Debugging 101

[falc0nhoof]

As I'm cheap and have "some" knowledge soldering together guitar pedal kits/stripboard layouts I made the move to eurorack modules. As I'm putting them together I find my success rate is about 50/50. Some have worked fine straight away, while others I've found a simple fix (wrong cap, incorrect IC direction etc.)

For others ... no such luck. They're still relatively simple modules but I just can't get them to work.

Does anyone have any simple guides on how to debug a DIY project, be it modular or otherwise?

[NoiseWiki]

Well other than what you've described I'd grab a volt meter and use the continuity check to try to make sure the connections were being made through the traces. Note that if both sides of a path connect you won't know which end is open.

The other is use the voltage meter to check if voltages are what you are expecting or present at specific points in the circuit.

The other is to carefully recheck each component to make sure they are correct value.

[crochambeau]

Debugging/troubleshooting is a developed skill. If you're tired or frustrated walk away, don't worry, it won't fix itself; you can return to it later.

I say that right out the gate, because in my opinion a fresh and open mind is the most important tool in the process.

I believe there are books on this subject, so all I'm going to put here is a little freewheeling spitballing. It is by no means a complete document.

Defining your failure mode can be helpful in deciding the order of operations, but bear in mind that process takes place in a pond full of red herrings. Never conclude, always suppose - this extends to a "fixed" object. Conclusions cloud the mind.

• Trace power. Follow the power rails from inlet to destination (a voltmeter should be the most primitive tool under use in this process). Do not forget "common/ground/0v" as a power rail, it is equally important and often the reference from which every single stage and aspect (such as signal) operates.

• Trace signal. I like to use an oscilloscope for this (and tracing power), but most people are not always within reach of such an instrument - so an audio probe will work. An audio probe amounts to a NON-POLARIZED capacitor affixed to an audio cable or jack on one side, and a pointy metal probe on the other; there is also the common/shield/ground connection which should amount to a simple wire with alligator clip you can connect to the COMMON on the circuit under test. Then it's just a matter of plugging the audio output side into a functional amplifier of some sort and listening to various parts of the circuit.
  
  I like to start in the middle (from the schematic or signal path standpoint), that way I have an idea of which direction a given fault lies.
  
  Do note, again, that there are pitfalls to making any conclusions at this point. Example: the input node of an inverting summing amplifier built upon conventional operational amplifier technology will often read as "0 volts" and not produce a sound through our snazzy audio probe. You'll read signal up to a resistor and then nothing on the other side, but changing this resistor solves nothing - that's because this node is held at 0 volts by a flow of current, the intensity of which changes in relation to the signal - and is in essence invisible without the current to voltage conversion that will happen across a resistor. Simply jump that point and read downstream, etc..

Start by analyzing operational circuits, so you know what to look for. Stuff like bias voltages are fundamental to operation, yet are not a pure reflection of the original power rail. There are patterns that you will learn, with familiarity of such patterns it becomes easier to spot points in which typical standards of function are not engaged, then troubleshooting simply boils down to answering "why?" and resolving the equation.

Hint: it's usually something incredibly stupid or simple.

All parts can fail.
With polarized parts (actives/semiconductors & capacitors) orientation of polarity is important, especially with caps. Many types of electrolytic caps will operate in reverse (so long as the voltage potential is not too large) and only fail over time.

If you're doing a lot of discrete transistor work, get a tester. There are some part numbers in existence that are EBC and BCE depending on either one trailing alpha character (or maybe even era of manufacture, I'm speculating but I know what I've seen).

Anyway, hope that helps.

[falc0nhoof]

I sort of posted this and forgot about it for a few months but I must say the two responses were pretty awesome.

First thing I did was get an oscilloscope and second thing I did was step away from the projects for a week. When i returned I found two glaringly obvious problems that I must have stared at for ages only to find it within 5 minutes of sitting down after a break.

For many of the SMD projects (I had about 5 non working eurorack pcbs by then), these were fixed by simply reflowing the joints on the board especially around the main firmware chip. So that only left 1 which I used the oscilloscope to trace back to a bust up electrolytic cap.

What a rush!

[Indeterminacy]

Debugging/troubleshooting is a developed skill.

Finding out shit that defies logic theoretically but real world reality teaches you a lesson.


Fuck these fucking little pieces of shit.

Viewed 1849 times

[NoiseWiki]

Fuck these fucking little pieces of shit.

What is it?

[crochambeau]

Fuck these fucking little pieces of shit.

Bit by an octal packaged silicon rectifier?

Debugging/troubleshooting is a developed skill.

Finding out shit that defies logic theoretically but real world reality teaches you a lesson.

Illustrates malformed or incomplete logic, yes. Lesson learned? That is on the student.

[Indeterminacy]

Bit by an octal packaged silicon rectifier?

Debugging/troubleshooting is a developed skill.

Finding out shit that defies logic theoretically but real world reality teaches you a lesson.

Illustrates malformed or incomplete logic, yes. Lesson learned? That is on the student.

But when your signal out the waveguide/cable is spec on, you then have to go figure out what has happened
to the transmitted signal strength.

Boss you are not going to fucking believe this...........

[crochambeau]

Boss you are not going to fucking believe this...........

I had a similar experience (albeit on a smaller scale) with mice, cat food, and a tape machine.

[Indeterminacy]

Be alert for things that may not be causing an issue but quite possibly may.
If it looks wrong it quite possibly is.

Possibly is the operative.