NOT ALL THAT long ago, almost the only items of domestic equipment operating on a frequency above 1GHz were microwave ovens, all of which use a magnetron operating at 2.45GHz (the frequency which causes maximum heating of water molecules). But nowadays all kinds of equipment transmits and/or receives at frequencies above 1GHz. For example many cordless phones operate at frequencies around 2.4GHz, sharing these frequencies with wireless CCTV cameras, AV transmitters and receivers, security systems, remote access locking systems and baby monitors.

Other items using frequencies in the 2.4GHz region include "WiFi" (802.11b & 802.11g) computer networking gear and "Bluetooth" wireless links for computer peripherals (802.11a wireless networking equipment operates on even higher frequencies, at about 5GHz).

The UHF Prescaler circuit is housed inside a standard diecast aluminium instrument case which provides the necessary shielding from stray signals.

Then there are wireless internet service providers, which mainly use frequencies around 1.9GHz or 2.6GHz and there are "3G" digital mobile phones which operate on frequencies of around 2.1GHz in metropolitan areas. We mustn’t forget GPS receivers either. These operate on frequencies of 1.57542GHz and 1.2276GHz.

So how can you check the operating frequency of any of these devices, when the range of most reasonably-priced frequency counters only extends up to 1GHz? Well, you can either fork out the dough to buy another counter that is capable of measuring up to 3GHz or so, or you can build yourself the UHF Prescaler described here. This simply connects "in front" of your existing counter and divides the frequency of the signals you want to measure by exactly 1000. So 1.5GHz becomes 1.5MHz, 2.45GHz becomes 2.45MHz and so on, allowing you to read the incoming frequency directly and without any mental arithmetic.

The Prescaler uses some special high speed ECL (emitter-coupled logic) ICs to perform the 1000:1 frequency division and these are able to operate at input frequencies up to at least 2.8GHz. And because the output frequency of the Prescaler is still only 2.8MHz for an input of 2.8GHz, this means that it should be suitable for extending the range of just about any counter. In fact, it would be a good companion for the 50MHz Frequency Counter described in the October 2003 issue of SILICON CHIP.

So if you want to be able to measure frequencies up to at least 2.8GHz with your trusty old lower frequency counter, this project is for you. All of the components and circuitry are on a single PC board and although there are quite a few very small surface mount parts to fit on the board, this isn’t unduly difficult providing you take it slowly and carefully. You will need a soldering iron with a very fine chisel-shaped bit, plus steady hands and an illuminated magnifier to help in seeing what you’re doing.

We’ll also give you a few tips on manual soldering of SMDs (surface mount devices) in the accompanying panel.

Circuit description

In terms of its basic operation the Prescaler is pretty straightforward, as you can see from the block diagram of Fig.1. The incoming UHF signals are first passed through wideband input amplifier IC1, to make the Prescaler reasonably sensitive. The boosted signals then pass through a high-speed divide-by-four stage using IC2, which is basically a pair of very fast ECL flipflops in cascade.

Fig.2: the block diagram for the UHF Prescaler. The incoming signal is first amplified and then divided by 1000 using IC2, IC3 & IC4. It is then fed to two separate output sockets via transistors Q1 & Q2.

The output of IC2 then passes to IC3, which is another very fast ECL counter programmed to divide by 125. So the output from IC3 is a signal with a frequency 1/500th that of the UHF input signal.

Because the output of IC3 is in the form of very narrow pulses, we then pass them to IC4. This is an ECL JK flipflop, connected here not only to divide the frequency by a further factor of two but also to provide square-wave outputs so they’re more suitable for triggering low-frequency counter input circuitry. Then to make the outputs even more compatible with virtually any common frequency counter or scope, we finally pass them through a simple logic level interface stage using transistors Q1 and Q2.

For a more detailed understanding of the Prescaler, let’s refer now to the main circuit diagram – see Fig.2.

The UHF signal to be measured enters via CON1 and first passes through an input termination and overload protection circuit formed by two 100Ω resistors and diodes D1 & D2. The two resistors are in parallel to provide an input termination of 50Ω, while D1 & D2 are 1PS70SB82 very low capacitance Schottky barrier diodes, having a very low forward voltage drop. Because they’re connected in inverse parallel, they limit the input signal level to no more than 2V peak-peak.

The signal is then coupled to the input of IC1 via a 10nF capacitor. IC1 is a Mini-Circuits ERA-2SM monolithic broadband amplifier device, with about 12dB of gain up to over 5GHz. IC1 is fed with DC power via its output (pin 3), with the 47Ω resistor chosen to set the correct operating current. As the power feed is effectively in parallel with the output of IC1, choke RFC3 is used to provide a reasonable load. This choke is a Mini-Circuits ADCH-80A, a special very wideband device chosen because it has a very low parasitic capacitance and is therefore not self-resonant at frequencies below about 8GHz.

