Posts by dj0abr

    Before doing the final layout, we started some experiments with the BLF2425M9. Mounted on a homebrew prototype PCB this Mosfet generated 22 watts out from 500mW input with almost 50% efficiency.

    (nice for use in portable 60cm dishes).

    The plan for the final PCB is now as follows:

    1) 2.4 GHz SAW filter at the input (as recommended for the Amsat TX board)

    2) MHT1008 amplifier

    3) LP filter

    4) --- CUT ---

    5) Attenuator

    6) BLF2425M9 amplifier

    7) LP filter

    at step 4) there are simply a couple of holes in the PCB, the PCB can be cut here into 2 parts. So if only the MHT1008 is needed, the PCB is broken at 4) and the other part is thrown away (the PCB is very cheap :-) ).

    If one needs both stages (i.e. for higher gain), then he uses the complete board with a gain of up to 34 dB.

    Some applcations (i.e. the Amsat Mixer board) need a gain of >17 but less then 34 dB. To adjust the gain to the needed value a simple attenuator (0,5 watts SMD resistors) can be mounted.

    This week we are doing the measurements with the BLF2425M9 board and also design the low pass filters.

    vy 73
    Kurt, DJ0ABR

    (Board have been sent to DO1CTL, DD0KP and ON6AJ)

    The last measurements were take in my Shack with very simple equipment.

    This week we have been invited by Rohde&Schwarz to measure the PA with calibrated and expensive instruments.

    The measurements were take by Alex, DG5RAQ and Johannes, DC1RJJ.

    This picture shows the max. output power with 12 volts supply.

    this is the spectrum (12 volts)

    12V was measured just for fun. The datasheet specifies 28V for this transistor, so the next pictures shows the 28v operation.

    The PA was driven with 25 dBm:

    and generates 10 watts output power. Supply: 28 volts and 0,709 amps (=19,85 Wdc). This is about 50% efficiency, thats not bad for a 2.4 GHz PA.

    this is the spectrum at 10 watts output. The measurement was taken using a power attenuator.

    the requirement for spurious emissions (> 1Ghz) is -50dB. We measured 40,88. If the antenna has at leat -10dB at 4.8 GHz then we can use it without any filter. Otherwise some filtering is needed.

    vy 73, Kurt, DJ0ABR

    I have the SF8008 also running with the Amsat Downconverter.

    Everything works fine when I select the channels manually.

    In the link to dd1ku is a reference to a channel-list: QO-100.tar

    This file is maybe old ? It does not work in my SF8008 because the channels are set to wrong frequencies and also to DVB-S instead of DVB-S2, DVB-S2 needs additional parameters which are not in this list.

    Is there an update to this list available ?

    I have prepared a couple of board, tomorrw we will measure if the input matching is the same for all boards.

    Taking measurements on 13cm is a realy challenge :-) One SMA <-> N adapter has 2dB loss, another identical adapter only 0.5dB.

    Also the temperature of the transistor has some influence on the input match.

    But these are only details. The most important thing is, driven with the Amsat-TX mixer it generates a couple of watts output and can be used for QSOs.

    I found one important issue: the maximum input power should never exceed 300 mW.

    With 400mW at the input I measured a degration of the transistor gain, so we have to be careful not to overdrive it.

    In the short wave Amps, I simply put Zener diodes on the gate for protection. For 2,4GHz this is not possible due to the diode capacitance.

    Any idea how to protect such a circuit from being overdriven ?

    vy 73


    just did the first QSO with this PA with DK1KQ. He reported a good signal.

    120cm dish and the homemade dual band patch feed (according the NL and G document).

    The previous measurements were taken with a continuous carrier from a signal generator.

    In SSB the power is maybe half of that because we need linearity. With 24V and the Amsat board as the driver (abt. 50mW, maybe up to 100mW) the SSB output power was about 3 watts. This made an average signal on the NB transponder like most other stations are doing.

    The current board was just made to prove the concept. For a real PA a few things are missing, like a PTT circuit to switch on the quiescent current during TX only. Also I want to change the size of the board so it fits into the usual Tin cases.

