Fullscreen HTML5Med Res (??MB)HTML4

SERVICEMAN'S LOG The unfit Fitbit that was made fit I’m not really into gadgets such as smart watches, although the technology behind them is quite impressive. Recently though, I was given one to repair and it was quite a challenge to make the Fitbit fit again so that its owner could keep fit. Many people these days are into gadgets. Actually, it’s often not so much the gadgets themselves but the fact that they are connected to the internet that people find so appealing. Then there is this trend to wearable technology. A few years ago, we bought a watch just to tell the time. But these days, if one is up and coming, one must have a “smart” watch. In addition, we have now been reintroduced to activity trackers. I say “reintroduced” because personal pedometers were all the rage among a certain set not that long ago, though like all exercise fad gadgets, they usually ended up gathering dust under the bed. However, the latest wearable exercise gadgets, typically futuristic-looking wristbands, measure all sorts of human activity, such as steps walked, sleep 66  Silicon Chip patterns, pulse rates and other crucial data we simply can’t live without. All this data can be uploaded via WiFi or GSM networks to the cloud where users can plot everything on impressive-looking graphs and spreadsheets in order to track their overall fitness, food intake, calories burnt, hours of sleep (and sometimes even stages of sleep) and other (more or less) useful stats. More importantly, “Generation Me” can share this information online with their friends, colleagues and competitors. It’s a great idea and also a great motivator, encouraging users get off their rear ends and go and crank out some more data to upload. So where’s all this leading? Well, someone in this household (not me!) has bought herself a Fitbit, one of the fancier, Dave Thompson* Items Covered This Month • • • • Dave’s Fitbit repair Church audio system repair Digitech ultrasonic cleaner Intermittent electrical fault in Holden Berlina wrist-borne activity trackers out there on the market. It’s actually a very high-tech little gizmo, from its supple, purple, rubberised moulded body to its relatively small, high-resolution OLED display. Hers is a middle-of-the-range Fitbit and for what it cost to purchase, it should be flash! Depending on the model, a Fitbit can include a heart rate monitor, an accelerometer, an altimeter, the usual clock/watch functions and a long-life lithium-polymer battery. And without trying to sound like an advertisement for the manufacturer, it really is a nifty little gadget and is easy to like. They cram a lot into the small case and while it’s quite rugged, they can break down. Internet forums are awash with adopters complaining about this or that, as with any product, but as these devices usually cost a fair bit, users expect high-end results from them. The display on the Fitbit is impressive. The resolution is fantastic and the figures extremely sharp and clear and easy to read, even in the brightest sunlight. They really are “cool” to use and work very well for counting steps, which is essentially what my wife bought hers for. Fitbits are very reliable but as stated, they can have problems although it isn’t always the electronics that fail. External parts can take a hammering and they need to be very hard-wearing to stand up to the punishment active users give them. However, this particular Fitbit was a bit unfit in some respects. First, the material used in the strap siliconchip.com.au and body on my wife’s version feels durable but splits very easily if bent the wrong way. Then there is what I consider to be a design flaw. In order to charge the Fitbit, it has to be connected via a short USB cable to a computer or to an optional plugpack power adapter. This cable has a proprietary fitting at the Fitbit end and this clicks solidly into place in the bottom of the unit, through a U-shaped hard plastic bracket that wraps around three sides of the case and hides the charging port inside. The fourth “side” of the rectangle formed by this U-shaped strip is the display, with the rest of the case being there simply to hold that bit in the right place on the wrist for the sensor. And here’s the design flaw; the other day, when Nina went to put the Fitbit on after charging it, that U-shaped bracket stayed behind, still securely clipped to the end of the charging lead. It had completely come away and when I looked closely at it under my microscope, it was easy to see why. As I said, that bracket forms part of the “back” of the unit and it is held in place with four tiny plastic pillars. They are so thin and fragile, I was surprised that they’d lasted as long as they did! I know that everything has to be small in gadgets like this but given that the charging lead clips soundly into the charging slot in the plastic bracket, it wasn’t going to take much to wrench the bracket from the body, as those tiny bits of plastic were all that held it on. They could have made those plastic pillars bigger and still had room for other things. However, they really should have gone down the road that Apple went with their charging leads and used a magnet to hold it in. Or maybe the charging plug could have clipped into something built more securely into the body of the case, rather than just the hard plastic