The magical audio filter refined
and adapted for the Yaesu FT-301
by sv3ora


On the November 1983 issue of QST, Jim Pepper W6QIF has released the "Magical audio filter". By reading at the article, it seemed suitable as a companion AF filter for my Yaesu FT-301. This circuit, while not the sharpest filter possible, offers quite a lot of combinations to shape the AF signal accordingly and reduce noise, while still being able to recover the audio intelligence. For a detailed read of how the filter works, read the original article. Here I present my modifications, which add more valuable features to the filter and solve a series of problems in it that I have detected.



The complete modified schematic of the filter is shown above. After finishing building the original circuit, I found that the bandwidth of the peak filter can be varied too, so I replaced the switch and the 1K in the original circuit with a 10k potentiometer. 10k is enough to ensure the same minimum bandwidth as in the original circuit. When you set this potentiometer to the maximum bandwidth, the peak filter is essentially disabled. It's audio output volume is then the same as that of the incoming signal. Based on the above findings, the peak switch and the series capacitor are not needed anymore, I just set the potentiometer to the maximum bandwidth to disable the peak filter. This is a similar action to the one used on the notch filter.

Some potentiometers have been replaced with logarithmic versions so as to make the settings more linear. You have to set these logarithmic potentiometers to the correct "polarity" to benefit from this modification. I have not checked the polarity on the schematic, so they may appear backwards connected.

The filter works fine (no differences in volume or settings) with 13.5v instead of 16v. This voltage is convenient because you can take it directly from the FT-301 without the need for an extra power supply. On other transceivers 13.5v (or 13.8v) is more convenient, as every shack with a transceiver has one.

I have also connected one or two 8-ohm speakers in parallel at the output, no problems driving them.

I have also added another capacitor at the input, to ensure that there is no DC at the input potentiometer, since the FT-301 has some internal resistors connected to the top of this potentiometer and I do not want to interfere with them.

There is one thing I really did not like about the filter and this was the hiss that was heard even when no input signal was present. Most of this hiss comes from the filter stages, not the AF amplifier stage. When I set the notch at high frequency this hiss is reduced of course. I have replaced all the 741 with TL071 and the hiss was there. I have tried 33nF in different places to ground as a crude hiss filter, but it did not make any difference. Instead of altering the circuit, I decided to use a stereo potentiometer for the gain control. One gang of this potentiometer is connected to the input of the filter and the second gang to it's output. That way, when the audio volume is set to low levels, the hiss is also attenuated and becomes almost unnoticed. At high audio volumes, the received noise overcomes the filter hiss anyway.

The bass response of the original circuit was not that good. This is mainly the AF amplifier issue. I changed some capacitors in the AF amplifier and now the bass response is much better, close to that of the FT-301. I have also replaced the 470nF capacitors to 1uF, since this did not make any audible difference, so one has to order only one value of capacitors instead of two, when builting the filter.

When connecting the filter to the FT-301, power is taken from J11 on the FT-301 and the audio source is taken from the front panel phones plug. Initially I used pin 1 of the J4 connector on the FT-301 to take the audio out. However, this had two disadvantages. First, the internal speaker of the FT-301 was not disabled when the filter was connected. Second, I noticed some weird audio hiss when I switched from TX to RX. By taking the audio source from the front panel phones connector, both of these problems were solved.

During TX, in some bands and at high powers, the filter/amplifier chain was picking leaked RF (no it did not pass through the audio cable). This RF was rectified at some point and the demodulated audio was heard on the speaker. Instead of shielding the filter/amplifier chain, which might be more unpredictable, I decided to add a mute circuit, which would mute the audio circuit during TX. The simplest mute circuit and one that worked very well, was to disconnect the speaker during TX using a relay. The relay is controlled from the FT-301 by the pin 5 on J4.

