Well-Engineered RF Amplifier Powers Ham Radio Contacts

Typically, amateur radio operators use the minimum power needed to accomplish a contact. That’s just part of being a good spectrum citizen, and well-earned bragging rights go to those who make transcontinental contacts on the power coming from a coin cell. But sometimes quantity has a quality all its own, and getting more power into the ether is what the contact requires. That’s where builds such as this well-engineered 600W broadband RF amplifier come into play.

We’re really impressed with the work that [Razvan] put into this power amp. One of the great joys of being a ham is being able to build your own gear, and to incorporate the latest technology long before the Big Three manufacturers start using it. While LDMOS transistors aren’t exactly new – laterally-diffused MOSFETs have been appearing in RF power applications for decades – the particular parts used for the amp, NXP’s MRF300 power transistors, are pretty new to the market. A pair of the LDMOS devices form the heart of the push-pull amp, as do an array of custom-wound toroids and transformers including a transmission line transformer wound with 17-ohm coax cable. [Razvan] paid a lot of attention to thermal engineering, too, with the LDMOS transistors living in cutouts in the custom PCB so they can mate with a hefty heatsink. Even the heatsink compound is special; rather than the typical silicone grease, he chose a liquid metal alloy called Gallinstan. The video below gives a tour of the amp and shows some tests with impressive results.

35 thoughts on “Well-Engineered RF Amplifier Powers Ham Radio Contacts

  1. Wow. It says a lot about build quality and attention to detail when I had to take a second close look to realize that the photo is not a 3d model render.
    That is a fantastically clean build!

    I am concerned about the life of those SMA connectors though.
    I have not seen an SMA connector rated for such a high power at ham frequencies because they dont have the thermal mass to dissipate the heat and get hit by derating pretty hard. The only ‘high power’ one I have seen is only rated for ~200W.

    I worry that the dielectric breakdown if often used at full power will cause issues down the road.

      1. The GHz-ratings only apply if the connector is part of a transmission line where the energy flows in the field between the conductors. This is not the case here at the end of the line and with wavelengths that are some magnitudes greater than the mechanical dimensions, so RF-current will flow in the conductors and the skin-effect makes it even worse. IMHO screw terminals like the ones you used for DC input would be the better choice here.

          1. I hadn’t really thought about SMA power ratings before, despite having used these connectors for many years. Most of my designs have been a few watts at most so it has never really mattered.
            I looked around a bit and found a couple references to SMA power levels, but none of the references actually explain how they arrived at the numbers.

            This page suggests >600W below 1GHz: https://www.centricrf.com/power-handling-connectors/

            450W @ 100MHz: https://www.ducommun.com/pdf/mwcat/mw6.pdf

            Note that power handling will drop dramatically at higher VSWR. The second page mentions those numbers are at 1:1. The first page probably assumes this but it doesn’t explicitly state the test conditions. Aside from VSWR, temperature rise at the connector is probably also a concern at high powers but I haven’t found any references on this.

            I’d love to see a more definitive reference that also includes other connector types (like type-N, which I suspect would be more suitable for high power levels).

          2. Fire it up and let us know if and when the connectors fail. Otherwise, excellent work Razvan. I’m looking forward to these new chips sets in use for boosting qrp systems.

      1. Isn’t galinstan liquid above -11 degrees C? That doesn’t sound like a great thing to use for a heatsink compound, no? How do you make sure it doesn’t seep from under the transistor and go where it shouldn’t? You mention it in the article but this bit isn’t very clear to me. Or does it just “wet” the surface and stays put when you screw the transistor down?

        1. Galinstan’s melting point can be between -20C to over 10C, depending on the exact ratio of Gallium / Indium / Tin. However, it wets the surface* (just like hot solder does when it touches clean copper) so it wouldn’t go anywhere (unless you use excessive amounts).

          * to wet the surface = at a molecular level, the Galinstan mollecules are more attracted by the copper / zinc surfaces than by the air. This force is strong enough that makes the Galinstan stick to the transistor’s metallic tab & the heat spreader and not wander around. See https://en.wikipedia.org/wiki/Surface_tension

  2. Hi,
    Would be nice to have a little more specs / performance info:

    Frequency range? HF is 3..30 and 6M is 50-54, so 3..54 Mhz? Quite a range.

    Efficiency? Linear or NOT… ?

    Any needed setup adjustments?

    How about a Schematic???

    Not learning much from just photos….

    Regards, Terry King
    …In The Woods in Vermont, USA

    1. One iof the captions says hf/6m, so it stsrts somewhere, presumably 3.5MHz or maybe even 1.8MHz and goes to 54MHz.

      It’s a linear, not class C. The broadband transformers give it away, class C generally uses tuned circuits, besides, the schematic shows biasing of the gates, thst’s not class C.

      No tuning. It’s broadband. Besides if there was tuning, we’d see a bandswitch, ultiple coils and a variable capacitor or twI don’t know abiut adjustment. Maybe bit if fiiddling with the gate bias. No tuned circuits to adhpjust.

      The schematic is on the builder’s page.


    2. Terry: Dan’s first paragraph has a yellow hyperlink to the original article, which has the schematic, 1.8-54 MHz range specs, linearity & harmonics graphs, etc., and a more complete description & Q&A by Razvan.

  3. This reminds me a lot of the QRP Labs 10w PA, but about 60x the output! I wrote up the whole thing on my site with pictures:

    The impressive thing about that PA is the clean output and the robust design using 50 cent IRF510’s. The OP’s layout looks very similar, but perhaps that’s just because there are only so many ways to do a good job with the layout :)

  4. These excellent designs far exceed my NXP challenge effort, but if you’re interested take a look at sspafun.blogspot.com. My take was to get running quickly by using the PCB and some parts from a common eBay amp kit . The MRF101’s power, frequency range and efficiency justify $20 for me. 35v or 50v operation is no big issue with $25 LED strip supplies on Amazon. You need at least 25v on the IRF 510 anyway. Hopefully I can find time to fix 50mhz :)

  5. I wonder if multiple of the 600W RF decks could be combined for higher power output. That’s how my SPE 2K-FA is designed – 4 x 600W modules plus a combiner circuit.

    Love this design. Very clean and well thought out with good protection circuitry.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.