June 28, 2016

Elmer 101
and the Small Wonder Labs SW30+ Transceiver

Installment #3 ...
Keying & Transmit Mixer

A step-by-step analysis and build-up of the classic 2-watt 30-meter superhet transceiver

 from designer Dave Benson K1SWL of Small Wonder Labs.


Hope all you new SW30+ Kit builders are making it through the assembly phases.  In this episode, we'll analyze the Keyer and Transmit Mixer circuits, and discussing its components, testing and operation.  At the end of this "step 4" in the assembly manual, you'll be seeing the first evidence of the 10.1 MHz signal that ultimately gets transmitted!

BTW, did you say you missed the last episode? You can download the podcast and catch up on things at any time!  The whiteboard material and audio recordings for all shows are listed right there on our home page! (www.cwtd.org)

73, George N2APB and Joe N2CX


Chat window from the night of the show ...

<20:04:55> "wheelw": when gathering parts for pcb build, do we need to match crystals
<20:06:35> "Mike WA8BXN": for best results matching helps, but random crystals on the same frequency will work too
<20:07:16> "wheelw": Thanks Mike- that was my thought, but wanted to make sure
<20:09:22> "Bruce - N1RX": if you have the ability to match the crystals, choose the ones closest together for the IF filter, and the "outlier for the mixer. If you can't match them, don't sweat it.
<20:11:10> "Joe N2CX": The crystals provided with the kits weree matched to within 40 Hz to get the best tx/rx offset and best if filter performance.
<20:17:25> "wheelw": I have an extra one of these little Arduino DDS kits (featured in QST magazine project a couple of months ago). Might be perfect for SW+ upgrade. http://www.farrukhzia.com/k2zia/
<20:18:37> "George N2APB": That's a real nice combo! Arduino "Nano" + inexpensive offshore DDS board (~$5 or so).
<20:20:10> "wheelw": Do you recommend hot glue or something to "Fix" toroids, once set? I haven't done this, but have heard others have.
<20:20:41> "Mike WA8BXN": DO NOT attempt to strip the toroid after putting it in the board, strip first!
<20:22:35> "Mike WA8BXN": the only coil I might "glue" is the VFO coil, but I probably would not do that
<20:25:25> "wheelw": could you provide URL for the site you mentioned with toriod tips?
<20:25:28> "Bruce - N1RX": For a VFO coil, I have often dripped candle way on to secure the windings to prevent freq shifts.
<20:25:42> "Bruce - N1RX": theats wax, not way...
<20:26:42> "Joe N2CX": http://kitsandparts.com/wtoroids.html
<20:31:23> "Clint - KA7OEI": On a kit like the KD1JV ATS I RTV'd the windings/cores in place as the plug-on LPF modules are not protected by a chassis. Hot glue will eventually break loose with temperature cycling.
<20:32:24> "wheelw": RTV'd is??
<20:33:06> "Bruce - N1RX": room temperature vulcanizing rubber, like pbathroom caulking
<20:33:32> "Bruce - N1RX": look for non-acidic RTV
<20:33:41> "Mike WA8BXN": some types of RTV are VERY bad to use
<20:33:59> "Bruce - N1RX": exactly most give off acetic acid while curing
<20:34:23> "Joe N2CX": Use non-contaminating RTV. The common type of RTV gives off acetic acid - smells like vineagar and can cause component corrosion. RTV sold for gluing engine gaskets is usually non-contaminating.
<20:39:00> "wheelw": would this be considered "full Break-in" QSK? Why?
<20:39:50> "Mike WA8BXN": pretty much not much delay in switching, TR switch is electronic
<20:41:22> "Mike WA8BXN": C110 puts in a little delay
<20:42:24> "Joe N2CX": C110 also adds a little keying envelope shaping.
<20:47:10> "AC2GL_Dave": In the VFO Transistor Q2 collector connects to Vr. Is that the 8V supply or am I missing somethine?
<20:48:27> "George N2APB": Not quite. Vr is derived from above U3 (center-upper part of the schematic.
<20:49:17> "Obe - KC4VZT": it's more like 7.3 to 7.4 because of the diode drop from 8V
<20:50:32> "Bruce - N1RX": the key is that is is regulated better than the "raw voltage" that the final power stages may draw down on TX
<20:53:09> "AC2GL_Dave": won't that track 8V minus a diode drop anyway?
<21:01:48> "Bruce - N1RX": AC2GL- Yes, but the goal was to isolate it from the main supply, and that is accomplished
<21:04:20> "Bruce - N1RX": we learn more from failure than success.
<21:04:56> "Joe N2CX": BTW Mike said "602". The original design used NE602 devices, now the SA612 is what is available. For our purposes they work identically.
<21:07:36> "Obe - KC4VZT": My kit came with 602's in it.
<21:09:31> "Joe N2CX": No problem with either device.
<21:20:45> "Joe N2CX": One difference between hollow state and small signal semiconductor devices is that with small signal semiconductors you shouldn't blister your fingers!
<21:20:52> "wheelw": I missed call, but is there a way to verify VFO operation with a DVM only (I would need to check out an oscillosope at work)?
<21:21:20> "Pete wb2qll": you can verify the oscillator by listening wth a receiver tuned to the VFO frequency.
<21:22:02> "wheelw": just push SW+ close to the antenna lead of the receiver?
<21:22:02> "Obe - KC4VZT": I'm up to step 9 and taking my time, having fun and learning lots.
<21:22:29> "Armand WA1UQO": Complete through step 6 with no problems!
<21:23:10> "Pete wb2qll": Take the antenna lead from the receiver and put it very close to the VFO circuit.
<21:24:24> "George N2APB": Wayne: Yep! Tack-solder a wire onto the output of the U5 mixer and you'll be able to "tune in" the two frequencies that we've been mentioning (the sum and difference). One of them (10.1 MHz) will be the one that we're interested in and will be using going forward.
<21:24:30> "Pete wb2qll": The antenna lead should end with a short, perhaps an inch long, wire that's beyond the shielded coax.
<21:25:45> "Obe - KC4VZT": I used a short coax cable with an alligator clip lead to the center pin. Then put it near the VFO
<21:36:02> "Obe - KC4VZT": this is a good place for information about toroids http://kitsandparts.com/mtoroids.html