piece on the back. It’s a poor design in my opinion, considering everything else on the unit appears to be well thought-out and implemented. The first job was to correctly refit the back to the body of the device. Just below where the bracket sits, there is a sensor. This is designed to sit on the top of the wrist and two bright-green LEDs flash away, monitoring the user’s heart rate through the skin. The problem was that the bracket siliconchip.com.au had deformed slightly as it came away and when placed back into position, it didn’t fit properly. The broken-off plastic pins didn’t line up to where they should have and the sides had flared out, so that was going to be a problem. Another tricky problem involved the button that controls the display functions. This is a separate assembly that passes through one side of the bracket and has to line up with a tiny microswitch beneath. This would have to be held in-place when the bracket was finally finagled into position. I had to gently tweak and manipulate the bracket’s plastic until it sat back where it should. This was a bit awkward but the plastic behaved itself and I eventually got it to sit in place. If the bracket had broken, it would have been game over. The next challenge was to devise a method to securely hold the bracket in place. Gluing the original plastic pins was out of the question as there was virtually nothing to glue anything to. When a part has pins that break off, I can usually just glue them back into place and they are then strong enough to hold the part securely. However, with the Fitbit, the broken pins were so tiny that even if I could successfully glue them, the assembly was highly unlikely to be strong enough to withstand the stresses of the charging lead. As a result, I initially considered simply gluing the edges of the bracket to the unit but again I doubted that it would work. None of the glues I had on-hand would adhere to the rubberised body and I wasn’t sure if there was such a glue available anyway. In the end, there was really only one thing I could do; screw the bracket back on. This involved some risk, as I wasn’t absolutely sure whether or not there was anything vital to the operation of the device beneath the holes for the pins that originally held the back on. If I went poking about in there too deeply, I might hit something critical and that would surely be the end of the device. Well, sometimes a serviceman has to make a bold decision and since there is next to zero information about the insides of these things anywhere on the web, I figured that I’d just have to take my chances. If I couldn’t attach the bracket, the thing would be useless anyway. I wasn’t about to break out my taps and dies because the set I have didn’t go anywhere near small enough for this job. Instead, my plan was to use a tiny drill bit to clear out as much of the remains of the pins as possible, then use one of my dental picks to get the rest out. I’d then be able to use tiny screws to self-tap into the holes left behind. I was a little wary about using a drill but winding it by hand in my pin vice gave me enough control to ensure that I didn’t go too deep. I also wrapped some tape around the bit to prevent it from going much deeper than the length of the screws I intended using. The four screws to be used were gleaned from my spares tray and once Servicing Stories Wanted Do you have any good servicing stories that you would like to share in The Serviceman column in SILICON CHIP? If so, why not send those stories in to us? In doesn’t matter what the story is about as long as it’s in some way related to the electronics or electrical industries, to computers or even to car electronics. We pay for all contributions published but please note that your material must be original. Send your contribution by email to: editor<at>siliconchip.com.au Please be sure to include your full name and address details. September 2016  67 Serr v ice Se ceman’s man’s Log – continued It wasn’t divine intervention that got the audio system in the church cry-room going again. Instead, rather than working in mysterious ways, B. C. of Dungog used old-fashioned, down-to-earth sleuthing to track the problem down. And he got the T-loop system (for those with hearing aids) working for good measure. Here’s his story . . . Our local church is celebrating its 100th anniversary this year and all stops are being pulled out to get the necessary preparations under way for this big event. One of my tasks was to investigate the lack of sound from the church’s cry-room loudspeaker and from the T-loop system. Before starting, I was given a brief run-down of the sound system upgrade that had been done a few years ago. This had involved the installation of a new mixer, a multi-core cable, a large stereo power amplifier and two large speakers. During a past building renovation, a cry-room (where crying children are taken) had been added inside the back corner of the church, along with a T-loop system for the hearing impaired. Inside the cry-room, I found an old column speaker box mounted on the wall. This had been fitted with a 5-step attenuator control, while a figure-8 cable had been run back from the box to the sound system control cabinet. Upon inspection of the sound system installation, there appeared to be only four microphones on the stage area (at the front of the