Breadboards - How bad are they at RF?

[JimB]

Over the years many threads have revolved around the poor operation of a circuit built on a breadboard, the type with lots of strips where components can be plugged in to make the connections.

The usual problem is when the circuit is expected to operate at more than a few kilohertz, strange things start happening.

When someone asks a question about this, all the experienced guys here will jump in and tell that the breadboard is totally unsuitable for use at "RF".
What I have never seen is a comparison between a circuit build on breadboard and the same circuit built using a more robust RF technique.

So, here is my attempt to demonstrate the scale of the problem.

I knocked up a circuit of an amplifier using a BFY90 transistor which is intended for VHF/UHF working and has an Ft of 1000Mhz (yes 1 Ghz).
The circuit is not exactly a masterpiece of RF design but it does work when build correctly.

To test the circuit I used a spectrum analyser and tracking generator, this effectively draws a graph of the gain vs frequency.

I first built the circuit on breadboard and got a bit of a surprise, something was very wrong, it would not amplify even at low frequencies, I then discovered that the thing was oscillating quite well at about 700Mhz, even with the emitter bypass capacitor removed reducing the gain considerably, it still oscillated.

Still working on the breadboard, I replaced the BFY90 with a 2N3904, and re-installed the emitter bypass capacitor.
The 700Mhz oscillation was gone and there was some useful gain at low frequencies, dropping to unity at about 5Mhz.
But now there was something odd at low frequencies, the "amplifier" was also oscillating at about 200khz.

I now built the BFY90 circuit on a piece of copper clad circuit board.
No etching or anything like that, just plain "ugly construction", (see the photo) a very effective way of building RF prototypes.
The support with the BNC connectors is a test jig which I use for things like this when I want to test a circuit or a component at RF, the BNCs make a convenient way of connecting the test equipment with low stray impedances.

The amplifier was now stable, no spurious oscillations, the gain was good, rolling off to unity at about 500Mhz.

Have a look at the pictures, I am sorry that some of them are not very sharp but I think that they show the analyser screen sufficient well enough to make my point.

So, what can we learn from this?


1 A circuit which works reasonably well when built with a good RF technique, will give problems on a breadboard.

2 RF circuits don't work well on a breadboard.

3 Even "low frequency" circuits can have high frequency spurious oscillations.

4 If you dont have the test equipment, how can you know what the problem is?

 

[MrAl]

Hi Jim,


Very nice presentation. I like to see things like this.

 

[MRCecil]

JimB,

A GREAT presentation. Gonna promote this to article! GAUD, I miss my old HP141T w/ tracking gen. from way back in the '70's. I just loved seeing that solder-ball in the sixth panel! A bit of nostalgia there when I was just a tech.

 

[Sceadwian]

There's only one thing majorly bad about breadboards that I'm aware of, and that is that at RF frequencies they have a lot of capacitance per strip and a lot of capacitance between strips at higher frequencies. At the really high frequencies the breadboard can actually be looked at as an antenna array, soldering deadbug style or wirewrapping is quiet practical and easy to do though.

By the way, that built 'ugly' is built right if you ask me =) Circuits don't have to be pretty to work well, and if yout put a power plug on that thing and put the case on it's about as good as it gets for isolation.

 

[transistor495]

That so called 'ugly' construction is indeed looks good for me. They looks pretty especially on VHF solid state circuits. I mean Manhattan style.

I work on all my RF stuff on breadboards only. So I've a neutral opinion on that.

But I think I need to re-think about that, probably finding a better way to experiment on the VHF stuff.

 

[Nigel Goodwin]

That so called 'ugly' construction is indeed looks good for me. They looks pretty especially on VHF solid state circuits. I mean Manhattan style.

I work on all my RF stuff on breadboards only. So I've a neutral opinion on that.

But I think I need to re-think about that, probably finding a better way to experiment on the VHF stuff.

You certainly should, as RF stuff won't work on breadboard.

 

[Ubergeek63]

1 A circuit which works reasonably well when built with a good RF technique, will give problems on a breadboard.

2 RF circuits don't work well on a breadboard.

3 Even "low frequency" circuits can have high frequency spurious oscillations.

4 If you dont have the test equipment, how can you know what the problem is?

1) of course
2) never
3) of course
4) you can't, sometimes even putting a probe on it is enough to kill your readings!

reason: at GHz frequencies every one of those component wires is an INDUCTOR!!! Inductance of a Straight Wire: A Calculator

there is a reason ALL circuits at these frequencies are "breadboarded" using commercial artwork with transmission lines in multilayer boards. Even the capacitors are just artwork in the PCB.

if you want some idea of what you are dealing with enter the whole thing into spice INCLUDING your wire inductances and resistances. it will not be accurate but perhaps give you a bit of understanding as to what is happening. you are talking about a circuit that can be tuned by moving a solder connection a tenth of an inch one way or the other.

 

[BeeBop]

Excellent! Thanks for taking the time to do this JimB.

 

[webpower]

This is awesome. I love the "ugly construction" I have just begun the adventure in electronics study and been doing stuff on breadboards and been wondering why nothing will work properly. This has helped. What is the network analyser you are using. I want to find one that is low cost? I am going to quit breadboard! "Manhatten Ugly Construction" here I come!

 

[JimB]

What is the network analyser you are using.

Beware of the terminology.

The device I used there was a Spectrum Analyser, which is basically a receiver which tunes between two frequencies quite quickly and displays the signal strength as a vertical line on the screen.

A Network Analyser is a very different animal.
Again it is a swept frequency device (usually), but it measures impedances (Resistance and Reactance) and displays them on a screen.

