I have been hand-making printed circuit boards since I was a teenager. In the late 1970s I started with ‘proper’ PCBs, hand-drawn, hand-drilled, then painted with resist (bituminous paint) thinned with turpentine and applied with a tiny brush to form the copper traces, drilled holes on the copper side, etched in Ferric Chloride, cleaned and sprayed with a clear enamel to stop tarnish. This was the way all circuit boards looked at that time, including commercially made ones in kits, and those inside transistor radios, digital clocks, televisions and ham radio transceivers.
Sometime in the 1980s I tried a ‘copper side up’ board, dispensing with the drilled holes, free-hand drawing on the copper side, painting and etching, then soldering the components directly to the copper pads. I’ve been making them this way ever since.
I may have been trying to reproduce the copper-side earth-plane of VHF and UHF preamps and converters of the day. For some reason, the exotic nature of homebrew VHF-UHF modules legitimised the use of copper-up construction. I used it for an HF superhet receiver, for the dual gate mosfet RF amplifier and mixer stages. I liked seeing the components and the copper traces connecting them, without having to turn the board over. You could see the circuit and its parts right there, in front of you. I liked bolting the boards down to the chassis with 1/8 Whitworth nuts and bolts, allowing the nut to make clean contact with the copper earth plane, and feeling reassured that the board was well bonded to the earthed aluminium chassis. I liked that way I could remove or add components with the board in place. For experimenting, it seemed like a smarter way to build. And I loved not having to drill those 1/16th and 3/32nd inch holes, breaking the bit on every tenth hole.
‘Copper-up’ style had a sense of ‘quick and dirty’. I used to think that once the circuit was proven, I would rebuild it on a proper drilled board. Of course I never did.
Before long, the Americans were all talking about ‘dead-bug style‘ construction in which the leaded components are soldered to an unetched PCB ground plane, resulting in mid-air solder joints. It was rougher than my etched copper-up method, but obviously far quicker. But seeing 8 and 14 pin dual inline ICs glued to the copper board with their little legs up in the air was too much for my nascent sensibilities! I could see it was effective, particularly in regard to being quick, and to encouraging rigorous grounding, but to me, it looked like a mess (apologies to esteemed dead bug aficionados, of whom there are many).
In the Manhattan variant, little glued-on pads made from scraps of PCB afforded landings above ground, relieving the need for some mid-air joints. It was better, but the technique seemed to result in widely spaced layouts, particularly when builders tried to preserve the right-angle lines and joins of the circuit diagram they were reproducing. Manhattan has become a mini-industry now, and the results can look much neater.
Back to my etched copper-side-up construction method. Chuck Adams K7QO calls the technique MUPPET (Manhattan, Ugly and Precise Placement Experimental Technique). I like that he came up with the same approach, dispensing with drilling holes, making the copper side right-side-up for experimenting, tinkering, de-soldering, and substituting. I expect that he, like me, found dead-bug style too ad hoc to give the maker a sense of pride of workmanship. However I do find myself wishing he had come up with an acronym other than MUPPET which the Oxford Dictionary defines as ‘a foolish or incompetent person’. Name aside, the technique is what’s important. Here’s what he says about it:
After years of building, experimenting and writing about Manhattan and PCB construction techniques, I have just come up with the term MUPPET board. MUPPET standing for the words Manhattan, Ugly and Precise Placement Experimental Technique. You can think of the following as allowing you to use all of the building techniques together as a hybrid and could even add SMT (surface mount technology) into the mix.
I use printed circuit board material as the basis for the building of a circuit. But unlike making a regular printed circuit board that has to be drilled with a large number of holes to place components, the MUPPET board allows you to place and solder components upon the top of the board without having to drill large numbers of holes. So in that respect it is like Manhattan construction, but without the tedious task of placing pads using super glue to hold them down. It is like ugly construction in
that components are soldered to the copper layer, but if I think ahead enough I can save a lot of wires from one part of the board to another by doing traces, i.e. lines of copper material that form a flat wire from point A to point B.
