Sunday 18 November 2018

Es'hail 2 and my Spectrian amp

2 MS/s @ 2.4 GHz 80 watts

Spectrian bare bones amplifier

Thermal image of Spectrian

Now that Es'hail 2 is launched I thought it might be time to see how much
power I will be able to generate when it comes time to use the asset.

It didn't go as well as I had planned. I can generate about 80 watts of DVB-S2
at 2.405 GHz that looks reasonable but it requires almost 800 watts of input power.
The amplifier is the working parts of a Spectrian unit minus the original heatsinks
(because of shipping costs from the U.S). It is not obvious from the thermal image
as to why the efficiency is so low. The MRF286 devices have a claimed efficiency
of 28 - 32% at 2 GHz obviously they will have less efficiency at 2.4 GHz and there
will be losses in the PCB and the combiners but 10% is pretty low.

So it looks like I will have to try a different approach to saving us all from
global warming. Currently I am looking at the MRF7S24250 device which has
250 watt output at 2.4 GHz a gain of 14.5 dB and an efficiency of about 55%.


12 comments:

  1. Looks interesting. The efficiency is certainly very frequency dependent. I must have a look at Es'hail 2, I read that it had been deployed.

    ReplyDelete
  2. Class A amp so runs warm driven or not.

    ReplyDelete
    Replies
    1. Wrong, the driver on the left is Class A but the main amplifier modules are Class AB. With no drive the main amplifier draws about 2 amps driven to 80 watts output it draws around 38 amps.

      Delete
  3. Why do you need 80watt? When 20 will be enough? Are you going to use 60cm dish?

    ReplyDelete
  4. Well I hope I need a lot less than that but that 80 watts was at the output of the amplifier. What actually gets radiated will be considerably less than that due to feeder losses and poor dish illumination. Until we have some actual figures in February rather than estimates we won't know. Someone in the industry told me that the estimates satellite manufacturers give are usually optimistic.

    ReplyDelete
  5. I am not surprised at the rolloff in Spectrian performance at 2.4GHz. I used mine on 2304MHz (and 2302MHz to work VK ) on eme in the USA and it worked as well as could be expected. The performance was awful on the JA eme band at 2424Mhz.

    When I brought it back to the UK there was a noticeable dropoff in efficiency and output at 2320MHz.

    ReplyDelete
  6. That's just over 25W out per board!

    ReplyDelete
  7. I'm interested in building an amp. I downloaded the PCB .dxf files for the MRF7S24250 amp from the NXP site. Stopped short of having them converted to Gerber. Did you find a Chinese fab house that is able to directly use the .dxf files? Price and minimum quantity?

    KM7MH

    ReplyDelete
    Replies
    1. I was waiting until I have some positive results to share before doing an update to my blog. I looked at getting some boards made using the material specified by NXP but as it is not a stocked material by most board houses it would have been very expensive. Instead I used RO4035B material which is electrically similar to the required material but does not have as good thermal characteristics. The boards were purchased from www.pcbcart.com and cost $191 for 10 boards, the set up fee was $73 and each board cost $9. The first attempt didn't work too well and I had to modify the board to get any power out. I then discovered I had inadequate grounding on the input and output. So I replaced the hacked board with a new board. With proper grounding, I am getting about 60 watts out. The unsaturated gain is very close to the 14.7 dB claimed by NXP. Obviously the output match is not correct. There are a couple of transmission line stubs on the drain that can be tuned with ATC capacitors. I plain to model the output match using a Smith Chart program to get some idea as to what effect those stubs have. The input circuit seems to model very well, basically it is a 1/4 wave transformer that matches directly from the device impedance to 50 ohms. The output match on the other hand is more problematic. If I were doing this again I would solder the ground plane side of the PCB directly to the heat spreader as good grounding seems to be imperative at these frequencies. I will do a fuller post when I am happy with the results.

      Delete
  8. That should of course have been RO4350B

    ReplyDelete
  9. Thanks for the information.

    I was hoping for a turn key approach for a change. Sounds like a ground up design may be the better option. I have designed many microwave amplifiers using Eagleware Genesys: https://www.keysight.com/en/pc-1297125/genesys-rf-and-microwave-design-software?cc=US&lc=eng

    What bias voltage are you using?

    KM7MH

    ReplyDelete
  10. This is another potential candidates as it comes as a module but the bias would need looking at https://www.ampleon.com/products/rf-energy/pallets-and-modules/BPC2425M9X2S250-1.html

    I am using a bias voltage of around 1.75v (current of about 1 amp).

    ReplyDelete