6cm Project
So, time to try to realize a new project. I have done the 3cm project and it is working. From my QTH it is quite hard to get long distance contacts. SM7ECM is 222km away, and is about the distance for 10GHz I make at the most. Some of the Danish amateurs has also been worked. That said, the cost / QSO is high. Transverter, PA and parabola cost.
6cm is even less to work here. I doubt I will hear many stations. BUT I want to try.
SM6VFZ is an old radio amateur friend (he is not old), he is a very good RF-designer. He has written in DUBUS magazine many times, and he also work in the RF industry. He had an idea of building a 6cm transverter. It is not (as for now) an available kit. But I have the possibility to try to put things together. He made it easy for me to get started.
So, LO, mixer, front-end and PA is needed. My IC-9700 is my IF working at 144MHz. To lower the IF images 432MHz could have been used as well.
Daniel has constructed a very versatile LO board. This board is now locked to 2808MHz. To be able to use it on 6cm the LO needs to be doubled. It is possible to use a frequency doubler to generate the needed 5616MHz for 144 IF, and then a suitable mixer.
Daniel preferer’s a sub-harmonic mixer. I have not made one before, so let’s try.
The schematic of the sub-harmonic mixer:
144.4MHz resonate frequency yields 100pF and 12.15nH inductor. The inductor can be made 2 turns 6mm diameter 9mm long.
If the mixer is constructed well, it can be very efficient. I used RO587 0.8mm laminate, which is low loss. However, my layout is very basic. The losses in my mixer are about 15dB. But my parashoot in this is my small Chinese LNA. Despite the loss I can add gain and move on. Of course, would a perfect mixer be better.
This means that I could go only with Chinese LNA’s and omit Daniels front-end and only use the 2808 LO. But his front-end is very nice, and has a nice low noise figure. One of the Chinese LNA’s in the picture has a NF < 1 dB. For my 6cm work that will do.
I have mounted the basic set on a double-sided PCB board. I made a double 5760MHz BPF of plumbing cups. The filter work well. The loss @ 5760 is about -2dB.
LO, sub-harmonic mixer and filter on this side:
The other side consists of the RX/TX front end and a dc-dc step-down module.
With everything connected have 0dBm IF in and 2dBm 5760 out. IF rejection is about -30dB.
I guess I will receive comments that this can be better. However, I will try this approach.
The PA:
Well nowadays it is not really hard to find PA’s in the 50 =>100W range for 6cm. The one I found is the YPM5850C. It is a jammer for 5.8GHz drones. It is available on different sources on the net. The specification is 47dBm out with 42dB gain.
The jammer function is the VCO to the upper left. Of course, needs this to be removed.
From data sheet:
Innotion’s YP40601650T is a 50-watt, internally matched gallium nitride (GaN) high electron mobility transistor (HEMT) designed specifically with high efficiency, high gain and wide bandwidth capabilities, which makes the YP40601650T ideal for multiple applications with frequency from 4400MHz to 6000MHz.
I could not get more than 45.08dBm out (with my measurement uncertainty) which yields 32.2W out with about 2-3dBm drive. Which makes the specification fair, but not 50W out.
Update, I now have 47dBm out see lower down.
RX is not fully tested. However, the close beacon SK6MHI @ 5760.800 could be read 59 from tree reflections.
A small video:
A simple block diagram:
Antenna:
My antenna is a small dish measuring 400x300 and has a gain of 24dBi. This is not so much gain, but the small size makes it easy to install or to bring.
Some success was made and a local beacon was heard. However high loss in the mixer and added LNA started to generate problems in NF and overall gain.
Ideally, I would like one of this antenna big brother. And yes, I bought one 30dBi is probably what is needed to reach other stations with my QTH. 900 x 600 is still a quite small antenna.
I need to make a new hole in the roof and mount the 10GHz and 5.7GHz on the elevation rotator. Now it is located close to the other antennas and it is not possible to add one more parabola. However now it is -7 outside and snowing, so this has to wait. (3 January 2026).
Second step:
The Bulgarian company SG-LAB has recently made a transverter for 6cm. Their latest version is 1.4.
I have seen tests done comparing this against DB6NT version, and the conclusion from that test was that it is equal good or even better. I say nothing about this since I have not done this test myself.
Pricing is very good for this transverter, currently 288 Euro (late 2025). That is from my own perspective a good price for a transverter @ 5760 with 2W out. (DB6NT 815 Euro, 250mW out).
