A Multiband Modular Amateur Station Based on the Rick Campbell R2/T2 modules

by John Seboldt, K0JD
Last update 6/8/2001 - (Motorola links updated)

A constant work-in-progress... my approach to using the marvelous single-signal direct-conversion R2 receiver design in a multiband, modular amateur station.

What you see above is how it was laid out, and used, for several years. Yes, a little haywire, but quite functional. The boxes are made from double-sided PC board material. The BNC jacks are a special PC board type that I found cheap at my local surplus house. They mount with 4 ground pins and a center conductor pin, saving space.

Around January 1, 2001, the driver, PA, and R2 went  into a large flea-market cabinet with a hinged lid, so it looks neater and yet can still be shown off and easily fiddled with. Microphonics and other glitches are much reduced with a setup like this. The input boxes still plug onto the front of the unit (remember the HRO-5, you old-timers?). There will be LOTS of room for expansion. The T2 exciter is in there too, but still not hooked up (it's the nicest-looking board, though...) The digital age is dawning with a Hands DDS3 VFO coming my way; will this go in the cabinet or in a separate unit for a variety of uses? Decisions, decisions.  See below for some possible upgrade directions.

A neater block diagram of the old modular layout  (TX chain substantially changed since this was drawn)

Interconnections were a combination of Molex connectors, BNC's, clip leads, etc. In my cabinet, many connections are now made quickly disconnectable, with crimp-on pins and push-on contacts found at the local surplus houses.  For the future I keep eyeing a neat connector option available available from Mouser Electronics. They stock a mixed-contact D-subminiature connector system in various configurations. Depending on the specific model, you can get various combinations of small built-in pins for DC lines, and larger contacts (snap-in options) for 50 or 75 ohm ohm coaxial, high voltage, or high current leads.

The modification of the surplus Collins T-368 exciter is a separately documented project in itself! This unit was available from Fair Radio Sales in the early to mid 90's for about $45; many correspondents tell me this source has dried up. The basic PTO is 1.5-3 MHz, with multipliers to 3-6, 6-12, and 12-24 MHz with a mechanical digital dial. Once transistorized, it is the most stable analog VFO I have ever used, mostly because it is built so heavily. Others may wish to use various combinations of homebrew VFO's, multipliers, or pre-mixers, especially for portable use!

Input Modules
I started out with boxes that combined input filtering and phase shifting.  BNC connectors on both the input module and the R2 were on 3/4 inch center to center spacing, with a set of male-to-male BNC adapters remaining "resident" on the R2. I switched filters and phasing networks with one 4PDT slide switch. Rick's suggested "pi" phasing network in the R2 article is easy to deal with, tuned up predictably, and worked well. I think, though, with the haywired approach, there was more RF radiation than I might have wanted. I got by, but hum started to show up on the high bands (LO leaks out to an AC power line, gets modulated, and re-enters the radio). Microphonics and not-perfectly-noiseless connections also showed up, but usually was tolerable. The photo below gives you an idea.

LATER: I turned to the twisted-pair quadrature hybrid, with greater bandwidth, and needing one less piece of iron to wind on. In progress now is the "Quadrature Expressway", a surface-mount approach to making neat networks bandswitched with PIN diodes.  Details here.

The input filters are standard 2-resonator bandpass filters found in various homebrew books. They're adapted to tune a range, using dual 250 pF plastic variables (Mouser Electronics).

R2 Receiver Module
Rick's R1 article has some precautions for power connections. Because of the high gain on one board, any circulating ground currents generated by the output stage must be minimized. Therefore you don't just connect the DC negative lead to any old point on the chassis, but right to the point where the output transistors are grounded. If you go him one better, and return the SPEAKER ground lead to this point also, you get even more immunity to feedback. This means to insulate your output jack from chassis ground (easy with PC board boxes: simply scrape some copper away).

In my new cabinet setup, I chose to keep the R2 board completely insulated from the chassis, making it possible to experiment with various grounding options to see if it makes any difference.

Call it overkill, but I found some high-level mixers at a hamfest, the SRA-1H (+17 dBm LO drive, I think), and have used them for some time. See my 74HC240 CMOS buffer which drives my mixers.

