Challenge Thread: The fastest breadboard oscillator on the mudball

[BreakingOhmsLaw]

10k at 5Vp-p seems a bit odd.   Does this mean at DC, you read 10K with your ohm meter?   Don't suppose I could talk you into something a bit more like say 0dBm @ 50ohms?

There lies the hidden challenge. Selecting the right resistor for the job. Remember the difference between metal-film and carbon resistors?
But since we've already progressed into the GHz Area, i suppose it's okay to deviate from this rule within reason.

Can drift and have as much phase noise as we want?

Yes, we are building circuits on a yagi breadboard

For example, can contact be removed or cut down?

no

Is there a limit to how much money is spent on it?

no

Can we install more than one wire into a hole on the breadboard?

no, see rule #3

Thinking about other mods on the breadboard, can we add foil below to tie some of the contact bars together?

no


As far as i can see, your entry seems valid, so for now, you take the cake.
Looking forward to other members picking up the gauntlet.
I am surprised noone has mentioned snap-off diodes yet. 

[ejeffrey]

Does a discrete laser diode count as a ~600 THz oscillator?

I have trouble seeing how you are going to fulfill rule #4:

4. The output signal shall be sine(-ish) and able to drive a 10k load at at least 5Volts peak to peak.

But if you do manage to do that, be sure to give Stockholm a call for your Nobel prize.

This certainly gets a bit tricky to figure out.  Voltage isn't an easy thing to define for a propagating EM wave since there is a curl-E != 0 so there is no perfectly defined conservative electric potential -- although you can make an approximate one in many situations.  5 volts p-p into 10k is only 300 microwatts, so the raw power isn't the issue.  The impedance of free space is 377 ohm, which is the ratio of E to H for propagating waves in free space.  5 volt pk-pk into 377 ohm is 10 mW, so a moderate laser diode can actually have a big enough E field to be comparable to the specification you posted.

Of course measuring that electric field directly on an oscilloscope is... difficult but you could argue that is the fault of the oscilloscope not the oscillator.  There are actually techniques to make something similar to a sampling scope that can directly measure a repetitive electric field at optical frequencies, but they wouldn't work on a simple laser diode and in any case would look pretty silly connected to a breadboard oscillator.  You could certainly view the signal on an optical spectrum analyzer.

Anyway, definitely violates at least the spirit of the challenge, mostly just an interesting aside.

[joeqsmith]

Shown are three different techniques.

Left, using some semi-ridged.  Can this be used as long as at least one end is terminated to the protoboard?  For example could one end be soldered to a single pin of a single component as long as the other end plugs in?    Can multiple  ground wires be added to the shield and plugged into the board?   I would assume if this is allowed, that soldering two sections of coax together by their shields would not be. 

Center, showing some sort of make shift heat sinkish sort of plane.  Any problem stitching sections together like this for what ever purpose?

Can we install more than one wire into a hole on the breadboard?
no, see rule #3

Right, showing what I meant by plugging two components into a single hole.   They are still in contact with the connector.  I just want to make sure you understood what I meant. 

I'm not sure if ANY of these practices would be helpful or not and am just trying to understand the constraints we have to work with.        

Adafruit offers a 4X4 breadboard.  I was trying to find something for sale with something  other than the standard 5 pin connectors.  If something like this were say 4 rows of 4 inner connections,  would it be allowed?  Same for other breadboards?   As long as its something standard that anyone could purchase, is it allowed so long as it is that standard white board, proto board, plug board or what ever we call this stuff now days?   

https://www.adafruit.com/product/2463

**** Sorry, rescaled picture *****

[BreakingOhmsLaw]

Left, using some semi-ridged.  Can this be used as long as at least one end is terminated to the protoboard?  For example could one end be soldered to a single pin of a single component as long as the other end plugs in? 

What's the forums opinion on this?

Can multiple  ground wires be added to the shield and plugged into the board?   I would assume if this is allowed, that soldering two sections of coax together by their shields would not be.

I guess we need to allow this for mechanical stability.
Solder on Coax to coax, no.

Center, showing some sort of make shift heat sinkish sort of plane.  Any problem stitching sections together like this for what ever purpose?

That's perfectly okay and the kind of creative thinking we are looking for.

Can we install more than one wire into a hole on the breadboard?
no, see rule #3

Right, showing what I meant by plugging two components into a single hole.   They are still in contact with the connector.  I just want to make sure you understood what I meant.

Yes, understood, but that would be againts the spirit of the challenge.

I'm not sure if ANY of these practices would be helpful or not and am just trying to understand the constraints we have to work with.        
Adafruit offers a 4X4 breadboard.  I was trying to find something for sale with something  other than the standard 5 pin connectors.  If something like this were say 4 rows of 4 inner connections,  would it be allowed?  Same for other breadboards?   As long as its something standard that anyone could purchase, is it allowed so long as it is that standard white board, proto board, plug board or what ever we call this stuff now days?   

Encouraged in order to gain knowledge about good breadboards.

