Digital Troubleshooting and testing

[walters]

I'm just talking about digital troubleshooting in General to have good
rules to know and learn from but in General to get the concept down so i can use this to apply to any digital circuit that i test

Im just looking how to do testing,checking digital componts and circuits
just the basics but in general rules

Digital Troubleshooting rules:

1.) So first test the Voltages going to the digital components or Chips
2.) Test to see if the Clock pin on the chip is getting a clock pulse
3.) Check the inputs and outputs with a truth table with a logic probe
4.)

What else can i do test or check for?

 

[Thunderchild]

ok that is a very good line of action but if and this will probably often apply as a digital circuit in itself is useless unless it controls something there is something to be controlled like a motor via a relay ecc and say there is just no output then check all the power stuff like 1 the motor thats working ? then go to the relay that controlls it or the transistor. if say a relay is controlling the motor and the relay is being switch by a power transistor then check that often power transistors r not mounted on a sufficient heat sink and so if on for a while burn out if the power section is ok then move on to the difital control. if say the motor (or what ever else) is working but say at the wrong moment (eg the conrol circuitry is not doing its job right) then u may go strait to the digital stuff.
u see designing and making something is one thing - trying to find out why it wont work is: JUST A MUCH BIGGER KETTLE OF A FISH especially if u did not make it and so working out how it works (or was supposed to) is the first best thing and a schematic diagram is the best thing for that. i sometimes if it is not available draw it myself by tracing the pcb but that can get very complicated and i only do it on little stuff (or perhaps i should say would-yea i realy hate fixing a problem making the thing myself is just so much more peace of mind than trying to figure out what another guy got wrong :?: :!: :idea: :arrow: )

 

[ThermalRunaway]

In order to troubleshoot complex electronics equipment such as encoders and decoders you need to get hold of a logic analyser. Logic analysers allow you to monitor multiple data lines simultaneously by taking a "sample" of digital activity. Once a sample has been taken you are able to go back and view the activity that has taken place. Typically you can view this information as binary information, hexadecimal values, octal values or more commonly you can view the information in graphical form. The graphical form is usually a timing diagram which shows which data lines are taking high and low values and when the high-low transistions are occuring in relation to each other.

A logic analyser is invaluable for conducting these kinds of tests and is incredibly useful for troubleshooting your own digital designs. Logic analysers vary in complexity and price, ranging from small devices which can monitor low speed 8-bit systems to high speed equipment which is capable of monitoring in excess of 200 data lines at the high speeds associated with modern microprocessors. The price varies in accordance, from just a couple of hundred pounds for a simple analyser to a few thousand pounds for a good one. Top of the range analysers aimed at industrial and design applications can fetch price tags in excess of £20,000!

For basic applications such as those you are talking about, a simple logic analyser will do and will still make for an invaluable tool. You should take a look at the Pico series of logic analysers which interface to a PC. The PC interface means that you're not paying for a piece of equipment which incorporates all the control knobs, firmware and display screens typical of an "all-in-one" unit. A PC controlled analyser uses the PC's standard peripherals and screen for display and control which alleviates you of these costs. Simple analysers capable of monitoring 16+ data lines of speeds around 20Mhz can be picked up for a couple of hundred pounds, or even cheaper second hand.

Good luck!

Brian

 

[Thunderchild]

thanks brian u gave a better explanation of the logic analyzer than i could ever have given. i was unsure of the name because i studied electronics here in italy and realy wish it had been back in england so i sometimes have a specifics terms lingo difficulty

 

[walters]

Thanks alot for the information

Simple analysers capable of monitoring 16+ data lines to get timing diagrams

Digital Troubleshooting seems to be hard cause you have to monitor alot of clocks and timing diagrams, So i have 16 data lines and im looking at the timing diagrams and inputs and outputs of gates. How do i know if something does not look right beause there is so many data lines , clock lines, inputs and outputs to gate all i am seeing is On and off state at different pulses and waves they all look the same pretty much its very complex to me it just looks like a bunch of diffrent timing clock signals
on a 16 channel oscilloscope how do i troubleshoot from looking at the 16 channel oscilloscope looking at the timing diagrams?