CWTD Episode #80:  Elmer 101 and the SW30+ Transceiver  ...

                                        Keying and Transmit Mixer


First, a few Notes: 

  1. Kits are sold out.

  2. Enclosures are sold out ... but surprise coming soon ;-)

  3. Bare SW30 pc boards are sold out.

  4. You can get the FreqMite Kits now from 4SQRP ... http://www.4sqrp.com/freq-mite.php   ... And AA0ZZ shows here how to connect it to your SW30.

  5. Accessories in progress ... Keyer Kit, Display,  Arduino control, Spectrum output

  6. For the latest & greatest SW30+ Kit information, see/download the Updated Manual

Block Diagram ...Tx Mixer

Technical Discussion

Keying Circuit

I followed the wisdom of others in using the assembly step ordering of the various Elmer 101 renditions. They mostly dealt with the 40 meter version of the radio, I've tailored details for our 30 meter radio. If I had not had that previous foundation to follow, I probably would have first done the receiver construction and then work on the transmit side of things.

This ordering makes a lot of sense in terms of testing during construction. The next several parts of the assembly process will work on the transmitter up to but not including the final amplifier (which will of course be completed later). This clever approach will allow us to use the transmitter to provide a test signal for receiver testing.

The “Keying” part of this step in assembly, Q3 and associated components (take a look at the schematic), is part of the rig that handles changing between transmit and receive. Pressing down on the key causes Q3 to conduct, supplying 12 volts to most of the transmitter components (the final amplifier gets power all the time).  Pin 8 of U1 is a convenient place to observe the switching.

Key up should give little voltage there (I saw a fraction of a volt) while pressing down the key (a jumper connected between pins 1 and 3 of J3) should give around 7.5 volts there. Its not the full 12 volts because of the dropping resistor R19 and zener diode D11 which form a simple voltage regulator to supply proper voltage to U5. This little regulator circuit is not precise, I got around 7.3 volts when I did my measurement, that' s plenty close. Seeing 12 volts here would be a problem to track down as well as not seeing a voltage change between key up and key down.

Mixer Circuit  (reference:  http://epic.mcmaster.ca/~elmer101/swgilb.html )

The SW-30+ radio employs three of these chips: one to heterodyne the incoming small-signal (10.1 MHz.) RF down to the I.F. frequency of 7.68 Mhz., another to heterodyne that down to audio, and another to heterodyne the VFO frequency of 2.4 MHz. up to 10.1 MHz. for the transmitting amplifier.

