Copperless perfboard + wire wrap for RF prototyping?

[aneevuser]

I've been trying to play around with some simple RLC circuits, measuring resonant frequencies, Q factors and the like. I'm putting them together on a Wisher breadboard, and have been getting some very strange behaviour, which I think I've finally tracked down to inter-pin capacitances (which seem to be about 10 pF). This has been frustrating and it's taken me a long time to figure out.

I'm now wondering what is a more suitable low cost method of putting together an RF circuit that is sensitive to stray L and C and I'm thinking of putting them together on copperless perfboard and wire wrapping the components instead, my reasoning being that lack of copper track will minimise stray C, and I can wire wrap up to the edge of the components to minimise stray L. However, I don't want to buy the kit just to find out that this doesn't help much, or that there's some other problem that I'll run into.

Does anyone know if this is a good way to go, or if I should take another approach?

[ledtester]

A technique called "Manhattan construction" is often used for RF circuits.

Have a look at this slide deck. There are examples of the technique starting at slide 13.

http://s-iihr64.iihr.uiowa.edu/MyWeb/Teaching/ece_55195_2015/Lectures/Construction.pdf

[Howardlong]

I use these extensively for prototyping RF & mixed signal.

https://www.busboard.com/surfacemountpcbs

In particular:

https://www.busboard.com/SP3T-50x50-G

and

https://www.busboard.com/SP3-100x100-G

They have a solid copper groundplane and non-plated through holes. They also do PTH versions but I find that is a bit restrictive on layout at times, and it doesn't take more than a few seconds to run a through wire.

Edit: I wouldn't ever use wire wrap for RF, wire wrap is an antenna.

Edit2: Apart from some baseband and LF applications where S/N is unimpotant, for most RF applications, solderless breadboard is a problem waiting to happen. Embrace and hone your soldering skills, and preferably switch to surface mount!

[Ian.M]

Soldering components to brass pins stuck in shellac or paraffin wax impregnated balsawood (or other low density kiln dried  softwood you can set pins in with a pin punch or just a pair of pliers) was the origin of the electronics term 'breadboarding', from literally building a circuit on the significant other's wooden breadboard!  Its still viable for point to point wired circuits using leaded components at frequencies below roughly the middle of the HF band.  Panels for controls etc. can simply be screwed to one edge of the board. It can also look pretty good (1920's retro-tech) if you take care to make an aesthetic layout, use hardwood, rout and fine sand a molded profile on the edges, highly polish the shellac before assembly, and pre-drill the pin holes under-size so you don't mar the harder wood setting the pins.

If you need to work at a higher frequency go Manhatten Island which with careful layout is good up to the low UHF band.  Its also *MUCH* better suited to SMD prototyping, (dead bug or on small breakout boards)  and even if you don't want to use SMD for the bulk of the circuit, its still worth using SMD ceramic caps for sub-uF decoupling - just tombstone them directly on the ground plane and use the top end as a circuit node.  0805 or larger 100Meg 10% SMD resistors make quick, good and cheap low capacitance standoffs for 'air wiring' parts of the circuit  where you need to minimise capacitance to the ground plane.  If you need low leakage standoffs,  then its expensive pins in Teflon bushes in drilled holes.


Above the low UHF band, the layout *is* the circuit reactive components, so you need to cut carefully designed and dimensioned striplines etc. on double sided specialist low loss PCB material, and 'fence' round the edges of ground islands with copper rivet vias to stitch them down to the ground plane, then edge solder the assembled boards into screening cans.  Its not too bad if you've got a good CNC mill with a precision high speed spindle and a supply of small conical carbide cutters, and small carbide drill bits, and can go straight from CAD to PCB ready to copper rivet the via fences (and don't mind the effort containing the abrasive PCB dust so it doesn't get into the mill's ways), but is a royal PITA if you are using hand tools.

