This past weekend, I started on the process of laying out the 50W QRP Amplifier project as a PCB. Small PCBs can be remarkably inexpensive these days – $10-$15 for 5 pieces of say 4″x4″, shipped in 2-3 weeks. I’m treating this amplifier project as a chance to experiment with different, similar FETs to learn about critical power MOSFET properties, and also as an opportunity to brush up my layout skills that I haven’t used in awhile.
As the first step of PCB design, I captured the schematic of the amplifier as built in AutoDesk Eagle. I did this on a livestream on YouTube, the first time I’ve tried such a thing. It was great fun! Kenneth W6KWF stopped by to lend advice – he deals with prototype PCBs as part of his day job, though he has team members to do most of the actual layouts when needed. We’ve had a great deal of fun over the years, including building a cloud chamber for seeing charged ions in high school.
This post is cross-posted to my more general-purpose nerdery blog, jeff.glass/blog.
As I alluded to last week, I’ve been working on a simple “QRP Amplifier” to kick my power up from 5W to something a little more punchy. Specifically, an amp I can still use when portable. There’s something wonderful about achieving a contact with only 5W, but there’s also the frustration of getting into the field and having band conditions just wreck your day. It’d be nice to have the power to crank up the juice for special occasions.
While I have awhile to go before this project is wrapped up with a bow and ready for field use, here’s a brief video about my first successful test. 5W in, 50-60W out when run off two 13.8V sources in series:
More technical details to come, but for now, I consider this a really successful validation of the idea! Like I say, a few more critical steps to come, including an input 50-ohm pad, a low pass filter, and a case, but this is enough of a proof of concept to move forward.
Today I tackled the audio amp section of the Beach 40. And when I say tackled, I think perhaps the player was already down, and I just had to fall on top of him.
The audio amp configuration I’m currently using is the bog-standard, straight-from-the-datasheet LM386 20x or 200x configuration. A SPST switch puts the 10uF cap between pins 1 and 8 of the IC, or takes it out of the circuit.Currently, I don’t even have a volume control pot in place. Peter VK3YE later modified the LM386 configuration for greater gain and less hiss, which is something I’d like to tackle once I’ve gotten the rig working.
Part of what made today so simple is that I’d already built this audio amp section for another project. I’d been using a bit of scrap copper-clad to experiment with coupling between an NE602 and an LM386, to see if passive filtering would be a useful option for future receivers without introducing too much loss. But since I haven’t touched that project in a couple months and this one’s exciting me now, that board hit the chopping block.
Extracting just the audio portion of the board was simple enough – now that I’m looking at the pictures and seeing the two big bypass capacitors on the red power line reminds me that I need to think about bypassing the power input at the source, probably with something beefy. I’ve had enough LM386’s turning into oscillators, and I’d rather not subject my ears to that too much if I can avoid it.
With the VFO, mixer, and audio amp “completed,” I now have the bare bones of the receiver portion complete. Of course, there’s exactly zero filtering, either at the front or the audio portion, so it’s not like this would make a particularly reliable bit of kit. But it’s got enough parts that I can string them together and test.
While I didn’t exactly succeed in keeping the leads short, using clip leads was just the fastest and simplest way to tie the modules together. RF from the ‘antenna’ (read: long bit of wire strung around the office) is brought into the mixer via the white clip lead at center. The VFO drives the mixer and converts the RF down to audio, where it’s amplified and sent out to the headphones.
To generate a test signal, I used my SI5351-board without an amplifier, with just a short bit of wire hooked directly to the output of the Si5351. I set the output somewhere around 7.150 MHz. Tuning the VFO across the band with the polyvaricon, I found my strong local source no problem. The receiver works!
I can see why Peter later added a fine tuning control to this project – with half a turn of the variable cap, I’m covering around 200Khz of the band, which makes tuning in to an individual station tricky. I do have a rather large panel knob in mind for the cap, which will help, but we’ll see if it’s enough. If I can’t get the VFO drift a little more under control, that fine tuning cap might be necessary to help stay on frequency.
This week I finally got around to improving the transmitter/amplifier I started work on back in March. With a shipment from KitsAndParts, I replaced the J110s that had been part of this amplifier and replaced with BS170s. The differences between the parts are striking – the J110 is a general-purpose JFET (which is by definition a depletion-mode FET) with a rated power disipation of a few hundred milliwatts. The BS170 is a fast-switching enhancement-mode MOSFET with a rated disipation of 800 mW.
Here’s how the schematic looks now. Pretty similar to the last time, with a couple key changes:
I stripped the heat sinks that had been on the J110s off and threw them on the BS170s. These puppies still do get hot, and even with the ability to dissipate almost a Watt, I think they’d not be pleased too much heat. The heat sinks, plus some thermal goop from Microcenter, are easing my mind a bit.
As built, the amplifier had an output impedance of about 10 ohms at 7 MHz, as measured by the method suggested by W2AEW in this Youtube Video. To bring that up to the standard 50 Ohms, I built a little L-match, using the values suggested by this L-Match calculator, about .43 uH in series with the amp and a little over 1 nH shunted to ground. Because I don’t know the voltage ratings on most of my miscellaneous caps, I put for 4.7nF caps in series to form the fun capacitance. The inductance is 12 turns on a T37-6 (I did some experimenting to find the ideal number of turns). Re-measuring the output impedance (again with the W2AEW method) showed an output impedance of around 47 ohms, which is close enough for my purposes, I figure. I
Now, being driven by the Si5351, the amplifier puts out about 1W on a 5V supply, and about 5W on a 12V supply. Not a particularly clean signal, mind – the 5W is as measured in the power meter of an MFJ-949 – but it is an actual 5W!
It was only a matter of time until something blew up. And last night, it was TWO things.
The first one was my own darn fault – I’ve been playing around with some simple transistor amplifier circuits, and mis-read one of the transistor data sheets. When I hooked it up a 12V power source… BANG!
You can see the TO-220 package there, literally split in half by the power of electricity. Zam zam! Turned out I had grounded the emiter and applied 12V to the base of the transistor. It blew apart in my face in a shower of sparks.
So, I desoldered that transistor and replaced it with a little J110 FET I had a pile of from my last trip to California. While messing around with the supply voltage, I kept switching a clip lead back and forth between 5V coming out of the 5V rail on an arduino an 12V directly from and SLA battery. Unfortunately, at one point, I disconnected the clip from the circuit and ended up connecting the Arduino directly to 12V…
You can’t make radios without breaking a few toys. Thankfully, the Si5351 breakout and the nice LCD screen I had hooked up were unscathed. Another sacrifice to the radio gods.
Update 2/29: I’ve toasted another one! I’ve been using my old Duemilanove to work on a transmitter project, and apparently relying on its little 5V regulator to power an LCD screen, an Si5351 breakout, and provide bias current to the finals was just too much for the little thing….