I want to find one that is low cost?

Good luck with that!
Good ones are not cheap.

The one I have there is a Hewlett Packard 141T mainframe, with an 8554L RF plug-in and an 8552B IF plug-in.
There were various plug-in units produced to cover different frequency ranges.
I also have an 8444A Tracking Generator, which is an external unit which connects to the spectrum analyser with three cables.

My analyser was built in the late 1970s and still cost me £900 when I bought it about 20 years later.

There are plenty of different type for sale, but for a good one expect to have not much change out of £1000 or more, a beaten up one will probably set you back £200 to 300.

JimB

 

[audioguru]

I use Veroboard (stripboard) with a compact layout and all tracks cut as short as is possible. No intermitternt contacts because the parts and a few short jumper wires were all soldered. Some circuits used many high speed 74LSxxxx ICs and many 74HCxxxx ICs and some were used at 100MHz. They all work perfectly.

 

[Mikebits]

A Network Analyzer is a very different animal.
Again it is a swept frequency device (usually), but it measures impedances (Resistance and Reactance) and displays them on a screen.

They do have S-21 function which works like a spec analyzer and Tracker. We would test filters and such this way. Our HP Net Analyzer was around $50K.

One place I use to work at in the 80's had 141T's they did work pretty well, and we had one UHF plugin for our 500MHz stuff.

 

[webpower]

Thanks for the info JimB. I initially thought they were the same thing. Now I have learned something from your reply. I now get it that one is for testing impedance matching at different frequencies. And the other is the analyser/Frequency domain receiver as what you have here. Many thanks for the info on that, as is it got me searching and was good fun googling about it. I have just done a theory course but as yet have next to no experience of building anything (LM386 amp only so far). So have a basic understanding of electronics, but I am an extreme green horn when it comes to the build process. Tons I want to do with lots of questions. How difficult would it be to build a spectrum analyzer. I am fascinated by frequency filtering with caps and inductors. So that would seem like a great relevant first major project. But I don't want to do it, if it is way over my current ability. Value your opinion on this. I would like to be able to hook up some output to a PC for example for the graphics to plot the frequency domain. So, the project would be to build some kind of fancy filtering and get that output into digital form to send to a PC. Is it possible to build something like this?

 

[Rbeckett]

This is awesome. I love the "ugly construction" I have just begun the adventure in electronics study and been doing stuff on breadboards and been wondering why nothing will work properly. This has helped. What is the network analyser you are using. I want to find one that is low cost? I am going to quit breadboard! "Manhatten Ugly Construction" here I come!

I scored three wire wrap guns on Ebay and found square pins by the hundreds and thousands so thats the direction I am going. May not work well for RF, but garden variety home projects like my brew troller will be just fine and save me from burning a PCB and a lot of soldering too. Dead bug, Manhattan, and wire wrap worked in the beginning, we just got lazy with the new tech that became available later. Sometimes a return to old skool helps a ton.
Bob

 

[Mikebits]

I did a lot of wire wrapping in my day. It is great for digital circuits. They sell boards made specifically for wire wrapping, ground plane one side, power plane other side.

View attachment 66913

 

[JimB]

Ever seen a backplane from a PDP11 mini-computer?
It is all wire wrap. I cannot remember if there was anything special for the power busses, but the address, data and control lines were just a sea of wire-wrap.

I have been doing some stuff with PICs recently and used a matrix board with holes and copper pads.
I used the continuous rails at the top and bottom of the board for the supplies and then just went point to point with soldered 30awg wire for all the signals.

It worked well.

Would it work well in an RF situation?
Most unlikely!

JimB

 

[unclejed613]

i used to work at a prototyping/ATE shop that used a LOT of wire wrap. be aware that if you use ground plane perf board with wire wrap, you also need a tool that clears the copper from around holes that contain wire wrap pins. it's kind of like a wire wrap tool, except it has teeth at the end that cut away the copper around the pin. we used to do power wiring by soldering #22 wire directly to the pins where power was required. several wire wrapped prototypes we made were used with ECL logic at or above 100Mhz. high frequency signals were routed using a special wire wrap coax, or rg-174 coax. some times with the ECL we also used twisted pair wire wrap wire, since many of the signals were differential.

 

[KJ6EAD]

...a tool that clears the copper from around holes that contain wire wrap pins. it's kind of like a wire wrap tool, except it has teeth at the end that cut away the copper around the pin...

You mean a pad cutter like this:
View attachment 66944

It's still available from Vector through Mouser and is useful for other prototyping work using copper clad board, not just for wire wrap.

https://www.electro-tech-online.com/custompdfs/2012/09/New20Page2075-78.pdf

https://www.mouser.com/ProductDetail/Vector/P138C/?qs=sGAEpiMZZMtArvHj40ttL8wa8MDisQN5C9ZLDsHpzu0=

I read somewhere that a typical breadboard is good up to about 10MHz if you use good strip line technique such as a ground plane base, short leads and grounded interstitial terminals.

 

[Mikebits]

If you use proper wirewrap board, you do not need a pad cutter. When connecting a pin to the bus you solder little round doughnuts from the pin to the plane.

 

[Grossel]

Hi.

I do agree that it is a nice article and a good test. However I noticed that the resistors and capacitors on the breadboard looks to be put together in a hurry. All the component legs have huge bends as well as the legs way longer than they need to be.

I bet if you've used some time and had only sharp bends on the component legs, then it would solve some of the problems.
Therefore I thing the breadboard cannot be that guilty