2009 ~ The 40M CW Transceiver.
It all started with receiving my GQRP SPRAT late in September 2009. In that wonderful publication was a project to build a simple regenerative receiver. Ahh Haa a chance to relive the thrill of the good old days.
I built the unit and even included an RF stage to isolate the antenna from the regen part. Upon application of power I heard the wonderful swishing noise and after a bit of tinkering got the unit to tune the 40M band. Then it all came back to me about the simplicity and the shortcomings of a regen receiver. Once you have tasted modern receivers -- it is hard to go back. I really did want something better in about the same size of package.
Then I remembered I had purchased some MMIC amplifiers that I spotted in a MOUSER catalog. Hmm this might be a good chance to try these out. A half hour spent with some pencil and paper convinced me I could build a better mousetrap.
About a year ago I had purchased a large quantity of crystals for some SSB transceivers and these included frequencies such as 4.9152 MHz, 5.185 MHz, 6.144 MHz, 7.387 MHz, 8.0 MHz, 9.0 MHz, 10.0 MHz and 12.0 MHz. I zeroed in on 5.185MHz as I also bought batches of crystals for VXO's. In one batch I ordered 12.228 MHz and 12.352 MHz. If you do the math the 12.228 MHz results in the subtractive sum of around 7.040 MHz and the 12.352 MHz results in 7.170 MHz. Now we were cooking. I had previously built a crystal switched VXO and so now I could have a receiver that tuned both CW and SSB portions of the 40 Meter band. To my dismay the 12.228 MHz Frequency in the Mouser catalog was a misprint as the actual crystals were 12.288. Oh well I built it any way just to prove the concept.
Next I looked at using some SA602's in the project, one for the VXO and the other for the Product Detector and Carrier Oscillator. For the audio I used an LM386-3. Now to get really better performance something other than the SA602 should be used as a front end mixer. The next step up is a Double Balanced Mixer such as the SBL-1; but that comes with a price of requiring more drive to the DBM. That said I was on a roll.
The MMIC's which are manufactured by Watkins Johnson now a part of TriQuint, (P/N AG303-86G) are good for 20 DB Gain from DC to 6 GHZ. Sounds like they might work fine at 7.0 MHz. Only 6 parts are needed for a complete gain block amplifier stage. ( A bias resistor either 22.1 ohms or 51 ohms depending on whether the source voltage is 5 or 6 volts, a bypass cap of 0.047 Mfd,, two coupling caps 0.02 Mfd (in/out) and an RF Choke and of course the MMIC.) I didn't have any 1 uhy chokes so I used 2 turns of #24 on a FT-23-43 ferrite core which is about 1 uhy. (I have since purchased chokes and swapped them out as the ferrite cores are more expensive than the packaged chokes.)
I decided to use one of the AG303-86G MMIC's as an un tuned RF Amplifier ahead of the 7.0 MHz Band Pass Filter, a second would be used following the the SBL-1 as a post mixer amplifier and the third would be used as an IF amplifier. Keep in mind 18 parts gives you these three stages. The in/out impedance of the MMIC's is 50 ohms so there is no matching required to the band pass filter or the SBL-1. To match the in and out of my four pole homebrew crystal filter which is on the order or 200 ohms I used two more FT-23-43 Ferrite cores to give a 1:4 and then a 4:1 transformation. Five turns of #26 to ten turns of #26 gives the proper ratio. [5^2 =25 and 10^2 = 100 25:100 = 1:4]
There was no matching to the Product Detector so it is a simple capacitive coupling from the MMIC output to Pin 1 on the SA602. That perhaps could be looked at as a further refinement. The SA602 as a Product Detector and Carrier Oscillator needs but a few parts: five capacitors,one resistor, a crystal, a trimmer (0-65PF) and a FT 37-61 Ferrite core with 15 turns of #26. The latter two are used to set the crystal on the pass band of the filter. AF output is taken from Pin #4 on the SA602. Oh should mention that I operated the MMIC's and the SA602's at 6 Volts. I included an LM317L regulator on the board to precisely set the voltage at 6.0 volts under load.
