Nick M0NTV designs, builds and tests a 10W power amplifier for his latest homebrew SSB transceiver. The MRF455 Datasheet can be downloaded here: pdf1.alldatasheet.com/datashe...
Its nice to have a scope and a line sampler to monitor trace. with a Two-tone audio generator, adjust the bias current to the best RF envelope trace on the scope without exceeding the Current rating of the transistor. a current limited power supply is nice to have. an Am trapezoid envelope test is the best test for linear amplifier commissioning. Congratulations on your Homebrew linear amplifier. All the best from DU DV3SB
Hi Noy! Thank you very much for watching and for sharing some good advice. I haven't done the 2-tone test yet ... I was too scared of what I'd find!!! Many congratulations on your own homebrewing projects. I was looking at your QRZ page with interest. 73, Nick M0NTV
Interesting as always and good see a video where the result is not perfect first time and is more real world. You have generous friends, I would need to find a cheaper transistor. Thanks for taking the time to make this video.
Thanks Ace. To be honest I probably wouldn't have used that transistor unless I'd been given it. Personally I think if you are building for QRP(ish) levels then the Mitsubishi RD16HHF1 MOSFET takes some beating. I've had some great results with these and they don't cost the earth. Thanks again. 73, Nick
@@M0NTVHomebrewing Thanks Nick, the Mitsubishi RD16HHF1 MOSFET (also used in the Xiegu X1M and X108 units) at around £7 is easier on the pocket than £56 for the Motorola. Thanks for taking the time to reply.
Nick, try a broadband input matching transformer instead of LC inputs. They are too narrow-band even for this application. One little ferrite binocular will do away with the relay and most of the rest of those other components. You also need a stiffer bias supply. BJT's aren't as simple as FET's when it comes to biasing in class A/B, (or close to it.)
Hi John, I've tried several broadband transformers on the input but none of them were broad enough! The problem is the transistor presents quite a reactive load that the broadband transformers will do nothing about. To be honest in the past I've just stuck in a broadband transformer and not worried about it. But it would have meant a big reduction in signal of 10m which I wanted to avoid. I was probably making the perfect the enemy of the good! You are right about the bias network. I've spent so long using FETs in my PAs that I forget that BJTs need to be driven by a current source. Every day is a school day! Thanks for watching and commenting. 73, Nick
I remember my teenage self cloning a CB "afterburner" with this exact transistor. No clue about the schematic just 1:1 Copying the layout, coils, component values. 😆
If you monitor the 1000uF on the base bias network, with an oscilloscope, you will probably see the voltage collapsing under drive conditions (or even going negative). As others have mentioned a low impedance source is optimum, as the transistor sinks considerable DC current under RF drive. The 2 diodes and an emitter follower solution is probably do-able. Couple the lower diode to the vicinity of the transistor. A 2 tone test would help in seeing any issues. Keep up the great work.
To me, it seemed like you were slightly mistuned on the SDR for 10m; even seemed not centred in the SDR passband. So, I think its ok on 10m. Anyway, thanks for all your hard work! 73...
Yeah, where to start? You're not going to get anywhere expecting narrowband matching circuits to work across broad frequency ranges. The data sheet test circuit is 30MHz narrowband input and output matching. 21 MHz was never going to happen. You need to look into broadband matching. You're on the right track with the output binocular ferrite core, but look at the data sheet, the input and output impedances are like 2 or 3 ohms(plus reactive components) So that gives you the ball park of where you need to be with transformer ratios, 20-25 to 1. You are trying to bias the transistor as if its a FET or MOSFET, voltage bias, you need to adopt a current source type base bias. Essentially you just need a series variable resistor and not a potentiometer. Often there would be a power diode to ground and the forward conduction voltage fed to bias the transistor, the diode would be mounted on the output device to give thermally tracking bias. 73 de G0AFV
Thanks for your honest feedback and helpful suggestions. I suspected the bias network would be an issue! Anyway, I appreciate you taking the time to watch and share some good advice. 73, Nick
I have about a dozen of those MRF455 devices. Not a bad transistor. Notice how the datasheet says 30mhz characteristics... thats not suggesting that is its maximum frequency, more like a typical frequency of use and what to expect. Motorola themselves used them in their maxar 80 low band vhf transceivers at around 70mhz to give 50W output I think. They easily are capable of 70W at that frequency in a tuned class C configuration and give around 10dB gain there. There's a number of suggestions I could give you to make your power amplifier here much better... (if you're interested) a few watts should be more than enough to get that transistor hooting at 50W at 30mhz. My suggestions are surrounding the bias and input/output matching
