Do circuits built using PCB and using bread board perform the same?

1. Jul 29, 2012

royzizzle

I want to build a EEG and the site tells me to buy a PCB from some manufacturer. Will my product have competent performance if built with breadboard?

What is so special about PCB？

2. Jul 29, 2012

yungman

It all depends, for RF, it is most likely not the same, for high impedance circuit, it likely not the same. For run of the mill circuit that is a few hundred KHz, impedance in 100KΩ or below, it is likely the bread board is close to the pcb.

Circuit layout is very important. This is one thing school don't teach you. You learn signal but you thing everything just magically going to the ground and is all good. In real life, ground is the second half of the circuit and it is just as critical. If you know grounding, circuit is just that simple, the circuit in the text books usually work. The devil is all in the detail, that's where people lost the war.

Also, wires in breadboard are much more susceptible to EM noise pickup as they hang in the air.

That said, there are different ways of bread boarding also. For example, for high speed circuit, we use ground plane board and "dead bug" the IC or transistors. You can get to a few hundred MHz with bread board, but it gets harder and harder to do. Now a days, making a run of pcb is a lot cheaper than 20 years ago. I have a few circuits that it's just too much trouble to build for demo, I am actually laying a pcb and combine a few of them on one layout. I did some inquiry on ebay from pcb house in Hong Kong, for 4 layer 30cmX30cm panel, it's only about $200 a run. I love to do it in USA, it's easy to say than to do. I got a rough quote from a friend, the same run would be close to$700. It is easy for people that don't pay from their pocket to say keep the job in America. But it it is your money and there is a bid difference in price...........

Last edited: Jul 29, 2012
3. Jul 29, 2012

Averagesupernova

It will also depend on the current involved. The breadboard will typically have more resistance on connections than a soldered up tidy PCB.

4. Jul 31, 2012

Mike_In_Plano

These biomedical circuits have a lot of gain, but little bandwidth. Hopefully, you're not intending to tie it into an AC connected device - few are rated for patient connection.

That said, yes, you can use breadboard techniques - if you're neat and thoughtful. By using a 1/4" strip of copper as a local ground, you can get fair performance from the front end amplifier.

5. Jul 31, 2012

the_emi_guy

Depending on the nature of the circuit it can go both ways ...

Most circuits are more well behaved when they are hugging a nice solid ground plane on a PCB (nice low Z return paths, low crosstalk ...).

Certain analog circuits (oscillators, op-amp circuits) have been known to work beautifully hanging in the free air, then fail miserably when hugging ground plane after implementation onto a real PCB (parasitic capacitance).

Of course we want that nice solid ground plane in all cases, but need to manage parasitic capacitance on a case by case basis (clearing plane under certain pins of certain parts).

6. Jul 31, 2012

yungman

Don't forget one most import thing, if your ultimate goal is to put the circuit on the pcb, doing the first try on pcb get you much closer to the final product, less surprise. As I said, a run of pcb is not that expensive. You can layout with spare circuits that you can try. When I do a proto pcb, I put in more poles and zeros compensations than necessary, I don't have to stuff it if I don't need it. By careful planning, you can put in a lot of "hooks" so you can experiment even with a pcb.

7. Jul 31, 2012

gnurf

In addition to what has already been mentioned, I'd add reduced size as a reason to use a PCB (using SMT) over a breadboard.
Do you by any chance have an example schematic of a circuit that is so sensitive that even the capacitance (a few pF?) formed by a single pin and the underlying plane can be harmful? Thanks.

8. Jul 31, 2012

Averagesupernova

A place I used to work had the first LO in a measuring receiver air mounted. It was a negative resistance VCO that had a top end frequency of around 1200 Mhz. Placement of components would adjust it into range so the PLL would lock throughout the whole range. It was a pain to tweak those components position. I recall capacitor leads were coiled and the spacing of those coils would trim the oscillator enough to bring it into range. It used a couple of varactor diodes of course and if those leads were too long or too short it would not be able to be brought into range. A pain to calibrate but a much bigger pain to manufacture. It seemed to work very well once set up and had good stability over temperature and did what it was designed to do very well.

