What TX amplifier do you plan to use?

  • M0VKK

    Hallo... any question or note about the MRF7S21110HS board...

    i can not see any 50R resistor on the output combiner..maybe it is missing on the PCB and has to soldered on the open end of the stripline? Or maybe the stripline and the capacitor is used as the LOAD @ 2.2GHz?

    Michael dg0opk

  • dg0opk To be honest I hadn't noticed until you pointed that out. I suspect that the stripline/capacitor forms the load. Too much mismatch between the two devices could cause that to fail quite quickly. I might cut that track and replace it with a 20W 50 ohm resistor.

    I'm also going to get my 20dB coupler tested, it's a cheap Chinese one and I'm suspicious about the efficiency I'm getting. It can't be that easy to get this amplifier to work. I also changed the shunt input capacitors whilst initially matching and now I'm not sure I put the original ones back....

    I've got another one of these on order and I will test it properly next time. I just had two hours spare on Saturday morning and was in a rush to see if I could get it to work.

  • M0VKK

    Hi Richard...

    i have also orderd ... but no time to test it the moment... maybe a good next step is to cut the stripline on the output combiner and put in a 50Ohm load .... then more tests can be done?


    the anaren combiners are specified until 2.3GHz so may be some little more loss on 2.4... and the transitor datasheet looks not so bad... max output >150W at 32V P1dB (2.14GHz Page 8 of dataheet), S parameters are given until 2.2GHz and they look flat... maybe it is usable at 2.4G... i hope for that ....if not ...then maybe a good one for 2.32G contesting...

    so lets see ..

    73 de dg0opk

  • Those are about the main reasons I chose the board. The devices being rated at 2.2GHz gave a much better chance of success that the ones that are more suited to 2GHz. The Zs was pretty flat over the operating range and so could be assumed to stay reasonably flat over the extra 10% range. The output looked a little bit more tricky and I assumed I would have to snowflake quite a bit to get a good match . Which is the reason I'm a bit suspicious of my efficiency measurements, I was guessing that it would be closer to 20-25% without adjusting.

    As you say the xingers are specified up to 2.3GHz - but looking at the curves and extrapolating a bit indicated that the loss at 2.4GHz wouldn't be too bad - as long as the power was kept to a reasonable level. Also I found a replacement Xinger the same size rated for 2.4GHz that only costs £3 which could be substituted if required and rated at 150/200W.

    Model XC2500A-03S

    Sadly I'm unlikely to get any more time to test until next weekend.

  • Ok more progress - but not good news! I roughly built the amplifier into a case for testing - and tried running it from the pluto with DATV. The DATV output from the pluto was about -9dB so it was only putting out a few watts. I ordered another +18dB amplifier for the input and was testing it and it happily gave an estimated 40W out without getting too warm and was pretty stable. I ran it for about half an hour like that and could easily receive 500kbs with a good MER -so the signal looked clean.

    However - because I was experimenting there was no screening on the preamplifiers or driver . I disconnected the input but forgot to switch the PTT off. It broke out into oscillation and drew about 20A at 28V for a few seconds and one of the output devices died.

    The other device survived and I could limp out about watts - but the second device couldn't be set to a good idle current and now is only class C! Surprisingly the output xinger survived!

    I have a spare output module so I will fit that at some point - however work requirements mean I'm not going to be able to repair it for the next two weeks. I also need to replace the 28V-12V regulator as I forgot the extra draw of the extra preamp took it out of spec and it got very warm!

  • Whilst I haven't experimented with mine any more as I'm still tied up at work and will be for the next week or so, someone on facebook (James smith) has tried the same amp and said

    'Andrew amplifier module from Bison Electronics tested at 24 volts biased at 1A per device gives 55 Watts out for 900mW in at 2405MHz with no tuning yet. It will be the driver for my next big amplifier but it is already as good as a spectrian! Supply 6.4 amps for 55w out 3.8A for 20w out 13.6dB gain and it will also work on 32 volt supply'.

    So his results look pretty much in line with mine.

  • DB8TF thanks for the pioneer work…

    I ordered same Nokia Amp and checking the different posts as a preparation until the board arrives. I read the thread several times and I seem to miss some parts of the puzzle...

    Here what I got - StepbyStep

    1. remove and bridge the first stage (D100 = PreAmp?) and both filters (W100/W101) and Circulator

    2. change IDq for Driver- and End-Stage

    3. change Doherty Cirquit to enable 2nd EndFet

    4. put a Anaren Xinger in the output and cut the 90°combiner lines on the board

    5. Start Optimization/SnowFlaking as shown in the pictures with low Power

    (Probably I will interrupt connections to the Peak Amp and drop step 3/4, if one EndFET will give me enough power)

    Could you give me some details about following mods, please:

    1. IDq - you setup an IDq of 2A for the whole amp (incl. pre-Amp) or per End FET?

    2. you injected the BIAS from external 12v, could you post the circuit diagram?

    vy 73 de

    Alex, DK4FT

  • Hi, Alex and friends of BLD6G22L,

    I first tested my amplifier without any mods.