Hi Dazz, thanks very much. Whilst I'm probably not going to mess with the one I've built that is not to say I wouldn't build another. For a start I've got another MRF455 lying around doing nothing. By all means send me an email with some of your biasing and matching suggestions. My email is on QRZ.com and I'm always keen to learn more. Thanks very much. 73, Nick
@@M0NTVHomebrewing Ok mate, I'll have to bumble around to find you since I'm only using a mobile device, but I'll try to. The suggestions are fairly simple though, I almost could describe them with words. One is the bias... what you're using is a very high impedance bias effectively which is bad because it shifts under drive causing intermodulation distortion. Also you can match the input and output pretty easily by using binocular ferrites. Same as your using on the output for the output, go 2 turns to say 8 turns then play with a matching network if required and add some capacitance from collector to the other end of the transformer primary. Yes you're feeding the collector dc through the winding. Try feeding the base in a similar fashion, applying the bias dc through the winding. A simple bias approach is to have a diode going to ground that is similar to the BE junction of the MRF455 forward voltage and apply a dc bias current to that, increasing it until you get the quiescent value you want. 90mA should be enough
Been doing some similar experiments for 17m with CB Power Amps, and MRF433s; same form factor but only rated for 12.5W up to 30MHz. Some terrible feedback on some set ups, one really good one with a very chunky 5W (!) feedback resistor, from SPRAT 70. More work required, but 10W is looking good.
Hi Steve, glad to hear I'm in good company! Keep up the great work. I'm hoping to be with you at the GQRP Club Convention this year. It'll be great to meet up with a few folk. 73, Nick
Time to redesign your bias system. It would be good to give it a little of temperature compensation, even on 4x 1N4148 diodes. It should be sufficient and it will be more safe for MRF :)
Carefully constructed transmission line baluns are your friend on the input (and output of course). You can get 50 ohm coax that is less than a mm overall diameter, this is the stuff that I use to get the required wide band match to the low input impedance. Any resonant circuits in the input path will basically make it a two band amp. Also, and I am sure you are aware of this, a large transistor like that will achieve much greater linearity if biased using a constant current method (a bias drive transistor) but the method you employ here is much safer for the beginner and I guess that`s why you are showing it. You should be able to get a reasonable return loss if you are only interested in 14 to 30 mhz using the good old brass tubes in ferrite sleeves method uncompensated. One turn to the transistor (the brass tube) and maybe seven or more turns of teflon insulated wire. The same method used when driving a push pull amp, but a few more turns on the primary. In my experience that transistor will do at least 50w until you run into some gain compression at around 55 to 60w.
Thanks very much for watching and for your helpful thoughts and advice. The transmission line balun sounds very interesting. To be honest when it comes to RF power amps I'm very aware that I still have a lot to learn! But building helps me do that - as does the feedback and helpful pointers of others. So thank you once again. 73, Nick
Nice! An interesting transistor to look into for future designs is the MRF101, which is a 100 Watt LDMOS for up to 250 MHz. Just something I thought of when I was the price of the MRF455.. While the MRF101 is a device designed for 50V, you can get quite a bit of power at 12V as well. Anyway, great video! I like the "raw" style of your videos. Unfortunately something YT (or viewers?) don't seem to favor anymore.
I agree. I have been looking through new transistors and even tested some and I keep finding that the input impedance is sporadic when you get to the HF region and keep you from having a wide bandwidth. I was just looking for a few watts to 20 watts. I didn't want any higher. but the MRF101AN and the BN and also the MRF300AN and BN look like my next step. if memory serves me correctly I stumbled on to it because it is used in my Yaesu FT-891 and looked it up. the 891 does 5 to 100 watts on SSB and 40 watts on AM. it is $27.09 us on Mouser for the 101's. I think I will just jump to those there was a contest not too long ago and some information on what others have done with them. the only other step I can think of trying is to just use some regular mosfets I'm sure the technology nowadays may let us use some of those for HF but it's a lot of work guessing and testing. Have a good day guys!
@@curtstacy779 What do you mean by "the input impedance" is sporadic? Up to 30 MHZ, things are quite easy. Well, relatively easy. My observation is that many homebrewers can't get out of their mental block of seeing input/output impedance as just the magnitude and then treating it as if it were a real resistance. When the real and imaginary parts are viewed in the correct way, even broadband matching is no longer magic and common schemes for broadband matching can be used.
