electronic fault finding without schematic

[banjohamster]

Hi all,

I was wondering if anyone had any tips or techniques they use when trying to find a fault on a PCB without any circuit diagram. This is more out of interest, rather than me wanting help with a particular problem.

I've done a bit of reading on the internet, and there doesn't seem to be much on the topic, so I'm just interested in other people's ideas.

All the best,
Ross

 

[IRstuff]

"there doesn't seem to be much on the topic"

Is that really that surprising? How would that even be possible? Without a schematic, you wouldn't know the purpose of any given component, and therefore, you wouldn't know whether any observation is a valid or correct, much less whether the observation is of a fault or proper operation.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[MacGyverS2000]

What type of fault?

Dan - Owner

http://www.Hi-TecDesigns.com

 

[VEBill]

A significant percentage (probably not quite half) of the time the fault is visible upon close inspection. Burnt or overheated components; bad, overheated or cracked solder joints; dirty contacts, etc. So a *very* close visual inspection should be the first thing. Like 30 minutes or an hour under the microscope. I can't count how many times endless fault-finding led to a fault that was perfectly visible in the first place.

Checking 'the obvious' is quite easy. Correct voltages on the correct PS pins. Clocks clocking. Use an oscilloscope and examine the waveforms very carefully for anything unusual.

Depending on the complexity of the circuit, one can often create a schematic by reverse engineering the item. Obviously this is a non-starter for complex digital boards. But for small analog circuits it's pretty easy.

Looking for stuck outputs is also pretty easy. These are best measured with the card in situ and powered up. The fault symptoms can allow one to zero in quickly.

A 'Huntron Tracker' is a gadget that plots the IV curve for any pin. Damaged semiconductors show up very easily.

 

[VEBill]

"Without a schematic, you wouldn't know the purpose of any given component..."

Depending on the circuit, a good percentage of the parts have a self-evident purpose. E.g. 7805, filter capacitors, oscillator, driver chips, etc.

YMMV depending on the circuit. A "sea of gates" is hopeless.

 

[byrdj]

I had to verify calibration on a cabinet of cards with out drawings once. a lot of time was spent holding cards up to light so i could make my own drawings

 

[MikeHalloran]

If you can arrange full access to the card while powered, you can inspect for abnormally hot, or cold, components with a finger or an infrared temperature gun. ... after first verifying there are only low voltage circuits on the card, of course.





Mike Halloran
Pembroke Pines, FL, USA

 

[VEBill]

With miltary electronics, it seems to be a tradition to draw the schematics in such a convoluted fashion that one cannot recognize even the most-common circuit building blocks.

 

[automatic2]

First of all, you need to understand the circuit enough to know that there is a fault, and that's it not a mode of operation. I generally check the following unless evidence indicates otherwise;
- power, is the device powered.
- I/O, my experience has been that mechanics fail before solidstate. That means confirm that I/O devices are functioning as intended.
- I/O isolation. Do signals propogate thru isolation.
- I/O drivers. Are output devices being driven at ratings.

Beyond that will depend on the characteristics of your circuits. A scope is very handy for confirming the existance of a system clock, or tracking an injected signal through an analog stage, etc. Paper, pencil, and a good eraser are always required.

 

[banjohamster]

Thanks for the responses everyone!

I suppose my query comes from the fact that I'm often brought a piece of kit that's developed a fault with the expectation that as I'm the electronics guy, I can fix it. I'm sure we all realise how little technical documentation there is with small pieces of kit. Or at least this has been the case in the relatively short time I've been working in the field (I'm guessing some of the more veteran techie types would tell a different story).

In response to IRStuff, I think that is a bit surprising. I would have thought that my query was a common one, and that this sort of thing might have been a skill in itself. I guess those bits of kit with no service information are destined for the scrap pile!

Thanks all,
Ross

 

[VEBill]

See also 'Logic Probes' and 'Pulse Injectors' for digital circuits.

 

[IRstuff]

Aside from the fact there there is often sheer and complete ignorance about what a circuit might be doing, there is also often a safety issue.

Having a oscillocope is all very well and fine, but where to hook the ground lead? Without knowing anything about how the circuit is configured and hooked up to the AC mains, you could very well blow the crap out of the circuit, and at least, blow the crap out of a few ICs that might have been fully functional prior to one's ministrations (been there, done that). A 741 op amp whizzing by your eye is a very memorable experience and not one to be blithely repeated.


TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[VEBill]

One should start with a DVOM (battery powered, plastic case) and measure the power supplies *first*. In addition to confirming the basic PS function, this should also quickly and safely reveal if the circuit is wired-up strangely with respect to PS grounding.

 

[OperaHouse]

Form follows function. After basic power checks, define what the circuit is supposed to do. Start at a known output and work back. Always solve the simplest problem first even if it seems unrelated. That almost always will solve the problem. Almost nothing I work on has a schematic. Tracing out just a small amount of the circuit is sometimes needed.

