Carsten Groen's (OZ9AAR) personal place
Interests and experiences
10 GHz Divide-By-10 prescaler
Some time ago, Howard, G4CCH, was searching for a prescaler that would divide the frequency of an RF signal with 10. I offered to help and the resulting design is shown below. All the files for copying this design are available, as always, do so on your own risk! Howard and I did some quick brainstorming on the design, I did some CAD work of the electronics and the milled aluminium housing.
Next step was building and testing two devices, everything seemed to work exactly as predicted. Below you will see some screenshots and pictures of the device.
All design files are contained in this ZIP file, Gerber, STEP files for box, schematics, BOM file etc. With this, you can make your own copy of the prescaler.
Design
The design uses a HMC361 and a HMC438 device, both from Analog Devices. The HMC361 is a divide-by-2 that covers from DC to 10 GHz, the HMC438 is a divide-by-5 that covers DC to 7 GHz. By cascading the two devices we get the required divide-by-10. As the design is AC coupled, it will not cover to "DC", the lowest range is set by the capacitors used.
Datasheet for HMC361: PDF file
Datasheet for HC438: PDF file
The two devices are easy to handle and solder, the only real downside to them is probably their cost, HMC361 is around €19,- and the HMC438 is around €50,- (both at Mouser at single quantities).
There is not much to the design, power supply is via 5VDC, current consumption is around 170 mA.
PCB layout
The PCB is a 2 layer 0.8mm thick FR4 board (controlled impedance, JLC7628 type substrate). The bottom side is fully covered with ground and does not have any solder resist in order for it to make full contact with the milled aluminium housing. Board has ENIG surface.
One of the boards while being soldered on a hot plate (I use stencil and solder paste).
Box design
After the PCB design was done, I imported the PCB as a STEP file into my mechanical CAD program (Autodesk Inventor). The box was created by deriving from the PCB STEP file, adding walls, screw holes etc to the dsesign (the box design was a 30 minute job from start to finish).
The surface finish for the box and lid is "bead blasting", do NOT use anodizing as this will prevent good electrical contact between the box and lid.
All screws used are M2.5 x 4 mm (stainless steel). The power is supplied thru a 1nF feedthru capacitor (M3 threads), the GND post is a M3 size.
Finished results
I did some testing of the device using my SMU200A signal generator (max 3 GHz) and my TF960 6 GHz frequency counter). I tested various frequencies to see what the sensitivity of the prescaler was. I measured the needed drive level from the signal generator until the frequency counter would show the correct (divided with 10) frequency.
100 MHz: -2 dBm
500 MHz: -14.1 dBm
1 GHz: -19.6 dBm
1.5 GHz: -23.6 dBm
2 GHz: -22.7 dBm
2.5 GHz: -26.9 dBm
3 GHz: -26.3 dBm
The low(er) sensitivity at 100 MHz is probably due to the low capacitance used as AC coupling throughout the design. As the target was "10 GHz", I did not experiment further with these capacitors.
The sensitivity is generally much better than what the datasheet claims.
