making a frequency doubler circuit, want another set of eyes on it

[Kyle_from_somewhere]

Hello again all.

Today I've been looking at frequency multiplication. I found an article on the design of a frequency multiplier that seemed interesting, so I put it together in a simulator.

The original magazine says their example circuit is ran as a tripler going from 50Mhz to 150Mhz. I'm more interested in a doubler, so I punched in a value of 680nH for L1 (low pass filter?), and 160nH for L2 (output tank circuit).

CircuitJS doesn't simulate any inherent resistances to parts, and it gets fussy sometimes about "capacitor loops" formed by too many of them going to ground. So that's why I put a couple of 1 pico-ohm resistors in there. The 60mhz signal source can have voltage set, but not current, so I have put a 50ohm resistor on there.






I have some questions:

1) neither peak-to-peak voltage nor frequency is entirely stable. It likes to wander about 0.2V, and occasionally blips from 120mhz to 119.8 or 120.2. Is that normal? Would it cause any meaningful problem for AM modulation?

2) If the 60Hz sine wave input is replaced by 3.3V DC and the simulation reset, the output still oscillates at 120.5Mhz. It also increases in voltage until it hits 70V. Is this a quirk of the simulation? I assume I've not accidentally invented a new kind of oscillator circuit.

3) If I was to build this circuit in real life, is there any obvious design flaw that would stop it from working? Have I made any mistakes? I know that L1 and L2 will need to be tuned.


Thanks for your time.

[ledtester]

I would use LTspice instead of Falstad for this kind of simulation. It shouldn't have the same quirks like the "capacitor loops" problem you ran into. Also, you can use the model of a real transistor (like a 2n3904 or some other) instead of a generic model.

I've drawn up the schematic for "Figure 6" in the article trying to keep the same component ids.



and have also attached the .asc file below.

I changed the values of L1 and C3 because in the article it says "The tuned circuit of L1 and C3 selects the required harmonic." so it should select for 120 MHz.

As for your questions:

1) In the LTspice simulation the output is very strong with the output being 16Vpp (and this is with a 2Vpp input.) Have a look at the LTspice output and see if you're seeing the same kind of modulation.

2) I don't see this with LTspice.

3) Not sure! Build it and let us know how it works out!

Update: fixed the base connection of the transistor

[Benta]

I've drawn up the schematic for "Figure 6" in the article trying to keep the same component ids.

Missing connection at the base of Q1. Not sure what you're simulating.

[ledtester]

I've drawn up the schematic for "Figure 6" in the article trying to keep the same component ids.

Missing connection at the base of Q1. Not sure what you're simulating.

You're right! Fixed now in the above post. I should build it with the missing bias connection to see if it really works!

Btw, I think the reason I missed that connection is because I recently ran across this article:

https://www.analog.com/en/technical-articles/ltspice-connecting-the-dots.html

which explains that you can connect several components that are in straight line by just drawing a single wire through all of them, and I used this feature in drawing up this schematic. But I overlooked the fact that it won't make connections where another wire crosses.

[Kyle_from_somewhere]

Hello ledtester! thank you for taking a look.

I suppose you're right that really I should use LTspice over the javascript simulator. To be honest I only heard of spice here on this forum a few days ago. LTSpice is very difficult to use, I can't figure out how to do anything more simple than click "simulate" and see a scope of Q1 set to totally the wrong timescale. The developers should be told about all the exciting UI advancements that came along after the xerox star. Scrollbars would blow their fuckin minds.

Regardless I can see your version is doing something, and it's on me to figure out this program. So thank you.

I had read that L1 must be tuned to the harmonic, but circuitjs no-no'd it. of course that would make sense if it doesn't simulate harmonics in the first place.

edavid, How practical is using transformers for frequency doubling at VHF Frequencies? The link you posted describes it as big and fragile, and it's going to need amplifying anyway... Interesting idea though.

[DimitriP]

To be honest I only heard of spice here on this forum a few days ago. LTSpice is very difficult to use, I can't figure out how to do anything more simple than click "simulate" and see a scope of Q1 set to totally the wrong timescale.

A- for honesty
All sorts of LTSPICE tutorials everywhere.

The developers should be told about all the exciting UI advancements that came along after the xerox star. Scrollbars would blow their fuckin minds.

D for attitude

[Kyle_from_somewhere]

The developers should be told about all the exciting UI advancements that came along after the xerox star. Scrollbars would blow their fuckin minds.

D for attitude

When a program is used by laying symbols down on a worksheet that's bigger than the display, then lacking scroll bars is bad design, and I don't feel the need to be polite about it. We've known scroll bars are the best way to move the view around for almost 40 years. They are standard. It doesn't respond to the standard keyboard based controls either, so it's not even a matter of thinking early PC style.

I don't think it's unfair to say that, while it might be a very good computer program in the sense that it can take data in and then spit out answers, it's clumsy at being a user-interactive one. It's development apparently started in the 1960s, so this clumsy UI is probably "good enough" to those who started out with it by handing off stacks of punch cards to a receptionist and then getting the output printout back a week later.

People using a computer can get used to any program after a while, it's mostly just a matter of remembering where everything is. But unnecessary differences from the standard way of doing things makes that process take much longer, so should be done away with, even if it's developed in that way for historical reasons. Traditions are the deads' way of weighing down the living.

Edit: I discovered three quirks of this great program. Firstly, every third time I open it, it complains that it couldn't send a command to itself. This doesn't seem to mean anything. Second, when I looked up how to turn the frequency counter on I discovered that there isn't one. You have to zoom in very closely and take your best guess at where the peak of the sine wave is. When zoomed to 1 and 1/4 of a cycle, maximized on a 1080p screen, the absolute peak is still not visibly distinct. So you can only get a reading on a 120mhz signal with a margin of error of 4mhz. Thirdly, I apparently minimized it "too much" while switching windows to a tutorial web page, and crashed it.

