Rev 4 of a complex PCB and suddenly I have a weird problem I can't understand

Other Hardware General Electronics
1

I have been designing a custom in dash pod for my car for quite some time now, and over the course of my learning and building, I've slowly built up a reasonably complex PCB design to house 3 teensy arduinos to control input/output, motor servos, back lighting, OBD communication and other bits and pieces.

I had got to rev 3 of my main PCB and pretty much everything was working great aside from a couple of inputs. I amended the design by changing out the input/output circuits after some advice from this forum and then ordered what I hoped would be the final design.

Unfortunately what I thought would be a simple tweak clearly has an error in it as when the PCB arrived, I plugged it in to my power supply and noticed that instead of sitting on a few milliamps as the 5v supply warmed up, the current draw shot straight to 4 amps, with the voltage pulled right down from 12.7 volts, to around 3 volts.

So what appears to be happening is the voltage is being pulled down massively while the current shoots up - not good!

Sadly being a mechanical engineer, I'm struggling to understand why this might be happening to be able to rectify the design. I've pulled up my revision 3 and revision 4 designs and can confirm that only the input output circuitry has changed, so the problem must lie within these items:

problem
problem956×798 106 KB

The new items are 3 input variants which are repeated a number of times each. I've not yet fitted the teensy arduinos to the board, and have only connected the 12v and ground wires from the power supply, so the arduino input and outside inputs are all not connected in this instance. The problem therefore must be inherent in one of the input circuit arrangements.

Here are the 3 in a little more detail for clarity:

problem2
problem2718×997 85.2 KB

The 3v3 supply comes from one of the arduino boards which isn't fitted, so the 3v3 line is also considered not connected at the moment (though it makes no difference if I install it). The 5v supply comes from a PCB mounted DC-DC converter so is present. Looking at the 3 circuits, both 2 and 3 only have 3v3 supplied, and since these aren't connected I can probably say that the issue is coming from circuit 1..??

I'll attach both the rev 3 and rev 4 Eagle designs for reference. I would be very grateful if someone with more circuit knowledge than I could help identify the problem, or even give me some idea of where to look closer.

I appreciate that ordering the board without testing this new feature probably wasn't my finest idea but I figured it was a simple addition to a working system and therefore low risk (sod's law... doh!)

Rev 3 Design.zip (122.5 KB)
Rev 4 Design.zip (131.7 KB)

2

Sounds like a short.

You have rev3 and rev4 boards so is there a resistance difference across any of the power rails, PSU removed of course ?

3

Thanks very much for the reply! How would I go about tracing it?

If I put my multimeter across 12v and GND I get no reading on ohms, but also no beep on continuity (though it does show 1.007 - not sure what units this is though).

From 5v output on the DC-DC converter to GND I get 16k ohms and no beep on continuity but 0.434 on the screen (again don't know units for that).

How might I go about tracing the short?

4

So too much current draw. If not a short then you have the power supply connected the wrong way round.

5

Or possibly one of the 'new' components has reversed power connections.

6

Do you have bare PCBs spare, as in PCBs with no components?

Start by applying power to a bare PCB, is there a short?
If yes, then you can either carefully trace every track looking for the short or you can leave the power connected and wait for the tracks to blacken and possibly go up in smoke, then you will know where the problem is.

If the bare PCB does not have a short then solder a few components on and repeat the test. Start with power related components first. Keep adding components until you find the problem.

7

If the board has components installed, do not overlook the possibility of a bad part, new parts can be bad. Also check to make sure the correct parts were installed.

At that amount of current draw, something will be getting hot very quickly.

8

Did you test the 'new' bare board for shorts, before adding components ?

Did you build the 'new' board in stages as in adding components in a logical sequence, for example power supply first, and then testing after you added each component ?

(hint)

9

I have two boards, and both are fully assembled so I can't test the bare PCB. They both behave the same way so I'd say it's unlikely to be faulty parts unless it was a batch of faulty parts.

I tried running the power for around 30 seconds to see what gets hot, and no sign of black traces and I don't feel anything getting hot other than the 1N5400 diode that sits in between the incoming 12v and the DC-DC converter. The DC-DC supply and TVS diode both don't get warm at all. Does this help point any fingers??

The components all appear to be the correct way around and as per the previously working Rev 3 board...

EDIT: I tried powering the board up via 5v directly on the DC-DC converter output to cut out the 12v part of the PCB and the same problem occurs, though the TVS diode now gets hot... Trying to work out what that means??

10

Have you measured the voltage on this? If it gets hot it means it is being triggered, that is it has a voltage on it which is too high. Could be a wrong part.

Back in the 60s on courses for teaching people to mend TV sets, part of the practical exam involved replacing components with incorrectly marked components, maybe yours are too. I would temporarily remove the diode by cutting the track, it can always be bridged back in circuit later.

11

I would also suggest you start removing key components to see if the problem goes away.

And yes back in the day, I used to find such faults by cutting key tracks with a very sharp scalpel. You then slightly fold up one edge of the cut to break the track. You can make good afterwards by folding the cut track back down, remove a bit of the mask, and bridge with solder.

12

Hi, @450nick
Can you please post both your schematics and both your PCBs as EXPORTED jpgs.

Not everyone has Eagle.

Thanks.. Tom.. :smiley: :+1: :coffee: :australia:

13

Well done Mike you are right on the money! I looked very carefully at the part number on the TVS diode on the old boards vs new and it appears different. It should be a SMBJ170A TVS diode, though the old boards show a PN of B711 PR, and the new faulty one shows P836 PR - see below:

Solution

I've Googled both of these and can't find a datasheet for either of them so I can't be sure what the difference is. However I tried desoldering it from one of the new boards (kind of ripped off the track along with it but anyway it's gone. And the problem seems to have vanished so it looks like this is the problem.

I guess I'll need to go back to the manufacturer and see why they deviated from the stated BOM... If anyone is able to find out what the PN I have in the board actually is I would be interested to hear :slight_smile:

1 Like
14

Or the latest version.

15

If you dont have the kit for removing even simple SMT components, then to avoid damaging the PCB, cut the component up with a pair of very sharp cutters, this normally avoids damaging the PCB.

16

I went through that mill, in the late 70s.

C&G, Radio, TV, Electronics.

17

Hi, @450nick

What part number did you specify when you ordered the assembled PCB?

Tom.. :smiley: :+1: :coffee: :australia:

18

I specified a Bourns SMBJ170A (170V, 600W, SMB) in the BOM, though it looks like they substituted for an equivalent on the original boards which did the same job. It seems they switched it for a different one again on this version which is not the same function.

19

I had no problem finding datasheets for the SMBJ170A.

The first two datasheets I looked at (from different Manufacturers) both indicated that the "PR" marking indicates that the component is an SMBJ170A

Diodes Inc SMBJ Series datasheet

Littlefuse SMBJ Series datasheet

"PR" is also used by Bourns - the manufacturer you specified.
Bourns SMBJ series datasheet

Maybe some other manufacturer uses a different marking scheme.

20

That is helpful, but any idea what the 711 or 836 means?

EDIT: I've just realised it's also backwards... I guess the 711 or 836 doesn't matter - they installed it the wrong way round. Time to talk to the supplier!