Challenge Thread: The fastest breadboard oscillator on the mudball

[0culus]

Shahriar just did a review of their newest one that has 10Gb ethernet instead of USB and it can stream a full 160MHz IQ unlike the USB models. Pretty slick device, and crazy realtime analysis capability. I was surprised the list price was under $20k, actually. I suspect the largest market is probably TSCM, due to the portability and ease of running off a battery.

But yes, being able to use mixers externally by breaking out the 1st LO is mighty nice. The 8566B also has the IF broken out (and jumped by default with a piece of hardline coax). It can be used both with plain external mixers as well as a line of preselected mixers (which still cost big bux if they are working) up to 110 GHz from -hp- themselves and up to 325 GHz with mixers from third parties (I don't even want to ask how much for those).

[joeqsmith]

I saw where he had posted that review but I have not watched it.  20K would be a bit much for my hobby     but it would be fun to see if LabView could keep up with it.  It's no problem with the BB60C but that's far less data.     The BB60C doesn't provide access to the LO and would require an external oscillator and mixer. 

I can't think of a time I have wanted to experiment >> 110GHz.  I cringe just thinking about connectors and cable costs.     Most of my fun is below 1G where the cost are not too bad and get away with home made bits.  I've been wanting to play around above 2GHz for some time and finally picked up another old VNA to replace my old 1970s system.   The old VNA could run up to 2.6GHz but data was always questionable.   

[0culus]

When the coaxial connectors start to cost $1k or more apiece (1mm connectors), it's too rich for me. I'd like to work up to 40-50GHz eventually but that's it.

[BreakingOhmsLaw]

OP here, calm down guys, it's all fun and games.

[SilverSolder]

When the coaxial connectors start to cost $1k or more apiece (1mm connectors), it's too rich for me. [...]

You know, DC has a lot going for it! 

[joeqsmith]

For my next oscillator, I would like to use Kapton for the two capacitors.  The one around the coax is more just for safety in case the magnet wire chips and shorts.  I don't want DC getting into the equipment.   

Also shown is a small chip of Ferrite from a broken core.  I played with it off and on but have yet to use it.  Its interesting to just move it near the parts and see the effects.     

There are two longer wires that supply a sort of reference for the transistor.  There is another long wire that lays over the top of them.   Once the new parts arrive, the plan is to cut out a section of this wire to form a gap and solder the transistor to the two remaining legs,   one per lead.   

OP here, calm down guys, it's all fun and games.

They were just upset over another thread.  I've never seen them post anything technical and doubt very much you will see an entry unless you allow them to use their 555 timer IC.           

When the coaxial connectors start to cost $1k or more apiece (1mm connectors), it's too rich for me. I'd like to work up to 40-50GHz eventually but that's it.

I don't know any hobbyist  that could afford $1000 connectors, let alone the equipment that uses them.   What are you wanting to do up at those frequencies?   There may be some interesting physics experiments you could run.   

****
Added pictures of trimmer.  Stub will ride in the notch.     

****
Checking out the latest oscillator using an old transistor.

[SiliconToaster]

How fast can a 2N3904 go?

Here it is, a 2N3904 in a "colpits" oscillator running at over 460MHz. 
Not too bad for a transistor with a 300Mhz transition frequency.

The schematic is really simple, the BJT is biased in a common base configuration and it relies on the breadboard parasitics to get into oscillation.




As you can see the signal is coupled into the oscilloscope probe via the breadboard capacitance, a direct connection is too much load for the circuit.




For a moment I was able to get it running at 505MHz, I think that's the limit for the 2N3904. Next step is to get a better transistor.

[SilverSolder]

How fast can a 2N3904 go?

Here it is, a 2N3904 in a "colpits" oscillator running at over 460MHz. 
Not too bad for a transistor with a 300Mhz transition frequency.

The schematic is really simple, the BJT is biased in a common base configuration and it relies on the breadboard parasitics to get into oscillation.
(Attachment Link)

(Attachment Link)

As you can see the signal is coupled into the oscilloscope probe via the breadboard capacitance, a direct connection is too much load for the circuit.
(Attachment Link)

(Attachment Link)

For a moment I was able to get it running at 505MHz, I think that's the limit for the 2N3904. Next step is to get a better transistor.

Very cool!  -  How does the oscillator actually work?  I don't see a feedback path...

[0culus]

How fast can a 2N3904 go?

Here it is, a 2N3904 in a "colpits" oscillator running at over 460MHz. 
Not too bad for a transistor with a 300Mhz transition frequency.