From the output of IC1 the boosted signal is fed to the clock input of IC2 via another 10nF capacitor. By the way, it’s the value of the coupling capacitors at the input and output of IC1 which determine the lowest frequency that the Prescaler will work at. The 10nF capacitors as shown allow it to work down to below 50MHz. The reason why we don’t use larger values to extend the range even lower down is that larger value capacitors tend to self-resonate at frequencies below 4GHz – which we don’t want because it would lower the maximum frequency of operation.

IC2 is our first and most critical frequency divider and it’s an MC10EL33 device from On Semiconductor. This is an ECL divide-by-4 device with very impressive specifications. It can operate at input frequencies up to at least 3.8GHz and has a propagation delay of less than 800ps (picoseconds!). It even includes its own bias voltage source (Vbb, pin 4) which is used to provide the correct ECL bias for its two inputs (via the 2.2kΩ resistors).

IC2 has complementary outputs (pins 7 & 6) which both need to be tied to ECL low logic level via termination resistors of close to 50Ω. Here we use 56Ω chip resistors, because this value is more readily available than 51Ω.

From pin 7 of IC2 the signal (now 1/4 the input frequency) passes directly to the clock input of IC3, an MC10E016 ECL 8-bit programmable synchronous binary counter able to count/divide input frequencies up to at least 700MHz.

We have programmed it to divide by 125, by tying its parallel load inputs (P0-P7, pins 3-7 and 21-23) to the appropriate ECL logic levels. For division by 125, we set the parallel inputs to the binary code for 256 - 125, or 131: ie, 10000011. Note that the ECL high or "1" level is established by the 75Ω and 430Ω resistors, forming a voltage divider across the 5V supply rails.

The rear panel provides access to the two BNC output sockets and the DC power socket.

Fig.4: the mounting details for the PC board. Note the aluminium heat-sink under IC3.

The output signal from IC3 (1/500 of the input frequency) appears at the terminal count or TC-bar pin (19), which again must be tied to the ECL logic low level via a terminating resistor (here 51Ω, because it’s a standard leaded part). The ECL logic low level is established by ZD1, a 3.3V zener diode.

By the way if you’re wondering where the current for ZD1 comes from, to establish the nominal 3V level, it’s sourced from the various ECL outputs tied to it via the termination resistors, plus the inputs of IC3 that are connected directly.

As mentioned earlier, the output signal from IC3 is low in frequency (below 8MHz) but it’s in the form of very narrow pulses which would probably pose problems for the input circuitry of many low-frequency counters. That’s why we don’t program IC3 to divide by 250 (which is easily done).

Instead, we program it to divide by 125 and feed its output to a third ECL device, IC4. This is an MC10EL35, a very fast JK flipflop with its J and K inputs tied to ECL logic high level so it operates in toggle mode as a divide-by-two counter.

So at the complementary outputs (pins 7 and 6) of IC4 we finally get output signals of exactly 1/1000th the input frequency and, just as importantly, in the form of symmetrical square waves which are much more compatible with typical counter input circuits. The outputs of IC4 are again tied to ECL logic low level via 51Ω terminating resistors.

Since the outputs from IC4 are still switching between ECL levels (nominally +3V and +4V), the remaining step is to pass them through a level translation and output buffer/interface circuit, to provide them as buffered low-impedance signals referenced to ground. This job is performed by transistors Q1 and Q2, connected as a differential switch. This has the advantage that it allows us to easily provide the Prescaler with two independent outputs, so that it can drive either two different counters or perhaps a counter and an oscilloscope.

Because all the Prescaler circuitry operates from a single 5V DC supply, the power supply is very straightforward and involves only a 7805 regulator (REG1), driven from an external 9V DC plugpack. Although the total current drain is about 190mA, giving a regulator dissipation of about 800mW, the regulator is provided with a small heatsink so it keeps reasonably cool.

Construction

As you can see from the photos, all the Prescaler circuitry is on a double-sided PC board measuring 111 x 81mm and coded 04110061. This board has rounded cutouts in each corner so that it fits snugly inside a standard diecast aluminium instrument case, measuring 119 x 93.5 x 34mm. It’s actually mounted on the box lid, which forms the Prescaler’s base.

Fig.3: this is the full circuit diagram. IC1 is the input amplifier and this provides about 12dB of gain. The boosted signal is then divided by four in IC2, by 125 in IC3 and by two in IC4. Q1 & Q2 buffer the complementary outputs from IC4 and drive the output sockets.

