If you are experiencing bad MW reception then good news would be if the receiver has an external antenna connection point. It should also have a ground (GND) connection as it is rare such a point does not accompany an external antenna connection.
Before we start with the discussion of antennas: The problem with MW reception is the vast frequency range of the MW band being 540-1760kHz or just over 3:1 span. Using the formula wavelength = 300 ÷ f (in MHz) we see our MW band encompasses wavelengths from approximately 555 metres down to 170 metres.
Wire antennas, e.g. half wavelength, are only suitable for about 10% either side of the frequency for which they are cut (narrowing to only 3% when used for transmitting). With such a vast wavelength span having "a piece of wire cut to wavelength" (or proportion thereof) is just not feasible. Antennas, for MW, therefore have to be designed for wide frequency use.
Two rules need to be remembered with external antenna connections of this nature. Unless otherwise stated these inputs are usually high impedance. Also, most are simply capacitively coupled to the internal ferrite antenna and therefore the external antenna must not load the input as this will detune the input circuit thus defeating the object of using an antenna. Determining the input type, and thus the type of antenna to be used, is a fairly simple task.
First, touch the antenna lug with your finger. If the signal strength is drastically improved the chances are the input is high impedance. To confirm this connect a short wire (1.5 to 2m) to the lug, this should also make a drastic impact on the signal. Now add a second piece of the same length to the end of the first wire thus doubling the length of the wire. This should have a much smaller effect on the signal strength. If it appears to double the signal strength the chances are the input is a low (50 to 75 ohms) to medium (300 to 600 ohms) impedance type.
The next requirement is to determine if the input is capacitively coupled to the ferrite antenna system or not. First, remove any the antenna connected in the previous test. Tune to a station at the highest frequency possible and then short the antenna lug to the ground lug. If the signal is drastically reduced the system is likely to be coupled to the internal ferrite antenna system in the usual manner being a capacitor. What happens when you short the input is the capacitor is put in parallel with the ferrite antenna which detunes it. This indicates that the external antenna must not be large or of a design that will load the input circuit.
Having a signal on the end of an antenna without any reference point to work against renders it fairly useless. This is where the ground lug comes into play, "completing the circuit" so-to-speak. In order for the maximum amount of signal to be fed into the receiver the ground lug needs to be secured as best possible to Mother Earth. A wire (and the shorter the better) from a buried cold water pipe serves superbly for all wire antennas. Loop antennas will not require the ground lug to earth connection as the signal appears across the loop and the circuit is completed at the receiver.
Back to the main subject: There are basically three types of antenna available to you. The first is the only one usable on high impedance inputs, the first and second for medium, and all three for low impedance (the effectiveness being the only deciding factor here).
Although there is extra signal this amount of capacitance will undoubtedly detune the input, the extent of the detuning determining the amount of benefit of the extra signal. There is a point at which any extra length, which detunes the input through the extra capacitance, would counteract the extra signal available.
A good starting point, using a station at the top end of the band (as tuned circuits are more affected by the capacitance at higher frequencies), would be a 5m (16ft) antenna. Then double this to 10m (33ft) and noticing if the signal improves dramatically. If it does then extend this by 5m a time until 20m is reached, each time listening to the extra signal received. If no significant improvement is heard then remove the last extension made and call it a day.
If the very first extension from 5 to 10 metres seems to not improve things much then halve the length to 2.5m and notice if there is any serious degradation. If so then 5m is the length required. If no degradation is noticed then even this small length is already affecting the receiver's antenna tuned circuits. Unfortunately there is no golden rule here and experimentation will be necessary.
Long-wires unfortunately require a very large back yard as they need to be over a half wavelength in length before being effective enough to warrant their set up. Setting up a long-wire, however, is simple. Just point it at the station required and you're done. Long-wires are suitable for vertical polarisation although they do work better with horizontal, which makes them slightly unattractive when trying to blot out interference from power lines (being horizontal makes them emit any interference in a horizontal plane). Long wires are directional devices so best reception tends to be limited to one direction.
Loops are also directional but their size allows for the direction to be changed as required. Usually some experimentation is needed but a good starting point is a frame made about 1m X 1m (3 X 3 ft) wrapped with 50 turns of fairly thin insulated wire, with connection points made available at every 10 turns. Connect the 'start' to the ground lug and then experiment with which point (10, 20,...,50 turns) works best when connected to the 'Ant' input. Usually one 'tapping' works fairly well across the whole band. Using this loop is a two fold process in that the station should be found then the antenna turned for best reception.
This is a derivative of the loop known as the "tuned loop", but we believe the scope of this paper is for simple alternatives and have steered away from such complexities.
One feature of a quarter wavelength antenna is that it also operates at odd multiples of frequency i.e. that which works at 550kHz will also work at 1650kHz, so only one piece of wire will cover both ends of the band.
There is one antenna that is open to the more adventurous (and those who have an electronics store available to them) and that is the "travelling wave". As said previously MW stations transmit using vertical polarisation which makes this antenna a little simpler as only one leg needs to be constructed (using a horizontally orientated would mean twice as much work. Simply using a quarter wavelength the lowest frequency (in our case 540kHz) and inserting a 100 ohm at the one-third point and a 470 ohm at two-thirds point we effectively create an antenna with about 3 to 4:1 frequency ratio. Dimensions are shown below. As this antenna is only for reception we require nothing more than ¼ watt resistors.
Mount this antenna between a high tree with the other end directly connected to the receiver. Should there be a long distance between the antenna base and the receiver (i.e. a large proportion of the antenna lands up in the house) then use a piece of 75 ohm TV coax from the receiver to the antenna base, not forgetting the earth also has to move to the end of the coax as shown.