Some amateurs dislike buying ready build things, and look down on others that do not build everything from scratch. However, I’m not one of those guys. If it is available, why not.
When running without 30V (PA) it will be QRP. The output from the transverter is lowered to about 27dBm, later after TX relay attenuated to about 5dBm which the PA needs. This means an ERP of 300W is available, with no 30V applied.
Simplified schematic, I omitted the bias tee. The transverter operates nice with the VOX and the PA is 30ms delayed.
A small antenna comes with the transverter, with this connected to the now ready outdoor box the beacon SK6MHI could be received in the shack with closed windows and wrong direction. It is located 6km from my QTH. What to say, not bad.
As usual, the common trap is the connectors/pig-tails used when wiring all together. I assumed (totally wrong) that one of my Chinese pig-tail that did not work on 10G would on 5.7G. Well, it did work but the loss was high -1dB for 15cm length.
After the first poor output power measurement, I measured and replaced the worst ones. I needed to cut and re-solder one of them. This can most probably newer be a EME station, but I don’t want to lose any dB unnecessarily.
Now 47dBm out -30 dB attenuator connected. Maybe hard to see on the Giga-tronics display.
A close-up pic of the box. The small antenna can be seen.
I only trigger the PA from the sequencer and turn 32V on. This is risky business if the bias circuit has flaws.
Giovanni, IN3HOG and IW1EPJ Gianni has done lots of measurements on a PA module. They have seen that the bias circuit isn’t perfect. The 32V is engaged before the negative bias has reached -5V.
I quote Giovanni;
Unfortunately, this system isn't perfect; in particular, the +32 volts applied to the GaN are enabled when the negative hasn't yet reached -5 volts, but rather when the negative Dc-Dc converter is at approximately -2.5 volts.
Therefore, at power-up (even if done with the VEN/PTT), a strong and worrying current peak flows through the GaN for more than 2 msec (<10 A, which is the limit of the Hall Effect measurement system we used to measure the drain current).
At power-up, there is also another very short-duration current peak due to the Miller capacitance between the GaN drain and gate. That is, the +32 volts applied to the drain briefly raise the gate positively, and consequently the GaN experiences a dry short.
To visualize this peak, otherwise masked by the previous peak, it is necessary to power the GaN with a correct and stable bias and only then apply the +32 volts to the drain.
This peak lasts about 1.5 microseconds. Unfortunately, this inevitable effect is usually mitigated by keeping the impedance of the negative bias source very low. In this PA, the impedance appears too high (about 2 ohms), and this exacerbates this problem.
So, here is my attempt to try to mild these effects, I’m sure that building a new bias and feed externally is probably best.
32 V soft-start for cheap 5.7–5.8 GHz GaN PA modules (PTT switches the supply) Many low-cost Chinese 5.8 GHz GaN PA modules work well, but when 32 V is switched on at every PTT press, the module can see large turn-on transients. The internal DC/DC converters (including the negative gate-bias supply) need a few milliseconds to stabilize, and a fast drain-voltage step can also couple into the gate through Miller capacitance.
A simple improvement is to add a high-side P-MOSFET after the relay output, with an RC gate network. This forces the PA supply to rise gradually (typically a few milliseconds, depending on the PA input capacitance and wiring), instead of an abrupt 0→32 V step, reducing:
- capacitor inrush current
- stress current through the GaN device at power-up
- Miller-induced gate kick
- relay arcing and contact wear
Key parts (starting values):
- P-MOSFET: IRF4905 (TO-220). Note: tab = Drain; isolate if mounted to chassis.
- Gate series resistor (at the MOSFET gate): 100 Ω
- Gate clamp: 12–15 V zener from Gate to Source (cathode to Source) to limit |Vgs|
- RC soft-start network (self-driven from the same 32 V line):
- o R (Gate-node to Source): 10 kΩ (sets ramp time)
- o C (Gate-node to GND): 220 nF … 1 µF (larger = slower ramp)
- (Optional) Gate pull-down: 47 kΩ to GND (mainly to define OFF when floating)
- TVS diode (unidirectional) across PA supply (after the MOSFET): 36 V recommended (e.g. SMCJ36A class)
It is a very simple circuit, but with a load of 22 Ohm the ramp time is about 4ms.
Note:
I have not left the first building step, it is fun to build things. However, even SM6VFZ pointed out that it is very difficult to get this right. A good LO and then all other things take time to get right. I’m still testing.
This project will go on, it has no real time plan. Nice weather will hurry the time plan...
Welcome
Welcome to sm6why.n.nu.
My Newsletter
Links