All the connectors, power switch S1 and the power indicator LED (LED1) are mounted on the top of the board, along with the regulator (on its heatsink), transistors Q1 and Q2 and the other leaded components. The surface-mount ICs and other components are mounted on the underside of the board.

There are quite a few connections between the two copper layers of the board but these aren’t likely to pose a problem even if you don’t get a board with plated-though holes. Some of the connections are achieved simply by soldering the leaded component leads on both top and bottom, while the others are mostly "vertical links" between the upper and lower groundplane copper areas. These links are easy to make using short lengths of tinned copper wire (eg, resistor and diode lead offcuts).

The location and orientation of all the parts on both sides of the board are shown clearly in the two PC board overlay diagrams of Fig.3, so you shouldn’t have any problems if you use these and the photos as a guide.

Since there are quite a few surface-mount parts (SMDs) to fit to the board as well as the leaded parts, we recommend that you assemble everything in the order set out below.

First, fit the various connectors to the top of the board, beginning with CON1, which is a reverse polarity SMA socket. Follow this with CON2 and CON3 (the BNC sockets) and finally the DC power input socket (CON4). That done, fit the DPDT power switch (S1).

Fitting the SMDs

Next, turn the board over and lay it "bottom copper up" on your workbench, using a small block of wood or plastic if necessary to support it. This will then allow you to fit all of the surface-mount devices, with a minimum of obstruction. Fit the chip resistors first, then the chip capacitors and finally the input protection diodes (D1 & D2), the ICs and RFC3.

Fig.3: install the parts as shown in these two diagrams. The red dots show where you have to solder to both sides of the board and where to install vertical wire links (but only if your board isn't supplied with plated-through vias).

We have prepared an accompanying 2-page panel with some diagrams to guide you in manual assembly of the various SMD parts. There’s also a photo of a small rotary "SMD work table" which you might like to duplicate. We also recommend the use of a magnifier lamp – ie, the type that’s fitted to an articulated, spring-loaded arm.

After you’ve fitted all of the SMD parts, the board can be turned over again and the smaller leaded parts fitted, including the resistors, RFC1 and RFC2 and the small capacitors. As mentioned earlier, some of the leads of these parts are used to make connections between the top and bottom copper – so remember to solder the leads concerned on both sides. They’re identified with a red dot on the PC board overlay diagrams of Fig.3.

If your PC board is not provided with plated-through hole vias, there will also be quite a few "vertical links" to fit, to provide low impedance links between the top and bottom copper. These are also identified on the overlay diagrams with a red dot, so don’t forget them. They can be made using resistor or diode lead offcuts – just don’t overheat or dislodge any of the SMD parts nearby when you’re soldering them in place.

Above: the top of the PC board carries all the leaded components, along with the sockets, the power switch, the indicator LED and the regulator and its heatsink. Keep all leads as short as possible.

Next fit LED1, the Prescaler’s power indicator. This mounts in the front centre of the board, with its leads bent forwards by 90° so that it lines up with CON1 and switch S1. Position it so that it will later protrude through its mating hole in the front panel.

The final parts to fit are power diode D3, the two electrolytic capacitors and regulator REG1. As shown on Fig.3 and in the photos, the regulator mounts flat against a small 6073 type TO-220 heatsink and this assembly is secured to the board using an M3 x 6mm screw and nut. Tighten the screw before soldering the regulator’s leads, to avoid stressing the solder joints.

Functional checkout

At this stage your Prescaler should be electrically complete and ready for a quick functional checkout before it’s fitted into the box. To check it out, place the PC board assembly on a clean timber or plastic surface and connect 9V DC supply (eg, from a 9V 250mA plugpack or similar) to CON4. The positive input should connect to the centre pin of CON4.

Right: the surface-mount devices all go on the reverse side of the board. Refer to Fig.3 and to the 2-page panel in this article for the details on mounting these.

Now turn on power switch S1 and you should see LED1 light up. This will confirm that LED1 is fitted with the correct polarity and also that REG1 is providing a +5V supply rail to the Prescaler’s circuitry. To make sure that the supply voltage is correct, you can check it with a multimeter or DMM, connected between the centre and output pins of REG1.

You can also check the voltage across zener diode ZD1 which should measure about 3.1V if the ECL circuitry is working correctly.

Self oscillation

If all seems well so far, try turning on your frequency counter and connecting its input to one of the Prescaler’s outputs (ie, CON2 or CON3). You may well find that the counter shows a reading straight away, even with no input signal applied to the Prescaler as yet. That’s because IC2, the Prescaler’s input divider, tends to self-oscillate when there is no input signal. So if you connect the second Prescaler output to a scope, you’ll probably see a squarewave of about 1.6MHz.

