going beyond dozens of MHz

[cmrincon]

Hi!

Im trying to design a very basic colpitts, ~40 Mhz oscillator ( or above), the problem is that i can't make a realizable air core inductor (i can't make a inductor with less than 1 loop) or maybe i choose the wrong architecture.

Do you have any suggestion for me?

[Yansi]

Different topology will be better.

Good up to hundreds MHz, easily.

[coppercone2]

if you need to make small coils, you need very thin wire. What I like to do is wind it around bobbins (like you get yourself a miniature machinists vise and put a drill bit in it and wind it around the drill bit). Stripping tiny wire precisely can be difficult because its so fragile, some people will actually dip it into carefully molten sodium hydroxide and then clean it in water/alcohol. you need good tweezers, good fixtures (weigh it down) and careful hand eye coordination

with some insulations you can also dip it carefully into more hot then usual solder to melt it away sometimes

you can also get glass or ceramic rods that you can snap after prescoring after winding if you want a more real/tough part.

nylon tweezers (the ones that come in first aid kits) might be helpful if you are worried about abrading the very thin wire

[JDubU]

What about just buying an SMD inductor?

https://www.digikey.com/products/en/inductors-coils-chokes/fixed-inductors/71?k=&pkeyword=&sv=0&pv2087=u250nH&sf=0&FV=-8%7C71&quantity=&ColumnSort=0&page=1&pageSize=25

[Yansi]

And what about stopping nonsense?  Even VHF oscillator can have a small air coil with like 6 or more turns.

40MHz is a piece of cake. I have even posted 50MHz Colpitts for reference. Easy. Just need the right scale for the caps.

[szoftveres]

There circuit and the topology is just fine up till several hundreds of MHz, however there are serious problems with your component values.

C1 and C2 (200p and 400p respectively), are HUGE, the resulting series capacitance (~160p) is in parallel with your LC tank and C4. I would reduce them to at least an order of magnitude lower (20p and 40p).

Rule of thumb: The resulting series capacitance of C1 and C2 should be in the range of 1/3 of the tuning capacitor (C3 in your schematic).

Here's a schematic of a local oscillator from an FM radio (~100MHz range), see the capacitances.

[cmrincon]

For the coils i use 0.5 mm wire and drill bits of  a few mm (3 - 5 mm), which, let me make coils of ~100 nF. Probably i will need to buy thinner wire if i want to go to the vhf upper limit.

Yansi, i make your design but i made it in my breadboard and it was a mistake, parasitics of the breadboard reduced the oscillation frequency to ~14MHz. PRobably i will try to make a board with this oscillator and an class A amp, Probably it will be useful for another proyect i want to do (gilbert cell).

Thank you all

Cesar.

[vk6zgo]

And what about stopping nonsense?  Even VHF oscillator can have a small air coil with like 6 or more turns.

40MHz is a piece of cake. I have even posted 50MHz Colpitts for reference. Easy. Just need the right scale for the caps.

You took the words right out of my mouth!
This seems to be a continuing problem with not only Noobs, but sadly, even among those who should know better.

Because tne resonance formula works with many different values of L & C, they assume that  circuits with huge capacitors & miniscule inductors (& vice versa) are viable ways of building resonant circuits.

[Yansi]

I also suggest this video, as the basics for oscillator design.

[Wolfgang]

Hi!

Im trying to design a very basic colpitts, ~40 Mhz oscillator ( or above), the problem is that i can't make a realizable air core inductor (i can't make a inductor with less than 1 loop) or maybe i choose the wrong architecture.

Do you have any suggestion for me?

Some hints:
- your capacitive divider is wrong. The upper cap must be larger (3-5 times) than the lower.
- your LC is wrong, too. L should be a lot larger, and C accordingly smaller. Then making a reasonable coil is a cakewalk.
- Try a Clapp (series resonator) instead of a Colpitts (parallel).
- Buy a good book about oscillator design theory. IMHO, a Colpitts or Clapp at 40MHz is absolutely no rocket science.
  ARRL has some info, or Rohde, Oscillator Design, ...
- If you want low output noise, decouple thru the resonator. You will need an isolation stage anyway.
- The transistor fT should at least be 10x the oscillation frequency. Not true for your 2N2222A.

Have fun experimenting
  Wolfgang DL1DWG

[Yansi]

Hi!

Im trying to design a very basic colpitts, ~40 Mhz oscillator ( or above), the problem is that i can't make a realizable air core inductor (i can't make a inductor with less than 1 loop) or maybe i choose the wrong architecture.

