The best part about engineers is that if I'm ever stranded with a dead battery in the middle of nowhere all I need to do is say "So I'll just connect up the red to the black and the...." out loud and within 10 sec one will be on scene to a) call me out on the error b) do the work for me /shrug
I am afraid there is something peculiar about the atarting point of your calculations - step 4 outlined above. Using Fo = 1/(2*PI*sqrt(L*C)) LC = 4*PI*PI*Fo*Fo Now assuming a standard value for L, e.g., 1nH C=4*PI*PI*Fo*Fo*1.0E+9 Now if C is the equivalent capacitance, i.e., C = C1*C2/(C1 + C2), C1*C2/(C1 + C2) = 4*PI*PI*Fo*Fo*1.0E+9 If Fo = 5MHz, C1*C2/(C1 + C2) = 100.0*P*PI*1.0+12*1.0E+9 A very big number, i.e., the equivalent capacitance capacitance values is huge !! Most likely the expression for the equivalent capacitance is a bit strange.
Yes of course, those ceramic filters, or should I say SAW filters, are everywhere, for example in TVs as IF filter. But I meant for frequency stability, for example these
formatting link
the greenish round things on the right are 9.75 GHz and 10.6 GHz oscillators (smallest one highest frequency), but precisely the reason I am using the latest LNB with a crystal reference and a PLL to get those same frequencies, as the crystal is much more stable, even without temperature compensation. This is needed to receive 10.4 GHz single sideband signals where a few hundred Hz at 10.4 GHz makes the speech un-intelligible. That is also where the Rubidium reference comes into play. For less precise things such as some filters ceramics are OK, those are also OK for local oscillator in LNBs for digital TV. Important for all this in my case is that you can electronically adjust frequency, say pull the crystal, so you can PLL lock to some reference. Look at the feedback coupling of those oscillator transistors to the ceramic resonators: ( O ) b c
1nH is 31mohms at 5MHz, certainly less than the load resistance (50 ohms for instance), but you'll never realize a pure 1nH without more nH's in the other components, nor a Q factor probably much over 20 at that frequency, nor a transistor that can oscillate into a 0.03 ohm (reactance) * 20 Q = 0.6 ohm (resistance) load.
Following the steps I outlined, say for a 50 ohm load, a voltage ratio of say 10, and a Q factor* also around 10, then Zo = 50 / 10 = 5 ohms, and L = (5 ohm) / (2*pi*(5MHz)) = 0.16uH and Ceq = 6.4nF. C2/C1 ~= 10 so C2 ~= 64nF and C1 ~= 6.4nF, but actually 10% higher for each, and give or take the exact ratio (just reverse Ceq = C1*C2 / (C1 + C2)).
*The transistor's base load will be low ~kohms. The impedance, looking into the resonant tank from the base, is on the order of 1 / (2*pi*F*C2), and the resistance is lower. We should be able to ignore this for most cases; a very low power or low distortion oscillator (gm limited), may not be able to ignore this, which is why I'm not completely setting it aside.
It would take further analysis to determine if a particular transistor can drive this load, and what C ratio is required for that selected transistor. After a few steps, you'd have something quite worthy of breadboarding, since it is a poor theoretician that does not verify their results in the lab. :-)
Note that this is a power oscillator, or at least a very low voltage oscillator. You don't use 5 ohm tanks with small signals! If the intent is to have a little voltage into 50 ohms, not a large fraction of Vcc -- one should use another impedance matching network to derive that. A tapped or coupled inductor, or a capacitive divider (or taking the output from the base side of the tank, or the emitter circuit if unbypassed) can be used for that purpose, in which case all the circuit impedances can rise accordingly, the inductance and capacitance get much more familiar, and a general purpose transistor is suitable (say, anything fT > 50MHz, which, truly, is a hell of a lot of "general purpose" transistors).
I use a varicap. A CCRO is actually a shorted transmission line (and TDRs like one) but you can pretend it's a parallel LC.
A CCRO is different from a SAW. And different from a mechanically resonant ceramic filter. I guess you knew that.
I'm not very interested in sine waves or tuned circuits; we work in time domain. But lots of microwave parts work great off-label in pulse circuits. They'e just not well characterized for that.
--
John Larkin Highland Technology, Inc
lunatic fringe electronics
Apex used to use an AT&T HV cmos fab process that made possibly the noisiest transistors ever invented. Nowadays they just put parts on boards and charge a lot.
--
John Larkin Highland Technology, Inc
lunatic fringe electronics
OK, you don't know, it seems, or overlooked the question. I looked up the Ft of my OC76 and it is listed as 1 MHz. Those early Ge transistors were usually slow, that is why I asked.