There’s no cause for concern about this self-oscillation because as soon as you feed in a "real" UHF signal, it stops. The Prescaler’s output changes immediately to a square-wave with a frequency 1/1000 that of the input signal.

If you have a source of UHF signals like a wireless CCTV camera or an AV transmitter module, try connecting its output to the Prescaler’s input via a suitable SMA cable (note: you may need an SMA/RP SMA adaptor at one or both ends of the cable, depending on its own connectors). The counter should immediately begin reading its carrier frequency or strictly, 1/1000 of its frequency. So if the camera or AV transmitter module is operating at say 2.432GHz, the counter will read 2.432MHz.

Final assembly

If your Prescaler passes this quick checkout with no evident problems, you’ll now be ready to assemble it in the box. This assumes that your box and its lid have been prepared, with of the holes shown in the diagram of Fig.6 having been drilled. If the box hasn’t been drilled yet, then now is the time to do so.

Fig.5: these full-size artworks can be copied and attached to the front and rear panels of the case. Cover them with wide, clear adhesive tape before attaching them, to protect them from damage.

Note that the holes for the BNC connectors in the rear of the box are extended to form slots, so the box can be slipped down over the connectors.

As mentioned earlier, the PC board assembly is mounted on the lid on 6mm-long untapped metal spacers. It’s then secured using six M3 x 15mm countersink-head machine screws, as outlined below.

Before the board is fitted, attach the small aluminium heatsink plate to IC3, the PLCC28 device. This IC gets fairly warm in operation and the plate helps keep it cool by conducting heat away to the box lid.

The plate is first prepared by smear-
ing it thinly on both sides with heatsink compound. That done, press one side to the top of IC3’s body, sliding it around a bit so any air bubbles are worked out. Then position it squarely over the IC body, where it will tend to stay put until you fit the board assembly to the box lid.

Attaching the board assembly to the lid is straightforward if you first fit the six countersink head screws through the lid holes and then turn the lid over and place it on the workbench. You then fit one of the 6mm spacers on each screw before lowering the inverted PC board assembly into position. Be sure to press the board down gently just over the position for IC3 (see Fig.3), so that the heatsink compound on the lower surface of IC3’s heatsink plate is partly transferred to the box lid underneath, to form a good thermal bond – see Fig.4.

After this, you can fit an M3 star lockwasher on the top of each board mounting screw, followed by an M3 nut. It’s then just a matter of carefully tightening each mounting screw and nut to secure the board and sandwich the aluminium heatsink in position.

Fig.6: the drilling details for the metal case. Drill pilot holes for the larger holes first, then carefully enlarge them to size using a tapered reamer.

The final assembly step is to fit the box over this assembly. To do this, first remove the nuts and lockwashers from BNC connectors CON2 and CON3 and also remove one nut, the keyed flat washer and the lockwasher from power switch S1. Thread the remaining nut right down to the switch body and then refit the keyed flat washer with its locating lug facing towards the switch body. This washer should also be down against the nut.

Now you should be able to bring the inverted box down over the PC board/lid assembly, at an angle so CON1, LED1 and switch S1 can be mated with the matching holes in the front end of the box. The box can then be lowered at the rear end and moved back at the same time, until the slots in its rear slip down around the threaded ferrules of CON2 and CON3. The box/cover will then be fully mated with the lid, allowing you to invert the complete "shebang" and fit the four box assembly screws.

After this, all that remains is to fit the front and back panel dress stickers to the box (see Fig.5) and finally, refit the remaining nut to power switch S1 and the nuts to CON2 and CON3 at the back. Your UHF Prescaler should now be finished and ready for use.

One final tip: when you’re screwing SMA cable connectors and adaptors to the Prescaler’s own input connector, be careful. These connectors are designed for precise mating, so they can operate reliably, with low losses up to about 8GHz. As a result they’re small and have a fine thread, making them easily damaged by rough treatment.

