Layout doubts on prototype project

Hi everyone,

I´m building a project that I tested on a breadboard and it worked fine. But when I built a prototype PCB it didn´t work well.

I wonder if it was due to a wrong schematic or tracks layout.

Could anyone give an opinion?

I´m attaching prints of the screen and the original schematic for the project as well as my version on Kicad. I´m also attaching a zip file with my Kicad files.

This is the prototyped board. Do the copper tracks look ok? Or is there one or another that´s shorted or incomplete?

Prototype (890.5 KB)

Quick note (before heading out for the day…)

The Copper tracks on hardware look terrible! I suspect some discontinuity here and there… did you test them with meter?

1 Like

I used 0.3mm for signal and 0.5 for power.

I tested it out with a meter and some are showing O.L.

Should the tracks be thicker? What tickness do you usually use @BlackCoffee?

1 Like

It looks to me like all the problems here are with the manufacturing process, not the design.

Whatever process you are using does not look like it can handle traces as small as 0.3mm (which are not small at all by modern standards). If you can’t change your process, I would suggest using 1mm or larger traces.


Thansk for the suggestion @craftyjon. I´ll try it with the 1mm traces.

I believe it should work like this.

It’s been a while but it looks to me like you left it in the etching solution to long. Once the copper is gone, it starts to eat under the etch resist and into the traces. You end up with a lot of open circuits. Wider traces will help but also removing it sooner from the solution should help. Once the copper is gone, it needs to be removed pretty soon. Sometimes even minutes can make the difference.

This brings back old memories. :wink:


Just tack bus wire down on all the traces and get this board running before spinning a new one. Yeah, it is not a pretty board on the bottom, but you spent a lot of time making this so you should give it some love and bring it to life. If there looks like any hairline copper shorts scratch them off with the back side of an xacto blade.

BTW, it is so cheap and pretty fast these days to get just a few boards from jlcpcb, or pcbway, or myro, or other Chinese fabs, it is hard to justify etching (or even routing) a one-off proto. I did a lot of board etching decades ago, so I realize what goes into that.

Also, plop a ground polygon over the entire board – even a single-sided board. Will keep the signals much quieter, especially the high-impedance input. For sensitive signals like your input, the run should be as short as possible, and if ground plane fill encases it it will be very well-behaved. If the opamp output is oscillating (typically MHz range) you need a small cap across R3 (100pF is a decent size), as we discussed in your other thread.


Thickness and Width are not the same… (you know that)

For the past 20 yrs, I’ve CNC milled my PCB’s - several folks on various forums have asked me about doing it so, my CNC mill was $3k so, for recommending to others, I bought a very low-cost CNCmill and posted about it… thus, FYI Link to video

The Track Thickness is dictated by the PCB stock I buy (on Amazon).
The Track Width’s I typically use are shown on this example board

I tried the 1mm track width but it didn´t work.

However, I do see that the signal stops coming in on the GUI when I turn disconnect it from the computer.

Are the tracks any better now?

Tracks are certainly good enough now.
It is too simple circuit to not work if mounted on breadboard worked.

Where to start – wow, different board layout, no top side shown, no new cad layout shown, looks like some parts connect to pads with no traces (unless it’s now double-sided?), unused opamp is open (short 6 and 7, and ground 5, as pointed out in previous thread), your schematic improved dramatically in the last thread but is back to contorted version again here… It is difficult to help when progress goes backwards.

Signal gone from gui when disconnected? What does gui signal look like? Is it overdriven? Some soundcard thingy? Does it have level controls? Do you have a scope? Maybe opamp is oscillating. Once again, add that 100pF across the feedback resistor R3 to quiet it down. I realize that you are putting a bunch of time into this, and folks want to help but have limited time – perhaps peruse some resources such as:

1 Like

Don’t know where the Power is coming from - sounds like you’re using a control board (Arduino, ESP…etc) and did Not pay attention to it’s Requirements for External Powering (generally can’t mix external and internal-USB). They always have different Pins for USB and Vcc External.

