Even though I’m insanely busy, it seems I just can’t help myself when it comes to cheap Chinese kits available on eBay. Of course, having built most of the cheap radio-based kits already, there isn’t really that much to build anymore. But just as some people like to build model planes and trains, I find the whole electronics kit-building process to be a form of hobby that can be both relaxing and rewarding, and in the case of the Chinese ones, cheap and sometimes frustrating.
If you’re feeling a sense of deja vu, that might be because I did build two insanely cheap PIXIE QRP CW transceivers already. With those, you got half-a-watt for your (just under) AU$5, so about 0.1W/$. But what if I told you that there was an upgraded three watt version for AU$12.29, for a nice 0.24W/$? In the words of Jeremy Clarkson …. powwwwaaaaaaaahhhhhh!
Unlike the other kits built to date, this one is branded Sainsmart, and actually has a product page, although the English used on the page is rather amusing at times. The company seems to have a variety of kits and focuses on “open hardware”. Terms used in the description include an “American ring” (otherwise known as a toroidal core, local stuff isn’t “good enough”), “hearing frogs” (in reference to the audio from a competing kit known as “Frogs”), “suites” (which means kits) and “… the restaurant also has a client transmission” (which I have no idea what they mean).
Just like other kits, this one arrived in a bag, but this one was a rather large anti-static shielding bag with a resealable edge. Very nice.
Inside the package, components are grouped into individual resealable bags – from the left, we have a bag containing all capacitors; then resistors, inductors and ferrite cores; diodes, transistors, crystals and regulators; and finally plugs, sockets, enamelled wire, and integrated circuits. IC sockets are provided which makes for a “safer” construction experience, especially for beginners. While sockets and connectors are provided for power and antenna, no plug is provided for the key, so users should provide their own key switch and a 3.5mm plug.
The board itself is an attractive red colour, and is a double-sided tin-plated-through-hole board with solder mask and screen-printed top. Component values are not screen printed – it’s already quite cluttered as it is, so reference to the printed material is required. Because of the numerous components, the physical footprint of the board is noticeably bigger than the PIXIE. The good thing is that no SMD components are used in the design of the kit, making it easier for beginners to construct. The difficulties lie in the use of a hand-wound inductor (so a little work is necessary), and the use of mixed 4-band and 5-band resistors with vertical mounting. Overall, I would think this kit is suitable for beginner-intermediate level constructors.
The underside has a large ground plane with tinned strips at the edges for “connection” with a suitable metal shielded case, which is not included. The ground plane is well implemented with thermal-pads, which make soldering a lot easier and more consistent.
The unit is supplied with three pages of photocopied print – two pages are component listings and board layout with the final page being the schematic reproduced above. No information on circuit theory is provided, and although some Chinese is present, most of the text is also written in English which makes this a bit easier to construct.
At a brief look, the circuit comprises several segments. The NE602 is a double-balanced oscillator and mixer chip, which forms the heart of the device and derives its input power through a 78L08 8V regulator. It receives input from the antenna through C21, back-to-back surge protection clamping diodes D1 and D5, and crystal Y1 which I suspect forms the beat frequency in the mixer on pin 4 which drives the LM386 audio amplifier. The input to the audio amplifier appears to be switched during transmission, probably to attenuate the receive path which would have a very strong signal. A second crystal Y2 is used, which appears to be configured as an oscillator. The key controls a number of transistors which amplify the oscillation and send it to the antenna, where a number of inductors are used to keep the oscillation from leaking into the power and out-of-band oscillations from leaking into the antenna. This design is more costly and complicated compared to the PIXIE, with quite a few more transistors, inductors, an NE602 IC and twice as many crystal oscillators. Accordingly, I expect this to be a superior performer.
As with most modern kits, you’ll have to BYO solder, desoldering braid (in case you make a mistake), hobby knife (to scrape off the enamel) and side-cutters. In this case, you’ll also need a suitable 12V DC power supply.
The whole construction process took two hours from start to finish while not rushing myself to complete it. One thing I noticed was the use of an ESD shielding bag on the outside, but all the components inside (even ESD sensitive ones) were packed in regular plastic bags. This wouldn’t have helped them against potential ESD damage, although that being said, most of the chips are fairly hardy.
I started construction by putting in the sockets, then capacitors, inductors, resistors, diodes, transistors, crystals and then finally the ICs. The clearances between components can get quite tight, so dexterous hands are essential. I found the board easy to solder, however, the wide holes also meant an embarrassing amount of solder flow-through occurred at many joints. The components also appeared to be of mixed batches, with the size of identical valued ceramic capacitors occasionally varying, and the print quality of the values being mixed. The resistors were mixed 4-band and 5-band resistors of 5% or 1% tolerance. The JST connector orientation silkscreened was incorrect for the pre-made wires. On the whole, there wasn’t anything show-stopping, and the components supplied were pretty much exactly as required.