But what is handy about the MUPPET board is that you form the pad area from the copper layer and create traces for signal and power paths to other parts of the circuit. You can leave a number of areas of the PCB plain (unetched) for room to construct modifications or additions at a later time. Think of this additional area as an area to ‘tinker’ in.
I go further than just etching power and signal traces. I draw up all of the copper pads and traces needed to realise the entire circuit, just as you would find on a traditional PCB. Here’s how I do it.
First, I draw up the board layout on an actual size piece of paper. It helps to lay down the larger components and the parts that demarcate areas on the board first, such as potentiometers, trim-caps and trim-pots, and ICs. Trace their pins onto the paper. Leave the four corners clear for mounting holes. Then join up the component landing spot marks with pencil lines, reproducing the circuit diagram onto the (positive) paper template.
Next, I cut, file and clean (with steel wool) the piece of PCB blank board to be etched. Sitting at the shack bench, and with plenty of light, I copy the pencil sketch with a marker pen on the copper board. The marker pens I use are ordinary permanent ink markers. A whiteboard marker will rub off with your fingers and won’t survive the etch bath. I use a local brand called ‘Sharpie’ (VK only). Use a marker with a narrow tip, not the ones with a chiseled or broad tip. These latter types come in handy for filling in large areas of copper ground plane later. Have your personal choice of 160 meters AM or 40 meters CW running quietly in the background for ambiance. You’ll concentrate better.
Marking up the board. Below are two boards, a 160 meter transmitter and the receiver, marked up and ready for the Ferric Chloride bath.
Bath time. After using Ammonium Persulphate I reverted to good old Ferric Chloride. I had very mixed results with the former, it etched slowly and some undissolved powder and other refuse in the solution abrased the resist, resulting in scratched and even cut traces. Maybe I wasn’t using it correctly. The first time I went back to FeCl3 I got an excellent result.
As a bath, I use an opened up plastic milk bottle. It is perfect for a board of approximately 15 by 4cm (6 inches by 1.5 inches), my preferred size for homebrew radio modules. I have used ice cream containers for larger, squarer boards in the past. I rest the bottle in a basin of hot water to increase the reaction rate, holding the bottle and gently rocking it back and forth so that the FeCl3 is always washing over the board. I try not to let the solution sit stagnant. I always wear glasses or eye protection to avoid getting a splash of FeCl3 in my eye! Any FeCl3 that comes into contact with my skin is immediately washed off.
A word about the basin. don’t use the house’s prime bathroom sink as FeCl3 will stain the porcelain. Similarly, don’t use the laundry stainless steel sink as this may be irreversibly marked as well. At times I have used a bucket of hot water in the back yard.
With a fresh solution a board typically takes 5 to 7 minutes. When done, I remove the milk bottle cap and pour the FeCl3 solution back into its storage bottle. Rinse out the milk bottle with plenty of water.
Next, I put some methylated spirits on a piece of paper towel and rub off the marker ink, then rub down the copper side of the board with steel wool, wash and clean on paper towel. Here’s a finished board.
Take it back to the workbench and inspect it for bridged pads. These can be opened or widened with a Stanley Knife (Australian for box cutter). Your new PCB is ready for solder. You can optionally spray the copper side with a matt acrylic clear spray, such as are used for sealing prints or artwork, or other art and craft projects. (I also seal the rub-on lettering on front panels this way). A few light sprays always ends up looking better than one thick coat. A thin spray dries in under 10 minutes. This stops the copper from tarnishing. When you apply the soldering iron, the heat burns off the thin enamel layer without harmful side-effects.
Here’s the boards assembled, a crystal oscillator, driver and a PA for 160 meters, and a superhet AM receiver.