Next I focused on the VXO. To move the frequency about I used a varactor, an MV209, operated with a 10K linear pot from a regulated voltage of 8 volts. My initial attempt used four 12.288 Crystals in parallel ala Super VXO which gave about a 40 Khz swing --too bad the frequency was not 12.228 as that would have been perfect .I only had two of the 12.352 Mhz crystals and there it was a disappointing 8 kHz swing. I think the fact that the crystals were cylindrical may have been part of the problem. The HC-49U seem to work best in VXO's. The output from the VXO was taken from pin 5 and showed about 0.49 Volts PP.
This is not enough to drive the SBL-1. The SBL-1 is a 7 dBM device which if you do the math says you need 1.414 Volts PP. [10Log(1000[(1.414)^2/(8X50])) = 7dBM] Thus there needs to be a boost in signal to the SBL-1. I used a 2N2222A amp running hot so you need a heat sink to boost the signal to 1.0 Volt --not quite 7dBM but certainly works! One of the problems with a 12 MHz injection frequency is a sideband inversion since you are using the subtractive mix of 12 - 7 = 5 to feed the IF. The ladder filter tends to favor LSB so with the subtractive inversion you get USB. Given the wrong tuning range and the inversion problem -- a need for a better mouse trap.
In looking at the SA602 with a few component changes, it would be possible to convert the VXO to a Varactor Tuned Oscillator (VTO). A little computational time indicated that with a T-68-2 core (61 Turns #26 ~ fills the core) and larger value coupling capacitors (both 680 PF NPO) I could have the VTO tune from 1.8 to 1.9 MHz. In parallel with the coil is a 150 PF NPO capacitor along with a 12PF Ceramic Trimmer to provide fine tuning of the band edges. In addition I also used back to back MV209's which actually halves the tuning range but allows connecting the anode of the second MV209 directly to the tank coil. Connection with the tuning tank network was made through a 180PF NPO capacitor. This resulted in about a 60 Khz spread with a full rotation of the pot.
A regulated 8 volt supply is used for the tuning voltage going from 0 to 8 volts. The frequency spread can be reduced further by connecting fixed resistors to either end of the linear 10K pot thus setting a smaller voltage range of adjustment. The 12 PF trimmer enables futher setting of the endpoints. I am amazed at the frequency stability of this arrangement which I am sure is in part due to the low frequency of the VTO. Other possibilities include using two pots for coarse and fine adjustment. I have not plotted pot rotation versus frequency change but it probably is not linear!
I also added a tank circuit to the SA602 so it is a balanced output across pins 4 and 5 of the device. This was done so that a second output could be take from the VTO. The tuned network for the balanced output uses 35 turns of #26 on a FT-37-61 core. This entirely fills the core and the tune capacitor is 470 PF. This resonates around 1.85 Mhz. The secondary is 8 turns. The third winding is 3 turns which is used to feed the transmitter mixer.
Well the performance of the receiver is certainly a huge step up from the regen that I started with several weeks ago. I heard many DX signals and no touchy circuits! There is no AGC but for CW that seems OK. I am sure there are some possibilities for adding an audio derived AGC that would drive an attenuator IC somewhere in the loop. (Another opportunity). So here I was with the relatively small complete CW/SSB receiver all on a 3X4 inch copper board. There was no attempt to make it smaller; but now that I have proven that it works -- there are possibilities to shrink it down to perhaps 2.5"X3".
Right now I am using a Radio Shack 10K linear pot for the tuning. A high grade 10 Turn pot with a mechanical display would really make for very accurate tuning. There are some circuits for linearizing the tuning even more so than with the 10 turn pot. The varactors offer an opportunity for PLL tuning and frequency stabilization using huff and puff techniques especially if a different range is used. [This is the very first time I used varactor tuning and I am impressed.]
Now what? How about a companion transmitter? Again use of the MMIC and the SA602 had to find its way into this part of the project and they did. A third SA602 is used as mixer stage where the 1.850 MHz VTO signal is mixed with a 5.185 MHz crystal oscillator frequency (shifted 700 HZ in the proper direction) to produce a 7.0 MHz output. A balanced output across pins 4 and 5 of the SA602 is tuned to 7.0 MHz. Here a T-44-6 core filled with 35 turns of #26 for the primary and 3 turns for the secondary is used. The resonating capacitor for the primary is 100 PF NPO. This balanced output circuit along with the 7.0 MHz Band Pass Filter in a subsequent circuit is to assure that the output is the SUM frequency.
