ad: ProAudio-1

600 watt HF Linear Amplifier Project

Discussion in 'Homebrew and Kit Projects' started by KD8UYQ, Feb 9, 2017.

ad: L-HROutlet
ad: l-rl
ad: l-BCInc
ad: abrind-2
ad: L-Geochron
ad: Left-3
ad: Left-2
ad: L-MFJ
  1. KD8UYQ

    KD8UYQ Ham Member QRZ Page


    Last year I bought the Icom IC-7300 transceiver along with Icom's IP Remote control software (RS-BA1) with the intent of operating this radio remotely from my laptop via an internet connection. In addition I wanted to boost my signal up a bit by incorporating a linear amplifier into the mix. Because I would be operating my station via remote control, the linear amp also needed the same remote control capability. Another requirement was this linear should be a solid state amp.

    There are several commercially available options in the $1,600 to $2,000 and on up range that would fit the bill but I decided that it might be fun to build my own. Research on the internet suggested that a 600 watt MOSFET based linear amplifier based on Motorola's EB104 Engineering Bulletin could be a good match for my needs and it just so happens that Communications Concepts out of Dayton, Ohio sells a EB104 kit along with several usefully accessories.

    Now this amplifier circuit design is not new. It has been around for almost 25 years and many Hams have built linear amplifiers based on this Motorola circuit with varying degrees of success. Thanks to a lot of research and experimentation from other individuals who have worked on improving several of the shortcomings inherent to the EB104 design as implemented by Communications Concepts, Inc (CCI) I was able to build the amplifier board and power it up without any adverse far.

    The first order of business was to gather components and sub-assemblies for this build.

    Above: The 600 watt EB104 kit from CCI and one of their accessories, a 3/8" copper plate heat spreader. You need this heat spreader or something very similar plus a finned heat sink because the 4 power MOSFETs get very hot when operating at their power output rating.

    Above: This build turned out to be a project incorporating many internationally sourced parts. The 48 volt 21 amp switching power supply on the left was made in Taiwan and is sold by Jameco Electonics. The six circuit boards came from in Russia via eBay. They are:

    Top left; Input protection and attenuation board
    Middle top; LCD Power output and SWR display board.
    Middle second board down; Tandem match that connects to the display board
    Middle bottom board; High SWR protection board
    Bottom right next to the power supply; Power supply protection board
    Big board to the right of the picture; Low Pass Filter diplexer board covering 160 to 10 meters

    I also have three blue LED displays for voltage, current, and temperature. They came directly from China via eBay.

    After I received everything the next step was to mock up this project on a piece of plywood. This makes it a whole lot easier to test and troubleshoot the sub-assembles and components before I move them into the metal cabinet in which will be there final home.

    Above: You can see that I started wiring the AC power portion of this amplifier. Also note the little circuit board under the low pass filter board. That little guy is the interface to my IC-7300.

    To be continued:

    KX4OM, K9RDU, W1BR and 3 others like this.
  2. KD8UYQ

    KD8UYQ Ham Member QRZ Page

    As I understand it the Motorola EB104 circuit was never meant to be a production ready design. Instead Helge Granberg, the Motorola Circuits Engineer that authored the EB104 Engineering Bulletin was touting the benefits of their MRF150 Power FET over bipolar transistors for HF push-pull/parallel based power amplification applications. His circuit offered a good starting point for further development. This is where Don Solberg, (K9AQ), Andreas Duessler, (DL6EAT), Corlin Darby, (M0OTT) and others took on the challenge to improve the performance and stability of this design.

    Don built a linear amplifier some time ago based on the CCI's EB104 kit but found that it had stability issues, high internal SWR on the higher bands which contributed to significantly lower power output on those same bands. What Don did, and I adopted in my build, was to improve on the design and better yet document in great detail exactly what he did. Armed with that information this is how I built the EB104 PA board.

    Firstly, the circuit cutouts for the MRF 150 power FET doesn't give them much clearance and could contribute to a flash over or a short to ground during soldering. The recommendation is to open the holes which I did with a couple of Dremel tools with two different diameter drum sander bits.

    Routers to make the holes bigger.JPG

    Below: Note that the top hole has been opened up a bit compared to the two below it.