Put the breadboard (of whatever construction technique) in a biscuit tin (or sweets tin - Altoids is the classic) and spot solder the lid on a couple of points on each side (or use PC case panel grounding clips) for screening.   You can also build screening boxes by tin-smithing from bent and soldered tinplate.   Another method is using copper clad PCB, with the seams soldered.  Double sided PCB is stronger with a slightly oversize baseplate and end walls, with all seams soldered both sides.

[bd139]

Important things are a solid, low inductance ground plane, keep all the leads short and keep everything rigid. Short leads kills stray L as much as is possible. Stray C is inevitable and unavoidable so you tend to design around this. The biggest problem I find is stray variable C.

I did an example here of why you need a solid ground plane by building a really naff prototype HF LC oscillator on a "gouged PCB" method popular in the 60s and 70s and tuning it on my HF radio. The stray C here is between the traces on the board and the board underneath it. As you can see it detunes things terribly. Thus I don't use anything with isolated pads on it.



Inevitably this leads to "dead bug" construction, at least for me. Typical example. Looks like s**t but from a performance perspective it's pretty excellent. Never correlate neatness with performance - they are 100% independent.



So what I do:

1. Take a piece of solid single sided FR4
2. Solder all the parts to the top (and BNC connectors) keeping all the leads short.
3. Make sure it is rigid.

The circuits built are high performance, reliable and cheap to build. This is good to at least 250MHz or so (I haven't gone higher)

Edit: I still use stripboard for low frequency/audio stuff though. Does the job and Roth sell some nice stuff with FR4 substrate. And I still use (Wisher) breadboards but mainly for bias testing stuff where a bit of experimentation is required.

As you're in the UK, Rapid electronics sell half decent FR4 single sided stuff under the RVFM brand. It's pretty good for the money. I cut it with some aviation shears I got from Machine Mart.

Also a final note, worth a read, particularly the construction section: https://www.analog.com/media/en/technical-documentation/application-notes/an47fa.pdf

[MarkF]

[Ian.M]

That 70's 'gouged PCB' oscillator would have been fine on double sided boards with 'hot' side non-ground pad area minimised, and copper foil tape soldered round the edges to bond the bottom ground plane to the topside surrounding ground pad.

Another cheap and dirty option is copper foil tape on the backside of non-PTH matrix board.  Punch through the tape (from its side)  for ground leads, soldered both sides (use two adjacent  holes when you need to reduce inductance and tie them together topside) and solder all 'hot' nodes top side only without putting the lead down the hole.  Cheap SRBP is even lossier than FR4, so I probably wouldn't do that for VHF band or higher work.

[bd139]

That's true indeed. The illustration was a purposeful exaggeration.

Foil tape works but getting it to stay on the board through rework is a bastard from experience. Much better someone glued it on there at the factory 

[fourfathom]

I did an example here of why you need a solid ground plane by building a really naff prototype HF LC oscillator on a "gouged PCB" method popular in the 60s and 70s and tuning it on my HF radio. The stray C here is between the traces on the board and the board underneath it. As you can see it detunes things terribly. Thus I don't use anything with isolated pads on it.

I do dead-bug for RF, but I also use the "gouged PCB" method on double-sided board.  This gives me a full groundplane under my gouged pads, so I drill strategically-placed through-holes and solder wires connecting top and bottom groundplanes.  The top pads do have some capacitance to ground, so where this might be an issue I keep the pad dimensions small.  I also occasionally use copper foil tape along the board edges to solder the top and bottom planes together -- it depends on my layout and the frequency. 
To make the gouges I use a woodworking chisel like this:

https://www.amazon.com/gp/product/B001NIC1XM

(Take care that you don't stab yourself -- I use a steel-rule and clamps when I think there might be blood.)

[aneevuser]

Thanks for all the ideas - I wasn't expecting so much detailed information, which will take me some time to go through.

