Electronics > RF, Microwave, Ham Radio

SI5351 as VFO up in to microwaves?

(1/3) > >>

rwgast_lowlevellogicdesin:
Im currently using the SI5351 in a project where it produces the carrier wave which is then mixed togather with a signal output in GNURadio, to make a transmitter. Is there any reasonable way to push the SI5351 up to a few ghz, maybe with some sort of PLL or something?

rwgast_lowlevellogicdesin:
Is this just kind of a stupid question? There are a lot of clock generation chips out there but none quite as neat as the SI5351 becuase it can create any frequency it doesnt appear to be locked by a divider.

Now I have never used a PLL chip before but Ive always assumed its purpose is to lock on to the incoming signal and then multiply it to create a higher value clock? If a PLL is the answer Im not even sure where to start.... I have some jelly bean pll chips and CB pll's laying around but I know those arent the answer obviously I would be looking for somewhat of a more "designer chip". Im not even sure if a PLL would work at all.... I would like to be able to generate any frequency up to at least 2.6ghz. I guess I basically want a DC-2.6ghz or higher VFO. Am I totally going about this all wrong?

hendorog:
I've been working on a 6GHz synthesizer, bit of a learning curve but that's why I do it:) It uses the MAX2871 chip.

That chip has dividers which allow output down to 23.5MHz, so IMHO you would take two of these PLL's and mix the outputs together to get down to DC.
The spurs and mixing products would be an issue though and so filtering and carefully choosing frequencies to simplify the filtering are going to be important.

Or you could take one PLL and mix it with your SI5351 to do the same thing. The filtering would be more difficult as the difference in frequency between the wanted and unwanted outputs of the filter would be much closer.



evb149:
I can't speak about that particular chip since I haven't looked at the specifications.
But you should look at it in conjunction with possibly similar offerings from Analog, and Linear and see what the benefits and disadvantages of the products are relative to each other.

* When driving a mixer, the mixer characteristics in conjunction with the related system design and filtering design will dictate what the appropriate type and level of LO signal is.  Some LOs can be low level sine waves.  Some LOs can be high level sine waves.  Sometimes something more like a square wave could even be used depending on the mixer and filtering and frequency spectrum you'll be dealing with.  But you must know what kind of wave (sine, square) and at what amplitude (0dBm, 3dBm, 7dBm, -3dBm, whatever) you need for the LO.

* All clock synthesizers that take in a reference frequency produce various noise and distortion products at their outputs.  That will show up as broadband noise with various spurious frequency peaks "spurs" present in their output spectrum.  At some frequencies and harmonics the spurious outputs could be quite strong relative to the capability of your signal chain to be affected by those signals.  Whether they are a problem depends on your filtering and spectrum usage in your application.

* Generally fractional-N (e.g. Fout = Fin * Ninteger / Minteger ) synthesizers tend to be noisier and noisy in more complex ways than integer N (Fout = Fin * Ninteger) synthesizers so you often have to decide what is best for your needs.  Obviously you have a lot more choice of output frequency with a fractional N synthesizer.

* The input reference clock will have certain phase noise and broadband spectral purity, and the output of a synthesizer will just multiply that noise and add to it producing generally a worse "close in" phase noise at the synthesized output and worse overall noise spectrum with added spurs etc.  Depending on what you do with the synthesized output (e.g. clocking an ADC) you may find that a certain level of phase noise XX dB + Fcarrier +/- XX (where XX = some number of Hz, kHz, 10s of kHz, 100s of kHz, MHz) is acceptable or not for your needs because it will certainly degrade your output signals somehow.

* When there is noise and drift or frequency tolerance error of an input reference frequency, that will generally just be multiplied up by something like the ratio of the synthesizer's input or output frequency.  So if you take a 1MHz +/- 100ppm input and multiply that up to 10GHz that'd be +/- 100ppm of 10GHz or +/- 1MHz accurate  of an output frequency maybe drifting somewhere in that range over temperature and time.  Not very good.  Good enough?  Up to your application.  And the noise will be proportionally worse too.

So yeah compared to say 1960s technology you could do a lot of at least somewhat practical things with some $10 synthesizer chip and some slapped together mixer and ADC or whatever but if you're looking for something that is relatively well performing you will probably find that you canot get truly good performance without seriously constraining the "just make any old frequency I want!" part of the design and optimizing it to perform well with particular bands and bandwidths.

Most RF systems these days do use a synthesizer chain of some sort and various kinds of filtering and mixing to implement the system so the overall idea is very practical, but you may find that you may need multiple stages of IFs or higher quality references or more limited instantaneous bandwidths to really work well.  The example of "bad design but that is just barely adequate" are these SDR TV receiver chips that are sold for a few dollars each.  It isn't low noise.  It isn't that accurate.  It isn't high fidelity.  But it is just barely adequate under "relatively good" digital TV conditions for receiving your TV.  If you have weaker signals or other signal quality problems in a crowded spectrum you'll quickly be looking for better antennas, filters, etc.etc. to try to make it work at all.


rwgast_lowlevellogicdesin:
Well the SI is so cool becuase it will drive 8khz to 200mhz any old frequency, it is square out which is perfect for me all mixers I use do better with square switching. I guess I said DC to 2.6ghz just to say it... DC is not that important lol. I just want a wide band (usp to 2.6ghz) square wave VFO that can be stepped 1khz and also has adjustable drive current. I do realize what I am asking for may not just be in a chip like the si5351... I could look around and figure out how to do mixing of the si5351 with a hi frequency PLL, but maybe there is better ways... I have a few VFOs around for different bands, although these are sin output they give me an idea. Instead of looking at one chip or two chips mixed to give me the frequency I want maybe I I could find square wave synthesizers that act like the 5351 but cover different bands and then use software to stitch them all together making one wide VFO, or does this sound more ridiculous? Im sorry if this sounds tot?lly stupid, I have never given frequency generation a thought past what my micro could do until recently when I started with RF.

Right now the SI will work well for a DC-200mhz VFO, getting to 2.6 ghz would be the ultimate goal and may take a bit more study and smaller steps to get there. What could I do to get to at least 500mhz? It would be nice to get at least a dual band VHF/UHF set up going. Is there some way I could much more easily do this using my si chip, run two VFO's at one 1-200 and 201-500?

If any of you gurus were to make a digital VFO from 50~100mhz up to 2.6ghz that would allow 1khz steps and current control what chips and methods would you most likely start with?

Im going to be using a Pi3 to host my SDRs remotely after I get rtl_tcp running. I know there is a program called PI_TX now which allows CW,WFM,NFM,AM,OOK,PSK and a few others modulation modes to be broadcast very dirty using square waves and it says as long as you filter your transmissions quite a bit then amplify them the pie will work well to transmit up to 700mhz!!! Im stumped and cant wait to play with it, I mean the GPIO pins cant even flip at 700mhz so I have no idea what kind of software majick they may be using to get 700mhz square carriers out of that board!

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version