Fullscreen HTML5Med Res (??MB)HTML4

Build this superb 1GHz Digital Frequency Meter This month, we complete construction of the 1GHz DFM and describe the calibration procedure. There's also a useful troubleshooting guide - just in case you strike trouble. By STEVE PAYOR Last month, we left off after describing the assembly of the two main printed circuit boards. The next step is to solder the two boards together at right angles. To so this, carefully align the two boards, using the arrows at each end of the display PCB as a guide, and solder tack them in several places. Now inspect the assembly carefully and make any adjustments that may be necessary. When everything ls correct, solder all the matching pads together to create a permanent assembly. Finally, check that there are no solder bridges between adjacent pads. The meter is now ready for its first test. Temporarily install an insulated This close up view shows the mounting details for the two BNC sockets. Note that two earth links are soldered to the 500 input socket. 48 SILICON CHIP link across the switch connections to the mains terminal block (the mains switch is installed later), then slot the counter module into the front of the case. Connect the power supply leads to the board with a multimeter set to measure DC current in the + 5V lead. Plug in, switch on and verify that the current is between 300 and 500mA (depending on how many digits are lit). If all is well, try pressing the front panel buttons. On power up, the display should read .000kHz and the second Function LED should be lit (frequency to 10MHz). When the Range buttons are pressed, the leading zero on the display should be directly above the selected button, and the decimal point should be immediately to the left or right of the leading zero, depending on the setting of the Function pushbuttons. The function indicator LEDs and the µsec and kHz light bars should change with the Function pushbuttons. If you strike problems here, switch off immediately and check for wiring errors. Assuming the power consumption of the counter module is within the specified limits, you can now go ahead and make the + 5V supply permanent. Twist the supply leads together and trim them so that they are just long enough to reach the counter PCB. The µsec and kHz transparencies can now be glued to the light bars. Use only water-based PVA glue (eg, Selleys "Aquadhere"). Only a tiny bead around the outside edge is re- , Se 12,8 1·. Above: the new 1GHz DFM is shown here displaying the frequency generated by the SAB6456 prescaler IC (see text). quired. If you make a mess, clean up with a damp cotton bud and try again. Testing the counter module The counter module should rest comfortably in the second PCB slot from the front, leaving a 10mm clearance between the display PCB and the front panel. At this stage, you will find that the lkO multiturn trimpot will be resting on one of the PCB guide rails, so carefully cut away the plastic at that point. Do not drill a hole right through the plastic case, as this trimpot is not meant to be adjusted from the outside. With the counter module and rear panel module in place, slip on the top half of the case, then turn the whole unit upside down. The bottom half of the case can now be removed to expose the underside of the counter and display PCBs. Next, solder a shorting link across the lMO input pads on the display PCB and plug a 1. 7V red LED into the Molex pins at the test point. The anode of the LED should be connected to the + 5V track, which is the thicker of the two. Switch on - with no signal input, the LED will either be on or off, depending on the state of the Schmitt trigger (IC2a). If the LED is off, turn VRl 1,28 ,128 QUENCV METER Hints on Drilling the Perspex Panel The greatest hazard when drilling thin perspex sheet is the possibility that the drill will take one bite and shatter the panel. To avoid this problem, the drill should be as sharp as possible, for a clean finish, but the cutting edge should have no rake . Specifically , the leading edge should be at 90 ° to the cutting direction as shown below. It only takes a minute to convert a normal drill bit, using a grinder or pocket stone. The other problem often encountered is a poor finish, caused by the drill overheating and melting the surrounding plastic. Again, keep the drill sharp. There are two other styles of drill grinding which are useful when building a project like this. The Wshaped bit is for cutting clean, perfectly round holes in sheet metal. This drill actually punches out a thin disc at the end of the cut, leaving no burr on the far side. It is perfect for drilling ventilation holes in aluminium or thin sheet steel and will even cut a clean 1 2mm hole in clockwise until it just comes on, or vice versa. Tapping VRl with a metal screwdriver will inject enough noise to trip the Schmitt trigger, so keep the blade in steady contact (or use a plastic screwdriver). NORMAL DRILL FOR SHEET PERSPEX METAL SPOTFACING tinplate. Using a normal drill for this job would leave you with a mess. The spotfacing drill is similar, but the "wings" are straight instead of W-shaped, and less rake is used on the leading edge. This type of drill will soon get rid of any unwanted plastic pillars, cutting them flush with the inside of the case. Having found one of the trip points, slowly turn VRl in the opposite direction, half a turn at a time, counting as you go. When the Schmitt trigger changes state again, you have found the other trigger point. On the prototype, the JANUARY 1988 49 The 100MHz preamplifier circuitry is adjusted with a 1. 