Thanks for contributing!
Where are the contenders? I'm out myself, my counter just goes to 1Ghz :-D

[joeqsmith]

Using the same setup, except using that section of semi-ridged for the output rather than the flimsy SMA, I made an attempt to tune it.  The problem is the amount of capacitance of those stupid connector blocks.  While your rules will not allow removing or cutting them, we certainly can skip over one to cut the capacitance down.  This helped.   

First picture showing the fundamental and first harmonic.  This is with 10dB attenuation.     The second picture, zoomed in to the fundamental with the attenuator set to zero.  We lost a little gain.    The fan blowing the power leads makes a difference.   It's really unstable. 

Also shown with it attached to the old LeCroy.  Still have not found the missing knob.   I'm surprised we can see anything with it being well over double the 600MHz. 

I did find an old perfboard that had similar 2 row connectors like what you see typically for the power rails but instead they were 2pin pairs.  If you could make a custom perfboard with these attached to the edges, we could drop some of that capacitance. 

[RoGeorge]

For a schematic with a diode that generates avalanche noise near reverse breakdown, followed by a filter and an amplifier, would that be cheating?

[BreakingOhmsLaw]

For a schematic with a diode that generates avalanche noise near reverse breakdown, followed by a filter and an amplifier, would that be cheating?

If you can pull that off with discrete components, go for it.
I would expect that you will have a hard time getting a narrow enough discrete filter to pick a single frequency from noise. But i'm always ready to be proven wrong.
Therefore, my advice would be to build the fastest oscillator you can, use a snap-off diode to clamp it, and filter one of the fundamentals that are created by the brutal distortion of the diode. Harmonics will have enough separation to have a chance of selecting one without an opamp.

[BreakingOhmsLaw]

While your rules will not allow removing or cutting them, we certainly can skip over one to cut the capacitance down.  This helped.

It's really unstable.

Hint: The rules don't forbid shielding. Recycle some tinfoil hats.

[joeqsmith]

3. All electrical connections between individual parts must be made by the breadboard contacts.
...
5. Self-made discrete parts are allowed and encouraged.

Please provide a clear definition of "individual parts" and "discrete parts".     

The wires used to connect the components to the breadboard, at what point they are themselves considered an individual or discrete part, requiring each lead to be inserted into the breadboard?   

In picture10 for example, which of these are considered inductors and require both ends be inserted into the breadboard?

In picture11 we have two wires that came a little too close to one another and have some strange twists.  Is this just interconnection wire, or is this an LC filter with some series resistance?

[BreakingOhmsLaw]

All of these are fine - the rule is in place to prevent that people will do all the connections free floating with solder and bypass the breadboard.
Your examples are fair game. Twisting wires to get an inductive transfer is allowed and encouraged ( and definately needed where you are headed  )

"individiual parts" as in separate parts that are not physically joined. A transformer may be two or more coils from an electrical point of view, but they cannot be separated.
"discrete parts": That definition is hard to make.
Classically: An electronic component with just a single circuit element.
As for transistors that have internal protection circuitry (diodes, resistors): Allowed, as long as that protection element does not become an integral part of the oscillator.
I also would not have anything against resistor or capacitor arrays. They may have an internal common pin, but you still need to wire each one individually.

[BreakingOhmsLaw]

So, a brief summary so far:
We had some good suggestions from users how to approach this. Joeqsmith has set the bar high right from the start, honoring the old RF engineers mantra, that everything below 1GHz is DC.
Hopefully some other members are busy cooking up circuits and reaching for the title.
If you have built a circuit, but haven't posted it yet because you are below the currect record, i still encourage you to post it for the educational value and to get suggestions from other members on how to improve it.

Generating RF from scratch is a fun way to touch a lot of topics: oscillators, filters, shielding, parasitics, analog circuit design and many others.
What inspired this thread is that i remembered engineering school nearly thirty years ago. One of the teachers, in order to make studies more exciting for his students, always had us do stuff like this instead of written homework. I remember many hours hunched over a stripboard, fiddling current mirrors, amplifiers, opamps and whatnot. In hindsight, I learned more about electronics doing that, than in all the other classes combined. Sadly, that teacher passed away two years ago. I have much to thank him for.
So, Rüdiger, if you are up there reading this: Thanks, you were a legend!

Ok, enough nostalgia. Carry on!

[joeqsmith]

All of these are fine - the rule is in place to prevent that people will do all the connections free floating with solder and bypass the breadboard.
Your examples are fair game. Twisting wires to get an inductive transfer is allowed and encouraged ( and definately needed where you are headed  )

"individiual parts" as in separate parts that are not physically joined. A transformer may be two or more coils from an electrical point of view, but they cannot be separated.
"discrete parts": That definition is hard to make.
Classically: An electronic component with just a single circuit element.
As for transistors that have internal protection circuitry (diodes, resistors): Allowed, as long as that protection element does not become an integral part of the oscillator.
I also would not have anything against resistor or capacitor arrays. They may have an internal common pin, but you still need to wire each one individually.

The problem is even something as simple as a wire can be fairly complex.   I didn't spend much time trying to optimize the LC until I make sure you understand that the whole idea of using it is to isolate the base from that large capacitor.  I've attached a circuit diagram that I hope will explain it.  It's a three wire device.  Two wires are plugged into the breadboard, the other is soldered directly to the base.  The twisted section forms the capacitor, so it's not directly touching anything and for DC is an open. 