    I will publish the PCB data this evening, mainly for people doing tests and measurements. If you want to use it for regular sat operation you should better wait for the next layout which is hopefully available in 10 days from now.

    Also, FR4 is not FR4, if a board is made from a different manufacturer, it may be required to check the input match.

    vy 73
    Kurt, DJ0ABR (DC1RJJ)

    as promised I give you an overview on the progress for building a low cost PA for es'hail-2 NB transponder.

    We have the prototype finished and tested. It is running fine.

    The total cost of material of the amplifier is about 20 Eur only.

    The performance data are:

    at 12V/850mA

    output power is 5 watts

    with 100mW input (nice fit to the Amsat TX mixer board)

    at 24V/680mA

    output power is 10 watts
    with 200mW at the input.

    You can read more detailed information here:



    since this prototype works so well, we will continue with a second version including a PTT circiut and a few other parts. But we keep it simple so everybody can build it easily.

    If more gain is required (i.e. for SDR transmitter), simply two of these boards can be chained, giving a total gain of up to 34 dB. I have used two boards in series for the power tests because my generator has only 10mW max output.

    All parts are available worldwide from digikey or mouser.

    I have 4 PCBs left, if anyone wants to build it, just write me a forum-message.

    The part list, circuit and gerber data will be published on above web pages during the next days.

    vy 73
    Kurt, DJ0ABR

    super ! Danke für die Info,

    hab das Filter gleich bestellt da ich nächste Woche erste WB-Empfangsversuche machen möchte.

    Hab den TX und RX Mixer jetzt 10 Tage im NB Einsatz und bin außerordentlich zufrieden, spielt perfekt ! Ihr habt für uns tolle Platinen entwickelt.

    Nur ein kleiner Hinweis: für die TX Platinen habt ihr zu enge SMA Buchsen bestellt. Vielleicht die Stückliste mal checken. Aber mit einer Feile kann man sie erweitern dass sie auf die Platine passen.

    73 de Kurt

    thanks to all answering my request for building a low-cost PA for QO-100.

    We now have a small group of hams, and also access to R&S test equipment.

    We started the development beginning this week.

    Circuit and layout is ready, components arrived.

    Next week, when we get the PCB, we start assembling the prototype and testing.

    Designing the matching using PCB-traces was not easy, so there is a high risk that it maybe will not work as expected and we have to make some changes in the layout. We will see. If it works I will publish all documents to the community.

    More information end of next week.

    vy 73, Kurt, DJ0ABR

    PH2M , i know the circuit in your link. It looks good except one thing.

    What looks strange: L4 ... 10uH.

    These SMD parts (with 10uH) have self resonance frequencies of usually below 100 MHz. So it will be just a capacitor at 739 MHz and pull down the LNB's output signal through C5.

    And L5 with 10nH has no effect.

    I think L4, L5, C5, C6 should be recalculated.

    I have opened and modified the Megasat Quad-LNB (see PH2M's link above) to connect an external reference (25 MHz):

    This is the original LNB:

    first the black plastic cover was removed, then it looks like that:

    Now I removed the silicone:

    and removed five screws (Torx No.8):

    the 25 MHz Crystal is at the right. What a funny layout, the traces from the crystal go over half of the board to the chip, but it works so its fine.

    The external 25 MHz must be fed into pin14 of the chip. It did not unsolder the crystal (which is very hard to do), instead I cut the trace close to the chip:

    the wire of the connector was desoldered from the board, and via a 1nF capacitor a thin wire goes to pin 14 of the chip.

    The connector is now marked "LO". This is where I connect the output of the leobodnar GPS. Via a simple 25 MHz low pass LC PI-filter, 2x 220pF and 680nH in series). This filter reduces the level to abt. 6dBm.

    Three LNB outputs are available. I am using one for NB and another for the WB transponder.

    vy 73, Kurt


    Heiner's link shows a very good paste.

    You should set a reasonable quiescent current. There are some descriptions in the net talking about 2A per mosfet. This is too high and results in a very hot board needing a huge heat sink.