My box has a switchable broadband preamp added. I find no need for it on 40 meters and below, but it definitely is essential on 20 and above.

I've used two preamp approaches so far, and these are not the last I'm sure...

1) Mini-Circuits MAR-6 MMIC amp -- 3dB noise figure, 20 dB gain, 1 MHz-2 GHz bandwidth. It works well, but that's a lot of gain, and its output 1 dB compression point is only 2.5 dBm. But if you really want to listen down to the noise on a quiet band, or need something that at least gets you a start on VHF premplification, it's great. A little kit is available: the WBA-6 ($19.95) from Electronic Rainbow, 6254 LaPas Trail, Indianapolis, IN 46268, tel. 317-291-7262.

2) A 2N4416 grounded-gate stage (fig. 3). I can't remember where I read these specs, but I believe it has about 4 dB noise figure, higher overload margin, and a more manageable 10 dB gain. A better choice for HF, though I found the gain at the high end rolled off (maybe I need a better-selected transistor...)

My amplitude balance pot is put on the front panel, because readjustment is required for each band. You will most often get adequate rejection of the opposite sideband with a reset from memory. Best adjustment requires a steady carrier or VERY strong signal. Your phasing networks should only need to be adjusted once, unless you change your LO drive in some way.

If you hook things up as specified, you get UPPER sideband reception. Rick's suggested NE5532 sideband subtractor in the R2 article would then give you lower sideband. This circuit is built on a small piece of experimenter PC stock. I have only a trimpot on this board, set for 40 meters where I am likely to need it most at this point. You may wish to use a dual section pot for your front panel amplitude balance control so either sideband is easily adjustable.

I have switching for a CW and SSB lowpass filter -- and can switch them both out! It's an interesting hi-fi sound to hear signals above 3 kHz. My ears hear to 10 kHz or greater in my hi-fi headphones, serving as a mini spectrum analyzer. Strong shortwave broadcasters sound stunning this way. At high gain settings, though, you may get some oscillation, sometimes supersonic.

Switching out the highpass filter is another possible option -- you have to cut some traces on the board to implement this. This is especially useful if you make this modification: change the highpass filter caps to 0.47 uF for better CW reception. It is not very sharp for QRM rejection, but helps eliminate low frequency noise. Because of the gentle slope, SSB reception is acceptable with the filter in -- and excellent with it switched out!

The capacitor in the mute circuit should be reduced to .03 uF if you want fast recovery for full break in CW. The 0.1 uF originally specified means a good half-second recovery time.

You need plenty of LO power to drive the diode mixers and overcome the loss in the splitters. The original mixers need +7 dBm minimum, and I have even gone with SRA-1H mixers (+17 dBm drive!). So once you split it for transmit and receive (3 dB loss), and then again for phasing (another -3 dB), you need to make up that loss with higher output from your buffer.. The mixers generate harmonics, and even prefer a square wave, so don't bother with the low-pass filtering featured in some buffer designs.

Any VFO you use needs a varicap for CW and/or RIT offset. See the Lewallen Optimized QRP Transceiver (1994 ARRL Handbook, QRP Classics) for a good basic circuit. I implemented it this way in my T-368 exciter:

Pick the capacitor for the right RIT range for your needs. Mine had to be big because the PTO operates at 1.5-3 MHz. Also be creative on your varicap diode. The NTE612 is designed as a varicap. But many diodes will work as varicaps. Lewallen suggested a zener diode. Try some old rectifier diodes in your junk box -- something that is likely to have a larger PN junction than a fast switching diode.

Lewallen's RIT switching is also OK as far as it goes. However, you can get offset above and below your frequency just like the big rigs with this circuit.

Pick the capacitors for the right delay for your needs. Remember that when you unkey, the VFO needs to stay at the transmit frequency long enough for the PA output to decay.

