phemt colpitts: beads

112 MHz is slightly above the FM broadcast band. It is in the aircraft navigation band (108 - 118 MHz), and one the very worst to spill spurious radiation into.

--

-TV

On a sunny day (Wed, 3 Feb 2021 16:10:46 +0200) it happened Tauno Voipio wrote in :

Indeed, I meant to say around broadcast band... With those rtl-sdr (also 100% integrated) USB dongles you cannot only receive that, but also the 1090 MHz plane info and the AIS ~1090MHz info, and the 433 MHz weather station and ... GPS even.

records 24/7 to AVI -rw-r--r-- 1 root root 262936110 Feb 3 15:38 bp49.avi 'bp' stands for boats and planes

mine are 1ppm TCXO controlled ...

As to that FPGA, just wonder, should be possible to make an internal oscillator using some gates feedback, wire one to a pin, and do some capacive loading on the pin to affect frequency?

That one is dangerously approaching readability.

I could look for a diode or a transistor c-b junction that makes a suitable varicap. It wouldn't be designed to be a varicap, namely it wouldn't be hyperabrupt doped, which would be an advantage in my case, very narrow tuning range.

I'm thinking that digital techniques account for most varicaps being gone these days, especially higher c values.

--

John Larkin      Highland Technology, Inc 

The best designs are necessarily accidental.

The LVDS inputs to FPGAs are pretty good, so can be used as comparators to make time delays. But there is a lot of noise from all the digital stuff, and delay tempcos are an issue.

--

John Larkin      Highland Technology, Inc 

The best designs are necessarily accidental.

On a sunny day (Wed, 03 Feb 2021 07:48:35 -0800) it happened snipped-for-privacy@highlandsniptechnology.com wrote in :

Note that in that circuit Ccb and Cbe are in series, and the transistor is a PNP, so driving its base positive via a very high impedance (for RF) decreases capacitance. You can get as small as you want by choosing say a VHF transistor.

Yep

Am 03.02.21 um 16:51 schrieb snipped-for-privacy@highlandsniptechnology.com:

It might be possible to abuse the PLL section of one of these GTX transceivers in the FPGA. They are designed to lock quickly on an incoming data stream at 10 GHz. Frequency dividers exist.

Gerhard

I need instant (say, 1 ns) oscillator start and picoseconds of jitter after that. A regular PLL can't do the instant start thing, and the prop delays inside an FPGA vary with temperature and especially supply voltage. An FPGA is a noisy place.

There are still places where discrete component design are needed!

--

John Larkin      Highland Technology, Inc 

The best designs are necessarily accidental.

Isn't the startup time of an LC tank inverse proportional to the quality of the tank? This will either give you long startup times or high frequency deviation.

--
Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de 

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt 
--------- Tel. 06151 1623569 ------- Fax. 06151 1623305 ---------

I'm using a Coilcraft surface-mount inductor.

I'm used to an inductor increasing diameter with increasing temperature, which increases L, which decreases frequency. Cooling the Coilcraft increases the frequency. That helps explain the unexpected tempco effects I've been seeing.

It's a sort of transparent plastic block, nice part, but the plastic or something seems to be changing the tempco sign.

Decreases it of course.

I said it had parasitic oscillation with a bead in the gate but works with a resistor. That's fine by me.

The differential equation tells the story. If the initial conditions are right, an LC starts making a perfect sine wave instantly when you kick it on.

The trick is to precharge the tank by passing a current through the inductor.

When you want to start the oscillator, disconnect thecurrent from the inductor and let the oscillator take over. Here's an example using the .ic command in LTspice to set the voltage across an inductor in series with a resistor:

Of course, in the real world, you have to figure out how to switch the applied current off very fast. There's a bunch of ways to do it.

This trick also works well with high-Q tanks, such as in crystal oscillators. You can bypass the long startup time and analyze the circuit operation right away. But in the real world, you have to wait for the oscillator to start up and settle.

--
The best designs are no accident - sw

It doesn't work nearly as well, though; the high Q of a crystal is a measure of its tiny coupling to the external (lossy) components. It does NOT start ringing at its resonant frequency in the first cycle, but an LC tank does. The turn-on transient puts lots of harmonics into a crystal, only slowly damps the out-of-resonance ones.

Right, but your inductor is not just a pure inductance. You can inject a charge into it and different bits of it will start working at different times, against their own TCs and internal resonances. That's gonna fight the purity of your initial response. How much? I have no idea. But you're not going to get the right answers from a simple lumped element model, no matter how good your impulse generator is.

CH

Or a zener diode. Maybe even a PIN rectifier.

A mate had to tune an HF front end (1-5MHz) with a very small sense coil, to measure micro-ohm dynamic impedances in a coupled coil (with high-Q resonances spaced across the frequency range).

Big enough varactors were not available, so he wound up using 480 (!) varactors in series/parallel combination.

Some of my code is in that unit and the bench-top version. Interesting technology.

Clifford Heath

Holy cow. A DDS wouldn't have worked?

At 1 to 5 MHz, I'd think a simple V-to-F converter would be OK.

-- john, KE5FX

You snipped the rest of my statement:

"You can bypass the long startup time and analyze the circuit operation right away. But in the real world, you have to wait for the oscillator to start up and settle."

The trick only works in LTspice, where you need to speed up the long startup and settling time. This can allow analysis in the first few cycles.

The only way I know of the speed up real life crystal oscillators is to inject a high level signal close to the correct frequency to get the crystal started. I believe this trick is used in high volume production testing of new crystals.

--
The best designs are no accident - sw

Sacred bovine indeed.

I'm thinking of a capacative DAC, namely a modest bunch of caps and some relays.

It's not the frequency, it's the size of the coupling coils and the incredibly low impedances. The chip being read has a string of 52 MEMs resonators (some damped) spread across the frequency range, spaced about

2% apart, all in a cavity inside a 1mm^2 chip attached to a 4mm diameter multi-layer coil antenna. At cryogenic temperatures the Q of individual resonators can reach 5000. The whole thing is placed in a magnetic field and the reader antenna couples a pulse of energy in, then switches to receive to do a spectral analysis of the ring-down of the resonators.

Tiny amounts of energy, very high Q, and micro-ohms of dynamic impedance at RF. This company spent a decade developing this patented technology, and it's amazing they got it to work at all.

How would you have done it?

And if they aren't right, it doesn't. The interesting questions are how close you can get to getting them just right,and how long it takes for the sine wave to get close enough to perfect to serve your purpose.

You aren't talking about what you do as an engineer would - in terms of how close you can come to the ideal - but sound like a sales guy telling the customers that the approach can produce a perfect result.

--
Bill Sloman, Sydney