7V red LED plugged into two Molex pins on the back of the display PCB. range was five full turns. This is a convenient way of checking the DC gain of the amplifier circuitry each turn is approximately lmV. Check the trip points once more, then set VRl exactly half-way between them for maximum sensitivity. Once VRl has been set, remove the input short and the LED. Slip on the bottom of the case and turn the instrument right way up again. The instrument is now ready for its first frequency measurement. Select a lMQ resistor and clip one lead short. Switch on and touch this lead to the 0.047 µ,F input capacitor while holding the other lead in your fingers. The DFM should show a steady reading of .050kHz. Note: the lMQ resistor, in conjunction with the input capacitance, acts as a low-pass filter for the 50Hz signal picked up by your body capacitance from the mains wiring. The 7 10 prescaler can now be checked by pressing the F3 button. The display should change to 0.05kHz (ls gating time]. Now switch to period mode. You should get a fluctuating reading of around 20,000µ,s (20ms). The reading will be constantly changing because the short-term stability of the mains is nowhere near as good as that of the crystal. Removing the signal source in Period mode will freeze the reading since the 7216A counter chip will be waiting for the required number of input cycles to 50 SILICON CHIP be completed before it updates the display. Checking the 1GHz range is easy. Press the F4 button and select the minimum gating time (R4). You should now have a rapidly updating reading of around 1.2GHz. This is because the SAB6456 prescaler oscillates at this frequency when no signal is present. Now press the R2 button and wait 12 .8 seconds until the display updates. The leading "1" will now be in the overflow position, with the rest of the digits much the same as before. The above procedure checks out virtually all of the functions of the DFM. If your unit passes all the above tests, you can be sure it's fully operational. The front panel The front panel artwork shows the drilling pattern for the panel. Red perspex is the material of choice and the colour to ask for is "Red 502". This is a deep ruby red colour which looks quite dark but transmits the wavelength of the red LEDs extremely well. The thickness of the panel should be 2mm, with 1.6mm as the second choice. Unfortunately, these thicknesses are not universally available, although 3mm sheet is quite common and can be pressed into service if you have no other choice. The 3mm sheet will fit in the wide gap between the front panel slot and the first PCB slot. The PCB should be moved to the third rearmost slot (don't forget to cut the relief for VRl ). Provided none of the components are higher than the height of the 7-segment displays, everything will still fit. If you go for this option, note that you will have to trim an extra 1mm from the top and bottom of the panel. The width remains the same. Once the panel has been cut to size, the holes can be drilled to accept the switches and input sockets. This can be done by first taping the front panel artwork (or a copy) to the panel, then marking out the holes with a sharp scriber. Pilot drill all holes with a small PCB drill to start with, then progressively enlarge the holes, checking the centring as you go . The final hole sizes are 9.5mm for the BNC sockets and 10-1 lmm for the pushbutton switches. You 1.0~--~~--~~--~ 0.9 0.8 I·e o., i 0.6 TIME (mi nutes from switch-on) Fig.12: this graph plots the warm-up drift of one of the prototypes. The total drift is better than 1ppm. POWER • cycles counted 1000 gating time, sec 10 10 128 100 1 1 12.8 10 .1 .1 1.28 1 .01 .01 .128 I 10MHz 100MHz 1GHz PER~ENCl 1Mn/10pF \llfl \llfl 50n DIGITAL FREQUENCY METER 0 C041-1187 Here are actual size artworks for the front panel, printed circuit boards, and kHz and µ,sec indicators. may wish to ream the last mm or so of the pushbutton holes, particularly if the drill is not cutting evenly. Drilling thin Perspex is not an easy task, so we have included a panel on hints for drill sharpening for those readers who are not used to working with this material. Because the inside edges of the pushbutton holes are visible, you may wish to polish them. Alter- natively, blacken them with a marking pen. Front panel assembly Affix the Scotchcal label to the front panel, then cut out the three holes for the power switch and input sockets using a sharp utility knife. A light touch with a reamer will improve the appearance of the hole for the mains switch. I!