Adobe quality at a linux price.

It's the most common method.  It's not big and fragile at VHF, and doesn't necessarily need an amp (you can adjust the transformer turns ratio).  All you need is a small toroid and 2 diodes.

Here's a commercial version: https://www.davmar.org/TE/HP10515A

Interesting, interesting. I looked at that mini-circuits company you suggested too. Pretty expensive compared to self build, but I guess it comes with a proper warranty.

Thing is, if transformer doubling is so good, what makes people come up with transistor based methods? There must be a trade off somewhere.

[DimitriP]

Traditions are the deads' way of weighing down the living.

"All  generalizations are false including this one"
Perhaps you can take a breath. You are probably a very nice person in real life but coming into a forum and start blasting LTSPICE and minicircuits is not a way to get a co-operative spirit going.

I don't want to hijack the thread, I much prefer reading about frequency doubling to arguing about software interfaces in a frequency doubling thread.
You might want to visit https://groups.io/g/LTspice
As for scrollbars if I zoom in enough, suddenly, scrollbars appear !
 
Regarding minicircuits, by the time you design test , correct and produce  "a few" pieces of the "thing" versus ordering from minicircuits the price difference starts to fade.

...and now we'll return to our regularly schedule programming ...

[Kyle_from_somewhere]

You are probably a very nice person in real life but coming into a forum and start blasting minicircuits

I've done no such thing.

I have blasted LTSpice's bad UI, but unless LTSpice is your mother I don't see why you need to take personal offense.

That you have to zoom in so far that one symbol barely fits the screen before it lets you use a scroll bar is, to be honest, worse than not having them at all. That's a setup for a totally unrealistic use case.
 

Regarding minicircuits, by the time you design test , correct and produce  "a few" pieces of the "thing" versus ordering from minicircuits the price difference starts to fade.

The BOM for the circuit the thread is about is about 75 eurocents, including sales taxes. A frequency doubler from minicircuits is over $50US. Do you have anything useful to contribute?

[TimFox]

You are probably a very nice person in real life but coming into a forum and start blasting minicircuits

I've done no such thing.

I have blasted LTSpice's bad UI, but unless LTSpice is your mother I don't see why you need to take personal offense.

That you have to zoom in so far that one symbol barely fits the screen before it lets you use a scroll bar is, to be honest, worse than not having them at all. That's a setup for a totally unrealistic use case.
 

Regarding minicircuits, by the time you design test , correct and produce  "a few" pieces of the "thing" versus ordering from minicircuits the price difference starts to fade.

The BOM for the circuit the thread is about is about 75 eurocents, including sales taxes. A frequency doubler from minicircuits is over $50US. Do you have anything useful to contribute?

Apparently, you clicked on the first item in the MCL website, which is a $65 device in a proper hermetic metal package.
In SMT, you can find  https://www.minicircuits.com/WebStore/dashboard.html?model=AMK-2-13%2B  and others for $6.95 US each, but they have a minimum order quantity from the factory.

You may be able to find a distributor:  MCL has a huge catalog and make very good devices. 
One of my surplus dealers has a hermetic MCL doubler for $11.00 US (singles) on this page  https://surplussales.com/Mini-Circuits/Misc.html

MCL's data sheets are very complete, and the devices are well-characterized.  Of course, with transformers and diodes they will have less than unity gain.  MCL's core competences are wide-band transformers and diode-based modulators/multipliers, etc., and they also sell couplers, amplifiers, and many other RF modules.

I don't use LTSpice, but an old-fashioned line-oriented Spice from AimSpice.  When I was using PSpice decades ago (before they merged), I had problems with .TRAN analysis with resonant circuits, especially with high-Q.  You need to adjust the Spice simulation parameters to get good results with time-domain analysis of resonant tanks in non-linear circuits.  The .AC analysis is only for linear circuits, and won't work with a frequency multiplier.

[jonpaul]

Bonjour: Easiest solution:

wideband center tapped transformer $0.50
2 x 1N4148 or similar small sig diodes $0.03

SIG>>Transf >>FW rect>>DOUB

NO simulation, no power


Jon

[iMo]

Any none-linear circuit works as a doubler/tripler/..
A linear circuit, like a linear transformer, LC, or RC, or a perfect amplifier will not work as a X-ler..

When doing it with the transistor stage as you did - you have to set the transistor working condition such it is (highly) none-linear.

Then in its collector circuit many "harmonics" of the input signal appear (ie. a square wave is good to have there - it is full of harmonics).

Then you put there an LC tank tuned to the second (or third, or X-th) harmonics of the input signal, and feed that harmonics out of it. Mind with a simple LC tank you still will get many harmonics in your signal, the second (or X-th) being the strongest. You have to use more LC selectivity (more LC stages) to filter out your specific harmonics only.

Hints:
1) learn basics of radio-electronics,
2) learn to simulate circuits in Spice or in the LTSpice..
3) it must be a "highly none-linear circuit" to generate a mess of input signal harmonics such you can choose a specific harmonics from..

PS: below an example - a tripler

[Kyle_from_somewhere]

Works absolutely fine. A little sensitive to RF interference as a point to point circuit though.

[RoGeorge]

Works absolutely fine. A little sensitive to RF interference as a point to point circuit though.

 

To make it less sensitive to interference, keep the parts closer, with smaller loops, because any loop is also an inductance, and the bigger the area, the more interference captured.

A can-like shield enclosure will have a big influence in stability to both interference and external parasitic capacitance (like for example when your hand comes closer to the circuit).  W2AEW has a nice video showing how much a shield can help, even when it's just a partial shield, not a full enclosure:  https://youtu.be/fpD_mDCViPE



Later edit:
-------------
I think the bypass C2 from the first schematic must be tied from the coil to ground, not to V+, like it is now.  That will reduce a lot the sensitivity to parasitic influences, because it will keep the RF signals close to the circuit.  Like it is connected now (in parallel with the power filtering resistor R4) C2 is doing the opposite, it directs RF currents to go through the battery and its long wires, and that is exactly what it has to be avoided.

[TimFox]

Here is a good general reference to many methods of frequency multiplication:  https://www.qsl.net/va3iul/Frequency_Multipliers/Frequency_Multipliers.pdf
The final section, pp 15-25, discusses "active" circuits (powered transistors).
Note that the "conduction angle" of the device has an optimum value for efficiency, depending on the multiplication factor.
Back in the day, when dual pentodes (e.g., 6360) were in vogue for RF, frequency doubling would connect the two devices as "push-push" (plates in parallel), while tripling would connect as "push-pull" (plates on opposite ends of output circuit), with balanced (push-pull) grid drive.  Again, this was to optimize the power in the desired harmonic, but the output circuit always was tuned to attenuate the other harmonics.

[GumpyGus]

Doublers and Triplers and higher are easy, it's bound to happen in any stage other than a perfectly linear amplifier.

So almost anything you do with a transistor or diode, biased a bit off-center, it's going to multiply.   Hey, you could even use a multiplier!

Your circuit with a transistor is one classic way, you're going to see this in the schematic of any 1960's VHF walkie-talkie.  You do need a high_Q collector tuned circuit to surpress the fundamental and second harmonic. 

Simulation apps may tend to do a less than perfect simulation as you're depending on the non-linear parts of a specific transistor transfer curve.  The harmonic output could easily be off by several decibels.  As could the real-world circuit, for similar reasons.

[Kyle_from_somewhere]

To make it less sensitive to interference, keep the parts closer, with smaller loops, because any loop is also an inductance, and the bigger the area, the more interference captured.

A can-like shield enclosure will have a big influence in stability to both interference and external parasitic capacitance (like for example when your hand comes closer to the circuit).  W2AEW has a nice video showing how much a shield can help, even when it's just a partial shield, not a full enclosure:  https://youtu.be/fpD_mDCViPE

Already understood. Was just thrown together quickly to test the idea of it.

Later edit:
-------------
I think the bypass C2 from the first schematic must be tied from the coil to ground, not to V+, like it is now.  That will reduce a lot the sensitivity to parasitic influences, because it will keep the RF signals close to the circuit.  Like it is connected now (in parallel with the power filtering resistor R4) C2 is doing the opposite, it directs RF currents to go through the battery and its long wires, and that is exactly what it has to be avoided.

Yeah that was my mistake, I made it while I was very tired.

[Kyle_from_somewhere]

Back at at it again. Got some PCBs printed with the circuit repeated across it three times. My goal is 480Mhz.

To double 120Mhz to 240 I shortened the coils to 40n. My frequency counter reads no output. In LTspice it runs, but it's not a sine wave anymore. Each peak is a different height.

Thoughts?

[Gerhard_dk4xp]

That's not unstable, there are still lots of the 120 MHz fundamental in the output
signal. What you need is more resonators in your output filter; maybe you should
also play with the filter part values.

The output of 2 parallel 74LVC125 gates + 68 Ohms each,  driven with a 100 MHZ
square wave contains lots of usable frequencies if you can find cheap filters to select
only the one you want.
Doublers and quadruplers are badly served when you start with a 1:1 square wave,
ask Mr. Fourier. Nevertheless that worked ok when I just dumped the the output of
the 2 parallel LVCMOS-gates @ 5V into an EPCOS/TDK/qualcom SAW-Filter.
A SKY 65017 MMIC in sot-89 recovered the level, but I needed another SAW filter
to get really clean 400 MHz. But it worked in one hopp from 100 to 400 MHz.

For 100-> 300 MHz I got no SAW and used 2 three-pole LC-filters with 0603 coils.
For 480 there is the ecs-d480a filter (Digikey), but in my notes I have tagged it as
quite lossy.

regards, Gerhard

[Kyle_from_somewhere]

I call it unstable because no peak is alike. It's going up and down constantly. It's also completely inefficient, gone from 2.4V peak-to-peak to microvolts.

Duplicating the "resonator" (I assume you mean L4 and C10) just makes the voltages lower with no improvement. Adjusting the values by little amounts results in a waveform with weird steps in it and a frequency that doesn't in any way resemble a multiple of 60mhz. Increasing the coils by large amounts requires a capacitor too small to exist, increasing capacitor requires a coil too small to exist.

[fourfathom]

I call it unstable because no peak is alike. It's going up and down constantly. It's also completely inefficient, gone from 2.4V peak-to-peak to microvolts.

Duplicating the "resonator" (I assume you mean L4 and C10) just makes the voltages lower with no improvement. Adjusting the values by little amounts results in a waveform with weird steps in it and a frequency that doesn't in any way resemble a multiple of 60mhz. Increasing the coils by large amounts requires a capacitor too small to exist, increasing capacitor requires a coil too small to exist.

Note that your "instability" repeats every four cycles.  What you are seeing is happening because the first doubler stage is kicked by the 120 MHz signal and it rings at 240MHz as it should.  But the loaded Q of the 240 MHz resonator is low and the signal is decaying rapidly.  You then drive the second doubler with this decaying signal and the 480 MHz output reflects the varying drive levels and it's own rapid decay.  You might have a more uniform output if you increased the loaded Q of your resonator tank circuits (by reducing the loading or by reducing the impedance of the L and C).  But increasing the Q of a single-stage resonator requires stability and accuracy of your resonant circuits, so there comes a point where they are just too difficult to tune. and you need to look at SAW filters or (and I suggest this) a new topology such as a transformer/diode doubler followed by an amplifier and filter.

[Kyle_from_somewhere]

The first doubler receives 60Mhz and outputs 120Mhz, the second doubler is to take 120Mhz and spit out 240Mhz, and that's where the problem is. A third doubler would give 480Mhz but we're not doing that yet.

I know that running multiple of this kind of doubler in series in this frequency range is possible and done in a lot of projects from old electronics magazines of the 80s. I'm looking at one right now which uses a very different component selection, including much bigger L and C values. The suggestion to give up and use a completely different methodology is unhelpful - if you were to ask me how to change the final drive ratio on your honda, I'm not going to tell you to ride a bicycle instead.

I think my fundamental problem is that I have no idea how to select components beyond using online calculators. Working with transistors is generally incomprehensible, people talk about biasing a transistor to be "linear" or "non-linear" or to have a certain "Q" but there's nothing in any datasheet which I can relate to those concepts, nor anything written online that teaches it.

[fourfathom]

The first doubler receives 60Mhz and outputs 120Mhz, the second doubler is to take 120Mhz and spit out 240Mhz, and that's where the problem is. A third doubler would give 480Mhz but we're not doing that yet.

OK, but my observation still holds: your "instability" is the result of the low-Q of your tank circuits causing the ringing waveform to quickly decay at each doubler stage.

I know that running multiple of this kind of doubler in series in this frequency range is possible and done in a lot of projects from old electronics magazines of the 80s. I'm looking at one right now which uses a very different component selection, including much bigger L and C values. The suggestion to give up and use a completely different methodology is unhelpful - if you were to ask me how to change the final drive ratio on your honda, I'm not going to tell you to ride a bicycle instead.

We *are* trying to be helpful.  Why don't you post one of these schematics that in your estimation *does* work and we can make some comparisons.  And suggesting an alternate approach can often lead to a more optimum solution.  The method you have chosen is intrinsically not a great doubler circuit, although it can be made to function with some care.  It makes a pretty good tripler though...  The transformer/diodes and several other circuits mentioned have by design a cleaner 2'nd harmonic output and usually require less filtering to generate a usable signal.

I think my fundamental problem is that I have no idea how to select components beyond using online calculators. Working with transistors is generally incomprehensible, people talk about biasing a transistor to be "linear" or "non-linear" or to have a certain "Q" but there's nothing in any datasheet which I can relate to those concepts, nor anything written online that teaches it.

In that case you should probably find a known-good circuit and copy that.  Trying to design an amplifier or doubler (or whatever) without understanding biasing or Q is going to be very difficult.  Now if you are trying to learn those things by building and testing circuits like this then I applaud you.  In that case, do look at the Radio Amateurs Handbook and similar "RF for Beginners" (no pejorative implied) types of books.  That's how I learned.

[Kyle_from_somewhere]

The example I'm thinking of is this TV transmitter project, Radio Electronics Magazine June/july 1989

Part 1: https://archive.org/details/radio_electronics_1989-06/page/n37/mode/2up?view=theater

Part 2: https://archive.org/details/radio_electronics_1989-07/page/n41/mode/2up?view=theater

I've clipped the relevant part of the diagram. For some reason the authors think themselves above specifying inductance (why is this so common?), but they're successfully using a chain of transistors to double the frequency. I tried to upload the coil table but there are too many artifacts in the image to make a file small enough.

Learning things like how to bias a transistor for a particular result is very difficult because the internet makes for a bad educator. It's easy to stumble across a hundred youtube videos with a title like "how to bias a transistor", but every single one just has an overcaffinated guy with a whiteboard start by pulling two resistor values out of his ass, calculating the last two based on them, and then declaring it done. No explanation of what makes it linear or non linear, in fact no mention of the concept of a transistor operating linearly or not, just "it's biased now, don't forget to click like and subscribe".

Trying to find something written down using google is even worse. You fall into forum threads where the youtube scenario is repeated almost word for word, followed by someone saying to use an op-amp instead - without even the courtesy of explaining how. Just like here I'm told to use a transformer and two diodes, without being told how.

I don't want to just blindly copy someone else's circuit because often they're ancient and the parts aren't made anymore, and even if they were available I don't learn anything. But It's also difficult to learn to design my own circuits when the only "education" I can access is people doing stuff without explanation, or people saying "do something else instead" also with no explanation. I cannot go to university.

[Kyle_from_somewhere]

Bonjour: Easiest solution:

wideband center tapped transformer $0.50
2 x 1N4148 or similar small sig diodes $0.03

SIG>>Transf >>FW rect>>DOUB

NO simulation, no power


Jon

I've attempted to simulate the transformer and diode approach. No combination of inductor values will accept 120Mhz and turn it into 240Mhz. The most I've managed is a very uneven 190Mhz output.

[iMo]

1. the 2N3904 is not a high freq transistor, you should use a different one (w/ Ft>1-3GHz)
2. the "impedance matching" between the stages is important in ALL your circuits above (I do not see any matching there)..

[Kyle_from_somewhere]

1) If this transistor isn't suitable, then what is?

2) To match impedance, you must know what the impedances are. I don't. I can't find out how to find out.

If you don't bother explaining your remarks, they are useless.

[iMo]

Do not expect people will be elaborating pretty obvious stuff.. 

PS: Input/output impedance of a circuit - for example do google "LTSpice input output impedance measurement"

[Gerhard_dk4xp]

removed the text.
no point in trying to help someone without basic knowledge
who holds others responsible for it.

[Kyle_from_somewhere]

There's no BAT15 in LTspice. Can't find anything to download to add one either. So I've left it at 1N4148 until a better option presents itself.

Adding 1uH from output to ground just slows output down to 107 Mhz.

[Kyle_from_somewhere]

removed the text.
no point in trying to help someone without basic knowledge
who holds others responsible for it.

When I ask a question and a man replies "i don't want to explain something so obvious" then it is his responsibility. I did my part by asking.

Sorry that I didn't come from the womb knowing all the things that you find are so obvious.

[RoGeorge]

Do not expect people will be elaborating pretty obvious stuff.. 

You're going on my ignore list for being a deliberately unhelpful dickhead.

And you go to everybody else's ignore list, starting with mine.

And if you don't have the knowledge to understand the answers here, then you should have asked for clarification, and not start calling names.  Adios!

[Kyle_from_somewhere]

And if you don't have the knowledge to understand the answers here, then you should have asked for clarification, and not start calling names.  Adios!

I did. As you know (because you can read), he refused to clarify.

Glad to no longer see you around, whoever you are.

[TimFox]

There's no BAT15 in LTspice. Can't find anything to download to add one either. So I've left it at 1N4148 until a better option presents itself.

Adding 1uH from output to ground just slows output down to 107 Mhz.

The 1N4148 has a reverse recovery time of 4 to 8 ns, depending on current.  That won't work well at your frequencies.  The BAT15 is a Schottky diode rated for GHz operation.

Maybe try an 1N5712 Schottky or similar.  This link has a datasheet and a Spice model.  https://www.farnell.com/datasheets/89202.pdf

[Kyle_from_somewhere]

The "spice model" is a table. How do I enter it into LTSpice?

[TimFox]

I don't use LTSpice, I use Spice.  Those are the normal co-efficients for a diode model in Spice.
Check the documentation on LTSpice on how to enter the model co-efficients.
Perhaps you can find what you need about these diodes somewhere by Googling.
This took me 30 seconds:  https://ltwiki.org/files/LTspiceIV/Vendor%20List/Diodes%20Incorporated/Spice/spicemodels_schottky_diodes.txt

[Kyle_from_somewhere]

I don't use LTSpice, I use Spice.  Those are the normal co-efficients for a diode model in Spice.
Check the documentation on LTSpice on how to enter the model co-efficients.
Perhaps you can find what you need about these diodes somewhere by Googling.
This took me 30 seconds:  https://ltwiki.org/files/LTspiceIV/Vendor%20List/Diodes%20Incorporated/Spice/spicemodels_schottky_diodes.txt

Okay, but you could have spent less than 30 seconds to say "i don't know, sorry" instead of posting a random file you didn't actually look at and pronouncing yourself the king of google.

If you'd looked at it, you'd see nowhere is something like "1.1 x 10E-9" able to go.

Hell, if you'd looked at the PDF you'd posted you'd know there isn't even Spice parameters for the diode you mentioned, only other ones.

[TimFox]

The very first entry in that link is the model for a 1N5711, which was referred to in my earlier link along with the 1N5712 and other similar devices.
Apparently you only looked at that link and did not understand it.

  ".MODEL DI_1N5711 D  ( IS=315n RS=2.80 BV=70.0 IBV=10.0u
   + CJO=2.00p  M=0.333 N=2.03 TT=1.44n )   "
is the Spice model.  What do you see when you examine your LTSpice model, which should be a text file.

I can't do all your work for you, does that make me nasty?
OK:  besides YouTube videos on this very topic, I found
https://electronics.stackexchange.com/questions/397354/how-can-i-convert-pspice-model-into-ltspice-model

[Kyle_from_somewhere]

The very first entry in that link is the model for a 1N5711, which was referred to in my earlier link along with the 1N5712 and other similar devices.
Apparently you only looked at that link and did not understand it.

  ".MODEL DI_1N5711 D  ( IS=315n RS=2.80 BV=70.0 IBV=10.0u
   + CJO=2.00p  M=0.333 N=2.03 TT=1.44n )   "
is the Spice model.  What do you see when you examine your LTSpice model, which should be a text file.

A completely different format:

Code: [Select]

.model 1N914 D(Is=2.52n Rs=.568 N=1.752 Cjo=4p M=.4 tt=20n Iave=200m Vpk=75 mfg=OnSemi type=silicon)
.model 1N4148 D(Is=2.52n Rs=.568 N=1.752 Cjo=4p M=.4 tt=20n Iave=200m Vpk=75 mfg=OnSemi type=silicon)
Copy and paste what you said into LTSpice and nothing happen. Copy and paste it into the diode file of LTSpice with notepad: nothing happens, no new diode is avaliable to use.
Save entire text file you linked and open it in LTSpice: LTSpice does not understand the file format.

I can't do all your work for you, does that make me nasty?
OK:  besides YouTube videos on this very topic, I found
https://electronics.stackexchange.com/questions/397354/how-can-i-convert-pspice-model-into-ltspice-model

You haven't even done your own work. For one, there are no youtube videos showing how to enter a table of spice parameters from a datasheet into LTSpice. I know because I looked. If you search for them, you get people modelling ICs from datasheets without spice parameters in them. Or you get a guy showing how to create spice parameters from scratch using very long physics formulas, but not how to enter them into the program.

Then you link stackexchange thread about converting ready working model files from pSpice, which is also nothing to do with it.

As I said, if you don't know, it is OK, just say you don't know. Don't give wrong advice and then get mad I say it doesn't work.

[Kyle_from_somewhere]

Let it be a lesson to us all: if you ever want to accuse someone of not knowing how to look stuff up, don't be like timfox. Don't fling random links you haven't checked.

[TimFox]

There are other Schottky diodes in the LTSpice library.  I couldn't find the 1N5711, but I gave you the Spice parameters from two sources.
Have you tried real Spice?
I have grown bored with this discussion.
I got along without you, before I met you, gonna get along without you now.