@@BalticLab Yes that could be part of it. but I was trying new transistors, they are designed for wideband RF some up to 1 ghz and some higher but all rated for radio frequencies. I could make an amplifier for 200, 400 to 600 and 700 to 1200 mhz just fine. but the input impedance is not very constant under say 50 mhz and I could only use a narrow matching system very simular to what Nick is running into on this build.
As someone else as mentioned the MRF101 is a cheap(ish) interesting device 50 odd volts opperating probably a bit awkward but you can get cheap PC supplies for that and you can always run it at a somewhat lower voltage
Hi! Yes, in my last transceiver (the Shelf 17) I used a device that gave me 80+W output and I fed that 50V with a heavy duty boost converter. It seemed to work pretty well. I was just going for something more modest this time. Thanks very much for watching and commenting. 73, Nick
Thanks very much Greg. Yes, I began with a wideband matching transformer and tried several different variations of this. The trouble was I still couldn't get it wideband enough with a low enough SWR (on both bands) for my liking. It is something I will return to in the future for sure. Needs further work! Thanks again. 73, Nick
@@M0NTVHomebrewing Interesting that a transformer worked on the output but not the input. I guess the RF amp's output impedances must be closer to 50ohms? I'll stay tuned for your to return wide band matching Nick!
Hi Nick. I decided to take a look at the MRF455 info on the web, and it quickly became apparent (to me at least,) that the device was designed specifically for American CB external amplifiers, and only them. They have max input specified at about 4 watts, coincidentally the FCC's AM CB output spec. Of course, they are also spec'd only for car battery voltages, and there is almost no info regarding them at anything other than 30 MHz, just above the US Citizen Band. I don't recall any other RF power transistors that are specified at only a single frequency, and show a test circuit at only that frequency, and at 1 voltage, and only 1 voltage. Other RF transistors will show Smith Chart impedances for a range of frequencies, as well as graphs of output vs frequency. These do not. Very suspicious. These also have an unusually good gain spec for a BJT at those conditions, and at those conditions a pair of them in push-pull will put out just a bit over 100 watts, a nice round number for amplifier advertisements. What that all boils down to is that they are a single band transistor designed and optimized for 11 meters, (but conveniently for hams, also on 10 meters,) with their input impedances apparently varying widely at other freqs, probably a tradeoff for the gain at 12 volts. That would explain the difficulty of getting an input match that would work at both 21 and 28 MHZ. The transistor was never intended to cover multiple bands, just 27 Mhz. I note too, that nowhere in any of the manufacturer's specs is the 27 MHz band ever mentioned, nor is a bias circuit shown in the test circuit schematic, even though the device is clearly perfect for a CB amp. They wanted to be able to deny everything to the FCC when the transistors started showing up by the thousands in illegal CB amplifiers, which they did. (I have 3 different push-pull amps with blown MRF455's that I hope to repair and tune up on 10 and 20 meters.) But what other market could the devices have been designed for? They certainly didn't design a 27/30 MHz-only transistor for ham 10 meter transceivers. Nope, it's a CB specific transistor developed in the hay-day of the US CB craze. I'd say you did well to even get a workable match with a single transistor configuration at 21 MHz. Your input circuit design is not an indication of poor engineering, but a testament to your excellent impedance matching skills in getting the device to work in a single device configuration on a band the designers never intended it to work on.
Hi John, thank you so much for taking the time and trouble to look into this - and to let me know. That's very good of you. Yes, it certainly wouldn't be a device of choice if I hadn't been given it for free! That's all very interesting to know. I did a bit of digging before I built the amp but I was surprised that I couldn't find more info on using the transistor in ham radio circuits. I'm beginning to see why! To be honest I nearly shelved the project a couple of times and built a different amp with a different transistor but it kept bugging me that I couldn't get it to work in the way I wanted. Time (and signal reports!) will tell whether it stays in there or not! Thanks again. 73, Nick
Heathkit used the MRF455 in radios like the SS-9000 so it's definitely capable of performing in multiband SSB operation. Motorola AN762 gives an example of the MRF455 in a broadband 1.6-30MHz linear amplifier for amateur radio and marine applications, it shows how to match a pair of transistors and gives a lot more useful design information with regard to matching, biasing etc. It was definitely a popular transistor with the CB community but I suspect that was, as with so much of the badly designed to a price CB stuff, a cost decision on behalf of the CB 'linyer amplimifier' manufacturers rather than an actual design decision by Motorola.
Hello Nick, I sent you a message using a form on your Nick the Vic page. I was looking for an email address so I can send you some diagrams. Hope this was your page, otherwise I have bored some poor man to death with talk about RF matching. Chris.