Do you have any suggestion for me?

Some hints:
- your capacitive divider is wrong. The upper cap must be larger (3-5 times) than the lower.
- your LC is wrong, too. L should be a lot larger, and C accordingly smaller. Then making a reasonable coil is a cakewalk.
- Try a Clapp (series resonator) instead of a Colpitts (parallel).
- Buy a good book about oscillator design theory. IMHO, a Colpitts or Clapp at 40MHz is absolutely no rocket science.
  ARRL has some info, or Rohde, Oscillator Design, ...
- If you want low output noise, decouple thru the resonator. You will need an isolation stage anyway.
- The transistor fT should at least be 10x the oscillation frequency. Not true for your 2N2222A.

Have fun experimenting
  Wolfgang DL1DWG

Errr... what?

[Wolfgang]

The argument runs like this:

- The base input represents a negative differential resistance defined by gm.
- The base/emitter cap divider splits the AC voltage. The part along the BE junction should be small,
  and the part across the emitter resistor should be large. If you do it the other way, you will earn a strongly nonlinear
  behaviour and large negative VBE swings, at a comparably low amplitude. Why do this ?
- Rohde (N1UL) oscillator book has a derivation of this inside.

regards
  Wolfgang

[Yansi]

In most oscillators I have seen, the lower part of the divider is the part that is bigger and definitely not 5x. (maybe 1.5 to 2x)

[Wolfgang]

I absolutely agree. Its still wrong, only most people did no calculations behind their designs.
Suggestion: Try it out, and watch the signals.

PS: I know a little about this because the PHD thesis I am working on right now is about USOs.

[Yansi]

I'm not saying what's wrong and what's not. My oscillator-fu is still child level. Just noticing the obvious. However what you say makes sense. Sometimes I will sober up and read more about it. Designing meaningful oscillators and VCOs might be a handy skill to have.

[cmrincon]

I built the common colector oscillator, it should be a 83 MHZ osc: 

I had some smt resistors and caps so i used them, now, i think that it was unnecessary and it made harder to test the circuit.

here are the photos:

There is a strange effect around 100 MHz i think that this is my oscillation. I cant clearly see it because my osciloscope bw is 100MHz. All voltages tested with my multimeter seems ok.

Cesar.

[cmrincon]

Today i have made the same circuit using leaded components, but i can't get an oscillation. You can view the output (transistor emitter) of the oscillator in the image below

I used 2n2222 because i don't think that the transistor is the problem becuse last week i replaced the mmtb2222 for the mmth10 and i get the same result. Also i used a shunt capacitor at transistor collector of 1n but nothing changes.

The layout it's the same that the previus post.
 What can be wrong?

[Wolfgang]

Today i have made the same circuit using leaded components, but i can't get an oscillation. You can view the output (transistor emitter) of the oscillator in the image below

I used 2n2222 because i don't think that the transistor is the problem becuse last week i replaced the mmtb2222 for the mmth10 and i get the same result. Also i used a shunt capacitor at transistor collector of 1n but nothing changes.

The layout it's the same that the previus post.
 What can be wrong?

The BE and E to GND caps are too small (C1,C2), IMHO. the negative resistance is gm*XC1*XC2 and this must exceed the circuit losses. gm is Vt/Ic, with Vt 25mV. Another thing is the L is a bit small. make it 1uH and C about 22pF.

[iMo]

For 80MHz and up the oscillator shown by Yansi above is a better way to go. Also you would need a better transistor, imho.
Btw what you want to achieve? What is your application?

[chris_leyson]

There is a good discussion of the Colpits in "Crystal Oscillator Circuits" by Robert J. Matthays, page 31.
https://bgaudioclub.org/uploads/docs/Crystal_Oscillator_Circuits_Krieger_Matthys.pdf
Personally the Colpits is not one of my favorite oscillators with regard to phase noise, much prefer Pierce.

[Wolfgang]

There is a good discussion of the Colpits in "Crystal Oscillator Circuits" by Robert J. Matthays, page 31.
https://bgaudioclub.org/uploads/docs/Crystal_Oscillator_Circuits_Krieger_Matthys.pdf
Personally the Colpits is not one of my favorite oscillators with regard to phase noise, much prefer Pierce.

I disagree. All world-class USOs are Colpitts/Clapp 5-10MHz SC cut crystal OCXOs., with PN down to -135dBc@1Hz.
Try RAKON or Wenzel for product data.