The RIT pot is on your front panel, of course. Pamper yourself and use a 10-turn pot -- you won't regret it. The transmit centering can be an internal trimpot, or a front panel control. If you wish, add resistors on either end of the pot(s) to limit the tuning to the most linear part of the varicap's range. The SPDT switch removes RIT by keying the circuit. On CW, you would turn off RIT (switch to ground) to spot an incoming signal to zero beat, then turn on the RIT and adjust the RIT pot for your preferred beat note. (With experience and an ear for pitch, you don't need to do this every time). On SSB, of course, you normally tune with RIT off, and turn it on only as needed. Remember also that the diodes at the output will be part of your VFO thermal stability concerns. As temperature increases, the forward voltage drop goes down, thus the voltage increases, and your VFO frequency will creep up. To some degree this compensates for the way the varactor diode behaves: it creates a frequency drop with increasing temperature (increasing capacitance).

For the fine points on tuning diode temperature compensation, get Motorola bulletin AN847.

Transmit Chain
The standard homebrew/QRP books give you plenty of circuit specifics here. Typically, your keying circuits use a positive voltage that is pulled to ground to transmit. 

In my first iteration, blocked out above, I started with standard gain blocks: a 2N5109 linear stage, a paralleled pair of 2N3553's for a one-watt linear driver, and the final based on the "10-watt linear amplifier" (2 NTE236's) described by W1FB in the QRP books, with a more elaborate bias regulator.

Funny, that PA never worked quite right. But eventually, I found my way to the circuit's roots: Motorola's application note AN779, now conveniently available as a *.pdf file!  Turns out Doug's circuit was the second half of this. I built it up in my own haywire fashion, and it gave out a good 15 watts... but was never quite stable. Then after a few years of using this, I hit on the ingenious and unique idea of using the board layout provided in the app note - and it worked flawlessly! Lesson:  good layout does the trick, and if somebody's done the work for you, don't re-invent the wheel.

It's a nice little 20 watt amp design - 2 MRF476 drivers, and 2 MRF475 output devices, both in push-pull. (These devices are no longer made, so I used the NTE235 and NTE236.) I found the biasing a bit problematic, wasting a ton of current to heat a resistor through a diode, and even letting the transistors go into thermal runaway! So I provided some adjustable bias via pass transistors for both stages, based on the Ten Tec Argosy once again. This is driven by a single 2N5109 linear stage.

QRO? Well, yes, sometimes... waiting in the wings is the 140 watt amplifier from Motorola app note AN762 - via the kits from Communications Concepts that you see advertised in QST. The amp is built and tested (without output filtering), but waiting for some more time/patience/parts to add output filters, switching, keying, ALC, etc. But the need for it hasn't been too pressing, I've had so much fun at less than 20 watts...

Diode TR switching is the very best way to go. This is a broadband circuit, but you need 20 to 40 mA of total diode current -- a concern for portable operation. (I went as high as 60 mA to eliminate bleed through from a 50 kW AM broadcaster 5 miles away, thus the paralleled 330 ohm resistors. 330 ohms, 1/4 watt may be sufficient for many needs.) 

The switching part is cobbled from some circuit I read somewhere, and modified. The IRF510 MOSFET is used for its high gate impedance and sharp switching characteristic, otherwise the capacitor (.47 uF) that holds the switch closed on key-up has to be outlandishly large. There are probably smaller MOSFETs available for such low-current switching, but this one you can get at your local Radio Shack.

The diode switch is drawn from the Ten Tec Argosy, and handles that rig's 50 watts with ease. 1N4007's are 1 kV rectifier diodes that have a PIN structure; they operate satisfactorily, though they are not rated for RF PIN service. The 1N914's rectify the transmitted RF to back bias the diodes on transmit. At 5 watts, I have also had satisfactory results with 1N914's in the switching path; in this case you do not need the back-biasing circuit, because the fast switching diodes rectify their own back bias! I don't use the output transistor now, but I may try it to cure some receiver audio pops at high volume levels. The chokes are small molded ones at this power level; actually, the one on the transmit end could be eliminated if your DC return path is through an output transformer winding.

For lower current drain, you can always revert to the popular "pickoff" circuit -- back to back paralleled diodes (1N914 type) to limit the receiver output to 0.7 volt, a capacitor with 400 ohms reactance before your transmitter lowpass filter, and a series inductor to resonate with the pickoff capacitor at the receive frequency. Remember that the capacitor becomes part of the filter network on transmit, so subtract its value from the first shunt capacitor on the input of your lowpass filter. This is obviously frequency dependent, so put this in your appropriate transmit filter box. Another disadvantage: you rob some of the dynamic range of which the R2 is capable.

