Through-hole Op-amp Circuit - Immunity

Thread Starter

Mallard

Joined Nov 13, 2011
5
Hi Everyone.

I'm doing a little ultrasonic proximity sensor project using components/materials that I have lying around. Attached is the schematic for the ultrasonic receiver amplifier which will be connected to the ADC of a micro.

It all functions properly. The micro sends about 10 cycles (of 40kHz) through a transmitter and the receiver section amplifies and detects the envelope ready for the micro to analyse. (The output is currently disconnected from the micro.)

So far so good. At first, I had it prototyped on breadboard to check the design with the actual transducers. Everything still functioned but my mobile phone rang and I was reminded about EMC as the output started to jump about!

I thought it would be interesting to see at what point the design would become immune to RF radiation from my mobile phone.

So now I have the circuit on stripboard! But I have laid out everything with the tracks AND components on top. I trimmed the pins of everything so that they don't poke through the board. Also, I have cut out a piece of copper-clad board (slightly larger than the stripboard outline) and stuck it under the stripboard as a ground plane.

All connections to 0V are now made via holes drilled to the unspoiled copper plane below.

Immunity is a LOT better. With my phone in front of the transducers there is no interference at all. With my phone above the components, the op-amps are still affected.
---------------------------------------------------------------------
ALL your thoughts on the following points would be very interesting:

1. Is it possible to make the circuit immune using through-hole components on a 2-sided board (WITHOUT sheilding!)?

2. Can it even be made immune with such an old bipolar, leaded op-amp in the design?

3. Is there anything you would do to improve the circuit design to make it more immune, whilst still using an LM324?

I look forward to your opinions very much. :)

Regards!
 
Last edited:

Thread Starter

Mallard

Joined Nov 13, 2011
5
Hi Bertus,

Thanks for your reply. :)

I have a 470nF and a 100nF ceramic (X7R) very close to the LM324 positive supply pin (return straight down to ground layer).
On the 3v3 regulator for the micro I have a 22uF tantalum and a 100nF ceramic (X7R).

One other point I did not mention (SORRY!) I have the op-amp running off the 9V at the moment (and not the 3v3, which is exclusive to the MCU).

Regards.
 

bertus

Joined Apr 5, 2008
22,270
Hello,

It can be that the powersupply leads pickup the signal.
A choke in the powerlines, as descibed in the PDF, might help.

Bertus
 

jimkeith

Joined Oct 26, 2011
540
Design a SMT board, keeping everything as small as practical--keep the back side for ground circuit as much as practical--put ferrite beads on input and output leads--my experience is that outputs tend to be far more sensitive to RFI than inputs. Post your schem and perhaps we can offer more suggestions.

I use the free layout software offered by http://www.expresspcb.com/
Very friendly software
Three boards cost $51 + shipping
After you do this once, you will never use thru-hole again
 
Last edited:

Thread Starter

Mallard

Joined Nov 13, 2011
5
Thanks for your considered input, SgtWookie.

I appreciate the high number of factors that influence EMC. After a while though, people tend to "get a nose" for such things.

I've implemented quite a bit of best practice into what is not a great starting-point (stripboard!)

In the past I have had circuits go through testing without problem. But they have used more modern, surface mount op-amps and resistors. In each case I have used bypass capacitors and a nice ground plane only. No shielding and no chokes (except for on inter-PCB connections).

-----------------------------------------------------------

Really, I suppose I was just wondering if anyone had ever put a leaded, bipolar op-amp design successfully through modern EMC testing, using two PCB layers only. And without shielding. :)

-----------------------------------------------------------

"At what point would the design become immune?" The big question; and I wasn't really expecting a definitive answer. More a discussion or throwing about of ideas. I just think it is very interesting.

For example, I have tried replacing those huge resistors with SMD versions (0805) and they made no 'discernable' improvement on their own of course.

-----------------------------------------------------------

I have read about bipolar op-amps being more susceptble than CMOS op-amps but does that rule out their use unless you have expensive sheilding?

Bipolar op-amps still have advatages over CMOS ICs. But could my stash of old LM324s be totally useless? Could a design with them EVER be immune using those two layers? and no sheilding... :)

I find it very, very interesting.

Regards!

PS shall I put you in the "NO" camp? (get a better, smaller, SMD op-amp)
 

Thread Starter

Mallard

Joined Nov 13, 2011
5
Hi Jimkeith,

Thanks for your input. All is welcome and interesting.

I have put a fairly crude but very full ground plane on it. This has certainly helped. When my mobile phone transmits towards the face of the plane, there is no interference. Even though all of the wire leads are still exposed. I wonder if PCB vias would be better than the bits of wire I have at the moment? Then there is the PCB dielectric material and width of course...

Regarding the output, yes. It is currently not connected to the outside world for the reason you have given. Just to the diode/capacitors of the envelope detector.

PS I have already included a schematic for the Receiver and also some nice photos!
-----------------------------------------------------------
So, I will put you in the "NO" camp too? Really need to use SMD.

(I recently downloaded DesignSparkPCB from RS. Seems OK for free. Thanks for the tip though.)
Regards. :)
 
Last edited:

MrChips

Joined Oct 2, 2009
30,714
If you are just making a one-of-a-kind prototype I can suggest a way to go SMD without having to etch a PCB.

Start out with a piece of double-side copper clad board. Use SMD components only, including the opamp IC.

You are going to retain the bottom side as a full ground plane.
On the top (component) side, cut channels, 0.5mm wide using a sharp Xacto knife. Basically, you are going to create square or rectangular islands of copper on to which you will solder your SMD components. Plan this out with pencil and paper first. Keep all the components tightly spaced.

You can get this done in a couple of hours (assuming you have all your SMD components at hand).

I have done this very successfully with high frequency signals in nuclear physics alpha particle radiation detectors.

In tough situations, I use single opamps per chip to avoid cross interference. I also find it easier to lay out single opamp chips rather than multiple opamps in a single package.
 

MrChips

Joined Oct 2, 2009
30,714
Here is an example of my handmade SMT board:



This is a charged particle detector with peak detect, sample and hold. Operating frequency bandwidth is about 100MHz.
Starting from the right is the BNC connector to a charge-coupled preamp. You can clearly see the supply decoupling capacitors, 0.1uF and 10uF in parallel across both positive and negative supply rails. The 100-ohm axial thru-hole resistors were added to help to decouple the preamp power from the rest of the circuit. Also see the decoupling caps at every IC supply. These are mandatory. You do not have a choice. We keep stressing the importance of proper power supply decoupling (or bypass) capacitors.

You can also see that I use single opamp packages because they are easier to layout and present less interference and crosstalk.

On the periphery of the layout is a wide ground plane. The entire bottom side is also a full ground plane with thru-hole jumpers connecting the two sides.

After first planning the layout with paper and pencil, I use a fine black pen to draw lines on the copper laminate to help me keep my lines straight. If you do not do this your cuts will be all over the place. After you cut the outline of the channels, you roll up the copper strips as if you are rolling sod off your garden. The copper strips peel off very easily.

I try to choose SOIC chips and 0805 components.
 
Last edited:
At the moment ? The lm324 wont run off 3v3 as it isnt rail to rail input or output.
It will work. You will get about 1.5V output signal swing. Which might well be fine for the application and the user.

You are all just deviating from the main points of the OP. For a change.... Why do people do that! Always with the irrelevant, "listen to me" deviations.

I haven't seen anyone actually answer the questions posed but everyone has to have their say.
 
Top