impedances cancelled, could non-integer ratios be obtained?

Wideband txline transformers do not have segments that are numerical wavelengths long. If they did, they wouldn't be wideband.

We usually have short windings relative to the bandwidth of our signals. We shoot a 100 ns pulse through a transformer with a couple inches of winding.

There are txline transformers with non-integer ratios.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

The normal sort of txline transformers have integer voltage ratios: 1:1,

This works perfectly at all frequencies in LTspice, but in real life most of them are short circuits at DC, so you need ferrite to improve the low frequency performance.

Barium/strontium titanates are ferroelectric and piezoelectric. Forming them into packed or sintered powder results in the piezoelectric response averaging to nearly zero. Piezo ceramics have to be *poled*, i.e. heated above the Curie point with bias applied, then cooled down.

That breaks the symmetry of the randomly oriented grains, and so makes the ensemble piezoelectric.

That’s why X7R and other high-k capacitors don’t exhibit piezoelectric resonances unless they’re mistreated.

Cheers

Phil Hobbs

The LTSpice txline is an impossible dream. It includes an ideal infinite bandwidth 1:1 transformer.

It takes three TXline parts to model a real one, one that shorts the ends line a real coax does. I have a model of a real-life transmission-line tranny like that around here somewhere. Well, it doesn't include saturation.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

STM (and probably others) were clever, they have an on chip 48MHz RC oscillator for USB that is dynamically trimmed using the 1ms USB start of frame as reference, so it can always run USB even without an xtal

I used MLCCs on a 7660 negative voltage generator. It made a very annoying sound. Desoldering one end of the MLCC and re-connecting it with a thin wire helped a bit, as it was less able to flex the PCB. They really should have made that chip run above 20kHz.

That doesn't mean that they don't move. It there's an electric charge moving through the substrate, charged particles inside that substrate have to be moving.

A time domain reflectometer wouldn't have told him anything useful. It hasn't told you anything useful either.

Run what? Dropbox serves up a bunch of different files, and what would any of them tell me about coaxial ceramic resonators?

They might tell me something about your ideas on the subject, but they don't seem to be all that productive.

If you don't want to think about transmission line transformers, you might try thinking about tapered transmission lines which can also be used for impedance transformation. The extension to getting you to think about tapered coaxial transmission lines is probably a couple of steps too far, but even you should be able to think about burying a tapered strip of copper between ground planes in a multilayer printed circuit board.

The aim is to get from the 5R impedance domain within your coaxial ceramic resonator and the 50R (or 75R) domain in the coaxial cables you use to ship the signal around. A tapered lump of buried strip-line might do it. Google threw up a Ph.D. thesis that used a series of quarter wavelength sections to do the job, as you can if you are operating at a fixed frequency, even though it looks a bit clunky.

For adding USB or Ethernet I use ebay modules:

The USB modules I have used in forexample GPS modules and other USB stuff.

I use one in a very low noise microphone and the noise took a bit of taming - but I am very happy that it was audible, so I could hear whether it was there (the human ear can hear a single tone at very low S/N ratios).

I learned that lesson with a home-made a phantom power unit which used an ultrasonic inverter. That gave no end of trouble because, unknown to me, there was enough oscillator signal getting into the output to intermittently cause the following amplifiers to distort or cut out completely. The problem didn't show up on bench tests but caught me out on a couple of occasions when I was using the equipment for shows and we started getting unexplained faults during the actual performance.

Is that limited to 10 MBPS?

The RP2040 does USB natively. We normally emulate a comm port, but it can also look like a memory stick, for uP+FPGA code upgrades.

We use the Wiznet chip for 10/100 Ethernet, and a Silvertel thing for PoE.

Some people didn't want to use the RPi stuff because "Kids can do that."

Yup.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Yes, but then my old Pi has only 2 USB ports But GPIO has a serial out that you can connect an USB interface module to.

For the Pi4s I use Sitecom USB 3.0 Hub 7 port usb-hubs Lots of stuff is connected to the Pi4, RTL_SDR sticks for RF like radio, ship traffic, plane traffic, spectrum analyzer,

I have a litte ebay module for 2 POE ethernet connections that gets power from a walwart.

Yes, kids have expensive smartphones too these days :-) But kids can do? Ask them to show some code they wrote, some project they did. Anyways if written in C it wil likely compile on Raspberry if it compile on a big PC with Linux. gcc is universal and ARM architecture is supported.

The first Raspberries were simple in a way, I had to separately buy a license for MPEG decoding of movies back then. the Pi4 is quite powerful and all licenses included it seems. There is now a Pi5, unfortunately they dropped the analog audio output... But I have an audio in / out USB stick I could use with it. I have no plans to buy a PI5 at this time... Maybe a bit overkill for the stuff I do.

Sure. I did that and hired him.

The guy had been programming PIs since he was a kid, which wasn't so long ago. Found him on LinkedIn.

We're using the RP2040 chip on boards, and the dev system is almost entirely RPi.

We went the opposite way. A 75-cent RP2040 chip on our boards.

The Pi people just announced that it's officially OK to overclock the RP2040 to 200 MHz.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Nice.

Hope you've got a good source control repository for your code though (I've just dipped toe into RP2040/Pico development, and it feels dirty to be developing outside my well-controlled dev. environment. There seems to be a culture of keeping all your eggs on one flaky SD card in the Rpi world.)

Sadly, at my age, I can't hear 600 MHz any longer.

My problem with motion-based (ie, piezo) devices is that in triggered oscillators they start up slow and stop slow. At shutdown time, they ring like a bell, for a long time, so when you want to start them next cycle, they remember junk from the previous shot.

Given the pictures and the data sheet and the Spice model, I'd expect you to learn nothing.

Agreed, all those "file" things can be confusing.

"Productive" indeed.

We use PCB traces. Much cheaper.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

That's nice!

I am just reading this:

I monitor my Raspberry temperature (posting this from a Pi4 8 GB) cstraspberrypi: ~ # cat /usr/local/sbin/temperature raspberrypi: ~ # vcgencmd measure_temp temp=45.2'C

runs continuously in a small window in one virtual desktop (I use nine).

This Pi4 his in a metal housing (so no WiFi) and has a fan, so is always cool. Used for internet and writing code. So no heavy load.

I have not used any of my big PCs for month, only use those for the DVD / M-Disc player. But most is on the Toshiba 4 TB USB harddisks now, 4 of these in use.. One on the TV for movies and music I recorded. Those Toshiba harddiscs perform excellent. Just do not drop one... I once dropped an other make from a bookshelf... That was a total loss. So, optical discs do have a backup value. Do not store in the 'cloud'.

I doubt if you could have heard it when you were younger. Even bats don't go higher than about 60kHz.

Any high Q oscillator is going to have the same problem. The answer is to use a continuously running clock and capture it's output at the start of your interval and work out the state that would correspond to the end of the interval over which you need to do your timing.

There's nothing to stop you converting a high-Q resonantor into a low-Q part for a short time immediately after you have used it, by - for instance - briefly switching in a damping resistance, but you don't seem to have thought of that.

It's a wise precaution, given the quality of the instruction.

The file extensions are informative, but the source pretty much guarantees that the content won't be.

Good enough for short term forays into niche markets.

But they don't get all the way to the customer's equipment.

I think the guys use github as an editor, but we require the full code package, with readme files, to be released to our internal server before it goes into production. We assign a document number and revision letter to software, like any other engineering drawing.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics