Cheapest way to generate 100Mhz sine wave

[freebil]

Hello. Is there any cheap way to generate 100Mhz sine wave? I have 2 ad9850 dds. Is it possible to add the frequencies of each 9850 sine wave and see the result to the oscilloscope?

[evb149]

A 100 MHz oscillator can be made with about $0.1 of parts, a single transistor and a few passive components could be used to make
a Colpitts type oscillator or Hartley type oscillator which produce sine waves when adjusted properly.
A cross coupled pair type oscillator can be made with two transistors and a few passive components.
https://en.wikipedia.org/wiki/Colpitts_oscillator
https://en.wikipedia.org/wiki/Hartley_oscillator
https://en.wikipedia.org/wiki/Electronic_oscillator
http://rfic.eecs.berkeley.edu/~niknejad/ee142_fa05lects/pdf/lect23.pdf
http://www.pld.guru/_hdl/2/-asic.uwaterloo.ca/files/vcotut.pdf
http://www.ece.ucsb.edu/Faculty/rodwell/Classes/ece218b/notes/Oscillators1.pdf

I understand your AD9850 has some frequency limitation though I am not sure what that is.  If it can produce a 20 MHz square wave or pulse then you could take its output probably through a transistor buffer stage and feed that into a series LC filter that is tuned to 100 MHz so that you get the 100 MHz fifth harmonic frequency component of its output.  It will not be as strong as the fundamental 20 MHz coming out of the DDS due to the DDS output frequency limitations but you should get a little signal at least that you could amplify with a transistor if you wish.

To increase the fifth harmonic strength you could  drive a clipping circuit that will increase the 5th harmonic content generated by the 20 MHz input.  You might as well look up "frequency multiplier" for that, you'd be looking for a x5 multiplier which should be fairly simple.

But I think just building a discrete transistor oscillator would be simpler than using a DDS that isn't designed to put out more than a fraction of the frequency you'd like.

You could also build a 100 MHz signal generator with logic gates (perhaps some kind of ring oscillator or crystal controlled one) and then buffer that and apply the signal to a series LC filter tuned to 100 MHz to get something like a sine wave from the fundamental frequency of the input square wave.

EDIT: Fixed typographic error.

Hello. Is there any cheap way to generate 100Mhz sine wave? I have 2 ad9850 dds. Is it possible to add one sine wave from each one and see the result to the oscilloscope?

[Fungus]

I understand your AD9850 has some frequency limitation though I am not sure what that is. 

The cheapo eBay modules go up to about 25MHz before they start falling apart.

[joeqsmith]

A 100 MHz oscillator can be made with about $0.1 of parts, a single transistor and a few passive components could be used to make
a Colpitts type oscillator or Hartley type oscillator which produce sine waves when adjusted properly.

Agree with evb149.  Really depends what you need.    Here's a simple 80 - 100MHz single 2n4401 transistor oscillator I stuck on some protoboard.   Cheap.

[joeqsmith]

You could also build a 100 MHz signal generator with logic gates (perhaps some kind of ring oscillator or crystal controlled one) and then buffer that and apply the signal to a series LC filter tuned to 100 MHz to get something like a sine wave from the fundamental frequency of the input square wave.

Again I agree with evb149's comment.  I had to search for it but I had posted a fair amount of data using 7400 logic.   

https://www.eevblog.com/forum/projects/waveforms-in-a-74ls04-ring-oscillator/

[bobaruni]

You could put the ad9850's output through a frequency multiplier with a tuned output for ~100Mhz.
A single transistor amplifier with a tank in the collector tuned to 100Mhz should do the trick.
Just feed a sub-harmonic of the desired exact frequency (near to or on 100Mhz) on the input and the tank will resonate at an integer multiple of the input.
If you want to combine the outputs of both ad9850's, you will need preferably a double balanced mixer with a similarly tuned output.

[evb149]

Thanks, nicely done with the breadboard prototype oscillator.
Is that custom made perfboard?  It looked like it might be something of the sort with isolated through holes surrounded by coper fill, hard to tell from the picture.

A 100 MHz oscillator can be made with about $0.1 of parts, a single transistor and a few passive components could be used to make
a Colpitts type oscillator or Hartley type oscillator which produce sine waves when adjusted properly.

Agree with evb149.  Really depends what you need.    Here's a simple 80 - 100MHz single 2n4401 transistor oscillator I stuck on some protoboard.   Cheap.