The term "heterodyne" refers to the mixing function, where two different frequencies are combined in a non-linear way to generate an output waveform of a different frequency. Keep in mind that the process is non- linear: there is no way to linearly add two frequencies to get a third.

An audio "mixer" is an entirely different animal: it simply adds signals together linearly. If it did output other frequencies, an audio mixer would be considered faulty, and in need of repair.

In the case of U5 (a SA612 chip), one sinewave input signal from the VFO at 2.40 MHz. comes into pin 2, while the chip generates the other input signal at 7.68 MHz. internally. The output waveform is available at pin 4 and/or pin 5. An internal Colpitts crystal oscillator generates the 7.68.0 Mhz sinewave (the crystal oscillator connections involve pin 6 and pin 7).

The heart of the mixer uses a circuit known as a Gilbert Cell.  A superb feature of the Gilbert-cell mixer is that neither 2.40 Mhz. nor 7.68 Mhz. signals appear at the output, provided that switching action is seamless. Ideally, only two frequency components will appear at the output: one at the difference frequency ( 7.68 MHz - 2.40 MHz = 5.28 MHz) and one at the sum (7.68 MHz + 2.40 MHz = 10.1 MHz. That's the best we can hope for. Usually, these two dominate over a mess of other mixing products of lower amplitude.


The tests for this part of the assembly verify that the mixer (U5 and associated components) appears to be working. We have a couple of things to look for. Part of U5 implements an oscillator for one of the two signals being mixed (the other comes from the VFO). Crystal Y5 determines the frequency with RFC2 slightly shifting the frequency from the marked 7.68 MHz on the crystal

At pin 6 of U5 we should see the crystal oscillator signal at around 7.68 MHz. Our test equipment will have some impact on circuit operation. Using a scope, I saw about of a volt peak to peak. Using the RF probe and DVM I got about half a volt DC on the meter. A receiver tuned to around 7.68 MHz should be able to hear the signal as well. It should go on and off as the key is pressed and released.

Finally, we should see a signal coming out of the mixer chip U5 on pin 4. What is useful to us is a signal around 10.1 MHz. Since the mixer produces both the sum (7.68+2.4 MHz) and the difference (7.68-2.4 MHz) of its inputs, there will also be a component around 5.28 MHz that we don't need. Viewed on a scope we won't see a nice sine wave but just seeing a signal there is enough for now. Using the RF probe I found around 1.5 V DC on my meter.

Using a receiver you should hear signals around 10.1 and 5.3 MHz with the key pressed.

Schematic Fragment:  Keying & Transmit Mixer

Test procedure:

Apply power, check for 0 volts pin 8 of U5. (Measured: < 0.5 V)

Connect jumper between pins 1 and 3 of J3.

Measure about 7.5 volts on pin 8 of U5. (Measured: 7.29 V)

U5 Pin 6: Measured: 750 mV PP 7.68 MHz using scope,  0.5 VDC using RF probe

Pin 4 of U5 should have around 10.1 MHz signal. (Measured: 1.5 V DC using RF probe)

Disconnect power.

Mike's "View from the Bench":

I completed this step and applied power, the voltage readings at pin 8of U5 were as expected for pins 1&3 jumpered or not on J3. But I did not hear a signal around 10.1 MHz as expected. I tuned my receiver to 7.68 MHz to hear the crystal oscillator and nothing. 


This is good! I get to troubleshoot the board. I decided to cheat and turned on my oscilloscope. I connected the probe to the top of RFC 2, there should be something around 6.68 MHz there. I did see a signal. I disconnected the jumper from pins 1 to 3 of J3. The signal was still there. Set scope to give rough measure of frequency and it said 2.4 Mhz or thereabouts. That's just the VFO showing up there. 


It should be working, my soldering looks ok. Remember I am now looking for the 7.68 MHz signal. The only components needed for it are C28, C29, RFC2, Y5 and U5. I checked RFC2 as soldered in place with an ohm meter, it was good at around 1 ohm. I carefully removed U5 from its socket. I really don't think it’s bad, but I do have a couple others in the kit and it’s easy to substitute another one there. 


Substituting one of the other 602's didn't fix things. I decided to clip the antenna lead of my test receiver to the case of Y5 and tune the dial around 7.68 Mhz and what do you know I hear something! But that something is not a nice single carrier (using the receiver in LSB mode). It’s a bunch of carriers slight separated in frequency. Maybe it’s from something else. I disconnected the jumper from pins 1& 3 of J3, it should go away. The carriers were still there. Just for the fun of it, I disconnected power from the SW30 board. The carrier frequency started drifting and getting until they were gone. Remember, we have a 220 uF capacitor on the 12 V line that takes a while to discharge. 