    Routing out power MOSFET holes.JPG

    Below: All of the openings for the power MOSFET have been made larger and still the tolerance between the transistor and circuit board is pretty tight.

    MOSFET in routed hole.JPG

    Another thing that I should bring up is copper is one tricky metal to tap. You have to be very careful when tapping the holes for the 4-40 screws that secure the FETs to the copper heat spreader. I was able to drill and tap all of the 4-40 transistor mounting holes, the circuit board mounting holes, and the holes to attach the heat sink to the copper heat spreader without breaking a tap. Then a few evening later I broke four taps in a row while making the holes to secure the thermal sensors. You must use a lubricant designed for taping copper and go very slow, backing the tap out on every full turn to cleat the chips. If not you will snap a tap in a heart beat.

    Once that task was done it was time to populate the circuit board with components.


    A couple thing to note here, the kit that CCI provides comes with 100K single turn pots to set the bias for each power FET. Setting the bias is a tricky procedure as is and it is highly recommended that those four trimmer pots be replaced with multi-turn equivalents. I bought 20 turn pots and used them instead. The gold pins do not come in the kit. I added them. Also, the .1uf decoupling capacitors (c13 and C14) in the kit are under rated, voltage wise and it was recommended that I go with .1uf caps with a higher voltage rating. You can see one of them in the lower right hand side of the picture. The ones that I used are rated for 100 volts dc.

    I will talk about that black block with the two holes in it a little bit later.

    Now note the transformer with the orange, yellow and red wire shown in the two pictures below.


    One picture is taken from the rear of the board and the other the front. The important thing to point out is that the orange and red wires are crossed. You MUST install T2 in this manor. The Construction Hints that CCI provides as a down load from their web site is confusing on the installation of this transformer. Install T2 with the wires crossed as is shown in my pictures and you have an amplifier. Install T2 with the red wires and orange wires directly across from each other and you have a high power, MOSFET eating oscillator.


    Attached Files:

    KB1THM, NL7W, KM1H and 1 other person like this.
  3. BG6GLL

    BG6GLL QRZ Member

    SD2933 very good
    MRF151 low efficiency
    AF6LJ likes this.
  4. KD8UYQ

    KD8UYQ Ham Member QRZ Page


    That may certainly be the case but I am too far down the MRF150 road to make changes at this point.


    AF6LJ likes this.
  5. KD8UYQ

    KD8UYQ Ham Member QRZ Page

    Hello all,
    One of the biggest changes suggested in Don Solberg's paper "Improving the Performance of the the EB104 Amplifier with a Transmission Line Transformer" was to to replace T3, the output transformer provided in the the Communication Concept EB104 kit with a TLT (Transmission Line Transformer). I never heard of such a thing until I read his paper. Take a look.

    Above: To the left is the T3 output transformer provided with CCI's kit. To the right is the TLT made from a Fair-Rite #61 multi-aperture core, part number 2861010002 available from Mouser Electronics and about 10" per side of 17 ohm TC-18 coax cable available from CCI. The output of this transmission line transformer is connected to a 1:1 balum made from a #61 torroid core wound with 3 turns of RG-188 50 ohm coax cable. See below.


    The net result of this change is a low internal SWR (1.2 or less across all bands and a Match Efficiency equal to or greater than 98%. I have attached Don's paper which goes into great detail comparing the T3 output transformer included in the the kit with transmission line transformer that I used in my build.

    One other suggestion that was passed on to me and I implemented on my EB-104 board was to added insulators for the four ferrite beads (L1-L4). The ground plane is very very close to them. Should they come in contact with that ground then the 48 volt, 21 amp power supply could come to a quick smoke and fire producing end.




    ANCELB, AF6LJ and KD2ACO like this.
  6. KD8UYQ

    KD8UYQ Ham Member QRZ Page

    Hello again,

    Another short coming of the EB104 is with the circuit board layout itself. Others that have built this amplifier stated that the RF ground, especially around the power FETs could be a lot better. All sorts of fixes for this problem have been suggested. Two that I thought made sense and were relativity easy to execute was to use soldering lugs between the drain tabs on the power MOSFETs and the copper heat spreader via the mounting lugs of the transistors themselves. You can see those soldering lugs in the picture below. Standard #4 soldering lugs are too big but I found some that are used in Lionel train set transformers that fit perfectly.