church) plus an old Teac DVD player connected to a Yamaha 18/20-channel mixer. This meant that there was plenty of scope for future expansion! There were also a number of figure-8 speaker cables (entering via a flexible conduit through the floor), with most of these having being decommissioned during the last sound system upgrade. I began by testing all these figure-8 cables with a multimeter and eventually found one that measured about 80Ω. To double check this, I disconnected one of the input wires on the back of the speaker level attenuator control in the cry-room and this confirmed that I had the correct cable. I had also noticed a small transformer on the back of this plate. When I removed the speaker grille and the bottom speaker, I found another transformer underneath the inner bond filling. This had all been meant to be run on a 100V line public address system. No wonder the cry-room speaker cable had been left disconnected after the upgrade! Initially, I considered rewiring the column speakers and fitting a rheostat to control the volume level but that would have meant tapping into an amplifier speaker output at the front of the church. It would also have been necessary to rewire the four column speakers to get the correct impedance. When I returned to the sound system control cabinet, I noticed an old CS A600 series mixer-amplifier down on the bottom shelf. I pulled it out and found that it had a 100V line output as well as a normal 8Ω output. There was also an auxiliary input with its own level control but this had seized from lack of use over the years. Spraying some CRC 2-26 onto the seized shaft soon had this auxiliary input level control working again. I now had to figure out how to connect a signal from the Yamaha mixer to the CS A600 amplifier’s auxiliary input. A quick inspection of the mixer soon revealed unused left and right channel “record-out” sockets so the next step was to come up with a suitable patch lead. Rummaging through the back of the cabinet soon turned up an unused 6.5mm jack plug and a spare stereo RCA-type audio lead. A pair of side-cutters, a hot soldering iron held parts of a smart-phone together. I didn’t measure them but they were just the right size to self-tap into the holes that had been cleared using the drill and pick. Digressing slightly, I never leave my dentist’s surgery these days without asking for any old tools they can spare. Many such implements are retired after a certain amount of time and service and while they’re no longer any good for poking around inside someone’s mouth, they are perfectly suitable for hobbyist use. They are cleaned in an autoclave and put aside for disposal but can come in handy for fine work and my dentist is always happy to let me rummage through their box of unserviceable tools. I always grab a couple of handylooking picks each time I go to my dentist and I then don’t feel so bad when it comes to paying their bill. Anyway, after clearing out the rivet holes, I used one of the screws as a tap and threaded each hole with it. It is always a bit nerve-wracking when brute-force tapping holes and in this case, I wasn’t sure that the material would stand up to the process. My luck held though and the four holes were soon boasting nice new threads. I went just a little deeper than I had to, taking into account the length of the screws and the thickness of the plastic they’d be going through. After all, I wanted to be sure I could tighten them down easily, to avoid breaking the plastic bracket. Once that had been done, I drilled four new holes in the bracket itself, using the old broken pin stubs as a guide. I made the holes a neat fit for the screws and used a larger drill to care- fully countersink the holes so that the screws wouldn’t protrude and cause any discomfort on the wearer’s wrist. It was then simply a matter of lining up the bracket, making sure the activity button was sitting in place, and gently driving the screws home. This is where my precaution of pre-tapping the holes paid off. If the holes hadn’t been pre-tapped, I would have had no idea as to how hard I was clamping down on the bracket. If I’d over-tightened the screws, I would have risked cracking the already-brittle plastic. And if they hadn’t been tightened sufficiently, the bracket could potentially work its way loose again By pre-tapping the holes, I could accurately judge just how much I needed to tighten them. As it turned out, the bracket is now held on much more tightly than the original ever was and No more crying in the chapel 68  Silicon Chip siliconchip.com.au and some solder soon produced the required patch lead. This was connected and a test CD played through the church sound system. The auxiliary input level control on the old CS A600 was then adjusted so that a parent in the cry-room could adjust the speaker box over a useful range of volume using the attenuator control. Now for the T-loop system. Also inside the cabinet was a rack-mount black box labelled “Printacall Hear All Powered Audio Induction Loop System”. This had a volume level control on the back panel, while the front panel carried green and red LED indicators. Sliding the box forward off its shelf revealed that the original figure-8 loop output cable was still connected but there was nothing connected to the input socket! No wonder it wasn’t working; you didn’t have