What do you think of this design? Note especially the slot coupling of the microstrip to cavity, As well as the varicap position.

Nothing unusual (Ничего необычного).

Well, regarding the actual oscillator circuit, its nothing special. What is of interest for me is the coupling of the RF energy to the cavity without the mechanical complexity of the RF probes or loops. It is done by a slot radiating from the ground plane down toward the cavity. Anybody did it like this? Does it worth anything?

I do not understand your idea with groud-plane (antenna?). I propose the following decision:

Part of enargy will be directed to waveguid. But please remember, that the frequency of the oscillator will depend of the load of the waveguid. This is why the better way is to make coupling through probes or loops taking energy from oscillator output, when the oscillator's resonance system is buffered (as normally, i hope).

First, you may write in Russian. I just do not have Cyrillic keybord, so i write in english.

I wish to tell you what was the trigger for such an "old fassion" attempt to build a cavity oscillator.

The problem is, that if you need low noise LO at elevated frequencies, typically one goes for CRO or a DRO:
1. CRO is pretty simple to manage exept at frequencies above 2GHz. There, the length of the resonator begins to be too small to have any appreciable Q, so you revert to OVERTONE oscillator. Its OK, as long as you manage to suppress the obvious tendency of such an oscillator to jump back to fundamental tone generation. But it is still a nuisance.

2. If we are working above 3-4GHz, we go for DRO. Here it all very simple, except the need for special glue for the pill (to avoid Q reduction of the pill, and work on wide temperature range without weakening).
A very big trouble, is that the DR pill is built from such a hard ceramic material, that it cannot be trimmed with any tools readily available to "normal RF engineer". Using a metallic or dielectric tuner, any DRO can be trimmed, perhaps +-250MHz, before severe Q degradation. You want a new frequency beyond that? Order a new batch of pills! So, while the pills themselves may be low cost at large quantities, below 20-30 oscillators, their cost is prohibitive.

A possible solution i seek is to revert to good old cavity configuration. But no waveguides, please! Have any adea how the 2.7GHz or 4.5GHz waveguide looks like? An elefant! Going thru various literature, including the Rhea book- "Microwave oscillator design and computer simulation", reading about HP8640 VHF-UHF instrument and dismounting vatious Phase-Locked bricks like California Microwave and CTT, i noticed a cavity L band oscillators with SRD multiplier and "cavity" combline filters for the output frequency. All is well, except they use 1/2-Lambda cavity and thru-hole packaged transistors, 70's style. The mount of those transistors as well as the fine tuning of their coupling to the cavity is labor intensive and very awkward to tune and accomplish. The big issue, is WHAT TYPE OF CAVITY to use, and HOW TO COUPLE the transistor and the Varicap to the cavity. I decided to try and go for 1/4-Lambda COAXIAL cavity (actually, it will be 10% shorter than that) and easy modern SMD technology with MAGNETIC slot coupling of the microstrip to cavity via a narroe slot. The design in the pictures still uses probe coupling of the Varicap, but i may change this to slot later, if the experiment will be succesful. I wanted to know whether of any other experienced folks around the globe ever tried such a design.

Вышло что я не понял вопроса. Щель, я думаю, подходящее решение. (в вашей конструкции она, мне кажется, несколько маловата по ширине) А вот сама схема генератора может быть разной.
Вообще, всё это ещё в 80-х годах (если не раньше) была описана в книгах. Суть, по-моему, в том что добротность колебательной системы определяется не добротностью "железа" или "керамики", а тем элементом перестройки который был применён (врикап, восновном). По крайней мере так было в то время. Если с тех времён качество варикапов существенно улучшено, то это плюс, это значит что бОльшую перестройку можно получить при одном и том же уровне шумов генератора.
А Вы, похоже, задаёте вопрос будет ли работать ваша схема? Точно сказать я не могу. Возможно - будет.
Я в настоящее время этой проблемой не занимаюсь. Еще в конце 80-х годов я собрал генератор на объемном резонаторе на частоту около 2 ггц на транзисторе КТ911А, использовал именно щель как способ связи коллекторной цепи с резонатором. Работал он или нет, - уже не помню. Должен был работать. Где-то валяется в гараже, а может и дома. А что если бы я туда варикап, хотя бы параллельно щели поставил, имя такого варикапа не могу себе предстваить, отстал от жизни. К тому же там и мощность около 1 вт наверное.

Actually, i am milling it as we speak... I read a lot of books, but never seen such a solution-> oscillator+microstrip+cavity+slot coupling. I also was able to read some Soviet era books, but they are total waste of good paper in my view. Reading them, just could not shake the impression, that the writer just want to impress everybody with his mastery of high mathematics, and nothing more. Even the book i bought: Microwave Transition Design by J.S.Izadian and S.M. Isadian does not show this. So this is what makes me unsure, whether this all will work...
Regarding the Varicap, i can tell you this- stick to high voltage Varicaps- and be blessed by good tuning and phase-noise characteristics. Types like BB857,BB833,BB837 are very good in this respect. The phase-noise will be poor on those, low voltage Varicaps, intended for battery operated equipment. I also must stress out, that the AMOUNT OF COUPLING is very deceisive in determining the phase-noise of the oscillator. If the oscillator is only intended to be phase locked at a fixed frequency, you need to couple it only slightly, to compensate for the cavity Fo drift with temperature. So the phase-noise may still be very good. As a general rule, the closer the coupling slot to the GND side of the cavity centerpole, the stronger the magnetic field intensity will be there, and the coupling is more stronger. On the other hand a probe coupling is considered electrostatic, so to have a tight coupling one must place the probe near the top of the centerpole. I may use also 2 slots, one for the transistor, another for the varicap. I have MWO and CST but decided to skip the very complex combined simulation that probably will be needed here, going stright ahead to the milling machine! What is KT911A? Can you please provide me some data? I am curious, as i am unaware of any Russian types. Can you please be so kind to provide a schematic (mechanical included) for the unit you built? Do you have a picture? Also, those days electronic tuning for PLL lock is a must. So i work at low power (NE68519).