The MMIC would be used to amplify that signal before it passes through a tuned 7.0 MHz Band pass Filter. The 7.0 MHz Band Pass Filter was built using packaged 10.7 MHz shielded IF Transformers (Green Core available form Mouser P/N 42IF123). These are padded with 68 PF NPO capacitors and the coupling is a 2.2 PF Ceramic Capacitor. Tuning to frequency is done by adjusting the cores.
The real trick is how to sequence everything so that the antenna is disconnected from the receiver, the fixed frequency oscillator is turned on and that subsequent stages are keyed so that there are no key clicks. W7ZOI in his Solid State Design Manual (now a vintage publication) had a circuit that does all this with a NE555 timer and a PNP Keying transistor. That is what I used for this application with two changes one of which was to increase the timing capacitor from 3.3 Mfd. to 4.7 Mfd and for the keying transistor I used a TIP42. Works Perfect!
For the Pre-Driver, Driver and RF Amp stage, I used a 2N2219A, 2N5109 and IRF510 in that order. The AG303-86G, 2N2219A and the 2N5109 are keyed via the TIP42. The IRF510 is at all times connected to 12 VDC. The output with 12 Volts is 5 watts and with 13.8 Volts close to 8 watts. In looking at the output signal on my Tek 100MHz scope --it is clean and no fuzzies. The signal reports from Iowa and Boston were both RST 459 --so nothing to get too excited about --but both were around 0100 UTC. Nonetheless solid communications were made and on 40M. The signal report from Michigan was stronger.
So far I have had six contacts with the radio since it first went on the air on October 27, 2009. Imagine my surprise when my first contact was with AD1R, Dave in Halifax, MA near Boston. That was followed by a local contact in Spokane, WA. (I say local because Spokane is about 300 miles from my QTH 50 miles NW of Seattle.) My third contact was with Peter, KG4PRK in Cedar Rapids, Iowa. That was in two days of operation. My fourth contact on 10/30/09 @ 2200 UTC was with Paul, NS8V in Menominee, MI where he gave me a 559. The fifth contact at 0240 UTC on October 31st was John, N0GBR in Omaha, NE. My report was 559 and John was 589. The 6th contact was with Phil, WA6SEU in Shingletown, CA on 11/01/09 at 2315 UTC. My report was 459. So far 6 states in six days.
Let me say that I am most enthused about the use of the TriQuint AG303-86G MMIC as they are easy to implement with only six components and they are very cost effective in single lot units. Their performance is outstanding and even my feeble eyes I can still see them! Stay tuned I am already working on a design for a shirt pocket sized SSB transceiver using the AG303-86G's. Thank You TriQuint Semiconductor for an outstanding product!
Overall the total elapsed time from start to finish was about two weeks. In large measure much time was saved by using the MMIC's. The boards were made using my mini-mill which I purchased from Harbor Freight about 6 years ago. The process I used for the boards is to lay out a grid on the boards using millimeter squares. Thus is done with a small ruler and a good square plus all markings are made with my trusty Pentel 0.5 mm HB mechanical pencil. The circuit can be easily sketched out using the grid squares and with a 1/32 Inch end mill I cut only what is required plus a few squares here and there for the "I forgot" components. I have found that I can lay out the board and have it milled all in about a two hour process AND no messy hazardous chemicals!
As Post Script, this project was undertaken using what I had in the parts bins. [I am one of these types that buys 100 resistor for a penny a piece, 200 capacitors for 2 cents a piece etc. So for $20 you end up with lots of parts!] In looking down the road I can see that Mini-Circuits Lab has Double Balanced Mixers that require only 3dBM for the LO and there are MMIC's from TriQuint that have somewhat better specifications as to IP3 intercept and cost only slightly more. Another note is that most of the higher performance parts are SMD/SMT and that means adapting my processes to work with much smaller parts. But that does have an added bonus of reducing the size and weight of the finished product. Although it should be noted that some of the really high performance MMIC's draw 0.1 Amp. So there also has to be some trade-offs between current draw, size and performance. Imagine a really small high performance radio receiver with 5 MMIC's and that means a current draw of 1/2 amp just on receive and that is only for the MMIC's!
Port Townsend, WA