The schematic is really simple, the BJT is biased in a common base configuration and it relies on the breadboard parasitics to get into oscillation.
(Attachment Link)

(Attachment Link)

As you can see the signal is coupled into the oscilloscope probe via the breadboard capacitance, a direct connection is too much load for the circuit.
(Attachment Link)

(Attachment Link)

For a moment I was able to get it running at 505MHz, I think that's the limit for the 2N3904. Next step is to get a better transistor.

Neat circuit. I tried reproducing it with a 2N2222A but so far have "only" achieved a 358 MHz fundamental. I need to play around with the inductor I think...I've been experimenting with winding my own with some very fine gauge wire I had on hand. I'll share some pics once I've worked with it.

To segue into faster circuits, I found a couple of mixers on minicircuits that should work for my needs. I will soon order them and some high speed transistors. 

[0culus]

How fast can a 2N3904 go?

Here it is, a 2N3904 in a "colpits" oscillator running at over 460MHz. 
Not too bad for a transistor with a 300Mhz transition frequency.

The schematic is really simple, the BJT is biased in a common base configuration and it relies on the breadboard parasitics to get into oscillation.
(Attachment Link)

(Attachment Link)

As you can see the signal is coupled into the oscilloscope probe via the breadboard capacitance, a direct connection is too much load for the circuit.
(Attachment Link)

(Attachment Link)

For a moment I was able to get it running at 505MHz, I think that's the limit for the 2N3904. Next step is to get a better transistor.

Neat circuit. I tried reproducing it with a 2N2222A but so far have "only" achieved a 358 MHz fundamental. I need to play around with the inductor I think...I've been experimenting with winding my own with some very fine gauge wire I had on hand. I'll share some pics once I've worked with it.

To segue into faster circuits, I found a couple of mixers on minicircuits that should work for my needs. I will soon order them and some high speed transistors. 

OK, so I made some modifications to your circuit. I added some extra bypass filtering on the power supply, and I substantially reduced the capacitors down to 4.7pF. I also turned the wick way up, and worked my way up to 25V, drawing about 33mA. I have achieved a very nice stability at about 444.3 MHz. I have a few other things in mind to try as well. Also FWIW, the 7104 trace is overexposed by the camera's sensitivity to the phosphor. It is *not* that bright in person!

[0culus]

One more update to this modification of SiliconToaster's circuit tonight: I found a few 1 pF ceramic caps in my stuff. After much experimenting with adjusting the inductor and the cap, I happened upon this configuration. Turns out, trimming off the legs of the 1pF cap stops the circuit from oscillating. So I left it to stand tall and bask in the glory of parasitics.  It is so sensitive now that I cannot even bring a near field probe near the circuit without completely detuning it. So I grabbed the spectrum analyzer data while carefully holding the probe a distance away.

[SilverSolder]

This is starting to look pretty fast!

I still don't get how the oscillator in the diagram oscillates without a feedback path.  Is there a parasitic path somewhere that I'm missing? 

[0culus]

This is starting to look pretty fast!

I still don't get how the oscillator in the diagram oscillates without a feedback path.  Is there a parasitic path somewhere that I'm missing?

Yes, i believe it is depending on parasitic capacitance from the breadboard to form the full tank circuit that sets the frequency.

[SiliconToaster]

This is starting to look pretty fast!

I still don't get how the oscillator in the diagram oscillates without a feedback path.  Is there a parasitic path somewhere that I'm missing? 

No idea of where the feedback path is. As 0culus said, it's a combination of the breadboard and transistor parasitic capacitance.
At these frequencies every wire becomes an inductor and every cap becomes a resonant circuit with it's own resonant frequency that varies with the length of it's legs. 

I tried to exploit the parasitics even more, so I took out the coil and replaced with three wire links in parallel to reduce the inductance.
I also raised the voltage from 12V to 20V.
Now it's running at 536MHz. 







 

[KE5FX]

This is starting to look pretty fast!

I still don't get how the oscillator in the diagram oscillates without a feedback path.  Is there a parasitic path somewhere that I'm missing?

Miller capacitance between  base and collector would be my initial hand-wavy guess.  I don't think it's the breadboard strays, but could be wrong.

[herc]

i just stumbled upon this paper talking about THz Oscillators: https://hangroup.mit.edu/wp-content/uploads/2019/03/IEEE_Filling_the_Gap_Published.pdf

did someone here experimented with spark gaps ? what would happen if i would put sparks onto a simple biquad-antenna,  let's say for the 2.4 GHz ? how broad, how dirty would the spectrum be ?

[joeqsmith]

i just stumbled upon this paper talking about THz Oscillators: https://hangroup.mit.edu/wp-content/uploads/2019/03/IEEE_Filling_the_Gap_Published.pdf

did someone here experimented with spark gaps ? what would happen if i would put sparks onto a simple biquad-antenna,  let's say for the 2.4 GHz ? how broad, how dirty would the spectrum be ?