Для регулировки связи генератора с резонатором есть старое проверенное решение из техниеи ЭПР спектроскопии- в отверстии связи (щели) делалось перпендикулярное отверстие с резьбой, в которое вкручивался винт из тефлона. Вкручивая- выкручивая винт можно было плавно регулировать связь в широком диапазоне.
А как планируется варикап подключать? Несовсем понятно на чертеже платы как расположены земляные полигоны, можно ли детализировать?
Вот кстати фотки промышленного генератора диапазона 2-3 ГГц с механической перестройкой и на транзисторе (он лежит на дне резонатора). Так же виден варикап (слева) и петля связи выхода (справа). Перестройка механическая, движением диэлектрической втулки с помощью микрометрического винта.

Well, Mr. Alex190, have you had a chance to review a SYNERGY team approach to very low noise VCO design even wide tuning range?
Please see attached Ph.D thesis research by Dr. Ulrich L. Rohde and Dr. Ajay Kumar Poddar.
Key points are -SiGe high Ft transistor, varactor tuned coupled microstrip line resonator and push-
push topology (two sub-VCO circuits in symmetrical coupling to half lambda common microstrip line resonator).
"..The measured phase noise show 5-7 dB improvement with respect to the fundamental oscillator sub-circuit...".
By this way it's possible to get very low noise level VCO (mainly like DRO & CRO varactor tuned oscillators have)
but with planar, low cost SMT VCO structure.
IMHO, cavity resonator will be reasonable only for the fixed frequency ( or with bulky mechanical tuning) oscillator for extreme low phase noise.

/Sorry -" poddar_ajay.part1,2,3.rar" - limit to attachment size/

Не нашёл я генератор, вспоминал конструкцию и пришла на ум несколько другая, возможно она будет работать, я надеюсь. Вот рисунок.

Thank you for the files that you submitted. I am all too good aware of Dr' Rohde work. I don't know what implications may be involved from copying his design, as it is patented. What their work is all about- a very videband fast tuning low noise oscillators to (probably?) directly compete with YIG technology. I have also read an article of Matjas Widmar, in which a 2-4.2GHz instantaneous bandwidth unit was produced actually in home laboratory. Its phase noise is not as good as in Dr' Rohde design, but the design is FREE. I mostly seek a DRO replacement, so my case is very simple, relatively. I only need enouth tuning range of the Varactor to compensate for the cavity drift with temperature. Extreme low noise in very narroband cavity oscillators you say? Thats my goal!



Thank you for the photo. An electrically floating coaxial center line is a very odd approach, which will be prone to vibration and general mechanical instsability. How do i plan to connect the varicap? In the design you seen- by a probe and into the peak of electric field (thats near the top of the center resonator pole, with an etched opening in the PCB gnd plane). If the oscillator will turn out as a succesful design i think about another slot loaded microstrip just for the varicap. Or a turning platform with a magnetic loop loaded by the varicap. I imagine that the beauty of this approach is that by mechanically turning the loop insert, you will control the exact coupling of the varicap to the cavity, so to reduce phase noise to a minimum. The screw tuning of the slot coupling is also a very nice idea. Just need to figure out how to design it mechanically... And i would think that the screw should be metallic, not teflon. What is EPR? Regarding the gnd planes in my try - all the bottom side is filled with a solid copper plane with a small etched slot when indicated.



Thank you for the pictures. It looks like a Colpits -R cell (capacitor on emmiter with Cbe working as the second capacitor) with the cavity coupled EXACTLY as i trying to do. Too bad you did not tried it.... By the way, in equivalent circuit, the resonator (the cavity) should have been represented as TRANSFORMER coupled to the Base and not by 3 "solder" joints. The varicap should be drawn also as a winding on this transformer loaded by the varicap to gnd. The M factors (the magnetic coupling like in normal transformers) are a function of the slot position ALONG the resonator length. L1 is very important as it prevents the emmiter resistor from loading the -R too much. So it is better not be ommited. The big difference between our designs are two:
1. I use "common base"- actually inductor from base to GND to push a low frequency RF transistor as high as possible (for phase noise reasons). This creates a -G instead of -R.
2. I do not like to couple the varicap as the transistor, means, use the same slot. My goal is to couple the varicap just enouth to be able to phase-lock the oscillator across wide temperature range. So in my opinion, the varicap should be located at a different spot.

Я теперь исправил это на рисунке.
То что я не пробовал, а говорю что будет работать - это плохо, но есть у меня внетренее чувство (или опыт) что работать будет. Возможно придется подбирать точку соединения эмиттерной цепи, красным цветом я нарисовал ещё несколько возможных точек подключения эмиттерной цепи (EC - emitter connection point).


Да, разумеется, можно варикап сместить влево, если не требуется большая перестройка. Более того, его можно удалить совсем, а использовать для подстройки частоты ёмкость коллектор-эмиттер, например изменяя в небольших пределах напряжение на базе. Будет меняться и HF мощность, но думаю не настолько много и если это допустимо (далее наверное стоит делитель с переменным коэффициэнтом деления?). Можно изменять и напряжение на коллекторе при относительно стабильном напряжении на базе (питать базу отдельно). Можно делать то и другое одновременно, возможно при этом удастся снизить нестабильность полезной мощности при перестройке. Вобщем - поле для экспериментов.

Ниже я изобразил примерно то что было сделано когда-то. Это не длинный резонатор, а ближе к оптимальному для получения максимальной его добротности.