Aha, yes, cool.
No, I do not know everything :-) Looks in that picture like just some ceramic tubes like you add crystals together to make a steep filter.. Something vibrates mechanically... Any papers on that?
Yes, some transformers will, diodes... But filters.. if it has high Q then it will show something like this:
| || ||| || | damped oscillations. At those high frequencies any piece of wire is one or more wavelength long.
So if you are into making fast pulses... of one or one half period, does not seem that easy to me. I have no experience with that at those frequencies, never needed it.
That is why to me that RafaelMicro RT320M is so fascinating, and also that RTL-SDR stick, the R320M a complete PLL oscillator and mixer at 9.76 and 10.6 GHz, and the USB stick a complete receiver up to 1.7 GHz with IQ outputs, The length of the wires in the chip sort of overcomes the wavelength problem - is much much smaller than a wavelength. But the sign on the wall is that smaller on chip integration is the future.
What is the fastest, I mean shortest, pulse you can make? How much power does it deliver?
CK722 was slow, maybe under 1 MHz, but the data sheets didn't spec it. I think it was a nasty alloy-junction part.
No, there is no motion. Each tube is a shorted coaxial transmission line. The dielectric constant of the ceramic is so high that the line is short relative to the round-trip frequency. In the filter, the CCRs are coupled somehow, classic multiple-resonator math.
It's fun to TDR one. It looks just like a shorted coax, but the impedances are low, 10 ohms maybe, and there is a lot of delay relative to the size.
Probably stuff from the manufacturers, although they usually work in frequency domain. Boring sine waves.
Pulses are even more fun!
There are cheap logic chips with rise times in the 100s of picoseconds.
formatting link
Step-recovery diodes can make sub 100 ps pulses easily. NLTLs can get down to rise times in the single digits of ps.
Gain-switched lasers can make optical pulses in the 10s of ps width, from fairly slow electrical drive.
Spark gaps can be impressive!
--
John Larkin Highland Technology, Inc
picosecond timing precision measurement
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
And slightly less cheap ECLinPS Plus parts (from ON Semiconductor)
formatting link
explicitly designed to drive 50R transmission lines (or 75R).
Not that easily. You do have to be careful about construction to minimise stray inductance and capacitance.
But erosive. And the glow-to-arc transition takes closer a microsecond than a picosecond.
Getting an arc discharge involves getting the electrode surface hot enough that it can deform into atomically sharp spikes, sharp enough to deliver field emission.
In the glow stage, positive ion bombardment knocks off secondary electrons and incidentally warms up the bombarded surface.
It took me half an hour googling to find the Ft of the OC76 online... Anyways I _do_ remember Philips had OC13 sold as LF transistor (for audio) and OC44 OC45 as 'IF' transistor for the 455 kHz or so IF amplifiers. RF GE transistors came later... in the sixties IIRC, I build a small FM transmitter with just 1 transistor also using the Cce change with voltage to frequency modulate it. Not sure that would work with a modern transistor. The old ones were so sensitive I did something this
+9V /// | | dynamic mike (few hundred ohm) === |--- | ( | ) ( === tuned to about 100MHz | ( | | |------ | c | b | | e | | | | | /// | - R-----
That thing was so sensitive you could here a clock ticking over the radio. The mike changed the Vce a few mV, was enough to get many kHz frequency deviation.
Not sure I grab that, will look it up.
The whole world uses sine waves to commie-nuke-aid....
Yea digital, but even there it is deliberately bandwidth limited, to get more info over the same link, in the same bandwidth, that is an art too.
For those who try the laser fusion perhaps... but that did not work out so well either? Still waiting for usable break even. Most stuff uses sine waves....
Spectra...
Na yaa, 100 ps is only 10 GHz.... so rise and fall times are much faster /\ /\ / \/ \ | |
100 ps period time | |
50 ps up, 50 ps down This one works at > 12 GHz... ;-)
Yea, I did that as a kid, when I asked the son of a local radio ham to ask his father to listen for my 'transmitter' I did not get friendly feedback... 'If I ever hear that thing [again?] I will...'
We had a lot of TV interference from things with electric ignition driving past the house, local TV was at about 65 MHz.. Only later did they use screened ignition cables and some measures against RF radiation.
Erosion is real; there's a molybdenum electrode with a few grams missing in my junk box. The platinum button in a spark plug, though, lasts quite a while with modest currents. Probably the metal ions just aren't part of that kind of arc.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.