The knobless LeCroy shows it at 22GHz which is obviously not close to correct.  The signal is too high for the 600MHz scope and it attenuates it too much to get a good reading. 

The antique DSO seems to do a decent job.   This scope has a 4GHz front end but the signal isn't stable enough with RIS.

If this technique is deemed legal,  the next step would be to tune it and add a similar stage to the output. 

[joeqsmith]

If the previous LC circuit is deemed acceptable,  I  would like to use this conical coil.   

This should provide some idea just how unstable it is.  As long as this isn't a metric being used and it is solely based on output frequency, I'm fine with it. 

[joeqsmith]

If the conical magnet wire LC circuits are alright to use,   I would then like to also use this same technique to couple to the output coax. I don't like putting DC into my test equipment if it can be helped.  I would use an external block but that seems like it opens a can of worms for other in-line devices.  So for now, I will stay with direct connection using coax.   Using the magnet wire to form the DC block would remove one column of contacts further reducing the loading capacitance. 

Someone had recently asked  about using nail polish for RF work.  I had ran some tests on this particular bottle and would like to introduce it as a method to stabilize the coils.   

https://www.eevblog.com/forum/rf-microwave/fingernail-polish-and-rf-front-ends-do-they-mix/

If these methods are still within the rules, I would like to submit the next entry. 

In realtime, my highest end scope can only sample at 20GS/s in realtime.  That's a bit too under sampled, so I captured it in RIS at 200GS/s.   It has a nice shape to it.   

With this last LC network, the base is basically decoupled from the column of contacts (AC wise) and the connection is purely to satisfy the rules of wanting the parts attached to the breadboard.   However, nothing else connects to that section except through parasitics.   It would be nice to remove that inductor.

As the speeds continue to go up the definitions become more important.   

[RoGeorge]

New rule:
Close-up photo of the breadboard required!   

[joeqsmith]

New rule:
Close-up photo of the breadboard required!   

Personally, I don't have any problems with that.  Here are a few.   

[RoGeorge]

That's a good start, waiting for the pics with the rest of the circuit... 

Meanwhile curios to take a closer look at this mystery circuit here, please:

[joeqsmith]

I've tried to clear the way for the real high speed designers on the forum that Tim mentions.   Surely they will topple any of my lame efforts with their 10GHz+ equipment. 

Also if you allow other types of breadboarding besides the solderless socket kind, you may find quite impressive results from some members here (10GHz+ test equipment included ).

Tim

Rather than try and follow, step up and show us what you can do. 

[joeqsmith]

Of course, if you really want to see the next revision up close, watch the start of this next clip.  I'm sure I am pushing the rules with this one but it's just hookup wire.   Up to the OP to decide if it's a valid approach.   If this is within the rules, I think 3G is on the table.   

[BreakingOhmsLaw]

Enamel wire makeshift-LC equivalents are okay!

[T3sl4co1l]

Rather than try and follow, step up and show us what you can do.

What if the problem is @Hugoneus and others, don't have a solderless breadboard?

Tim

[T3sl4co1l]

If the conical magnet wire LC circuits are alright to use,   I would then like to also use this same technique to couple to the output coax. I don't like putting DC into my test equipment if it can be helped.

How about using two adjacent rows?  Typically about 4pF.

Tim

[joeqsmith]

Enamel wire makeshift-LC equivalents are okay!

   This makes the rules pretty clear.  Now all we need are the challengers. 

Rather than try and follow, step up and show us what you can do.

What if the problem is @Hugoneus and others, don't have a solderless breadboard?

Tim

Good point but surely if you can afford 10GHz TE, you can pick up a $10 breadboard.   You can't let this old hobbyist with his slow equipment that the knobs won't even stay on, become the "incumbent master oscillator"     Come out and show us those honed skills of yours!! 

***************
Many are under $5. 

https://www.digikey.com/products/en/prototyping-fabrication-products/solderless-breadboards/638?k=breadboard

[joeqsmith]

If the conical magnet wire LC circuits are alright to use,   I would then like to also use this same technique to couple to the output coax. I don't like putting DC into my test equipment if it can be helped.

How about using two adjacent rows?  Typically about 4pF.

Tim

The problem are each row has a lot of parasitics that you need to consider.   Mine is mounted to some plastic to get it up off the copper.  Our host has been pretty generous with the rules  and I wouldn't call fowl if someone mounted their breadboard to a half inch slab of Teflon.

[T3sl4co1l]

Eh so what, you're just trying to get the signal out, not get a 1.05 VSWR.

Oh, on that subject -- I would suggest flexibility in that, because I have several breadboards with metal base plates.  They aren't particularly useful as RF ground planes -- there aren't any vias to them, let alone exactly where and when you need a via for a particular layout -- but I have found them useful from time to time by putting a binding post into the metal, and getting at least some electrostatic shielding that way.  (Often the binding posts are provided, they're just insulated with washers; removing the washers and scraping some paint is all that's needed to guarantee a proper connection.  Perhaps not technically stock anymore, but an easy enough mod?)

Tim