Let this be a lesson to all of you out there:  this guy can't be satisfied even with real data, if he doesn't understand it.

[Kyle_from_somewhere]

I have grown bored with this discussion.

No, I don't think you got bored. I think you're the typical tech expert elitist on the internet. You can't answer simple questions because you think they're below you, and you DEFINATELY can't admit that there are some things you don't know, because that would damage your self image as an electronics demigod (This is also why you brag about only using "Real" spice, presumably meaning the one that was released so long ago no modern computer can natively run it).  So instead you just scream the mantra of "google it" as if the person you're talking to didn't.

Here today you embarrassed yourself by throwing google results at me without bothering to read them yourself to see if they're applicable.

So, you do what everyone who embarrasses himself on the internet does: you're trying to play it off with an air of superiority. I'll give you some credit, you went with the faux upper class "i have grown bored of you", rather than the more typical "i'm not mad, I'm laughing actually".


Seriously, you could've just said "Sorry, I don't know how to do that in LTSpice", instead of this.

[TimFox]

No, I am not an expert on LTSpice.
I suggested that 1N4148 was not a good choice for your circuit, and recommended 1N5711, 1N5712, etc. instead.
I sent you two sites with legitimate Spice models for 1N5711-type Schottky diodes, but all I got back was rude abuse.
Since you use LTSpice, I assumed you knew how to use it.
Good bye.

[Kyle_from_somewhere]

I sent you two sites with legitimate Spice models for 1N5711-type Schottky diodes, but all I got back was rude abuse.

No actually that's not what happened, Tim. I asked you what to do with the data provided, then you snottily took on airs:

Perhaps you can find what you need about these diodes somewhere by Googling. This took me 30 seconds

The "this" in question was a text file that does not work in LTSpice, but you thought that it did because all you'd done is read the URL and assume. That's on you.

You really shouldn't act so smug about your superior googling ability when it gets you wrong answers.

[TimFox]

That text file in my second link was a data library containing another normal Spice file for the 1N5711.
You didn't read it, or you would have noticed "1N5711".
'bye!

[Kyle_from_somewhere]

I did notice 1N5711, you should know this because I told you I attempted to get LTSpice to load it.

You shouldn't go around accusing others of not googling when you can't google properly, accuse others of not reading when yourself haven't read, etc etc

Or keep saying "bye" and then coming back.

[TimFox]

I only do it to annoy,
Because I know it teases.

(L Carroll)

[Kyle_from_somewhere]

I only do it to annoy,
Because I know it teases.

(L Carroll)

So, you do what everyone who embarrasses himself on the internet does: you're trying to play it off with an air of superiority. I'll give you some credit, you went with the faux upper class "i have grown bored of you", rather than the more typical "i'm not mad, I'm laughing actually".

You blew it.

[Kyle_from_somewhere]

Right then, now that the ignore list has it's fourth member...

Anyone is welcome to participate in my threads, on the condition that questions are answered to the best of a healthy person's ability. "I don't know" is an okay honest answer.

If being smug about how the question is "too obvious to answer", talking nonsense because you haven't read the thread properly, making declarations of your own superior googling ability, etc are answers to the best of your ability, then you have a personality disorder and aren't healthy enough to participate. You will go straight on ignore.

Since we're on a new page, I'll repeat the last post made before the freakshow showed up and attempted to ruin everything.

[Kyle_from_somewhere]

The example I'm thinking of is this TV transmitter project, Radio Electronics Magazine June/july 1989

Part 1: https://archive.org/details/radio_electronics_1989-06/page/n37/mode/2up?view=theater

Part 2: https://archive.org/details/radio_electronics_1989-07/page/n41/mode/2up?view=theater

I've clipped the relevant part of the diagram. For some reason the authors think themselves above specifying inductance (why is this so common?), but they're successfully using a chain of transistors to double the frequency. I tried to upload the coil table but there are too many artifacts in the image to make a file small enough.

Learning things like how to bias a transistor for a particular result is very difficult because the internet makes for a bad educator. It's easy to stumble across a hundred youtube videos with a title like "how to bias a transistor", but every single one just has an overcaffinated guy with a whiteboard start by pulling two resistor values out of his ass, calculating the last two based on them, and then declaring it done. No explanation of what makes it linear or non linear, in fact no mention of the concept of a transistor operating linearly or not, just "it's biased now, don't forget to click like and subscribe".

Trying to find something written down using google is even worse. You fall into forum threads where the youtube scenario is repeated almost word for word, followed by someone saying to use an op-amp instead - without even the courtesy of explaining how. Just like here I'm told to use a transformer and two diodes, without being told how.

I don't want to just blindly copy someone else's circuit because often they're ancient and the parts aren't made anymore, and even if they were available I don't learn anything. But It's also difficult to learn to design my own circuits when the only "education" I can access is people doing stuff without explanation, or people saying "do something else instead" also with no explanation. I cannot go to university.

[Halcyon]

Do not expect people will be elaborating pretty obvious stuff.. 

You're going on my ignore list for being a deliberately unhelpful dickhead.

Whether you personally think someone is a dickhead or not, how is it appropriate, particularly as a new member to post such a comment?

[Kyle_from_somewhere]

It's simple honesty. I think even people who've annoyed me enough that I don't want to ever speak to them again deserve honesty. I also don't think being a "new member" or an "old member" counts for anything. Do you pay pensions?

I could ask how it is appropriate for someone who fancies himself a moderator to participate in derailing a thread after efforts have been made to put it back on track.

[Halcyon]

It's simple honesty. I think even people who've annoyed me enough that I don't want to ever speak to them again deserve honesty.

I could ask how it is appropriate for someone who fancies himself a moderator to participate in derailing a thread after efforts have been made to put it back on track.