CLARIFICATION: I do see you’re bringing Power+/- in via connector - and, your comment about no power/signal when unplugging from Computer suggests to me you’re using a controller/USB/etc. That’s why I mentioned it. Naturally, with correct power hookup, gizmo should run as Stand-Alone…

Check the Specification for your device.

Hi @teletypeguy, good to hear the board is better now.

I checked on Kicad and some pads that should be connected were disconnected, perhaps that´s why it didn´t work. I´ll try the new layout now and see because I know it worked on a breadboard. The filter capacitor was disconnected so it would not filter giving me a very noisy and unexpected signal. Also some of the power connections were disconnected but I still managed to power it with a battery or an arduino.

What about ground planes how should I use and design them? How do I know if they are needed?

This is what the GUI look like:

When disconnecting a constant line should appear or asystole. Yes, it uses the computer soundcard. No level controls. I don´t have an osciloscope.

In the moment I don´t have a 100pF, could I use any of the below?
100uF, 10uF, 270nF, 100nF, 22nF.

Should the capacitor be upwards or downwards in relation to the resistor?

If you don’t have a scope you won’t be able to tell if the opamp is oscillating. It may not be, but it is always good practice to add a cap across (in parallel with) the opamp feedback resistor – this provides a first-order high-freq rolloff. It is a simple calculation: the -3dB point is at 1 / (2 x pi x R x C). I explained this in your other thread so peek at that. If you think you can swap a 100uF in place of a 100pF then you need to learn some electronics fundamentals. Sorry, but this is elementary stuff. Also, your un-connected opamp could be oscillating (though is is likely just sitting at a rail. Dealing with an unused section has been mentioned more than once.

As for the ground plane, it it easy. Just make a rectangle polygon inside your edge cut (I come inside 0.5mm) and define it to be a ground net. Then go to Edit/Fill-All-Zones and badaboom-badabing it is just like magic. You can even be sloppy and make the polygon outside the edge cuts and the outer perimeter will get adjusted by the board margin.

1 Like

I see, so I´d need to calculate the C based on the f (60Hz) and R3, right?

I just didn´t connect the unused pins as mentioned in the other topic because I have a doubt about ground.
Ground would be BATT-, correct? If yes I can connect it correctly.
When creating a ground plane should I label it BATT- or GND?

I´ll take a look into ground plane tutorials, should it isolate my board better from noises?

Dude – this is getting old – said it all before.

1 Like

Thanks for replying @teletypeguy, I´ll refer to the other post.

I’m also missing a decoupling capacitor over the opamp, and a “decent” passivation for the unused opamp. Just leaving the connections of an opamp open is not a good solution. An opamp can do weird things when it’s not operated within it’s normal operating range.

1 Like

And I just gotta say, to everyone new to electronics, get a scope. If you don’t have a scope you are working blind on a circuit. Maybe that should be the first question to a new post – if you don’t have a scope to verify power, verify dc bias points, see if the signals look correct… you are wasting your time with any electronic circuit. Spend a few hundred bucks on an entry-level siglent or rigol, or even a cheesy hantek.

1 Like

You have to not understand what pico, nano and micro means.
If you were understanding you would say only one of those values.
100uF is milion times bigger then 100pF.
If some circuit with 100pF will allow you to see short pulse once per second the same circuit with 100uF will allow you to see milion times longer pulse once per milion seconds.

Not exactly.
Without scope - oscillating opamp takes more current than not oscillating one.
With scope - it is possible that when you connect scope to opamp its state (oscillating/not oscillating) changes.
In 1981/82 having no scope I designed and build chronocomparator. It is the device that when you put (only put) on its table electronic watch it shows its speed with a resolution of about 0.1s per day. My device was about 15x15x30cm in size. I got for it my first serious money I earned on electronics (2.5 times my father month salary).
Then I designed and build the oscilloscope because I really needed to have an oscilloscope and it was not possible for me to buy one (I still have it).
I was so proud of it that I send an article to electronic magazine and they published it.