    I used three soldering lugs to ground the four transistors. The second thing that I did was to wrap .015' copper strips between the ground plane on the top of the circuit board with the ground plan on the bottom of the board. You can see those small strips soldered in place on the upper and lower right hand side of the board. Previously to this the only connections between the two ground planes was via a few mounting and component plated through holes.

    It goes without saying that a high quality heat sink compound is a must between each transistor and the copper heat spreader. I used Arctic Silver 5 thermal paste. You also want to use a similar heat sink compound between the copper heat spreader and a finned heat sink.

    The PA board is complete with a few other changes that I'll point out. On the top left hand side of the board you can see a SMA connector that I added. It is the RF input to this board. Because it only takes about 6 watts of input power to produce 500 watts on the output I figured it won't hurt to use that type of connector. The two black squares which are screwed to the heat spreader and have wires running from them to the circuit board are 10 ohm 30 watt resistors used in the feed back circuit. The kit gives you two 10 ohm resistors rated at 2 watts. The EB104 technical bulletin suggests that you replace them with higher power resistors if you are going to operate on other modes besides SSB.

    Lastly, the electrolytic capacitor that comes with the kit is a 10uf 100 volt cap. I replaced it with a 100uf 100 volt capacitor. Once again, on the advice of others that have gone before me.

    OZ6TI, AF6LJ and KV4PD like this.
  7. KD8UYQ

    KD8UYQ Ham Member QRZ Page


    With the power amp board completed it was time to start wiring up the protection boards and testing them.


    Above: The power supply protection board (center left of the picture with the two white rectangle shaped resistors) protects the power supply from excess current draw, shorts, and high temperature conditions. It also provides the +15 volt regulated supply for the control circuits and a +26 volt supply to run two 12 volt cooling fans in series. A second thermal sensor controls the speed of the fans. I added a third regulated +12 volt supply for the Low Pass Filter diplexer and remote control boards. That third power supply is the small project board just to the left of the power supply protection board. In this picture you can see that the input protection board has a few wires run to it and the digital displays are monitoring the voltage and current levels from the power amplifier board that isn't in place yet. A third display will be added to monitor the temperature of the heat spreader right next to the over temp sensor. When the temperature at that sensor reaches 80 degrees C (175 degrees F) the 48 volt power supply is shut down which effectively shuts everything down. The 50 amp current shunt is that black bar with the the two silver squares right below the power supply protection board.

    One thing that I found out quickly is that the current passing transistors and the two regulator ICs needed heat sinks. 48 volts dropped to 26 volts then dropped to 15 and 12 volts respectively produces a good deal of heat with not much current draw.

    Above: All of the supply voltages are run to their respective boards. No issues so far.


    Above: Now things are stating to get busy. All of the control lines have been run between boards, the manual/auto band select switch has been added, as well as the LED light bar that shows which one of the six low pass filters has been activated. The left most LED lights up when 160 meters has been selected followed by 80 meters, 40 meters and so on. The Yellow LED is illuminated plus the selected band Green LED when the switch is turned to AUTO mode and my Icom IC-7300 tells this amplifier which band to use. Notice how certain heat sinks are getting bigger. At this point the LPF board, remote control board, and LEDs are drawing 222 mA from the +12 vdc supply. The other boards are drawing 65 mAs from the +15 vdc regulator IC.

    AF6LJ likes this.
  8. KD8UYQ

    KD8UYQ Ham Member QRZ Page

    Hello Again,

    The cool thing about the input protection board is the adjustable trip point. Not only does this board have 7 dBs of attenuation into a resistive 50 ohm load on the front end but if you should happen to forget to reduce the drive from your transmitter or transceiver the protection circuit kicks in and shuts off the RF to the PA board saving you from possibly over driving and killing your amplifier. A LED lights up and the RF drive to the power amp board remains turned off until power is cycled.