Then please keep it private. If you want to call someone a dickhead to their face, then do that, but don't do it out in the open for everyone to see. You just get everyone else off-side and then you become the dickhead (rightly or wrongly). I've read some of your posts and they do tend to be a little on the hostile side. It's not a way to get what you want in this kind of environment and as you have already seen, some people like to stir the pot simply to get a rise out of you for entertainment.

This goes for everyone, not just you: Let's be a little more respectful and keep the snide comments down.

As for my involvement, I did so because your posts have been reported by members of this forum and contributors to this thread. If you don't like the way I operate, tough luck. This is not a time for you to start pointing fingers at others, this is specifically about you and your conduct.

[Kyle_from_somewhere]

It's pretty hypocritical of you to come into my thread and talk shit about me publicly while also telling me to only talk shit to someone in private.

You would be a better moderator if you tried leading by example, rather than this "do as I say, not as I do" routine, the tired schtick of drunken domestic abusers everywhere.

If you don't like me discussing how you operate, tough luck. You asked for it, and I always give as I've gotten.

[Jay_Diddy_B]

Hi,
To solve one of the LTspice modelling issues. This is how to add the DI_1N5711 diode to your simulation:



This is how you add a SPICE directive:



This is the best way, in my opinion, the model is now part of the schematic.

I have attached the model.

Regards,

Jay_Diddy_B

[Kyle_from_somewhere]

Hi,
To solve one of the LTspice modelling issues. This is how to add the DI_1N5711 diode to your simulation:



This is how you add a SPICE directive:



This is the best way, in my opinion, the model is now part of the schematic.

I have attached the model.

Regards,

Jay_Diddy_B

Thank you, that helps a lot.

[Halcyon]

It's pretty hypocritical of you to come into my thread and talk shit about me publicly while also telling me to only talk shit to someone in private.

You would be a better moderator if you tried leading by example, rather than this "do as I say, not as I do" routine, the tired schtick of drunken domestic abusers everywhere.

If you don't like me discussing how you operate, tough luck. You asked for it, and I always give as I've gotten.

Firstly, this is not your thread. This is Dave's forum and he has empowered a few of us to ensure that his expectations and rules are upheld. Like most things in life, if you choose to breach those rules, there are consequences. I think most people here will agree that the moderators here are pretty easy-going and this is one of the least moderated forums out there. But when people come here and start causing trouble, then we need to step in (particularly as other users have asked us to). Because you don't agree with it, doesn't make it an abuse of power. Either way, you simply don't have a choice in this matter.

I think most of the people in this thread would have accepted an apology from you at this point, but you chose to keep carrying-on like a pork chop. Since you seem to respond to honesty, allow me to be blunt: Pull your head in, or you will be removed from this forum.

Now which will it be? The red pill or the blue pill?

[Kyle_from_somewhere]

I've never heard of any dave. Am I supposed to be impressed you know a guy named dave? That's possibly the least amount of power anyone has ever gone mad with.

Make those other guys apologize to me, if you can muster that power. I've done nothing wrong whatsoever.

[fourfathom]

.....

Kyle, are you here to learn (or at least get an answer), or are you here for a pissing match?

If the former, start with J_Diddy_B's circuit and see what an isolated diode doubler does.  Then, if you like, start adding amplification and filtering.  This will require some degree of impedance matching.

One thing about the diode doubler, it's going to require a stronger drive level and probably more power than the transistor doublers you started with.  If that's an issue, perhaps your first design is a good starting point.  But it still needs lots of work.

[Kyle_from_somewhere]

I came for answers. People interested in pissing matches asked me for one and I gave it to them out of kindness. One by one they made themselves look like retards, but they wanted that, so it's okay.

[Halcyon]

I came for answers. People interested in pissing matches asked me for one and I gave it to them out of kindness. One by one they made themselves look like retards, but they wanted that, so it's okay.

I think you need to do a little self-reflection and consider the way you speak to people, particularly as you are reaching out for help and assistance with your projects. Getting people off-side and continuing to have the last word as your own hole gets deeper and deeper is not the way to go about it.

In order to help facilitate some of that self-reflection time, I've given you a short ban from posting. Hopefully you return with a little more clarity in a few days.

For everyone else, I apologise for the disruption this user has caused. Thank you to those who reported.

[Jay_Diddy_B]

Hi,

Here is a link to the HP 11721A frequency doubler manual:

https://xdevs.com/doc/HP_Agilent_Keysight/HP%2011721A%20Operating%20&%20Service.pdf

Here is another link with photographs:

https://idewerks.com/2020/03/24/hp-11721a-frequency-doubler-lab-repair/



The circuit is shown as:



The manual says this works best with an input of 13dBm +/- 1dBm.

This can be modeled as:



I have attached the model.

This is not tuned, like many other frequency doublers.

Jay_Diddy_B
 hp 11721A doubler.asc (2.04 kB - downloaded 26 times.)

[mawyatt]

We were going to post a circuit we've utilized well into the many GHz region (with fast SiGe transistors), but because of the OP attitude decided not to provide support.

It involves differential In/Output and 4 transistors configured in a differential quad type. Here's a plot with 2N3904s of the output differential current at 100MHz 200MHz (Input 100MHz) without any tuning network, of course faster transistors as mentioned would be better but the circuit seems to extract the best of the transistors used and why we utilized such in many of our IC designs. The green trace is the differential input voltage and the grey trace is the differential output current without any filtering involved.

Edit: Corrected above and added another plot with Input at 200MHz, Output at 400MHz with 2N3904 devices.

Use a PM if interested in more details.

Best,

[fourfathom]

Here's a plot with 2N3904s of the output differential current at 100MHz without any tuning network

Sounds like a sine-squared multiplier?