I have one question in the meantime though: ideally I'm looking for an "RF equivalent" of breadboard i.e. a technique that allows rapid mounting and demounting of components (so no soldering) but that works reliably for RF. From what I've skimmed above, I'm guessing that this doesn't exist?

[bd139]

This does exist. In a professional environment you have “Lego bricks” for RF which are individual boxes with single functions (mixers, VCOs, filters, eval boards, test gear etc) and SMA cables to glue them together. This is large and beyond the realms of the average bank balance unless you go digging around on eBay and hamfests.

[aneevuser]

So what I do:

1. Take a piece of solid single sided FR4
2. Solder all the parts to the top (and BNC connectors) keeping all the leads short.
3. Make sure it is rigid.

Maybe I'm being dimwitted, but I can't see what you've done on the solid copper board in the image - it doesn't look like it's been etched in any way, so it looks to me like a continuous conductive layer of copper. This is the side that you're soldering onto? Or is it the ground plane, and you're soldering on the other side for some reason?

[aneevuser]

This does exist. In a professional environment you have “Lego bricks” for RF which are individual boxes with single functions (mixers, VCOs, filters, eval boards, test gear etc) and SMA cables to glue them together. This is large and beyond the realms of the average bank balance unless you go digging around on eBay and hamfests.

I guess this is the kind of stuff that Shahriar of The Signal Path often uses? If so, that's way beyond my needs (and pocket).

[aneevuser]

I use these extensively for prototyping RF & mixed signal.

https://www.busboard.com/surfacemountpcbs

Presumably it's the ground plane that gives the added goodness over straight veroboard? I don't see much difference otherwise.

Edit2: Apart from some baseband and LF applications where S/N is unimpotant, for most RF applications, solderless breadboard is a problem waiting to happen. Embrace and hone your soldering skills, and preferably switch to surface mount!

Hmm, surface mount. I've always been scared of soldering that. What's the benefit in the RF realm anyway? Shorter leads + possibility of a ground plane on the other side?

[tggzzz]

I use these extensively for prototyping RF & mixed signal.
https://www.busboard.com/surfacemountpcbs
In particular:
https://www.busboard.com/SP3T-50x50-G
and
https://www.busboard.com/SP3-100x100-G
They have a solid copper groundplane and non-plated through holes. They also do PTH versions but I find that is a bit restrictive on layout at times, and it doesn't take more than a few seconds to run a through wire.

Precisely, in all respects.

I've used the variant with some dedicated SMD pads https://www.busboard.com/SMT3U

Edit: I wouldn't ever use wire wrap for RF, wire wrap is an antenna.

Edit2: Apart from some baseband and LF applications where S/N is unimpotant, for most RF applications, solderless breadboard is a problem waiting to happen. Embrace and hone your soldering skills, and preferably switch to surface mount!

Precisely, in all respects.

Solderless breadboards are the work of the devil; you always end up spending more time/brainpower debugging the breadboard than debugging your circuit.

As for wirewrap, the OP might like to consider the standard construction of microwave filters and compare them with wirewrap pins. He will find the pins are stubs that will act as capacitors at certain frequencies. And don't forget that every mm of wire/pin is ~1nH; that adds up with modern digital circuits and with linear RF.

[bd139]

Maybe I'm being dimwitted, but I can't see what you've done on the solid copper board in the image - it doesn't look like it's been etched in any way, so it looks to me like a continuous conductive layer of copper. This is the side that you're soldering onto? Or is it the ground plane, and you're soldering on the other side for some reason?

It's not etched at all. All the non grounded connections are floating in the air. The copper is just ground and nothing else.

This does exist. In a professional environment you have “Lego bricks” for RF which are individual boxes with single functions (mixers, VCOs, filters, eval boards, test gear etc) and SMA cables to glue them together. This is large and beyond the realms of the average bank balance unless you go digging around on eBay and hamfests.

I guess this is the kind of stuff that Shahriar of The Signal Path often uses? If so, that's way beyond my needs (and pocket).