I've played with a gas grill piezo ignitor.   It's wave shape is not real clean sinewave but no matter.   I don't think I could press the button fast enough to beat a 555 timer and I'm not sure how to mount it to the breadboard.      Let's see what you come up with.

[mnementh]

This is starting to look pretty fast! I still don't get how the oscillator in the diagram oscillates without a feedback path.  Is there a parasitic path somewhere that I'm missing? 

   No idea of where the feedback path is. As 0culus said, it's a combination of the breadboard and transistor parasitic capacitance.

At these frequencies every wire becomes an inductor and every cap becomes a resonant circuit with it's own resonant frequency that varies with the length of it's legs. 

I tried to exploit the parasitics even more, so I took out the coil and replaced with three wire links in parallel to reduce the inductance. I also raised the voltage from 12V to 20V.
Now it's running at 536MHz. 

Now... now you're starting to get into the same realm of Black Magic as this thing:

joe, OTOH... This is all like watching Master Po and young Kwai Chang Caine discussing discipline in the rock garden... 

mnem
Carry on, please! 

[joeqsmith]

A package arrived today and it's time to get serious.. 

The cheap mechanical trimmer is out.  It would be very helpful to have but the breadboard needs to be mounted to a metal plate with a real stage mounted on the plate as well.  It seems like it may fall into the realm of modifying the breadboard.   Currently, this is a job for tweezers, very sharp dykes and the microscope. 

OP, can we use some sort of off board mechanical positioning system as I describe above?   The breadboard would need to be hard bolted to the plate and the stage would be something like this but smaller:
https://www.ebay.com/itm/Newport-562-LH-XYZ-Stage-ULTRAlign-with-13mm-travel-micrometers/293255578226?hash=item4447650272:g:fcsAAOSwt95dkpHw

I have no plans now to use Ferrite, but if someone wanted to use it, would there be any special provisions?  It seems like using a continuous wire to make an air coil is one thing but filling it with Ferrite may be another.   

Can we use the Kapton tape as an insulator as I have shown?   There really isn't a performance gain, at least the way I intend to use it.  It's really just for added protection.  Like the Ferrite, it's now more than just some bent up wire.   See attached post for detaileds pictures:
https://www.eevblog.com/forum/projects/challenge-thread-the-fastest-breadboard-oscillator-on-the-mudball/msg3098642/#msg3098642

joe, OTOH... This is all like watching Master Po and young Kwai Chang Caine discussing discipline in the rock garden... 

Playing with high speed parts is a lot of fun.  You should join the fun.

[BreakingOhmsLaw]

@Joe:
Breadboard: Fixing it is fine, just don't damage or modify it. It should be reversible to a pristine breadboard again.
Kapton: Yes, all that's going to do is act as an dielectric in a ghetto capacitor, so that's covered by rules 5 & 8.
Ferrite: Yes, Rule 5 & 8 again.
Positioning jig: Absolutely, if that's what it takes. Not sure if a lump of metal near your circuit is going to give you trouble.

@SiliconToaster:
Great entry! I will update first post to reflect you as runner-up! Keep it up!

[cj]

This circuit won't break the record but has some 'black magic' to it.
It has only two components a BFR91A transistor and 10k resistor.

[joeqsmith]

This circuit won't break the record but has some 'black magic' to it.
It has only two components a BFR91A transistor and 10k resistor.

That's the old SA I used to have, except maybe the A version.  Very nice!  Also, welcome to the 1GHz+ club!!

[BreakingOhmsLaw]

This circuit won't break the record but has some 'black magic' to it.
It has only two components a BFR91A transistor and 10k resistor.

And two creafully selected and hand-tuned marker pins! 

[0culus]

That was my first SA too...I got it for a song and the guy didn't pack it...like at all. Miraculously it survived shipping. Tough bastard.

@joe, can't wait to see what you cook up! I'm going to order stuff soon. 

[joeqsmith]

@Joe:
Breadboard: Fixing it is fine, just don't damage or modify it. It should be reversible to a pristine breadboard again.
Kapton: Yes, all that's going to do is act as an dielectric in a ghetto capacitor, so that's covered by rules 5 & 8.
Ferrite: Yes, Rule 5 & 8 again.
Positioning jig: Absolutely, if that's what it takes. Not sure if a lump of metal near your circuit is going to give you trouble.

I've been soldering to the plugged in wires, melting the breadboard every so often.   Then there was a lockup meltdown... Sadly this old breadboard will never be pristine again.       

Thanks for clearing up my questions.

I plan to change over to using a small 3XAA battery pack.  I worry that one swipe with Tim's tail (our dog) on the power leads will end it.   

Now if LeCroy would just send me a 40GHz scope.