From ideas to working hardware !!!
Prototyping techniques for RF&MW 6 cm band stuff 
Coaxial transmission line can be used for phase control and even for matching if you can find coax of different impedances. 25 and 75 ohm line is readily available in addition to the traditional 50 ohm line. Coax is also available in a wide range of diameters and the thinner cables can be used at pretty high frequencies in even short lenghts. In this picture you can see my 6 cm ( 5760 MHz ) pushpull mixer using a four diode package ( specified to 2 GHz ! ). The conversion loss is about 7 dB and you can also see a power splitter made using 75 ohm coax and a 100 ohm resistor ( a Wilkinson divider ). This mixer was intended as a prototype, but as often happens, it ended up as the final TX mixer for the 6 cm EME/troposcatter transverter described elsewhere on my homepage. The discrete 1/4 watt resistor attenuator is the TX path attenuator for the 144 MHz IF. The mixer uses a 3 dBm LO at half of the final frequency ( a harmonic mixer of sorts ).
9 cm stuff 
This is a similar mixer arrangement for the 9 cm band ( 3400 MHz ). Here the IF is 432 MHz to improve image and LO rejection with more elementary filtering. This mixer can run with - 3 dBm of LO at half the final frequency. With + 3 dBm of LO level, the conversion loss is about 6 dB. This mixer remained a prototype and a GaAs-FET mixer designed by DB6NT was used for the EME rig described elsewhere on my homepage. The mixer is very easy to build and I use it for mixing signals from LO's under work to a more suitable IF for my spectrum analyser and frequency counter. It also provides nice 3400 MHz SSB and CW signals for verification of correct operation of other equipment. The mixer diodes are also a four diode affair forming a ring, but are connected in a pushpull manner ( two in each direction ).
13 cm band stuff 
This 13 cm ( 2.3 GHz ) mixer was the first of this type for me to experiment with. The conversion loss is 6.5 dB and the mixer works very well. The mixer diodes are a diode ring of four specified for 2 GHz operation. This mixer served well in my first 2.3 GHz transverter. I used two mixers, one for the TX and the other for the RX chain. This mixer is very noncritical at this frequency as the tolerances are not very tight. The IF for this mixer is 144 MHz out of convenience. A 432 MHz IF would be better from the technical point of view. The starved LO level is about - 3 dBm at half the final frequency. This mixer takes about an hour to make if you have everything handy !!!
23 cm stuff 
When prototyping filters and matching circuitry for frequencies between 300 and 3000 MHz I use double sided fiberglass or PTFE circuit board laminate that I empirically "cut to size" with a pair of scissors dedicated to this. Striplines may also be made air dielecric for lower loss and higher frequencies. The filter striplines are soldered down directly onto a ground plane circuit board. This ground plane exhibits excellent RF return and is a very stable environment for this type of circuitry. In the engineering stage the striplines are soldered from each end only and when the final dimensions and distances/gaps are determined, the striplines are soldered down all the way along their length. The 50 ohm tap points are determined empirically and experimentally. The parasitic parallel element is very critical to tune due to high impedances. Be careful when tuning not to miss the very narrow resonant frequency ! The extensive availability of MMIC circuits makes building a complete LO chain and mixer with preamplifier possible in a single evening without any predesigning !
And of course the average good looking enclosure may keep within all kinds of murky secrets that may have indescribable appearance, but work like a charm... The innerds of this 0.7 dB max. preamp for 1.3 GHz is a three stage preamp built into a circuit board enclosure using the described techniques. The amplifier is stable and exhibits 34 dB of gain. The preamp is used in conjuction with a 60 Watt SSPA based on four Mitsubishi broadband hybrid amplifiers. The preamp is in a weather tight double enclosure with an N type connector for input and a less desirable BNC for the output. The preamp uses two GaAs-FET's and has paved the path for a similar P-HEMT preamp that has an even lower noise temperature.
More 23 cm stuff 
Here are some low power amplifier stages constructed using the descibed cut and glue technique. The driver amplifier uses a BLU98 transistor for 11 dB of gain at 1296 MHz. The image shows how the input matching has been accomplished. The input goes through a parallel resonant circuit tuned with a plastic foil trimmer. Bias is fed via a 1 kohm mass resistor with low series inductance at RF. The resistor cold end is terminated with a Hi-Q chip capacitor. The DC decoupling capacitor is part of the matching and there is also a low impedance series stripline terminated with the tranistor base and a parallel trimmer capacitor. The input return loss is approx. 15 dB which is adequate for the purpose.
The output circuitry consists of a series transmission line, also of fairly low impedance ( wide stripline ). The DC decoupling capacitor is also part of the matching scheme. This could also be a trimmer, as is the parallel trimmer. The capacitive voltage divider would then be easier to tune and the trimmers could be replaced with fixed value caps ( that is if you can estimate the stray components accurately enough !!! ). The collector voltage is supplied via an RF choke.
23 cm 1 W amplifier 
This is a photo of the input matching found to be useful for the BFQ34 input at 1296 MHz. There is essentially a low impedance series transmission line terminated with the base and a parallel trimmer capacitor. The bias is fed via an RF choke with settable bias current. The base voltage is regulated by a Si diode. the PCB is bolted down to an aluminium heat sink with thermal compound on the tranistor bolt. The transistor emitters are soldered directly to the PCB ground plane with shorting foil to the bottom copper laminate. The foil is wide so as to exhibit low series RF inductance at 1.3 GHz.
Matching the output of the BFQ34 at 1296 MHz is very simple too. The output matching is basically a PI-type circuit with two trimmer capacitors and and a series inductance ( wide stripline ). The collector voltage is supplied via an RF choke formed by one turn of thin copper wire terminated with a Hi-Q capacitor for RF bypass. The output of the amplifier is 1 W with a gain of 7 dB.
Even for 70 cm !!! 
The only reason why you could not use this kind of prototyping techniques is the practicality of of dimenstions. At 70 cm the sizes are still quite reasonable though as you can see from this stripline filter ( the striplines have not yet been soldered down properly ). After having found the coax/stripline dimensions, all you need is a slide rule to determine resonator sizes for etched versions of the PCB. It will take some practise to guestimate the the edge and stray effects of the prototypes...but this is all part of the fun !!!