Carsten Groen's (OZ9AAR) personal place
Interests and experiences
400W attenuator
I needed a simple attenuator for some testing purposes and also for monitoring the isolation port on my QRO 90 degree hybrids (23cm hybrid and 70cm hybrid) for generating alarms in case of errors (using my Dual RF Head and Dual RF Head - USB devices).
There are a few commercially available high(er) powered attenuators, but they are quite expensive, so I decided to design my own. This design works with a reasonable return loss up to around 1.8 GHz (measurements pending).
I can produce a limited number of the attenuator for anyone interested (or you can have your own made from the data linked below),
Assembled and tested attenuators (WITHOUT heatsink) is sold for €135,- each, a drilled (with M4 threads) heatsink as the one shown on this page is additional €55,- (requires extra cooling/limited duty cycle!)
When ordering, I can measure the coupling factor for a, by you, specified frequency.
This is plus shipping.
The design
The attenuator is based on a 450W terminating resistor from TTM Technologies (Anaren), part number G450N50W4. This one can be bought different places, one possibility is Mouser.
Apart from the resistor, the attenuator uses:
- N male connector from QAXIAL, part number N04-00-01
- SMA female connector also from QAXIAL, part number SMA14-2H-15E .
- 2 pcs M4 screws, 6mm long
- 6 pcs M2.5 screws, 4mm long
- 4 pcs button head screws, max 4mm long (for N male)
(plus heatsink/fan depending on usage)
Both the connectors can be purchased from RF Microwave in Italy (as well as other places).
The coupling port is done very simple in this case. It is only the center conductor of the SMA connector that picks up RF at a low level. The coupling factor depends on the frequency, generally the higher frequency the higher the coupling factor is ("more power on the coupling port at high frequencies").
As the coupling port is made like this, the return loss on the coupling port is not good. The coupling port does not look (at all) like 50 Ohm. In most cases that does not matter too much, one way of helping with this if needed by the equipment connected to the coupled port, is to insert an SMA attenuator (f.ex 10 dB). This will make the connected device see a much better return loss (two times the attenuation).
Mechanical design
The design of the attenuator is pretty simple. I tried to make the housing as compact as possible. The attenuator needs to be mounted on a suitable heatsink, 450W is a lot when dissipated for a long time!
For my initial tests I used a 20 cm long piece cut from a Wakefield 125410 heatsink (1 meter long). Thermal resistance for this 20 cm long piece is 0.45 K/W in natural convection (1.2 K/W for 3 inch).
If we assume 400W is dissipated, this means that the heatsink will be 450 * 0.45 => 180 degrees above ambient!
This will of course destroy the resistor (data says max 100 degC for 450W). So with a heatsink like this, you will need a lot of airflow if you intent to run the dummyload at such high power for a long time. For short (multiple seconds) of 400W carrier there will be no problems, but please observe the temperature of the heatsink/dummyload!
The aluminium box for the attenuator is 50 x 36 x 25 mm in size, it has four holes in the corners to fasten it to a heatsink (using M4 screws, 35mm long).
Pictures below is from the first prototype, the final version is as shown on the screenshots.
Measurements
Return loss for the input port:
Coupling factor from input (N connector) to coupled port (SMA connector). The coupling port had a 6 dB attenuator connected, this has been subtracted from the numbers shown (so this is the coupling factor at the SMA connector).
Below is the coupling factor from input to coupling port, this time without a 6 dB attenuator. This means that the coupling port is connected directly to the VNA (with resulting mismatch as the coupling port is not 50 Ohm):
Below is the return loss on the coupling port with a 6 dB attenuator (shows around -12 dB as expected):