& JANUARY 1988 51 Troubleshooting the 1GHz DFM Don't rush out and replace all the semiconductors if the counter module fails to work first go. The problem is almost certain to lie elsewhere unless you've done something silly, such as installing an IC or transistor back to front. The first thing to check is that all the parts are in their correct positions and are correctly oriented. For example, the switches will not work if they are rotated 90 °. Next, check the copper side of the PCBs for soldering faults, such as bridges between the closelyspaced tracks around the IC pads. Another thing to look for is open circuit tracks, and these are most likely to occur where too large a hole has been drilled through an IC pad. One area where soldering faults are both common and easily detected is in the multiplexed display. A short between segments is easily recognised from its effect on the display, as is an open circuit along one of the segment tracks. Some constructors of previous Mount the two BNC sockets and orient them so that the solder tags will fit between the parts on the display PCB. The 500 input socket should ,have its tag at 3 o'clock (viewed from the back), while the other tag should be between 7 and 8 o'clock. Bend the tags up by 90° after the nuts are tightened. Next, solder two short lengths of tinned copper wire to the 500 socket and one to the lMO socket (the two connections to the 500 socket minimise the series inductance). This done, solder two short lengths of tinned copper wire into the central hollow pins of the connectors. Remove the counter module from the case and test the fit of the front panel over the switches. Bend the earth links from the BNC sockets to engage the PCB holes, then slide the module and panel together into the case. The leads from the input sockets can now be soldered to the 52 SILICON CHIP DFM projects have experienced trouble in getting the 7216A's oscillator to start up. The symptoms are no display at all, or only one digit lit and very briefly at that. The problem usually occurs with very cheap crystals which have a large equivalent series resistance. Calculations based on the minimum guaranteed gain of the CMOS inverter show that the oscillator will only work with a crystal having a series resistance of less than 800 and 20pF shunt capacitance, or 350 and 30pF nominal shunt capacitance. If you are unlucky enough to have a crystal with excessive series resistance, the cure is fairly simple: use less capacitance to the +5V rail at pins 25 and 26 and maintain the correct shunt capacitance by installing a capacitor across the crystal. For example, removing the 39pF capacitor and 40pF trimmer leaves the crystal with a shunt capacitance of about 3pF, so adding a 27pF capacitor between pins 25 and 26 will bring a 30pF crystal back to the correct frequency. The circuit will now tolerate a series resistance as high as 1200 although the stability will no longer be quite as good. display PCB. Finally, the wiring between the mains switch and the terminal block can be installed. Be sure to use mains-rated cable for this job, and cover the switch terminals with spaghetti insulation or heatshrink tubing to prevent accidental electric shock. For good measure, we also covered the entire switch body and part of the leads with additional heatshrink tubing. Don't forget to remove the link installed across the terminal block in place of the mains switch earlier on. sharpening). This will quickly cut the pillar right down flush with the surrounding plastic. We drilled four rows of holes in the front left-hand area of the case bottom, and another two rows on the other side of the ribbed section. The ribbed section was left undrilled (see photos pages 39 and 43, November 1987). This allows cool air to flow up under the main counter chip and around the 10MHz crystal. A small hole must also be drilled in the lid of the case, directly above the 40pF trimmer capacitor. This hole allows external adjustment of the trimmer during calibration. It is located 26mm back from the rearmost PCB guide rail, and 47mm in from the inside edge of the case (note: if the display PCB occupies the third PCB slot, read 29mm instead of 26mm). Drill a small hole first, then ream it out to accept one of those little plastic plugs that Ventilation It is important to ventilate the case correctly if you want the best possible frequency stability. Before drilling the bottom of the case, you will have to remove some of the plastic pillars that are in the way. This is best done with a spotfacing drill (see panel on drill Signal tracing A pair of headphones and a multimeter are the best de-bugging aids you can have for this project. Assuming the 1 0MHz oscillator is running , the 7216A should be putting out 500Hz signals on the digit and segment driver lines. You can verify this with the headphones. Use a 1k0-1 OkO resistor as a probe (to avoid loading the circuit and to protect the headphones), and listen to one of the digit driver outputs. A "spiky" 500Hz tone should be instantly recognisable, and all the digit drivers should sound the same. The segment drivers will have the same pitch but a different timbre as the numbers on the display change. Check that a similar 500Hz signal is also reaching the clock inputs of the 4024 and the 401 7s. As a matter of interest, follow the signal through the 4024 binary divider's seven stages . The 7 .8125Hz signal (at the output of the sixth stage) will sound like a series of fast clicks. To check the passage of the 500Hz signals through any of the 4016 analog switches , connect the