to be a genius to figure that one out! I found a suitable mono RCA-toRCA audio lead in the back of the cabinet. This was then modified by cutting off the RCA plug at one end and wiring it instead to a 6.5mm jack plug. It was then just a matter of connecting it in place and adjusting the level control on the rear panel so that the red peak level indicator LED occasionally flashed briefly when the unit was being driven by the mixer. And that was it – the service is now available to those in the cry-room and those with hearing aids! the repair should now last for the life of the device. Digitech ultrasonic cleaner Ultrasonic cleaners are great for cleaning parts – except when they don’t work. B. B. of Northland, NZ seriously contemplated buying a new ultrasonic cleaner when his old one failed but eventually managed to get it going again . . . My Digitech CT400D ultrasonic cleaner usually sits at the back of a top shelf in my workshop, its bright blue colour making it easy to find for those occasional cleaning jobs. It’s easy to operate – just put the items to be cleaned into the tank along with a suitable liquid (a solvent or sometimes just water), the press either the 35W or 60W buttons. A 2-digit LED display then counts down from 99 seconds and then siliconchip.com.au the ultrasonic cleaning action stops. A job requiring its use came up recently and after putting the cleaning solvent and the bits to be cleaned into the tank, I plugged it in and tried to start it. There were no signs of life whatsoever, so I emptied it all out so that I could have a look inside. After undoing three screws, the case came apart to reveal a small PCB beneath the display and the buttons, a round resonator glued to the bottom of the cleaning bath and a larger PCB on the base that does the “heavy lifting”. There weren’t all that many components so how hard could it be to fix? Because it was completely lifeless, the first thing I looked at was the fuse. It had blown but not with any signs of violence. The first replacement fuse lasted until I pressed the 60W button, while a second fuse stayed intact when I pressed the 35W button but smoke soon started to appear from two 100Ω resistors. It was time to reach for my multimeter. A quick check of the two BUT11AF TO-220 transistors gave low resistance readings and after removing them, I was able to confirm that they were both indeed faulty. I looked for other signs of heat and damage but found nothing, so I ordered replacement transistors and put the unit aside until the parts arrived. When ordering replacement parts, it’s often difficult to know just how many to get. Was the failure caused by these transistors? If so, only one pair would be needed; if not, how many would die (probably in twos) before the real cause of the problem was found? In this case, the transistors were cheap and are general-purpose enough to be useful for other jobs, so I ordered six. With the new transistors fitted, I did a few more resistance checks and then plugged it in. The fuse held in both the 35W and 60W modes but the sound was wrong. I could hear some 100Hz hum but not the normal “fizzing” sound it makes when working. The display PCB was working OK and just controlled relays on the main PCB, so I was able to eliminate it from my investigation. A schematic was looking like a useful thing to have in order to figure out what might have “killed” the transistors and what, in turn, their failure may have affected. Resorting to Google to find one gave me a sense of the likely configuration but nothing close enough to this unit to be worthwhile. At this point, it was very tempting to simply buy a new unit rather than repair this one. Working on it “live” would mean dealing with mains voltages since there was no transformer, so I would need to take extra precautions and make sure that an oscilloscope was properly isolated. Alternatively, I could try working out the schematic by “reverse engineering” the unit. Another option was to try working out what was wrong simply by checking the components one by one. And since there were not many of them, this became my preferred option, especially after I noticed a small hole in the coating of a 1Ω 1W resistor. I checked this resistor and it measSeptember 2016  69 Serr v ice Se ceman’s man’s Log – continued This story relates to a car that was once owned by my wife. At the time, it was a near-new 1993 Holden Berlina which we purchased from a car yard in Auckland. It had a fuelinjected 4-cylinder Opel Vectra engine and was popular in NZ because of its good fuel economy. It had done about 5000km when we purchased it and I wondered why the original owner had traded it in so soon and why the price seemed so reasonable. Well, we were about to find out. After just a few days of driving it here and there, the engine suddenly cut out during a short trip. After the car had drifted to a halt, we attempted to restart it but it would simply turn over without even a hint that it would start. And then, after about 10 minutes, it suddenly started again and all appeared to be normal. The car then ran normally for a few days before doing it again. This pattern of engine cut-outs was then repeated over the next two weeks and each time the engine could be restarted after waiting for somewhere between three and 10 minutes. Alarmed by this, we took it back to the dealer and left