Evolution. Homebrew SSB transceiver

The basic SSB transmitter wasn’t too difficult. Up to the 5mW level anyway. Trying to get from there to the 1 or 2 Watts level caused considerable grief. There are lots of push / pull circuits on the internet to give guidelines for construction. I found a circuit using a pair of 2N3866’s. Since I had a handful of these in stock I decided to use this circuit, driven by an NE46134. The same transistor used in the Elecraft K3 front end. This produces plenty of drive for the push / pull stage. At least a couple of volts of RF. 

After the unexpected demise of a number of 2N3866’s I suspected that the output transformer turns ratio probably wasn’t correct. I tried optimizing the turns ratio, but only succeeded in blowing further pairs of transistors. I eventually concluded that the transistors were not up to spec. I had bought a quantity, from an eBay vendor based in China, at a bargain price a couple of years ago. A quick look around ebay revealed lots of negative feedback issues concerning fake parts. Once bitten ….. 

I modified the circuit to take some VN1210 FET,s, that I also had a quantity of. Without any changes other than to the bias circuit, this circuit worked first time. By mis-setting the bias, the output pair consumed in excess of 750mA at one point during testing. The solder joints melted. But the devices survived the experience. These transistors are in a TO39 plastic case with a small metal tab that is connected to the drain pin. Not easy to cool due to the shape. After some further fruitless heatsink experiments I shelved the whole transmit idea for the time being.

I arrived back home on 16th Sept. Maria returned from her UK trip on 19th. Over that weekend a search of my gear in my storage shed revealed an old Codan 7727. This has a solid state 100W linear PA module that only needs a few mW of drive, and it has ALC on the PCB. I have pulled this module, and intend to modify the output transformer in order to reduce the output power. 

(Vcc^2/2*Pout) gives the collector impedance. At 13.5v and 65 Watts the turns ratio will be very close to 6:1. Currently the windings are 4:1 so I will need to remove the existing turns and replace with 6 turns. According to theory this should work. I hope it will reduce the heatsink requirements. The PCB is currently mounted on a heavy duty ( 10mm thick ) angle bracket that will allow me to mount it vertically down one side, on the inside, of the transceiver case. Adding a heatsink would not be impossible, but it will have to be cut to fit.

Constructing the QRP Labs Ultimate3S with ZL4SAE. QRSS, WSPR, CW, HELL.

The WSPR Transmitter is multimode.
The Ultimate3S is a versatile WSPR, CW, QRSS beacon Transmitter that can transmit messages on up to 6 bands. The basic kit is single band. A clever mod enables up to 11 bands to be used.

Along with a few other popular mods found here. http://www.qrp-labs.com/ultimate3/u3mods.html

  There is also the option of the temperature compensated crystal reference for the SI3151a synth module, rather than the plain vanilla PCB, this can help to reduce frequency drift to 1 part per billion. The synth can produce up to 3 separate frequencies, on 3 separate pins, at the same time. One of these outputs is used to drive the transmitter. Another appears to be sampled by the processor. And the third is used as a ‘park’ frequency when the transmitter is NOT transmitting. I propose to program this frequency to be used as the VFO in my homebrew WSPR receiver. 


Partly built ( modified ) transmitter 30 Mtrs LPF.

My order included a single LPF for 40 Mtrs. I subsequently modified this for 30 Mtrs which appears to be a more popular WPSR band. The mod was simply a turn or 2 less on the toroids and changing 2 capacitors for the next lower value. A table of values is included in the LPF construction details. http://www.qrp-labs.com/images/lpfkit/instructions2a.pdf

   Completed LPF. The missing components are options for other bands.

The transmitter kit.

Some progress has been made. I replaced the cheap 28 way DIL socket with a pair of 14 way turned pin types.

Almost finished. Needs the synth module to be plugged in before testing begins.

Front and left is the single PA transistor, with places for 2 more. The option also exists to run the PA from 12v rather than the default 5v.

Front view. Protective film still attached. Thank goodness, as there was a small spit of flux, which hopefully has not melted thro the film and marked the display. Which is blue incidentally.

The OCXO SI5351A Synthesiser.

The electronics part of the kit is just a handful of parts. Constructing the enclosure for the temperature controlled oven looks a bit more daunting. It all needs to be square. Thankfully the only SMD part is already soldered to the board. Thanks go to QRP Labs ! 

The manual is 45 pages, as opposed to the transmitter kit which is only 15.

These were quite demanding on my patience !

Using a clipboard as a vice helped a bit.

Assembling the synth PCB. Not too difficult.

This is how it looks with the oven chamber attached.

The PCB will eventually fit into the box.

All components are mounted. Set the pot fully counter-clockwise.

The temp adjust pot should be below the access hole on final assembly, of course.

Powered up in DIAGNOSTICS mode, first time. The LCD contrast pot had to be adjusted in order to see the characters.

Completed the assembly, and some basic setup done for testing. Synchronising the kit clock with the PC clock was tedious. You have to do this every time you cycle the power. Fortunately it is the first item in the menu once everything is setup.

Programming other parameters is a bit long winded due to only 2 buttons and a menu system. For example, inside the MESSAGE parameter, the ENTER key is a character that you have to select with the LEFT button in order to save the message. It all works OK. Just read the manual, over and over.

The dummy load is a couple of 100 Ohm resistors in parallel, tack soldered to the output pins of the 30 Mtrs LPF. Idle current was 66mA. I adjusted the FET bias pot to increase this to 75mA. Thats 375mW standing power. At 80% class ‘E’ efficiency, I calculate the ouput power at approx 300mW. I hope thats correct, I’m no expert on class E amplifiers.

YES !! It works !

Receiving on my K2, and my Ultimate3S is being decoded. The receive antenna for the K2 was a 6″ terminal screwdriver. The drift looks good considering the temperature controlled oven is not fully installed or calibrated.


 The 3rd homebrew Direct Conversion receiver. With low pass ( for transmit but left in line for receive ), and bandpass filters.

It worked. But the performance leaves a lot to be desired. I have since built 2 other versions of this and none of them were very good. 

The 2 earlier NE602 / LM386 efforts were hopeleesly inadequate, including the ‘Improved’ version, even with the DDS VFO. Why do so many operators extoll the virtues of this ‘toy’ radio. Its as deaf as a post and as hissy as an Asp. No offence ………

Next step is a superhet. I will base the circuit on the Elecraft K2, itself based on circuits from the classic work, Solid State Design for the Radio Amateur, now updated as Experimental Methods in RF Design. I have both these books, so there is no shortage of research material, or inspiration !

Since I have a couple of 9MHz SSB crystal filters, the receiver will have a 9MHz i/f. The VFO will be the ubiquitous AD9850 DDS, controlled by the equally ubiquitous Arduino. Not exactly a K2 then.

  The receiver comes together, and works rather well. So well in fact, that it is on its way to becoming a transceiver.

  The receiver PCB, built using modular construction, based on Me-Squares and Me-Pads.

  More progress. Wired up for testing. It all worked first time. First station heard on 20mtrs was in France. Not bad DX from New Zealand.