As no construction guidance is provided, I will offer a few selected hints and tips:
- L4 is the thinner core where 16 turns need to be wound, and L3 is the fatter core where 11 turns need to be wound. The enamelled copper wire provided is folded into a bundle, so take some time to carefully straighten it out without introducing any kinks. Cut the wire into exactly half to make winding a little more manageable. Wind with a little “tail” hanging out from the beginning and keep winding until the necessary amount of turns is reached, keeping the turns moderately tight on the core and spaced as evenly as possible. Cut the excess wire off, and leave the tail exposed. Use a sharp hobby knife and a flat surface to scrape away the enamel at each end, before tinning it with a thin coat of solder before inserting it into the board and soldering. It’s not hard, but it is a tedious process. In the case where vibration may occur, it may be sensible to apply glue to hold the inductors in place on the board and provide mechanical support.
- Watch the orientation of capacitors and ICs – because of the compact design of the board, adjacent parts often have the opposite orientation.
- Crystals have a “can grounding” point just next to them. Because of the proximity of other components and the distance from the pads to the crystal can, it’s not advisable just to flow solder, as it probably won’t make the connection. Instead, mount a short piece of component lead to the pad first, then bend it over the can and solder it at the top, then clip the lead. This may not be absolutely necessary (and too much heat can stress the crystals), but the grounding of the can may help prevent any stray emissions or intermodulation/interference. Since the pads were provided, I went ahead and did it anyway.
- An extra 51 ohm 1W resistor is supplied as a dummy load. It’s useful for testing the circuit, but it’s not rated for continuous duty, so don’t run with the transmitter keyed with only the resistor attached as it may get so hot as to damage itself and set things on fire.
- CP9 is marked as a polarized capacitor on the board, but has been substituted with a ceramic capacitor according to the print. Based on the components supplied, this appears to be intentional – but it does cause slight confusion during construction.
This is what the constructed board looks like – a very dense set-up indeed, and I’m rather surprised the main transistor didn’t come with a heatsink. This may be necessary if operating at high powers for a long time … depending on how much power is actually developed.
As usual, there’s some flux spatter on the bottom, but otherwise, everything’s rather neat and tidy.
Over half the enamelled copper wire remained, and I’ve soldered the resistor to one JST header lead for testing purposes. The other is just left flying for my bench supply.
As I still don’t know my morse code, and I’m not allowed to operate homebrew transceivers on the air, I’ve just decided to run it into the dummy load for a few minutes to test it. The same momentary push-button key was used from my “hack” for the PIXIE, and the Rigol DS1102E was used to assess its output.
The board operated first time with no fixes or tuning necessary. With 12V supplied to the board, it consumed (approximately) 380mA (or 4.56W) when keyed, which is less than the 450mA the site claims. The output reached 9.96V RMS over a 51 ohm load, or about 1.945W of actual transmit power, which is only about 2W, or 1W less than the advertised value. This is slightly disappointing, but is still almost four times that developed by the PIXIE kit. The positive part is that there was no residual carrier leakage to the antenna when the transmitter was unkeyed, and the receive note and side-tone were nice and smooth.
The harmonic emissions seem quite present at 14.046Mhz, being about 24dB below the carrier at 7.023Mhz. This is only slightly better than the PIXIE, but considering the higher carrier power, the actual energy emitted at 14.046Mhz is higher, about 8mW or so compared to the 5mW of the PIXIE. However, higher level harmonics aren’t visible at all, seemingly effectively filtered by the toroidal inductor.
At 5V, the circuit still operated although the distorted asymmetric waveform is much more apparent. The ratio between the wanted fundamental and the harmonic emission has reduced as well. Output is only 75mW or thereabouts, meaning this circuit tolerates low-voltage worse than the PIXIE and is less suitable for 9V battery operation.
Using my more accurate U1241B (as opposed to the inbuilt panel meters on my Manson HCS-3102 power supply), when supplied with 12V, the quiescent current measured 11.75mA (less than the claimed 23mA) and keyed current was 373.3mA (also less than the claimed 450mA).
Listening to the output of the transmitter using my Winradio HF receiver, the output tone was mostly smooth with much less “rough notes” compared to the PIXIE. It’s a cleaner output which should make for more pleasant “crowded band” operation.
The Sainsmart Forty-9er is a beefed up variation on the classic QRP CW kit. While it is more than twice the price of the PIXIE, I think it’s well worth it even if it doesn’t meet the claimed 3W power output, as it has a much more consistent performing side-tone, a much nicer receive note, cleaner transmit signal and most importantly, close to four times the output power. It has superior filtering of higher-frequency harmonics, which should keep other radio users happy as well, although the output at the 14.046Mhz frequency is still somewhat high for my liking. It’s also a pleasure to build with a quality double-sided silkscreened solder-masked tin-plated PCB and had no missing parts. It does require a little bit of skill in winding the inductors, but that in itself is no great challenge.