    To adjust the trip point you set the maximum power level that you want to drive into the linear. Let's say you don't want to exceed 40 watts. Connect the output of the transmitter to a watt meter with a 50 ohm dummy load termination. Set the transmitter to CW, key the transmitter, and adjust the RF output so that the watt meter is showing 40 watts out. Leave that setting alone and now connect the output of the transmitter to the input of the protection board and the output of the protection board to your watt meter setup. Key the transmitter again and adjust the multi-turn trimmer pot counterclockwise until the LED kicks on. Looking at the watt meter you will see about 7 to 8 watts of drive (remember the 7 dBs of attenuation) then nothing when the LED lights up. That's it the trip point is set. Cycle power to reset the over drive protection circuit. This circuit is designed to allow for short duration spikes in the output waveform so that you minimize false triggering of over drive protection circuit.

    Here is a link to the website with detailed information about this particular board and some videos showing how it works.

    Below is a picture of that board as implemented on my mock up amp.


    The last thing to do from a wiring standpoint was to connect the coax cables from the output of the power amplifier board to the transmatch board, from there to the low pass filter diplexer board, and from LPF board to the output connector like so.


    Now that everything was connected and all of the protection boards tested it was time to apply power to the EB104 PA board and set the bias for the the four power MOSFETs.

    This turned out to be tricky and I learned something. You MUST terminate the input and output of the PA board or it will break into oscillation when you try to set the bias. It took me a little while to figure that out.

    You can see a 50 ohm terminator on the SO-239 input connection to the input protection board. The output of the PA board was connected to the low pass filter board and that worked fine for an output terminator.

    Step one is to adjust all four of the 20 turn trimmer pots so that their center wipers were at ground potential.

    Place an accurate current meter in line with the 48 volt power supply and the + side of the wire to the PA board. I used my trusty vintage Hewlett Packard DMM

    Step two is to adjust the bias voltage for 5.5 volts. That is the Blue single turn trimmer pot next to the voltage regulator IC.

    Now for the tricky part. The goal is to adjust each multi-turn trimmer pot so that it's associate power FET draws 150 mAs. You have to work fast because as each power FET comes up to idle current the copper spread starts to warm a little bit. There is a Thermistor on the heat spreader that controls the voltage to the bias voltage regulator IC. You can see it in the upper right hand corner of the PA board in the picture below. There is a white wire running to it with the other end connected to a Red crimp connector to ground.


    That thermistor is very sensitive so as the heat spreader starts to warm up it begins to dial down the bias voltage.

    To be continued:

    WA7PRC and AF6LJ like this.
  9. AF6LJ

    AF6LJ Ham Member QRZ Page

    This has been an interesting project Ripley.
    Just a few questions.....

    1. The MRF-150 part is a time tested part and there is a great deal of support for it. I am wondering what made you choose over one of the more modern parts that offer up two matched LDMOS FETS that come from the same wafer all matched and ready to go in one package. A single BLF188XR will loaf along at 600W with 3DB of headroom. There is also an active community supporting amplifier designs. I know of one ham that is currently running one, and another who is building a second generation design that uses two BLF188XRs for a legal limit amplifier.

    2. What are you using for input protection besides excessive drive monitoring?
    One issue some have had with other solid state amplifier designs is when the amp is driven on say 40 meters while the amplifier is on 20 meters this often ends up with one or more blown transistors even though the drivel level is within safe operating limits.

    This looks like a fun project, I've been looking at building an amplifier using BLF188XR parts, and liquid cooling. Thanks for posting your project.
  10. KD8UYQ

    KD8UYQ Ham Member QRZ Page

    Hello Sue,

    The simple answer is I stumbled on the EB104 designed and followed that road before I learned about the other options. Plus the MRF-150s are much less expensive to replace should you blown them up and I fully expected to be killing a few in this build based on what I was reading. So far I have been lucky.

    As for your second question there is a high SWR protection board that works very well as I found out later in the build. I will elaborate further in my next post.

    I am headed out of town for the weekend to visit another ham you lives south of Cleveland. We are are going to work on a boat anchor AM station that he wants to put on the air then on Sunday go to the Hamfest in Mansfield, Ohio.

    Thank you for commenting on my thread. I wasn't sure if there was much interest.

    W4IJ and AF6LJ like this.

Share This Page