tt109

Upload: mrfixitok

Post on 14-Apr-2018

219 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/29/2019 TT109

    1/4

    Fig. 1: The main thing that sets a computer apart from a controller is that the computgets new programs from a disk or tape. It also has external memory, and complexinputs and outputs.

    Testing Microprocessors with the SC61 Waveform Analyzer

    Many technicians think that micropro-cessors are difficult to service. Why?Probably because they think of micro-processors as computers. Yet, mostmicroprocessor systems are used ascontrollers, not computers. Examples ofcontrollers are found in VCRs, microwaveovens, TV receivers, and most micro-

    processor-controlled test equipment.Knowing this can make servicing micro-processors a lot less fearful.

    Lets start with a practical piece of advice:dont change the microprocessor tooquickly. Time and time again, techniciansadmit that changing a microprocessordoesnt help a problem which looks like itmight be caused by a bad micro. But,microprocessors rarely fail. They areprotected from static discharge andpower line surges by filtered powersupplies and by buffering transistors andICs. The best process is to leave it on

    your list of suspects, but be sure tointerrogate all the other likely culprits first.

    You can quickly isolate most micro-processor-related problems, using theSencore SC61 Waveform Analyzer andfive quick tests we will cover a little later.First, lets see how a micro-processorused as a controller differs from one usedas a computer, so that you can see whymicroprocessor servicing has very littleto do with computer servicing.

    The Computer vs the Controller

    The biggest difference between amicroprocessor used in a computer andone used as a controller has to do withprogramming. A computer can bereprogrammed as needed, usually byentering information from a magnetic diskor tape. The controller lives a relativelyboring life - playing the same programover and over. Reprogramming rarelyhappens once the system has left thefactory. This main difference leads toseveral other differences as well.

    Computers (whether desk-top personalsor large mainframes) handle largevolumes of assorted data. One batch maybe numbers for a payroll, and the nextmay be a document from a wordprocessor. Controllers, by comparison,deal with much smaller volumes of data.The data are repetitive and predictable,

    often representing inputs from simpleswitches and sensors within the system.

    The microprocessor used in a computerconnects to thousands or millions of bytesof external random-access-memory(RAM), each byte containing 8 memorylocations. This RAM may require dozensof external memory chips. The micropro-cessor used as a controller only needs a

    small amount of memory often insidethe microprocessor chip itself.

    Lastly, a computer has complex inputsand outputs. Inputs come from typewritekeyboards, disk drives, or modemsOutputs feed printers, plotters, CRTdisplays, or other computers. The

    controller only has inputs from a few ICsrelays, and a simple digital display.

    Servicing controller-type microprocessorsdoesnt need to be any more complicatedthan servicing any integrated circuit. Thelimited environment of the controllemeans you dont have to know manythings you might think you need to know.

  • 7/29/2019 TT109

    2/4

    Fig. 2: The microprocessor used as a controller is programmed one time at thefactory. It uses simple switches and sensors for inputs, and transistors, ICs,relays, and digital displays for outputs.

    Fig. 3: Most defects are caused byproblems outside the microprocessor.Test the microprocessor inputs andoutputs in this sequence to isolateexternal problems before substitutingwith a new one.

    Fig. 4: When testing the powersupplies, use the DCV function toconfirm correct regulation and theCRT to find noise or ripple.

    The Simplified System

    One of the biggest differences between

    servicing computers and controllers isthat you dont have to worry aboutsoftware problems in controllers. Youdont need to know programming orASCII codes. If you suspect a softwareproblem, you have only one option;change the program chip.

    Second, you dont have to sort throughrows and rows of memory chips. Thismeans you dont need a $20,000 logicanalyzer or an 8-channel scope to vieweach byte of data separated in order tolocate a defective memory location. If aninternal memory location is bad, you have

    to change the microprocessor.

    Finally, the controller has limited inputsand outputs generally no more than 8 ofeach. You can test each one separately

    to confirm whether the problem is comingfrom inside the microprocessor or from anexternal component.

    Once you stop worrying about software,memory, and complicated interfacesystems, the microprocessor takes on awhole new look. You can find mostproblems with five standard tests usingthe Sencore SC61 Waveform Analyzer.The tests are of: 1. The power supply, 2.The clock, 3. The input and output lines,4. The reset circuit, and 5. The grounds.Heres how to do these tests with theSC61.

    1. Test the Power Supply

    Always test the power supply(s) first,whether the problem is a totally deadmicro or one with erratic operation. Startwith the DC level. Press the channel ADigital Readout DCV button, so that youcan monitor the DC level, while you watchfor AC problems on the CRT.

    Touch the channel A probe to the powersupply pin. Glance up at the digitaldisplay to confirm the DC voltage (usually

    5 to 10 volts) is correct. Your voltageshould be within about 0.2 volts of thecorrect level.

    But, dont stop there, because noise oftenenters the microprocessor through thepower supply, causing it to act erratically.Look at the CRT to confirm the signal isclean. Press the PPV button to measureits actual value. You should see less than0.1 volts of ripple.

    You may see 60 Hz ripple from a badfilter or regulator. Or you may see high

    frequency digital noise from anothestage, which can cause the micro tofreeze as the noise pulses intermix withnormal input signals. If so, suspect a badfilter or decoupling capacitor on the powesupply line, or a bad IC on the same linewhich is loading the supply.

    If the microprocessor has more than onepower supply pin, check each one in thesame manner.

    2. Test the Clock

    A problem in the crystal-controlled clockcan cause intermittent operation. Watch

    for the following conditions as you probeone, then the other of the microprocessopins connected to the clock input pinsusually coming from a crystal.