[evb149]

Thanks for sharing the link, that's an interesting thread about the ring oscillator results.

That 74AHC type part (if I recall correctly) which you characterized in the other thread seemed to perform pretty well up in the 150MHz range.
I suspect that some of the modern AUC/AUP/LVC series parts (or comparable series from ON/FAIRCHILD) should easily be able to reach over 100MHz at 3.3V (or 5V for those series that are capable of that voltage) since several have 1.5 to 3ns typical tPD values with moderate capacitive loads and generally also 24-32mA drive capabilities per gate.  Some of the discrete FFs are rated at between 100 and 300 MHz toggle frequencies and ~1ns tPDs.

It should be possible to make a transmission line controlled oscillator with some twisted pair (a CAT-5 pair maybe) or cheap coax that would be fairly predictable and temperature / time stable in a few possible oscillator topologies (though with a bit of care in passive tempcos and biasing one should be able to get a pretty temperature stable oscillator with cheap lumped elements, too).  I suppose using a faster transistor would help to eliminate the frequency uncertainty due to its phase shift / propagation delay.  Then it'd just be mostly up to the velocity factor of the cable and its electrical length.

EDIT: Added some tPD / IOH / IOL / CIN / ICC related data for some other logic families as attachments.

Again I agree with evb149's comment.  I had to search for it but I had posted a fair amount of data using 7400 logic.   

https://www.eevblog.com/forum/projects/waveforms-in-a-74ls04-ring-oscillator/

[KE5FX]

All you need is a filter of some sort.  Your AD9850 will put out a healthy signal at 100 MHz in its second or third Nyquist zone. 

[freebil]

Thanks for the answers. Is it possible to use the built in ds1054z filters?

[evb149]

Well the built in filters will respond more significantly to the sub-100MHz fundamental than the 100MHz and above harmonics.
So ideally you would externally band-pass filter 100MHz and feed that into your scope.
But even if you fed in a waveform with mixed harmonics you could still isolate the 100MHz component in the waveform by a FFT on the scope or more flexibly by downloading the waveform data to a PC and using digital filtering there.

Anyway a 'problem' with DDS board outputs is that there SHOULD normally be an analog low-pass filter built into the DDS PCB and the cut-off frequency of that filter should generally be well under 100MHz if the DDS is only rated to 30MHz or less.  So already you will expect to have a significantly reduced 100MHz component output level vs. the fundamental but maybe your DDS PCB does not have an (or much of an) output filter.  Anyway it should still be well over XXmV so you could probably see it on the scope despite the DDS and scope attenuation at that frequency.

Thanks for the answers. Is it possible to use the built in ds1054z filters?

[joeqsmith]

Thanks, nicely done with the breadboard prototype oscillator.
Is that custom made perfboard?  It looked like it might be something of the sort with isolated through holes surrounded by coper fill, hard to tell from the picture.

A 100 MHz oscillator can be made with about $0.1 of parts, a single transistor and a few passive components could be used to make
a Colpitts type oscillator or Hartley type oscillator which produce sine waves when adjusted properly.

Agree with evb149.  Really depends what you need.    Here's a simple 80 - 100MHz single 2n4401 transistor oscillator I stuck on some protoboard.   Cheap.

Just some Vector board.  They offer many different patterns.  Living in the USA, I get mine from Digikey.

[Kalvin]

If you need a better frequency accuracy, you can add a simple and cheap PLL which will keep your $0.10 oscillator stable.

You can use the AD9850 as the reference oscillator for the PLL so you can fine-tune or adjust the output frequency easily.

[ADT123]

If you are more worried about the frequency accuracy than purity of the sine wave then a quick way would be to buy a 100MHz CMOS crystal oscillator (couple of dollars/pounds/euros).  They are available in 4 pin DIP packages (power, ground and output).

It will output a square wave in theory but in practice the output driver will not be fast enough to produce the 3rd, 5th and above harmonics so it will be pretty sine wave like anyway.  To improve check the output impedance from the datasheet and then choose a capacitor value to make a low pass filter (cap between the output and ground) with a corner frequency of about 80MHz.  That will attenuate the 100MHz signal a bit but will get rid of most of the 200MHz and above frequency components.  On a scope it will look like a sinewave, you might see small harmonics if you FFT it but for many applications its not an issue.