Welcome to our site! EDAboard.com is an international Electronic Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
I know that Alumina is good but is too expensive for me. I was thinking to do something on Teflon. Also I don't like the idea of free running LO, even is cheaper. When I said cheap, I was thinking do not use a PLL. I prefer to start from a crystal LO and use multipliers. But the phase noise performance it will decrease with each multiplier.
Question is how good frequency stability you need at 10GHz.
Nowadays it is relatively chip to by LNBs for satellite tv reception (I think of the "heads" put into the focal point of outdoor parabolic dishes), especially if you look for old or second hand (maybe lightning-broken) ones. You can use the 10GHz DRO part of these LNBs I think with good results or can design some kind of AFC circuit around it if the stability of the LNB oscillator is not enough in itself.
This is just an idea to obtain cheap 10GHz components and adjust them to one's needs.
Regarding your frequency multiplier solution, it can be made less noisy if you start with some hundred MHz oscillator (the higher, the better) in which you use SAW resonator instead of crystal. SAW resonators are fabricated for several applications like remote controllers for car-door openers around 300MHz or 430MHz etc.
This way you may trade for less number of multiplier stages, especially, if you couple out the signal with a grounded-base bipolar transistor directly from the SAW resonator.
I mean if originally one of the resonator's leg was connected to ground in an oscillator, you lift that leg up with the very low input impedance of a grounded-base linear amplifier stage to couple out the signal from the oscillator. This way you use the resonator (this can be a crystal as well) as a narrow-band band-pass filter and this in itself reduces side-band noise/harmonics. Of course the resulting loaded-Q of the resonator suffers a little but you gain a "free" narrow band-pass filter! (This idea is not original from me but I mention it as a very good idea to filter oscillator side-band noise)
Your replay was right to the point.
The stability what I need should be enough ok to do a voice communication in SSB, and to transmit high speed data in NFM. The only problem that I had, was to make a high frequency SAW to oscillate, on the right frequency :smile:
if you plan to use ssb or fsk on nbfm your LO need to be very stable. DROs and saw are totally unuseful for those apps. You should point your attention on xtal or pll even considering the opportunity of an harmonic mixer as first conversion. Some useful items can be found on Hittite catalog, they have low cost oscillator in this range that included a splitter and a prescaler all in a chip. This will simplify a lot the pll design.
Is true, SAW oscillator didn't give me a good stability, and X-tal multiplier gave me a lot of problems to select the right harmonic. This using a minimum of BPF stages, to minimize the cost.
Unfortunately Hittite they do not make VCO's for 10GHz, but to build a 10GHz VCO is not big problem. HMC363 from Hittite, divider with 8 up to 12GHz and phase noise of -156dBc/Hz@100kHz, is really impressive. If the cost of this divider is not to high, could be a nice choice for a cheap PLL.
Building a phase or frequency locked oscillator at 10GHz is not such a hard task. For instance, a 2.5Ghz microstrip osc is easy to build, even purchasing an already built vco is an option. You can then multiply this with a simple x 4. I would use a FHX35LGT from fujitsu. You just phase lock the 2.5ghz vc0 using a PLL device such as the national semi LM series (i think they go to 3GHZ), Perigrine and analogue devices also do PLL devices up to and above 3GHZ.
The LNB DRO's are generally at 10.6GHZ (for Eurpoean universal high band osc) these will have a cavity that can be cut out and used on a brass plate. You will be very lucky to tune these DRO's to 10GHz as they are usually set to -170 Mhz low of the target frequency. I would use the low band oscillator which is 9,75GHz (this will tune to 10Ghz) if the tuning screw hits the puck you could grind the tuning screw down. Material is usually PTFE or now Rogers 4003C o.5mm material.
These dro's when fully assembled in their housings give a frequency stability over temperature (-40 degrees c to +60) of any where between 0.6 to 2 MHZ)
I know this because i design them for an LNB manufacturer.
You can now frequency push them a few hundred killohertz by varying the supply voltage. You will get a drop in power the harder you push it to the point where the oscillator will stop.
Depending on the amount of deviation you require you could try to use a coupled line and a varactor. the varactors are expensive and have to be carefully designed and simulated in the full oscillator design.
The frequency pushing is the easiest.
Now you have a super cheap VCO at 10GHz.
Free running you can expect -70 dBc/hz at 10K with a 20dB per decade roll-off. Not very stable for direct modulation.
To phase lock it. Take a Hittite divider or one from UMS. A divide by 4 will be best.
Try to get hold of a PLL that will run at 2.5GHz. Analogue devices ADF4113's are about the best device you can get. All you need is a microcontroller. You can now either modulate the reference oscillator (taking into consideration the 20 log N multiplication factor of the PLL) or ac-couple a signal onto the Vtune line.
The multiplication method is a little easier but does give a degraded phase noise in comparison to the 10GHz one.
Personally i would go for a 5GHz VTDRO and a x2. Phase lock the 5Ghz VTDRO using the new ADF4106 from Analogue devices, this will take 5ghz directly in, with no divider required. Use a bipolar siget (infineon BFP 420) in a shunt feedback configuration with a puck. Metallics and sanyo do some good low capacitance varactors. use a resistive splitter to divide the oscillator to feed into the ADF4106. Drive the fundamental into a x 2 using just about any xband fet (get one from a trashed LNB front end. There you have it.
If you need anymore help ask.