Yes that's the sort of stuff. You can usually pick it up cheap enough in surplus but you can't walk up and build something unless you have a large library of bits.

[tggzzz]

I have one question in the meantime though: ideally I'm looking for an "RF equivalent" of breadboard i.e. a technique that allows rapid mounting and demounting of components (so no soldering) but that works reliably for RF. From what I've skimmed above, I'm guessing that this doesn't exist?

Minicircuits https://ww2.minicircuits.com/homepage/homepage.html sell both components and the components mounted in small boxes - for a price. Not an easy website to navigate, unfortunately.

For many "building blocks", you can get naked boards with components in the form of evaluation boards - for a price.

There are many vendors on fleabay that take a manufacturer's circuit from their device datasheets, and put it on a board - for a small price.

[aneevuser]

Maybe I'm being dimwitted, but I can't see what you've done on the solid copper board in the image - it doesn't look like it's been etched in any way, so it looks to me like a continuous conductive layer of copper. This is the side that you're soldering onto? Or is it the ground plane, and you're soldering on the other side for some reason?

It's not etched at all. All the non grounded connections are floating in the air. The copper is just ground and nothing else.

Thanks.

By pure chance, I've just been watching a video by w2aew re: Smith charts etc. Towards the end, he shows the board on which he constructs an impedance matching network:

https://youtu.be/IgeRHDI-ukc?t=645

This looks similar to your board above but he seems to use some kind of socketing for the caps and inductors - this is somewhat closer to what I had in mind when I started this thread - do you have any idea what he is using here? Whatever it is, he seems happy to use it at 14.2 MHz, so it looks like a promising approach for quick-n-dirty RF prototyping.

[bd139]

Think he's using gouged boards there. You have to use double sided board for that as previously mentioned

[tggzzz]

Think he's using gouged boards there. You have to use double sided board for that as previously mentioned

The "sophisticated" version of gouging is to use a hand-held dremel with a spherical ball tip. In my limited experience it is more controllable.

[bd139]

I can't get on with that. I use a Stanley knife, metal ruler and peel the strips off with pliers.

[ledtester]

I'm considering trying out mini diamond hole saw drill bits to create isolated islands.

[fourfathom]

I can't get on with that. I use a Stanley knife, metal ruler and peel the strips off with pliers.

Before I got the gouge I used the knife, and usually just made two 45-degree cuts along the same cutline, giving me an adequate gap without the peeling process.  More fringing capacitance, but hey, it's a breadboard and I'm going to have to tweak the capacitors anyway. For microstrip you do want a bigger gap, and  the peeling method works for that.

I also have used a drill-press as a router for this type of breadboarding.  And I recently got an assortment of those cylindrical diamond bits (available cheaply on-line as "glass hole saws"), but haven't tried them yet.  I still don't have a Dremel tool, but ought to get one.

Finally, once you are getting close to having a usable design (or even before), do consider having some PCBs made.  I have a simple RF filter / preamp design that I used KiCad to lay out and sent the files to JLCPCB.  Insanely cheap and quick, and now I have a board with enough options that I can use it from 1 MHz up to over 1 GHz, with filter-shape and preamp options.  I did use the gouge-board method to experiment first though.

[aneevuser]

One other question: I have no single sided board at the moment, but I have some veroboard. Will I have any joy using the copper side of the veroboard as a makeshift ground plane by bridging the tracks with wire, and using dead bug construction on the other side? Or is there too little copper to make this worthwhile?

[bd139]

I wouldn't. It needs to be a solid sheet. Jumping it will create lots of capacitors in parallel.

I usually buy 20 sheets of this at a time: https://www.rapidonline.com/rvfm-copper-clad-single-sided-fr4-fibre-glass-board-100-x-160mm-34-0800

I cut with these: https://www.amazon.co.uk/dp/B00002N5KQ/ (also work on the fingers of clients who don't pay on time )