9. Jul 31, 2012

yungman

For transimpedance amp that can measure under a pA, the feedback resistor is like 10GΩ or more, one pF will make a day and night difference. Even for 1GΩ resistor, I had to cut out the ground plane around the junction and area close to the input.

Also, in microwave, 1pF is quite significant. At 1GHz, the reactance is only 159.23Ω!!! You tell me whether it is significant or not!!!

10. Jul 31, 2012

gnurf

Interesting story. I'm not sure coiling the capacitor leads would be the first thing that crossed my mind if such a setup failed! Until now anyway.
Yeah, that makes sense. I flipped through some transimpedance amplifier app notes and sure enough they recommended to remove nearby planes. I learned something, thanks.

11. Jul 31, 2012

yungman

You're welcome. Actually for 10G resistor, I don't even think pcb can do it. Not because of the capacitance, but the leakage on the surface due to moisture and dirt. We stand the input end on a teflon standoff. We use leaded op-amp and bend the -ve input pin out and solder onto the teflon standoff to the resistor.

12. Jul 31, 2012

Averagesupernova

The coiled capacitor leads were standard procedure. Every single one was built like that and the spacing between turns was adjusted to help bring it into range. It was part of the design.
-
I can also recall a design for a DVM that was integral to another piece of equipment that stood several larger resistors on end to avoid leakage current instead of having one single resistor laying down on the PCB. One end of each resistor was soldered to the PCB and the other two ends were joined in mid-air. It worked.
-
I also recall in similar DVM circuits having relays that switched taps on the main voltage divider on the input of the DVM. When the PCB went through the wave solder machine flux would get trapped under the relay and cause leakage. Now the input voltage divider no longer scaled the input correctly. A couple of volts off at 2000 volts was out of spec. Anyone care to guess how how we proved and tested for this besides just swapping out the relay?

Last edited: Jul 31, 2012
13. Jul 31, 2012

yungman

Don't need to prove, I can attest that too many cases of leakage that totally ruin the circuit. You have to be careful even for a few MΩ resistor. We have a special freon bath to wash the pcb and then use condensation freon to rinse the board.

14. Jul 31, 2012

Averagesupernova

Actually yungman it does/did need to be proven. The residual flux was only a problem in humidity so the fix for it needed to be proven well since we had no idea what type of climate it would end up in. It really wasn't a problem in the plant.

15. Jul 31, 2012

yungman

As I said, it depend on how high impedance. In my experience, some of the circuit will not work most of the time if you don't take precaution. Even through hole components, we washed a few times with flux cleaner and still it did not work. Try a 1GΩ resistor and you'll see.

16. Jul 31, 2012

Averagesupernova

What you actually said was:

And I'm not sure you are getting the point. So, forget it.

17. Jul 31, 2012

yungman

You said anyone can guess how you prove it. I reply that you don't need to prove, I've seen too many cases already and I am fully support what you said about the flux.

18. Jul 31, 2012

Averagesupernova

Yungman I don't understand you. I accept your last post and will not argue that you have probably seen alot of cases that are like what I described. I did not dare you to dispute what I posted. It seems to like you took my statement as that or that you were stating that such a thing does not need to be proven, which of course it does since there are many components that could be contaminated. I described what happened in the plant and asked if anyone wanted to take a guess at how it was determined that the problem was as I described. The exact wording was as follows:
I thought it would be assumed by the readers here that comonents involved with the front end of a digital voltmeter are very high impedance. Finding leakage at these high impedances is not such an easy task.
-
Is there some kind of language barrier thing going here? Did I not understand you? Did you not understand me?

19. Jul 31, 2012

yungman

We are going in circles, I repeatedly said I agree with you and said you don't need explain or prove. I back you up on your assertion from my experience as described.

20. Aug 1, 2012

Averagesupernova

You are right, you and I are going in circles. I have acknowledged that you agree with me which is not the point.
-
For anyone who actually gets the point:
Was wondering if anyone had an idea how it was proven at the factory what was really happening at the time, using instruments of some sort connected in some manner. Proving leakge that affects the front end of what I recall to be a 10 Megohm input impedance meter by less than a tenth of a percent is not easy. Sorry to say we didn't have yungman running around out there saying: You not need to prove nothing. I know it relays. Must be, I seen it billion times.