 

[IRstuff]

Seems to me that you're saying you creae the schematic on the fly, either on paper or mentally.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[sibeen]

It must be fairly easy, I see it done in the movies all the time.

Like I'd imagine most here have happened to them someone will hand you a piece of equipment asking for you to have a look at it. I've even fixed a few of them over the years and in the majority of those cases it has been visibly obvious or a power supply issue.

I have had equipment where I've been offered a job to fix xx number of said article. In that case, and if it is financially lucrative, I'll take the time and draw out the circuit, or as much of it as I suspect is required.

I still wish I was as good as those bomb disposal experts in the movies...I'd be rolling in it.

 

[BrianG]

As stated by others, apart from the basics like checking power supplies, form follows function. I have fault-found quite a few circuits from looking at the components fitted and knowing the expected function. If you have no experience of the type of circuit function you probably won't recognise typical circuit implementation or the likely types of devices to be used(e.g. special function chips such as D-A, A-D converters).

I'd say that ultimately it's really all down to experience: once you have an idea of function you can check around transistors and discrete components, pins of gates for activity - I can remember the pin-out of dozens of standard logic gates - or if you don't know the devices that well then you can always Google for data sheets.

 

[OperaHouse]

Lets take an example of something I just worked on, a dead LAMBDA LLS DC power supply with no illumination on the display. The steps I took demonstrate a process of analysis. It covers a lot of areas because the cause was exceptionally rare.

First looked for anything obvious, burnt or leaking. Then I looked to see if there was rectified line voltage, there was. Next looked for voltage on the control circuits because the display didn't light up. That would be supplied by small line transformer or small switcher chip. No power transformer so it had to be the smaller of the two switchers. 300V DC on the FET so the switching transformer was OK. It used a 2845 switcher, downloaded the data sheet to get pin outs.

I started out when almost everything was discrete. With major functions incorporated into chips, these datasheets are the quickie schematics with typical applications and voltages. Most designs don't vary much from the application notes. I sketch out a little bit of the circuit just to identify some test points like common and power. This tells me which ends of components to clip onto and gives me a place to jot down numbers. Without a paper trail you end up repeating the same tests and forgetting the numbers. It is just not possible to keep all the details in your head.

Checked gate of FET and no drive. Also checked gate resistance, bad FET almost always seems to have a shorted gate. About 8V for chip power, real iffy and about chip low voltage shutdown. Checked resistance V+ to common maybe a leaking cap. Going around the chip the voltage on the 5V REF was only 1V. That chip must be bad. I probably would have replaced it without thinking if it wasn't so hard to get to. I was still wandering about the chip power. Data sheet said nothing about 5V REF changing in a low voltage shutdown. There was also no sign of oscillation on any pin of the chip. Line switchers generally start with a little tap off of a little line voltage through a large value resistor and hen power themselves with bias winding. Those first fractions of a second at startup are a gray area in documentation.

Did a resistance check of diodes connected to that switching transformer. Besides powering itself it appeared to supply + and - voltages for op amps. Any shorted diode could dampen the output enough so the bias supply never kicked in. I even bridged in a supply cap in case the installed one had opened. I get all my LCD TV's and home entertainment equipment courtesy of manufacturers who skimp a few cents on electrolytics in power supplies. I never used to check power supply caps like I do now!

There was also a SMT op amp powered off the same V+ close by. Maybe that chip was bad and pulling down V+ power. It was used as a comparator to monitor line voltage. Why didn't they just use a two input switcher? I figured, pull one of the two inputs low and the output of the op amp will switch state. If the output flips, that will prove the op amp is still good. Pin out was the standard 2, 3 and 1. One pin had the 1V from the 5V REF of the switcher. Low but still high enough for what I was doing. The other input that I was going to bring to common was at nearly 0V already! This was connected to the 320V line power through a 562K resistor forming a voltage divider that should have produced about 8V. The output of the op amp was pulling the COMP pin of the switcher chip to common, a way to shut it down. This was to prevent the power supply for operating in low voltage conditions. Cut one lead of the resistor and made a resistance check. Replaced the failed open 562K film resistor and everything worked. Film resistor failure is very common in applications where surge pulses are common such as in series with a capacitor in a line operated device. Resistor failure is quite rare in a voltage divider with smoothed DC and conservative dissipation.

Defining a problem is a process. It is ok to make a few simple checks and follow some hunches initially. Most problems are the result of bad design causing components to operate on the edge. One can fall into a routine of going after the usual suspects. Totally random component failures require a good understanding of the circuit function. Breaking a circuit down into small sections will eventually end in results.

 

[itsmoked]

That is crazy rare unless that resistor ran hot. I'd probably bump it up to a 1/2W if it was a 1/4W, since it's running at about 80% of its power rating.

Thanks for the trip down Troubleshoot Lane Opera! Loved it.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[KiwiMace]

My experience is that a stupid amount of faults occur in the power supply, then mechanical failures come next. I usually look close, meter the supply, then manhandle the board until it gives up its secrets. Failing this and a Google search for known issues, I flick it on to real electronics folk or value for replacement.