How to manually solder SMD parts Many surface-mount components or SMDs are very small – the 0805 size chip resistors are only 2 x 1.3mm, while 1206 size chip capacitors are only slightly larger at 3 x 1.5mm. Many SMD IC packages have leads spaced only 1.27mm apart. SMDs are not really designed for manual assembly but it’s quite feasible to fit many of the more common types by hand if you take care and use the right tools. For a start, your soldering iron should be fitted with a fine chisel-point tip, which should be well tinned and kept as clean as possible. Ideally it should be of the low-power temperature-regulated type as well. You also need to use fine-gauge resin cored solder, ideally no more than 0.8mm in diameter. It helps a great deal if your PC board has the copper pads solder-plated, as this makes it much easier to fit the SMD parts. Manual assembly of SMDs is also a lot easier if the board is held horizontal and level, as they’re less likely to move out of position while you’re soldering them. In many cases, you can simply place the board flat on your workbench copper side up, although if there are leaded parts already mounted on the other side of the board you may need to support it using small blocks of wood, plastic or metal. Because it often helps to be able to rotate the board for easier soldering at each end or side of an SMD, I made up a small rotary work table by adapting a ball-bearing swivel base from an industrial castor wheel assembly. By removing the wheel and axle and then bending the upper ends of the fork sides outwards at 90°, I made a fairly sturdy rotating bracket (it even has a brake lever, which can be used to lock the table and prevent it from rotating). This rotary SMD work table was made up using a ball-bearing swivel base, an aluminium plate, some support blocks fitted with clamp brackets and a pivoting arm arrangement fitted with a pair of crossover tweezers. A thick aluminium block forms the base. The swivel flange was then attached to a block of aluminium to serve as a base, while a 120mm square of 4mm aluminium sheet was fashioned into an octagonal plate with a 6mm centre hole and 3/16-inch holes tapped in each "corner" for fastening board clamp screws. Two further holes were also drilled in the plate to line up with the former axle holes in the bent-over fork ends, so the plate could be bolted to the top of the fork to form the actual operating table, with its centre hole directly over the centre axis of the base swivel. You can see the basic construction in the photos, which also show three of the support blocks and clamp brackets I fashioned to hold boards in place. Also visible is a pair of modified crossover tweezers mounted on a pivoting arm arrangement, which can be used to hold some SMDs in place while they are soldered – a kind of "third hand". Such a work table is not necessary for all SMD work but it might be worth considering if you’re likely to be building up quite a few projects. Another useful accessory for manual SMD work is an illuminated magnifier – a magnifying glass about 120mm in dia­meter surrounded by a circular fluorescent lamp in a metal hood that’s mounted on an articulated, spring-loaded arm attached to a swivel base (so you can position it easily just above the operating table). They’re not cheap but if you’re likely to be doing a fair bit of manual SMD or just fine PC board assembly, they are a good investment. One at a time Before we go any further, here’s an important tip: when you have quite a few SMDs to solder to a board, handle them one at a time. If you try to tackle more than one at a time, it’s all too easy to accidentally send one or more flying off while you’re concentrating on soldering the first one in position. To handle tiny 0805 and 1206 size SMD chips and bring them to the board, use a small pair of stainless steel cross­over tweezers. They’re available in almost any Asian bargain store, either alone or in sets of tweezers for only $2. Having brought each part to the board, release it from the tweezers and carefully nudge it into position over its mating copper pads, using either the tip of the same tweezers or the point of a small wooden toothpick. That done, hold the part in position using either the toothpick or a pair of modified crossover tweezers as a clamp, while you clean the soldering iron tip and then melt a very small amount of solder onto its end. The tip is then brought up to one end of the SMD, at a fairly low angle so the tiny drop of solder comes into contact with both the board copper and the end of the SMD (see Fig.7). The iron tip is only in contact for about half a second – just long enough to allow the drop of solder to tack-bond the two together and hold the SMD in place. The toothpick or tweezers can now be removed and you can solder the other end of the SMD in the more "normal" fashion before returning to the first end and quickly re-soldering it properly as well. The sequence is shown in Fig.7. The same basic approach can be used with SMD diodes, transistors and ICs, with slight variations to suit the various packages. The idea is to hold the SMD in position using a toothpick or crossover tweezer clamp while you tack-solder one of its leads to hold it in place. That done, you can remove the clamp and solder all of the remaining leads properly – and finally, the first lead again. Doing this is much the same whether the SMD has flat horizontal leads emerging from underneath, S-shaped leads that bend outwards at the bottom or J-shaped leads that bend inwards and underneath. Fig.8 shows the idea. About the only kind of SMD package you can’t solder in this way is the type with no leads at all – just "solder bumps" underneath. These really aren’t suitable for manual soldering. One last tip: whether you’re soldering SMD chip resistors, capacitors or other devices like diodes, transistors and ICs, make all joints as quickly as you possibly can while at the same time taking care to make a good joint. The faster you make the joint, the lower the risk of damaging the SMD by overheating (which is very easy to do, since they’re so tiny). Also use the smallest amount of solder necessary to make a good joint – the less solder you use, the lower the risk of accidentally bridging between device leads with a blob of excess solder.