headphones across the switch, using two resistors as probes. It is easy to tell when a particular analog switch is closed - it should be almost silent when its corresponding controlling Range or Function button is pressed. Faults in the signal path selection logic can be quickly tracked down using DC voltage measurements. The logic levels at the ECL gates are described in the panel on ECL logic but note that some of the DC control inputs have a logic O level less than +3.4V which is OK since any voltage between +3 .4V and ground will do for a "O" input. You can use headphones to trace a 50Hz test signal through the ECL preamplifier, Schmitt trigger, and the +5 and +2 counters (but see note below) . cover the screws on 240V wall outlets. Finally, fit four rubber feet to the bottom of the case. Construction in now complete. Calibration Before attempting calibration, you should allow a warm-up time of at least 15 minutes with the lid on. This will give the instrument time to stabilise and will ensure the best possible accuracy. To properly calibrate the instrument you will need a signal source of known frequency, with a stability better than one part per million. For most of us, this boils down to one of two choices: either local colour TV transmissions or the 10MHz standard transmissions from WWV Colorado or WWH Hawaii. The latter is preferred by the author because it has the same frequency as the crystal oscillator. All you need is a domestic shortwave The logic levels throughout the ECL signal path will be +4.3V (high) and +3.4V (low) , and the first two differential amplifiers will be biased in between at +3.8V. A special note of caution here: any external connection to the signal path will make the 100MHz preamplifier oscillate. This is why we chose to use a LED to monitor the Schmitt trigger output during the setting up procedure. You can minimise your chances of disturbing the circuitry by using a resistor as a probe for your multimeter. A value of 1 Ok0-1 OOkO is suitable for 1 OMO digital or FET analog multimeters, however you must cut the "probing" end quite short. Even a centimetre of lead will upset the sensitive wideband circuitry. If you are using a low-resistance multimeter, switch to a low voltage range and choose a resistor which wil l provide a convenient "multiplier" for the reading. The resistor acts as a low-pass filter, preventing feedback between the multimeter leads and the ECL circuitry at high frequencies . receiver and seven metres of hookup wire as an antenna. Wait until evening, when the ionosphere over the Pacific Ocean will reflect a fairly good signal, and position the DFM near the antenna so that its radiated 10MHz clock signal is about the same strength as the incoming signal. Adjust the 40pF trimmer with a plastic screwdriver until you have a zero beat. If the shortwave receiver has a tuning meter, you can watch the -beats when the frequency is too low to be audible. Adjustment to within 1Hz is often possible. Most colour TV transmissions are synchronised to atomic clock standards and the 4.43361875MHz colour subcarrier oscillator in your TV receiver is phase-locked to these transmissions. The easiest place to gain access to this signal is right at the 4.43MHz crystal. Use a probe made from a 10MO resistor shunted by a small capacitor of between lpF and 3.3pF, depending on the capacitance of the shielded cable you are using. You can tell whether the subcarrier frequency has been disturbed or not by keeping an eye on the colours in the picture. The 15.625kHz line scanning frequency is also derived from the same accurate standard and you may be able to pick this up from the stray radiation around the line output stage, without making an actual connection to the TV. Unfortunately, there are less digits to play with in this case so, to obtain the highest accuracy, switch to Period mode and adjust the trimmer for a reading of 64.0000µsec. If you decide not to calibrate the instrument, just set the 40pF capacitor to about 10% less than full mesh. This places a nominal 20pF across the crystal and the frequency will be within 10 or 20ppm, depending on the crystal tolerance. Note: some 10MHz crystals will require a nominal shunt capacitance of 30pF. You can easily tell if this is the case because you will not get a zero beat, even with the 40pF trimmer fully meshed. The cure is to add some additional capacitance across the trimmer and the existing 39pF capacitor. A pair of 15pF or 18pF capacitors should be sufficient. These are best soldered to the underside of the PCB with very short leads. If this still doesn't trim the crystal to the correct frequency, then you may have been supplied with a series-resonant crystal by mistake. All capacitors used around the crystal oscillator should be high quality types; eg, polystyrene or Philips NPO miniature ceramic plate. Fig.12 shows the warm-up drift of one of the prototypes which used a cheap 10MHz crystal and an imported power transformer which ran quite hot. Surprisingly, the drift was less than lppm. Better quality components will give even better stability. Your new 1GHz Digital Frequency Meter is now ready for work. We hope that you find it a useful addition to your range of test equipment. ~ JANUARY 1988 53