 

[ThermalRunaway]

Well, unless you're actually the designer of the system, it's unlikely that you'll be able to look at a set of timing diagrams and say to yourself, "hang on a minute... the data at this point is not correct". If you're looking at a system someone else has designed, usually all you'll be able to determine is that yes, varying data is present on each data line or no, varying data is not present on a particular data line. Usually, if you have a set of varying data present on a data line it's safe to assume it's not faulty. You don't know that the data is correct, but the fact that data is there (and the pulse train varies) is usually enough to assume all is well. If you were talking about a system which you'd personally designed (which wasn't working properly) you'd be able to use the timing diagrams along with your own design data and IC data sheets to determine whether all the required control lines were being operated properly and with the correct timing. Timing problems can be amoung the most difficult problems to pin down for an engineer or hobbyist.

If a data line appears inactive for a prolonged period of time, it's possible (although not definite) that it's faulty. If it's stuck low for the entire capture period, it could be that this particular data line is short to ground - switch off the equipment and check with a meter. If a data line is stuck high for the entire capture period, it could be that the data line is short to VCC - switch off the equipment and check whether this is the case. On the other hand of course, it could be quite normal for that particular data line to remain in any one state for that particular capture period!

Digital fault finding is probably the hardest kind there is. Sometimes you find a fault (an i/o pin short to ground for example) but even then you're unable to pinpoint the exact IC that's causing the problem. For example, if you have IC1 which outputs information to IC2 via a set of i/o pins, and one of those pins is short to ground, how do you know whether you've got an internal problem within IC1 or IC2? The only certain way to tell is to remove one of the ICs (or at least remove the solder from the offending pin on one of the ICs) and see if the short dissapears. That technique is fine, but what if you're talking about a faulty data line which interacts with 3 or 4 different ICs? Which one is faulty then? And what if all the ICs are SMDs or worse... BGA packages? It really is very difficult

Brian

 

[walters]

Thanks alot this is helping me out alot to talk to someone about this
type of skill to so thanks


you're actually the designer of the system, it's unlikely that you'll be able to look at a set of timing diagrams and say to yourself, "hang on a minute... the data at this point is not correct"

Yes thats what i been doing because i don't know whats right or wrong
its really really hard to tell if the Clock or timing diagrams lines are shifted or has any varies

What would cause a Pulse train or Clock pulse to vary? or shift?

I know there is a Propergation Delay also from going in and out of the digital chips but would the chips make a clock pulse vary or shift ?

 

[Thunderchild]

that last post Brian must have made walter VERY happy hehe u could of spared him perhaps

 

[walters]

yea i really need the help about digital troubleshooting alot its kinda of complex and hard for me to track down by looking at the timing diagrams

 

[ThermalRunaway]

Ok so you're talking about a fault condition which causes a set of data to "vary or shift"? What exactly do you mean by that? Have you got a particular fault which you're unable to track down? If so, what equipment are you working on and what is the fault symptom?

Brian

 

[walters]

so you're talking about a fault condition which causes a set of data to "vary or shift"?

Yes but just in general so i can apply it
when the clock pulse is offset and it throwing off the timing of the digital chips to act faulty or not give the result or function right

How to track down a faulty data or clock line?

Have you got a particular fault which you're unable to track down?

Yes like a Digital chip but just in general how would a tech. track this
faulty digital chip ?

What are some General digital fault symptoms?

 

[ThermalRunaway]

Well you can normally tell whether a clock is faulty by sticking an oscilloscope probe on it, assuming you have one and assuming it has a bandwidth high enough to display the clock signal. If the clock speed were to drift off slightly, it could give some quite weird fault symptoms. In general, the timing should not be effected by a clock signal which had drifted from it's specified frequency. If the clock was slower than that which is specified, it would mean that the CPU (or whatever type of IC) would run slower than originally intended. Provided this drift was only slight, the control timing should remain unaffected. If the drift was such that the CPU could not operate fast enough to act on the control data then yes, you would end up with screwed up timing and without doubt, the system would crash. If the clock were to run faster than the specified rating, you could end up with a similar effect except that overclocking a CPU is dangerous and could end up actually destroying the IC itself.

