Fig 1. (MK 1)	F 2. (MK2)

try but to our surprise we were unable to make any real improvement on the first circuit using the general principles. We could have reduced the package count by using an LM389 (which in~cludes three independent transistors plus an audio output amplifier) but that would have cost more with no real change.
In the original design the trans­mitter was modulated and the peaks of the detected signal were gated and enormously amplified (See How It Works and Fig la). Although we refer to the signal being modulated, it was actually switched on and off and this resulted in ringing in the tuned circuit.
After literally three weeks solid experimenting we decided to take another approach. We decided to dispense with a modulated transmitter and work with DC until the final stages. In the original design the audio frequency was fixed, being dependent upon the modulator and metal was sensed by an increase in audio level. However, our ears are highly insen­sitive to changes in level but they are however, very sensitive to a change in audio frequency. Once we had decided to tackle it from this side everything fell into place. For a long while our voltage controlled oscillator was a unijunction transistor and although we achieved excellent results we were not satisfied with the unit in practice and eventually adopted the circuit shown in Fig.3.

FIp. la (above) shows the block diagram of the Mark 1. In thus the peaks of the modulated signal wore gated and enor­mously amplified. Diagram below shows the new arrangement. the RF signal, which is unmodulated, is converted to a DC signal which delves a voltage controlled oscillator (VCO).

IRMk2

The Coil
We cannot emphasise enough that the search head is the key to the whole operation: be prepared to spend some time on this, our own workshop is full of discarded experiments.

The housing of the coils is not important. In the Mkl we adopted a circular head but this is difficult for the non-woodworkers to tackle so we went for a rectangular shape. The coils Li and L2 should be sandwitched between two pieces of hardboard or plywood separated by thin battens about 6mm thick. The top should be built first and the battens fitted for a better appearance you can then file off the corners slightly.
To wind the coils you’ll need to get hold of a cylinder about 140mm (5’Ain) in diameter. Using 32 swg enamelled copper wire, trap one end onto the former with a piece of tape and carefully wind 40 turns as close together as possible. Carefully remove the coil and then wrap tape around it at intervals to keep it from spreading.
Two identical coils are required.
Lay one of the coils into the dish formed from the top of search head and the battens as you see in the photograph and spot glue it into place except on the part near the middle. Lay the other coil next, again spot gluing it except near the middle.
A hole should be made in this piece of wood to feed through the connecting cable to the main circuit. This cable must be a four-wire type with individual screening the screens are not used at the search coil end but don’t cut them too far back: we still have a few experiments to try out on our prototype and access to this screening may be used.

The Control Box
The circuit should be built up next. Everything except for the controls, the speaker and the meter are on a single POB. Building this up should present few problems. Spacing is designed for eighth watt resistors and tantalums are used, again to save space though the control box has plenty of room in terminal pins to the points shown in the PCB overlay as this will make connections far easier to make later on.
Assuming you haven’t got the coil in exactly the right position by luck in the original setting, you should get an audio tone of about 700Hz from the speaker and the meter connected) will be hard over. If you don’t get this, adjust RV 1 and it should appear: Back off RV1 until the frequency falls and then increase it a bit so that the tone is slightly higher than the minimum.
Now gently and slowly bend the coils and adjust the overlap till the tone falls. Add a fewmore blobs of glue but leave yourself with some adjustment. Readjust RV1 again and repeat. Continue to do this until you can no longer get any lower adjustment on RV1. Now check that no metal is in the vicinity (don’t forget cuff-links, watches and rings) and continue the manipulation. If you use a scope, monitor the level of the signal of the collector of Q2: when you are near to a minimum the level should fall considerably. If all works as described, bringing a piece of metal near the coil should result in the frequency rising. If the frequency falls instead of rising, continue adjusting. Near the minimum you can reach a point where the metal firstly adds to the cancellation.
Don’t glue down the final tiny, tiny adjustments until you are quite certain that all is OK. The amount of final adjustment is extremely critical as you’ll find out.

General construction
The general design can be seen from the photographs. We used a Verobox to house the main circuit and cut a piece of broom-handle at an angle and fitted a bicycle hand-grip to this. The stem is made up from Marley 22mm cold water plastic tubing, available from many plumbers. The connection to the search-head was accomplished by softening a short length of the stem plastic in hot water and quickly clamping this in a vice. The connectors on the stem are also Marley fittings. The heart of the circuit is the search coil, L1 and L2, These two coils, which are essentially identical, are arranged in the same plane with a small overlap in such a way that there is is practically no inductive coupling between the two.
There is a minimum pickup when the fields generated in L1 are cancelled in L2 when in free air. Any metal brought into the magnetic field of L1 will distort the field, causing pickup in L2. Q1 is a straightforward Colpitt’s oscillator working at a nominal 130 kHz. This type of circuit is very stable and the use of polystyrene capacitors also help with stability. The supply to this stage is separately decoupled by R4 and C1. The pickup coil L2 is tuned by means of C4 and C5 and amplified by Q2 which feeds th the level control RV1. This controls the ‘free air’ state of the circuit and is set to the point where the later stages are just operating. The signal is further amplified by Q3 (here it is still an RF signal) and is detected by D1 and D2. When no metal is in the vicinity of the search coil and with RV1 correctly adjusted, a DC voltage of about 500 mV appears across C8. R9 increase the effective input impedance of Q4 as seen by the detector stage. Q4 is just held off by the voltage available but as soon as metal distorts the electromagnetic field, L2 produces a large RF signal, a higher voltage across C8 and a consequent fall (from 8V) in the voltage at the collector of Q4. This voltage is also monitored by the meter in parallel with the load resistor of Q4. The fall in voltage is dependent upon the proximity and/or size of metal near the search coil. It is necessary to ensure that the DC voltage fed to the next stage is clean and R12 and C9 act as a a filter to remove any residual AC even if this is at low frequencies.
IC2 (the next but one stage) is a voltage controlled oscillator – but to operate this so that metal is indicated by a rising note, rather than a falling one, the voltage at the junction of C9 and R12 has to be inverted and this is achieved by IC1: in ‘no metal’ conditions there is about 2 V at the output of this op-amp which rises when metal is near.
This stage quickly saturates to give about 7 V at pin 6. IC1 has unity gain. IC2 is a voltage controlled oscillator in ‘no metal’ conditions it gives about 70 Hz which rise to 500 Hz when metal is present. Diode D3 gives a rapid recharge to C12 and affects the mark/space ratio of the output which results in lower battery consumption. R20 and C12 can be altered to give a different range of frequencies if desired. The output is taken to the volume control and fed to the LM380 audio power amplifier which in turn feeds the speaker.
The levels of signal around Q2,3,4 are all dependent upon transistor gain, temperature and supply voltage but this doesn’t matter because the level control RV1 is adjusted until Q4 just begins to conduct. Current drain for the complete circuit is in the order of 50 mA.

Re printed from ETI TOP PROJECTS NO7 (1979)