I threw together (LTSpice) a push-push doubler using three transistors:  The first stage is a unity-gain amp with collector and emitter providing 0 and 180-degree phase outputs.  These drive two AC-coupled transistors with the collectors connected together and driving a common load.  The output is closer to a full-wave rectifier than your nice sinusoid.  I simulated with 2n3904s and the amplitude was beginning to fall off starting a bit below 100 MHz.  I didn't spend a lot of time tweaking it.  Resistors, capacitors and transistors only, no inductors or tuned circuits.

[Gerhard_dk4xp]

That looks MUCH better than the output doubler of the ADF5356 synthesizer chip.

... and I'm curious for the details     :-)

dk4xp at arcor dot de

Cheers, Gerhard

[mawyatt]

Here's a plot with 2N3904s of the output differential current at 100MHz without any tuning network

Sounds like a sine-squared multiplier?

I threw together (LTSpice) a push-push doubler using three transistors:  The first stage is a unity-gain amp with collector and emitter providing 0 and 180-degree phase outputs.  These drive two AC-coupled transistors with the collectors connected together and driving a common load.  The output is closer to a full-wave rectifier than your nice sinusoid.  I simulated with 2n3904s and the amplitude was beginning to fall off starting a bit below 100 MHz.  I didn't spend a lot of time tweaking it.  Resistors, capacitors and transistors only, no inductors or tuned circuits.

Not sure what it was called, but here's the LTspice schematic and doubler plots. Because of the well controlled current range, lack of device saturation and differential input impedance control this worked well into the many GHz range with fast devices, and easily cascaded using a center tapped transformer on the outputs. Long ago a colleague utilizing something like this for generating 90GHz, then another doubler (this might have been with SB diodes tho) to get to 180GHz, this was in an experimental early IBM SiGe BiCMOS process 9hp.

First plot is 200MHz Output, second is 400MHz Output with 2N3904s.

Best,

[fourfathom]

Not sure what it was called, but here's the LTspice schematic and doubler plots.

That's essentially a push-push doubler. Q4 and Q2 are being driven 180 degrees out of phase and the collectors are tied together (very similar to my PP Doubler, but I used a transistor for the phase inverter instead of the two generators, and a resistor instead of the inductor.  I'm not clear on what the inner transistors Q1 and Q3 are doing -- you can ground those bases and it still works (the two 50 Ohm resistors form a virtual ground), but if you remove Q1 and Q3 the doubled output amplitude drops significantly.  Some sort of V_BE compensation?

Here's a basic transformer-input push-push doubler using 2N904s, 100 MHz input.  Input and output Z *not* 50 Ohms.  It works with 200 MHz input, but the output has fallen by almost 20 dB.

[mawyatt]

I'm not clear on what the inner transistors Q1 and Q3 are doing -- you can ground those bases and it still works (the two 50 Ohm resistors form a virtual ground), but if you remove Q1 and Q3 the doubled output amplitude drops significantly.  Some sort of V_BE compensation?

We used this on many of the RF/MW SoC chips we did back starting in ~2000 as a Square-Law Detector for accurate on-chip RF signal measurements, then a little later as a frequency doubler and also used as a real-time analog signal conditioner where the output current closely approximated a hyperbolic-trigonometeric function (likely the Cosh function) to "equalize" a dynamic signal in real time.

It's been too long and I'm too old remember the finer details and too lazy to do the math

I'm sure one could reverse engineer this and produce the non-linear transfer function finer details if necessary.

Do recall the two center transistors were for compensation and vaguely remember the 4 transistors were located in an Erdi Quad arrangement for temperature and process gradient compensation. Q2 and Q4 did have series resistors to the bases (not shown in first schematic, but redone below with resistors) so the arrangement had all 4 transistors "seeing" similar source impedances, altho none were exactly 50 ohms. This of course made the differential output current almost zero without an input signal, which was a necessary requirement.

Anyway, hope this helps, altho don't feel the OP deserves any help from anyone with the attitude shown, and why we didn't show the schematic earlier.

Edit: Added a couple plots showing Differential Output Current vs. Differential Input Voltage. Fist plot is with Vsource +-20mV and second plot is with Vsource +-200mv. Note, input sources V3 and V4 include 50 ohm impedances.

Best,

[Kyle_from_somewhere]

Hello everyone. First of all I'd like to say that i'm not sorry, you all suck and I am the best. I figured it out on my own, I've got a healthy 6mW stable 480mhz sine. None of you get to see because you dumped shit into my thread and cried like babies to the virginal forum mod when I shoved it back in your faces.

 I found out who that dave is too. I'm surprised he became an engineer, if my son had a voice like that i'd have left him in a hot car and told the police it was an accident.

[Halcyon]

Hello everyone. First of all I'd like to say that i'm not sorry, you all suck and I am the best. I figured it out on my own, I've got a healthy 6mW stable 480mhz sine. None of you get to see because you dumped shit into my thread and cried like babies to the virginal forum mod when I shoved it back in your faces.

 I found out who that dave is too. I'm surprised he became an engineer, if my son had a voice like that i'd have left him in a hot car and told the police it was an accident.

Welcome back Kyle. Good to see you've had time to reflect on your own immaturity and attitude problem. It's a shame that your little ban didn't help in this instance. I guess some people will just continue through life being complete assholes, right?

Since you made the decision to continue acting like a complete tool, as opposed to apologising for your behaviour, I've made the decision to permanently ban you from the forum, so that we don't have to continue reading your garbage. Consider this a life lesson on how you speak to people (you're welcome).

[Gerhard_dk4xp]

You've forgotten a "not" just before welcome.

Gerhard

[Halcyon]

You've forgotten a "not" just before welcome.

Gerhard

I give everyone the benefit of the doubt. But I guess when you have the maturity of a 13 year old, any and all efforts are pointless. I could have been harsher, but why stoop to his level?