    I am using 0,6A per mosfet (1,8A in total), linearity and gain is ok, and higher current only slightly improves that.

    If the +12v- enable pin is connected to the PTT (via a relais), then you can use it for normal SSB with a little cooler in the size of the spectrian board.

    This may be different for long transmissions like sstv, but I have not tested that until now.

    73 Kurt

    I am now active on the satellite using the AMSAT converter boards and an Icom Transceiver, but the SDR still runs in parallel to show the waterfall display.

    Therefore I have extended this software with a synchronisation between the Icom and the SDR waterfall (ttyUSB0 is used by default, make sure that the user has access right to the dialout group, which is required for the USB/serialPort).

    A mouse click into the waterfall sets the Icom to this frequency and also the waterfall will follow if the frequency knob on the Icom is rotated.

    Only for the linux guys. I have it running on an Odroid-XU4. A Raspberry is probably too slow, at least if it also runs the browser.

    The program is pre-configured for the frequencies used by the AMSAT converter boards.

    I am using the model from your second link, the same but the twin version. I have also the quad version here and will modify it this week.

    It is a PLL model using a 25 MHz crystal. It has the usual drift like all models have, but it is very easy to connect an external signal source, in my case a leobodnar GPS which works excellent.

    The other link pointing to the multifeed looks to be DRO.

    If you have access to the cqDL 1/19, there is an article about this Megaset LNB and its modification.

    73 de Kurt

    there are many hints regarding bandwith and so on, but I'm not sure if this is just a math issue.

    Because the CW beacon (in the steady phase) is just one single frequency, I do the measurement by comparing the spectrum to a known source.

    First I connect a signal generator (single frequency as the beacon) and write down the dBm value for each dB-Line of the SDR software (I don't take care on the labeling of the spectrum's y-axis).

    Then I connect the LNB and read the noise level as well as the beacon level and look into this calibration table for the real value. This makes the measurement almost independent of the SDR software.

    To be honest, I have no idea if this is a good procedure or just nonsense, but it looks to work well. The CW beacon is abt. 28 to 32 dB above noise level.

    the SF2124E looks good. Is there a stock somewhere in EU ?

    for a pre-amp 0dBm input and 27dB gain there are many devices available,

    just as an example the SE2576L-R from Skywork Solutions, which has already an integrated input match. and needs 5v at abount 0,5amps.

    nice comments !

    Now we have a good list of available options.

    I have the Spectrian here which I will activate the next days, so there is no real need for a PA here, but I will do something this summer, just for fun. Building equipment is a great hobby.

    Heiner, mny thanks for my first QSO on QO-100, running 0,4 W only.

    it looks like the German (and EU) customs is selecting out the Chinese Wlan Amplifiers. It gets more and more complicated to get it . Two of my orders never arrived and aliexpress dealers do not ship into the EU any more.
    Whats still available on ebay is usually fake.

    Do we need the Chinese ? No, we can do that by our own, and we can do better !

    What about concentrating our knowledge to build our own QO-100 amplifier ? I cannot do it alone, but if we work together, we can succeed.

    What I can do: PCB Layout, purchasing parts, manufacturing the prototypes

    what is needed: experience designing 2,4 GHz circuits, calculating 2,4 GHz strip lines, designing snow flakes and similar. And we need someone with test equipment for 2,4 GHz who is able to test the prototypes (my instruments end at 1 GHz).

    Some suggestions as a starting point:

    available at Mouser and Digikey: the freescale MHT1008N…c/data_sheet/MHT1008N.pdf

    18,5dB gain and 12,5W max CW power. Would be a perfect match to the Amsat TX converter.

    We could start with building the reference design from the data sheet, unless somebody has a better idea.

    Using a Rogers PCB material the total cost of material could be < 50.- Eur (in qty of 20 or more).

    I am not interested in any profit. If such a project generates a little profit we could give it to Amsat for their excellent QO-100.

    So, thats my idea.
    What do you think about it ?

    vy 73