Quite often a faulty clock signal will be caused by the oscillator or crystal oscillator itself. I've often had situations where the clock signal becomes erratic once the crystal oscillator has warmed up. In order to diagnose this kind of problem you can spray the oscillator with freezer spray to see if your fault clears. If it does, it's a sure bet that the crystal oscillator is causing the fault. crystal oscillators are also very prone to damage from shock.

Brian

 

[walters]

THanks ThermalRunaway

So to test a Faulty Clock is put the oscilloscope probe on the output of the
555 timer or clock generator and watches its Frequency and to see if its
varys or shifts in time?

How do i test or check faulty logic Gates : do i check every pin with a truth table ?

How do i test counters up/down or 4-Bit Binary Counter?

Do i test the Reset pins? to see if its a high or low state?

 

[walters]

I just put my oscilloscope probes on the BUS wires or lines to see it i get
any High or low states or clock signals

 

[walters]

How does a Tech. know if the "time intervals" are right and not faulty on the bus lines or going to the digital components?

How does a tech. know if the "clock cycles" are right and not faulty on the bus lines or going to the digital components?

 

[dingo]

I did this for many years.

For the most part you read the fault discription and change D7 because the last 500 boards you repaired with that fault it was D7. Test it and tag it, on to the next. 80% of the time this is what you will be doing.

Ok for starters you have to understand the circuit, you need to know how it works otherwise you are wasting your time. If you know what the device is supposed to be doing you can rule out 95% of "what could be wrong" just by looking at what it is doing or not doing.

Oscilloscope would be my test instument of choice.

Step (1) read the fault discription
Step (2) is it a production fault? Knowing if it ever worked is worthwile as you could have incorrect component, solder shorts that would be very unlikely in a returned item. Production faults can be the hardest to fix at times or the easiest. Missing chip or shorted track under a chip etc.
Step (3) Check your power supply
Step (4) This is where you need to know the product and the fault. You just start ruling out possibilites and eliminating them one by one.

You have traced out your circuit and suspect C10. Change it, you most likley cannot test it in circuit anyhow. Does it work yes/no. If no go on to the next most likley component, rinse and repeat.

 

[walters]

read the fault discription

Where do i get that from the Date sheet of the digital component?

 

[Wrighty]

listen mate what i do is the half-split method. it is the best method you can use. you can use any test instrument you want cos all your looking for is a voltage or not, a 1 or a 0.

1 Turn the equiptment on

2 look at what the board is telling you. if it has LED's that should be lit, are they? if it should be doing somthing is it?

3 from the symptoms, deduce what elements of the system are working and what COULD not be working. write it down.

4 go to a area that is possibly at fault and look for a correct output. is there a correct output? if there is, try the other possibilities.

5 If you find a incorrect output then half-split the board between what you know is servicable and your incorrect output. Is the logic there correct? if it is then go betwwen there and the faulty output, if its not go between where you are and your servicable input.

6 you will eventually get down to one componant and thats when you test all its inputs and outputs to see if they are correct.

hope this helps.

 

[dingo]

read the fault discription

Where do i get that from the Date sheet of the digital component?

The reason someone sent you something to repair means that the person that sent it to you thinks it is broken. These people for the most part will write down what is wrong with the device i.e. a fault discription.

For examples:
DVD player - remote not working
Good place to start after reading this is checking the remotes batteries. Getting another remote and testing that on the DVD and so on. There may be not need to open the DVD player up at all.

Mobile phone - screen gone funny
maybe bad solder join on LCD screen? faulty LCD Screen. Check the inputs to the LCD.

PDA - not working
not too helpful here, sometimes they just say "it's broken" so you have to start testing it yourself. Often you will not find any fault at all.