"Retro Repair" key electronics skills?

[Swift Griggs]

On Tue, 10 May 2016, dwight wrote:
> In order to trouble shoot, one needs to know how it is suppose to work. 
> This often means studying schematics, data sheets and sometimes even 
> app-notes.

Cool. I'm good on this. I love schematics when I can get them. I can't 
always figure out what I'm looking at but it sure helps vs looking at a 
board!

> Avoid the replacing everything, until it starts working, type trouble 
> shooting. As well as being a waste of time, it is more likely that you 
> will introduce new problems in the process. Don't replace a part unless 
> you can prove it to yourself that it is the most likely source of the 
> error.

This strikes me as very good advice. I use the same type of mentality when 
I troubleshoot computer components and servers. I'm sure it's just as 
sound when dealing with components.

> This usually means running experiments. Since you are a coder, I 
> find that anything with a working processor can be used as a self 
> debugging tool.

I love that phrase "Self debugging tool."

> EPROMs or front panels are great for trouble shooting. Simple test are 
> best.

I'm a Unix guy. Simple is almost always better. :-)

> Many fear power supplies. Linears are the easiest because they are 
> always some form of feedback loop. You just follow the loop until you 
> find two points that are opposite directions from where the input point 
> predicts the output point should be( goes back to knowing how it should 
> work).

Okay, I'm one of those with a little fear of power supplies. It's little 
because I'm still ignorant (*grin*). Funny you should mention power, I'm 
studying this section on DC metering to build up to learning more about 
power circuits:

http://www.allaboutcircuits.com/textbook/direct-current/chpt-8/what-is-a-meter/

BTW, I'm growing pretty fond of this instruction on allaboutcircuits. They 
have some really great material with excellent explanations & drawings.

> If you can't find a schematic on the web, draw one. As a minimum, 
> have a block diagram.

I keep reading that. I'm taking it to heart, mainly because I can barely 
keep up with one part of a circuit before having to move on and figure out 
a different part. Writing it down is the only way I could do it anyway. :)

> I'm sure you have heard of the "scientific method". Trouble shooting is 
> just that. It is the repetitive process of making an educated guess as 
> to the source of the problem and then having an experiment to prove it 
> either true or false. Try to have experiments that are conclusive.

That's good advice, in general, I think. I try to do the same with any 
type of troubleshooting. Even the Monte Carlo method (which I think rocks) 
is usually framed as an controlled experiment.

> Learn to use a two channel oscilloscope with trigger and delayed sweep. 

Great! I've been looking into that. I'm guessing that helps when you are 
trying to look at a detailed edge/level and you can overlay the two wave 
forms? Also do you use variable holdoff much in real life? I'm reading 
about how it works, and it sounds useful in theory (and if I buy a scope, 
it *sounds* good). However, at my level, I'm not sure I'd ever use it. It 
just keeps coming up a lot when reading about scopes. 

-Swift