it with their service department for several days. At the end of that time, they told me that they couldn’t fault it and on top of that, no error messages had been recorded by the ECU. They gave the car back but the fault quickly reappeared, the engine regularly cutting out although it ran faultlessly for up to three weeks at one stage. On delivering it back to the service department for a second time, I noticed as I gazed over the counter that its rego plate had been recorded in their service log book multiple times. I’d only brought the car in twice, so what was going on? When I questioned them, I discovered that the original owner had brought it in with the same fault on multiple occasions, before giving up and abandoning it as a “lemon”. They had even replaced the ECU (engine control unit) but to no avail. I said to the service department manager “You sold my family this car knowing that it had an intermittent fault and that at any time it could stop on a motorway and place them in danger”. He looked alarmed and became very defensive. “No I didn’t, it wasn’t me. It was those guys over in sales”, he replied, as he gestured towards the showroom. I decided that since we liked the car otherwise, and since the dealer was incapable of fixing it, I would have a go at it myself. Thinking about the basics, an engine needs the FACTS to run: Fuel, Air, Compression, Timing and Spark. In this case, it was likely to be either a fuel or spark problem, as it was unlikely that the timing (either electronic or mechanical) would suddenly go haywire in a previously working engine and then suddenly fix itself again. It was also unlikely to be an airflow issue that was producing the abrupt engine stoppage although an air-flow meter fault is always a possibility. What’s more, the ECU would have detected an out-of-range input from a faulty air-flow meter and thrown up a fault code. Since there were no recorded errors, an output device of one kind or another in the fuel or spark system was most likely intermittent but the problem was just how do you go about finding it? The fact is, intermittent faults in an ECU-controlled engine can be a nightmare to track down. If anything stops the engine, the ECU detects that there is no engine rotation (because there is no signal from the engine rotation sensor) and it switches off the fuel pump, the injectors and the ignition system. So at that point, once the engine has stopped and you pop the bonnet to find the fault, there’s no way of knowing which of the basic functions dropped out first to initiate the engine failure without fault codes and computer diagnostics. What’s more, those various subsystems cannot easily be checked in the case of an intermittent fault that sometimes occurs weeks apart. Because of this, I quickly realised that what was required was a monitoring system with latches to record which part of the system stopped first. In other words, I needed an “event recorder” with a memory. Given that this was an urgent problem, I scrambled to the junk box to find some parts. I quickly grabbed some CMOS hex Schmitt trigger inverter ICs because they can be cobbled together in a myriad of ways and have a handy high input impedance. I also had some spare 4013 dual D-type CMOS flipflops and some diodes and LEDs. There were several likely fault possibilities: (1) a fault in the output from the ECU to the fuel pump relay; (2) a faulty fuel pump relay output to the fuel pump; and (3) faulty ECU ured much higher than its 1Ω markings indicated, so I replaced it with a 1.2Ω resistor, the closest 1W value I happened to have on hand. I then tested the unit again but there was no change; it still wasn’t working. As I continued component checking, I was contemplating what sort of test equipment I’d need to check the resonator and the coils when I found a second 1Ω resistor that had gone high. During this time, I was vaguely aware that our cat had come into the workshop and was sitting on the floor, not far from the bench. However, my awareness of his presence suddenly increased after I had replac­ ed this resistor, the cat taking off in a blur as soon as I turned the cleaner on. That was one bit of test equipment I didn’t realise I had: an ultrasonic detector that runs on cat food! Confirming the cat’s diagnosis was easy because the fizzing sound was back. I then reassembled the unit, put some water in the tank and switched it on. It was now back to normal opera- Fault detector solves difficult intermittent in a Holden Berlina by Dr Hugo Holden 70  Silicon Chip Giving it a go siliconchip.com.au outputs causing either the fuel injectors or the ignition spark to stop. Each of those would require a monitoring line. Since the latter two rely on pulse signals, they would have to be monitored using pulse detector circuits. Another possibility was that the engine rotation sensor itself was defective but I decided to hedge my bets on that one. As a result, I initially built a 4-input detector system. This was designed so that if an event occurred, then that channel would be latched and inhibit the other three recording channels. At the same time, one of four LEDs would light to indicate which channel was at fault. The basic circuit I used is shown in the 4-Input Automotive Fault Recorder project overleaf on page 73. Rather than using more logic gates to inhibit the other channels, I