At this point, things are going downhill a bit. The original test of the voltage on pin 8 of U5 was OK as pins 1&3 of J3 were jumped or not. Now I get the 7.5 volts all the time on pin 8 of U5. Are we having fun yet?


I can solve this. I think I can at least! I'm glad I am doing step by step testing rather than having built up the whole board before doing any tests. First to make my work a bit easier maybe I removed the 602 from U5 socket and put it back in the foam it came in. It’s good to simplify things as much as possible. I disconnected the scope leads, all that I have connected to the board now is power. Measure pin 8 of U5 again to make sure that problem is still there. I see 7.29 volts. It should be close to zero. 


Time to look at the schematic. My current problem centers on Q3, the 2N3906 that is supposed to be a switch to turn on and off power for U5 (and other stages when we get to them) as the key is pressed. Our jumpering pins 1&3 of J3 simulates pressing the key. I measure the voltage at pin 3 of J3 and see 13 V, so pin 3 didn't get accidentally shorted to ground somewhere. 


Now Q3 is sort of upside down in the schematic. It’s a PNP transistor and the emitter and base voltages are the same, so it should not be conducting. Either the transistor is now shorted, or I may have done some bad soldering. Disconnect power and inspect the board under the magnifier (again). It’s the right transistor there, 2N3906, flat side where it should be. R20 and R21 are the right values. I don't see any soldering problems. 


Check from E to C with an ohm meter. I get 9.8 ohms in both directions. Sure looks shorted. Center lead on the transistor is the base. What to do next. I could get out the solder wick and remove the transistor. I don't think it’s a bad transistor though. What else could be causing that short? Let me look some more. I just don't see anything on the board. 


Let me carefully remove Q3. That wasn't too bad. Checking the transistor with my ohm meter, it does look shorted. I am surprised!  From my junk box I found a new 2N3906 and it does not appeared shorted. I solder in the Q3 location.


With U5 still removed from the socket, I apply power. Pin 8 of U5 without the J3 jumper is .395 volts, not quite zero. My replacement transistor has some leakage. With pins 1&3 of J3 jumpered, I see 7.29 volts on pin 8 of U5, as it should be. With the jumper removed, its 5 volts???? Wait, that may be OK. C110 got charged up and with no IC in U5 socket there isn't much to discharge it. The voltage is slowly dropping. 


I put a 602 in U5 and now hear the 7.68 MHz signal (and nearby birdies, that could possibly be my receiver not liking the out of ham band frequency). I will worry about that later. The signal keys on and off properly with the jumpering of J3 pins 1&3. I then tried using the other two 602 ICs, and it looks like one is actually bad. 


Next I looked for a signal around 10.1 MHz on pin 4 of U5 using my receiver.  Yay, I found it at 10.146 MHz, close enough at this point. And it sounds clean and does key on and off with the jumpering of pins 1&3 of J3. So at this point my board seems to pass the tests. I have no idea why I had problems. Maybe I should have taken static discharge precautions and turned off the Van de Graaff generator while doing my building. But I do enjoy those sparks jumping around the workbench!

Spectrum View

Now we are looking at the transmit mixer pin 6 which is the crystal oscillator at about 7.68 MHz, indicated as Peak 2. What is that other stuff? Well Peak 1 at 2.4 MHz is some bleed through of the VFO. Peak 3 is somewhat near what we expect for the mixed output, but maybe not close enough to be that. Although the frequencies are displayed to many decimal places their accuracy is not know. Peak 4 is probably the 3rd harmonic of the oscillator frequency.


Here (below) are the signals present at the output of the transmit mixer U5 pin 4. Who would have expected all that? Peak 5 at around 10.06 MHz is the desired output signal. Also present are the two input frequencies noted as Peak 1 and Peak 4. Peak 3 is the difference between the input frequencies, an expected but undesired output signal. Peak 2 looks like the second harmonic of the VFO. Peak 6 is about the second harmonic of one of difference of the inputs. The other peaks are various mixing results of the signals present (and their harmonics). Now it is very clear why the Transmit Bandpass filter is needed!


Component Layout:  Keying & Transmit Mixer



Troubleshooting Schematic