I’m so used to this technique now that I can bang out a relatively simple one (such as a power supply or an oscillator and buffer) in about 90 minutes. Although each board is custom drawn for a specific circuit, I often experiment after the components have been soldered in, changing components, and sometimes adopting dead-bug construction to add a transistor buffer, or to alter an op-amp audio mic amp or pre-amp. I sometimes leave an un-etched area on a board as a play-space for additional dead-bug/Manhattan construction, if needed. Here is a quickie oscillator and buffer board I whipped up recently for a G3UUR crystal characteriser (for filters):
So this is how I make circuit boards. MUPPET, copper-up, whatever you call it, it epitomizes time-honoured radio craft in an era of surface mount and multi-layered phenolics. Artisan-like, you sketch your thoughts in copper on fibre-board, with pencil in hand you lay down the essence of your idea, cut, etch and shape it to suit, no two alike, rub it smooth, meld on the components, and let the free electrons bring it to life.
If you try it, let me know how it works for you. A good board makes the assembly of a homebrew radio module a pleasure. Good luck with your boards and projects.
Postscript: Here’s a chemist’s view of etching by Ralph VK3ZZC, including how you can make your FeCl3 solution last forever.
There is discussion online about how potent Ferric Chloride is. All I can say is, it eats copper, so if it gets on my fingers I wash it off immediately. And if you do yourself an injury don’t blame me!
Update 22 May 2018: I have no commitment to ‘commercial’ PCBs. If I have a choice to scratch-build or PCB-build a circuit I always choose my own board. ‘Commercial’ PCBs, including the ones that Makers design and get printed at OshPark (etc) cannot be changed and are not suitable for prototyping. And every scratch-built homebrew radio is a unique one-off prototype.
So I was amused by a recent thread on the EMRFD Yahoo group. Someone opined that they couldn’t start a project until they had a double sided, thru-plated, silk screened quality PCB. The legends of the EMRFD world ripped into this comment with a passionate defense of dead bug/copper-up bread-boarding. As the Yahoo group may at some point disappear, I copied three of the most noteworthy posts here for future reference.
Amen on dead bug. Wes and I nearly always do early/experimental prototypes without a pc board, and long ago I absorbed his wisdom on the subject: “when there is a performance advantage it is in favor of ugly construction.” Look at the photograph at the top of this web page, and you’ll see many lead lengths shorter and more direct than you could get with a thru-hole pc board, and a superior ground underneath. I started my home brewing career at 144 MHz, and the lore back then was that if you could see the leads on the components, they were too long.
Some folks seem to think that pc boards have a more “professional appearance,” but my background in scientific instrumentation, where pc boards are the hallmark of mass-produced gear, has left me with the expectation that behind the front panel of anything cutting edge or interesting I’m likely to find custom-designed dead-bug circuitry.
The steps from working high-performance dead-bug circuitry to a clean pc layout with equal performance are serious and non-trivial. It takes me about 4 pc layout spins from hardware that works well in dead-bug to a KK7B board layout that I’m willing to have duplicated and included in a kit. For good pc performance at 50 MHz and above, I must use some surface mount parts. Even in designs at 7 MHz, I’ve been surprised when thru-hole pc board traces and a 2N3904 combined for a parasitic oscillation near 300 MHz.
Unless you are specifically designing something to be easily reproduced, I strongly recommend “ugly,” “dead-bug,” or “NASA lab” construction–whichever term you find most appealing. My friend Jim K8RZ builds experimental prototype circuits that are works of art.
Study the examples in EMRFD, and enjoy the dead-bug experiments. — Rick KK7B
Generally for experiments and testing I dead bug it (manhattan or whatever)
and get to testing. It’s efficient and with care it can be very rugged and allow
things like shielding. That’s a bias on my part. — Allison KB1GMX
I also want to comment in support of the recent postings from Allison and Rick. The current “maker” movement supports printed circuit boards for anything that we might build. While printed boards are sometimes handy, they can get in the way of experimenting and learning. Building one’s own breadboarded circuit provides a more intimate connection to the technology than the experience of merely stuffing a board. As Rick emphasized the other day, there is rarely any difference in performance between a clean breadboard and that with a printed board. When there is a performance difference, the breadboard usually wins. — 73, Wes w7zoi
One thing’s for sure, I prefer “NASA lab” construction to “dead bug”.