simply used 10kΩ series resistors and clamping diodes which are driven by the Q-bar outputs of the 4013 flipflops. I didn’t bother adding refinements like a zener diode on the 12V rail and cobbled it all together on protoboard, with light-duty wire-wrap connections. I also used the same thin wire to connect to the fuel pump relay coil connection (at the output from the ECU) and to the fuel pump relay output (at the pump itself). This was done simply by pushing the wire into the spade connectors and the same was done for the connection to one of the fuel injectors. Spark monitoring was achieved by wrapping five turns of wire around the outer surface of a spark plug cable, to make a “gimmick” capacitor. I had previously realised that I should ideally be monitoring circuit currents instead of voltages. That’s because the correct voltage can be present at a given point but there’s no current due to an open circuit condition. In fact, I had this up my sleeve as “plan B” if monitoring the voltages didn’t bear any fruit. I also realised that if worse came to worst, I would have to fit a pressure sensor to the injector’s fuel rail but I hoped that I wouldn’t have to go that far. For the time being, I figured that voltage monitoring was the easiest approach. Anyway, I fitted the assembly to the car, ran the wires through to the engine compartment and taped the horrible looking mess with its four LEDs to the dashboard. I then started the engine, pressed the reset button and found that all LEDs were off, as they should be. Nothing happened during the first few days of driving and then suddenly, on the fourth day, the engine cut out. I looked at the panel and saw that LED2 had lit, indicating that although the fuel pump relay was on (LED1 off), the output from this relay had vanished. After a 5-minute delay, the car started again and I rushed back home and unplugged the fuel pump relay for inspection. It turned out to be a Bosch unit with a grey plastic case. Its base was sealed onto the case with silicone rubber and I removed this before prising the inner assembly out. Once it was out, I found that it mainly consisted of a relay bobbin assembly mounted on a small PCB. This PCB also carried a diode that looked like a 1N4004. It was in series with the coil, presumably to prevent the relay from turning on with reverse polarity applied. The wire enamel on the coil was discoloured, indicating that the relay had been running quite hot. I took a closer look at the PCB and the answer was staring me right in the face; a 360° crack around the soldering on one of the relay coil’s connector pins. As it heated up, it was expanding and going open circuit and it was probably being affected by vibration as well. Once it had gone open circuit, it then cooled down again until it eventually remade the connection and reapplied power to the fuel pump. This explained why the car could be restarted after a short wait. I measured the resistance of the coil and, using the formula P = V2/R, calculated that the relay coil was dissipating about 4W, assuming a supply voltage of 14V (as it typically was). That explained why the coil wire looked as though it had been overheated. Fractured solder joints like this appear to be more common when there are a combination of factors: (1) significant heating and cooling cycles of the pin which can harden and crystallise the solder; (2) a PCB hole which is larger than necessary for the pin passing through it; (3) a fairly sparse or thin sheet of solder bridging the gap between the pin and the PCB pad; (4) the PCB hole not plated-through; and (5) physical forces due to a weighty object (in this case the relay). Resoldering the faulty joint and covering the crack with a generous amount of solder cured the problem once and for all and the car ran without any further engine problems. I returned to the dealership a few weeks later and explained to the service manager how I was able to find the fault with my home-made fault recorder. He seemed to be astonished at the notion of a fault recorder and had never before heard of using such a technique to track down an intermittent fault. At the end of the conversation, he offered me a new relay for free so I took it to keep as a spare. However, the fact remains that the dealership should never have sold us that car without first fixing this potentially dangerous fault. tion, with the familiar ripples on the surface changing pattern between the 35W and 60W modes. So, as it turned out it, I only needed a multimeter to identify the faulty parts. The components markings were readable and the replacements readily available, so it wasn’t too much of a hassle to repair the unit. It would have been interesting if I’d had to check the resonator though but fortunately I didn’t have to. Looking at the schematics I found on-line, it appears that the two BUT­ 11AF transistors operate in a highpower oscillator. A ferrite-core coil and a capacitor are tuned to the same frequency as the resonator, allowing self-oscillation with low-gain, highpower transistors. The ultrasonic signal is superimposed on 100Hz of unfiltered, rectified mains to include both high and low frequencies in the cleaning “signal”. As for the cat, he spends a lot of time at the other end of the house whenever SC the unit is operating. siliconchip.com.au Nailing the fault September 2016  71