Transmitter Filter Output Module
This is just a box with switch-selected lowpass filters of standard 50 ohm design for the various bands.
Adding the T2 Exciter
My T2 exciter board is now built, and mounted in my cabinet, but not yet in use. Rick's suggestions for station integration are made with VHF mountaintopping in mind. Here is my brainstorm to provide all modes available, and full QSK switching.

You really should have a SEPARATE phasing network for transmit for best nulling according to the original article. I might do an expansion of the "plug-on box" system.

Because TRANSMIT amplitude balancing is harder to do on the fly when changing bands, I am at a loss for the very best solution. My thoughts:

1. Use a 10-turn pot for amplitude balance on the front panel, and log the settings on a turns counter dial.

2. Incorporate a pre-set trimpot in the input module. Convenient connections are the question here. Here is where the Mouser D-subminiature mixed-contact connectors would really come in handy.

I would make everything switchable to get every mode possible, even if only just to say you can do it! Q INVERT would invert the audio phase (with a unity gain opamp inverter) coming into the channel that feeds the 90 degree mixer. (I think this will work -- engineers?) IDISABLE and QDISABLE would break the audio inputs at X and Y on Rick's schematic -- or possibly at the input of the audio phase shift network. CARRIER INSERT would ground the key line, unbalancing the mixer that receives in-phase LO energy. (See switching circuits below.)

Here then would be your modes (check me, engineers!)

SSB/CW off USB, on LSB off off off
SSB/carrier off USB, on LSB on off off
AM n/a on off on
PM (NBFM) off on on off

Below are proposed board interconnections. Only the DC control lines are shown; RF connections should be obvious.

I hope all my diode isolating logic is correct! Strictly speaking, you may not need to key anything in your transmit chain but the TR switch. However, you will save idling current on receive (and heat!) in your linear amplifier stages if you key them and/or remove bias.
Articles and Kit Sources
 Rick Campbell, KK7B, “High-Performance Direct-Conversion Receivers”, QST, August 1992 (“R1”)
 Campbell, “High-Performance, Single-Signal Direct-Conversion Receivers”, QST, January 1993 (“R2”)
 Mouser Electronics, 2401 Highway 287 North, Mansfield, TX 76063-4827, tel. 1-800-346-6873
 Fair Radio Sales Co., PO Box 1105, 1016 E. Eureka St., Lima, OH 45802 tel 419-223-2196
 Doug Demaw, W1FB, “A Diode-Switched Band-Pass Filter”, QST,  January 1991
 Wes Hayward, W7ZOI, and Doug DeMaw, W1FB, “Solid State Design for the Radio Amateur”, ARRL
Doug DeMaw, W1FB, “W1FB’s Design Notebook”, ARRL
Motorola Documentation Library 
Motorola Literature Distribution Center - includes the archived old stuff
Motorola Semiconductor Products Sector - List of RF design application notes

Upgrades in the works
  • Time to go digital... a Hands Electronics DDS3 DDS VFO kit is on the way across the pond via Kanga, which can generate the LO and provide some bandswitching TTL signals for the stuff below
  • The "Quadrature Superhighway" - a neater and tidier way to bandswitch the RF phasing networks with a three-lane microstrip setup and PIN diodes. Inspired by my recent purchase of some SMA connectors and some semi-rigid .141 inch coax plumbing at a hamfest...
  • Input filter bandswitching and/or tuning with PIN and/or varactor diodes
The Author
John Seboldt, K0JD, began hamming as WN0QXG some 26 years ago. Music, electronics, and ham radio grew side by side in his youth, leading to work in the broadcast industry while studying music at Luther College, Decorah, IA, and The University of Iowa, Iowa City. Church music is his main field -- he served 15 years in the Twin Cities, and moved to Milwaukee in 1999 (but will remain "forever a zero", previous moves never having taken him out of "zero-land.")