    The original bias control works very well from about 15...30V operating voltage (left PA 10...30 mA, right PA 1.23A, driver stage 0.36A, preamplifier 90...50mA).

    The gain at 2.2GHz is about 41dB, at 2.4GHz about 10dB less.

    A single small piece of copper foil on the C between W100 and the driver stage boosts the gain at 2.4GHz to 39dB.

    It is important to screw the aluminum cover on again before each measurement. Without the cover I had up to 2dB gain change and there is the danger of parasitic oscillations due to unintentional touch.

    More than about 20Watt I couldn't generate yet, because my laboratory power supply is too weak.

    Some more remarks:

    W100/W101 are not filters, but also circulators or more precisely isolators. This is easy to detect, they are slightly magnetic. I wouldn't remove them.

    In my opinion the better decoupling of the stages is more important than the possible increase in gain. This could have been a reason for the preamp problems of DB8TF.

    The input socket is MMBX.

    There are matching plugs: HUBER+SUHNER 16_MMBX-50-2-4/111_NE, e.g. from Mouser. The amplifier cover must be filed off a little so that it fits.

    The second MMBX socket is connected to the directional coupler at the amplifier output. An RF power meter could be connected there. I haven't measured this branch yet.

    The output socket is AFI.

    With an „AFI Plug PCB Bullet“ (Amphenol 920-263P-51P, also e.g. from Mouser) and some 3D printing a cable plug can be improvised.


    Roland, DM5RM

  • Hi Richard,

    I confirm that Andrew amplifier module from Bison Electronics is by far the easiest to get working on 2.4GHz. I've made no changes apart from adjusting the input trimmers for best return loss:

    2400MHz 25dB, 2408MHz 35dB - lowest was on 2411MHz at 43dB

    With a 28V supply and a quiescent current of 1100mA on each device checking the gain on 2.2GHz, I got 17dB, and at 2.4GHz 15dB. With 3W of RF input it gives 100W RF output on 2.4GHz, current 7.75A and that includes 150mA for the heat sink cooling fan, so it's about 50% efficient. Heat Sink temperature just reached 40 deg C and stabilised running at this level for an hour.

    Reducing the RF input by 3dB gave me 50W output on 2.4GHz, current 6A.

    I've made up temperature compensated bias regulators using the LP2951 IC with a 1N4002 used as a temperature sensor, one glued either side of the heat spreader.

    The quiescent current is very stable over a wide range of temperatures only varying a few mA from the ambient level set of 1100mA for each device.

    I've attached a photo of the Andrew Amplifier setup and also a circuit diagram of the bias regulator in case you are interested.

    73, Dave G3XOU

  • G3XOU Thanks for the Bias Cirquit.

    DM5RM Good hints from your side.

    I received my Nokia Amp from Bisonelectronics and startet to rework the input/output (attached directly semi-rigid Cable with SMA-Connector instead of the very specific existing ones). Biases I will leave as it is for the moment and ordered this Supply to have enough power:


    Roland, could you help me with the following two questions, please:

    1. Did you change the Doherty Cirquit of the peak amp like Flo?

    2. Could you upload a photo of the copper piece on the C?

    73 de

    Alex, DK4FT

  • DK4FT

    Here is the picture of my one and only mod so far. The size resp. area of the selfadhesive copper foil depends a bit on the operating voltage.

    You should start with a tiny piece of foil and then gradually increase the area until it doesn't get any better.

    And don't forget the cover...

    Just by the way, with your power supply, you should almost be able to weld ;-)

    73+55 de

    Roland, DM5RM

  • Just tested the Amp with my Lab Power Supply.


    - the Doherty Change to enable peak FET permanently,

    - change input sockets to SMA

    - change Output

    - remove Isolator/Circulator at the output,

    nothing else, so far.

    IDQ is 3,2 A @28V.

    The Amp has abt. 48dB Gain on 2,4 GHz, not too bad.

    If the meassured gain is correct, there is no need for snowflaking...

    Due to my Lab Supply I am able to produce only 2,5 W at the moment, but the Amp is getting really hot with both End FETs enabled even in idle. (need a fan...)

    Now I am wating for the big supply (welding Source ;-) )

    Thanks Roland for the picture....

    Here is my one....