    First, reach over and press the FREQbutton. The SC61s autoranged frequencyfunction displays the operating frequencywith six full digits of resolution, so thayou can be sure of the results. If thefrequency is wrong, suspect a badcrystal.

    Next, press the PPV button, and look upat the digital readout to confirm the signahas the correct amplitude. Although thecrystal might be putting out the correcfrequency, the micro may not know itbecause the amplitude is just below thepoint that gives reliable operation.

    Lastly, check the CRT waveform to makecertain the clock does not include extraglitch signals. These extra signals maycause the microprocessor to intermittentlyskip a program step, or may cause thewhole system to run too fast. The clockshould be a clean sine or square wave.

  • 7/29/2019 TT109

    3/4

    Fig. 5: Test the clock for correctfrequency, amplitude, and cleanliness.A glitch riding on the signal, as shownhere, can cause the microprocessor toact like it is defective.

    3. Test the Input and Output Lines

    Inputs: Generally, input defects affectonly a few functions. Try every functioncontrolled by the micro and note whichones work correctly and which ones havetroubles. Then, determine which inputpins are associated with the badfunctions. For example, one or two

    switches might provide an input to asingle function and not be used with anyother of the micros inputs.

    Fig. 6: Use the SC61 CRT to confirmthat the signals have enough peck-to-peak swing to keep them out of thequestionable area, shaded in red.

    Connect your SC61 probe to the pinsassociated with the questionablefunctions and observe the trace as youcycle the input switches. Press the DCVbutton, and note the DC level with theswitch contact open and with it closed tobe sure the level properly changesbetween the one and the zero logiclevel. Be sure that neither level falls intothe undefined area between the twolevels, or the micro may not be able to

    decide whether a high or a low conditionexists.

    Check contact resistance or pull-upresistors if the levels are wrong. Watchfor noise or glitches which may cause themicro to interpret a single switchoperation as two or more separate switchclosures. Check the switch contacts, de-coupling capacitors, and switch buffercircuits to isolate noise conditions.

    Outputs: Next, test all data (output) linesto be sure one isnt stuck at logic high orlogic low. Touch your probe to eachmicroprocessor output pin, one at a time.Dont worry that the signal shows a blur oflines seemingly out of sync. This issimply because of the asynchronous(random) data coming from the micro.

    Fig. 7: The data at each pin will appearto be out of sync because it isconstantly changing. Concentrate onthe fact that each pin is toggling, and

    not stuck at logic high or low.

    Set the SC61s CRT to its DC-coupledmode to confirm that the low points on thewaveform are below the minimum levelfor a zero and that the high points areabove the minimum level for a one.Suspect a bad pull-up resistor or ICoutside the micro if the signals are fallingbetween logic levels.

    If the signal at a pin remains cemented toground or to B+, look at the schematic tosee when that pin is used. You mighthave to trace the pin to a relay or an IC tofind out which function(s) it controlsThen, force the microprocessor into afunction which uses this pin by pressing abutton or cycling a sensor.

    If the signal at the pin doesnt changeisolate the pin from the external circuits

    by carefully removing the solder betweenit and the foils on the P.C. board. Connectyour SC61 to the isolated pin and againcheck for toggling. If the pin toggles withits load removed, the problem is mosprobably outside the micro. An externacomponent is holding the pin high or lowIsolate each component on that line, oneat a time, until the line toggles. Thenreplace the defective part.

    If the pin remains stuck after beingisolated, its beginning to look more like adefective microprocessor, but donunsolder the other legs yet. Youve gotwo more checks to make.

    4. Test the Reset Circuit

    Microprocessors need an external resepulse at turn-on. Without the reset pulsethe microprocessor starts in the middle othe program resulting in totally unpredictable operation.

    Take advantage of the SC61s CRT tocheck the reset pulse. Set the TriggeSource switch to channel A, the TriggerMode switch to Norm and the Trigge

    Level control to the zero in the center ofits rotation.

    Connect the device containing the micro-processor to a switched AC outlet stripso that you can turn the power off and onDont rely on the devices power switchsince the microprocessor often receivespower independent of the power switchIn fact, many power switches are simplyone of the microprocessor inputs, anddont interrupt power.

    Turn off the power and connect the SC61channel A probe to the reset pin. The

    SC61 CRT should show no trace. Watchthe CRT as you apply power to thesystem. If you see the trace flash acrossthe CRT, you know a reset pulseoccurred and triggered the SC61 synccircuits. If there is no trace, repair thereset circuits.

    5. Check Grounds

    Now, the microprocessor is highlysuspect. But, dont unsolder it yet. Firstcheck every grounded pin. Each shouldshow zero volts DC and zero volts AC. I

  • 7/29/2019 TT109

    4/4

    any grounded pin has a signal on it, it willcause the microprocessor to act asthough the micro itself is bad. Thepresence of a signal tells you there is anopen in the grounded path either abroken P.C. foil or a bad solderconnection. Repairing the bad ground willprobably clear up your troubles.

    If the grounds are good, you are ready tosubstitute the micro. Youve alreadyconfirmed that all the inputs and outputsare normal. And, as we mentioned earlier,one of these other circuits will usuallybethe cause of the poor operation.

    SC61 Waveform Analyzer is a trademark of Sencore, Inc. Form #4437 Printed in U.S.

    Fig. 8: Watch for a reset pulse when you first apply power. Defective reset circuits make the microprocessor appear defective

    because it did not start at the beginning of the program.