While this mute circuit worked well, it introduced another problem, clicks to the speaker during the TX to RX switching. To solve this problem, an adequate relay switching delay was introduced, using a series diode and a parallel capacitor to the relay coil. The value of the capacitor has been chosen so that it allows for enough delay for the clicks to be muted, whereas allow for immediate audio un-mute during the TX to RX transition. In very fast digital modes, you may want to decrease the value of this capacitor if break-in operation is desired, at the risk of not completely un-mute the clicks. For SSB voice and CW the value chosen was proven adequate.

I like how the filter works and with the modifications I did, it is even better now (at least for me), as only 4 pots are needed and no bypass switches (no clicks introduced in audio and no more holes at the chassis etc). The very useful peak filter bandwidth control has been added, the bass response has been improved, the hiss has been made unnoticeable, the clicks have been removed and the connections to the FT-301 have been optimized.

The peak filter is not super sharp i.e. you do hear for example the rest of the signals close to the peak or further. However, the peak is pronounced, so you can use the gain control to reduce the peak back to the same level, whereas at the same time attenuating the rest of the signals. And when you use the notch you can reduce one of the unwanted signals even more. I am not sure which way is better, a super sharp AF BPF or a more "ralaxed" tunable AF BPF combined with a tunable AF notch filter? To my view, if sharp audio filtering is the goal, it might be better to be done at the IF stage, using a CW crystal filter. After all, that's why CW crystal filters exist. Since most commercial transceivers already have one, either built in or as an accessory, there is no need for a really sharp AF filter.

As far as concern the filter section, a bit of warning. A sharp filter, either IF or AF, may sometimes cause problems in operation. When for example you scan a band for CW signals with this sharp filter enabled, you have to do it really slow, so as not to miss out signals. Scanning the band really slow, increases your chances of missing out signals. The cure is to switch to the SSB or the wide AF filter and scan the band. When a signal is found then switch on the CW or narrow AF filter and focus on that signal alone. Alternatively, with the use of a filter such as the magical audio filter, the rest of the signals are not attenuated by the peak filter, so you can still hear them. But the peak is pronounced and when a signal is  centered to it, you can reduce the gain so that only this signal is heard. These are two ways you can achieve the same thing, so which one is preferred is really up to you.

However, the magical audio filter offers something that a CW filter or an SSB does not offer alone. This is the ability to shape the audio response, either when on narrow or on wide settings. The only shaping that can be done on the FT-301, is with the "Reject" notch control. When on narrow mode, the magical audio filter allows you to pronounce the wanted signal and attenuate the unwanted one. The frequency of the peak can be varied, in contrast to the IF crystal CW filter, where it cannot. When on wide mode and depended on the bandwidth adjustment, the tone of the received SSB audio can be varied, knocking out some interference (and usable AF) at the bass or the high region and make it easier to copy the signal. A similar effect is achievable with the IF shift that some transceivers have. On this wide mode, interfering tones can also be cut with the notch filter. For example there are sometimes you have an SSB voice QSO and a annoying tone appears in the passband. This tone can be cut with the notch filter. Also because the filter does not attenuate the unwanted signals but just pronouncing the peak, you can narrow the width of the peak quite a lot and still be able to recover the audio QSO well, at a much more reduced fidelity, but at a much lower noise levels as well.

Initially, I thought that a second notch filter would be a benefit in pile-ups. However, there is no need for a second one, because the FT-301 already has a noch filter. This notch filter works in the IF but the result is an audio notch. By combining the IF notch filter of the FT-301 and the AF notch filter of the magical audio filter, two notch filters are available.

Another feature added was the mono/pseudo-stereo capability. To use this feature, two speakers (or headphones) are required. Even with the switch set to mono, I have found that the usage of two speakers instead of one, is an improvement in how the operator understands speech. With the flip of a switch, the connections of one of the two speakers are reversed. This reverses the phase of one of the speakers (the speaker cone diaphragm is moved inwards instead of outwards). This phase reversal causes a stereo effect to the ears when the operator is sitting at equal distances from the right and left speakers. This is the most simple form of stereo simulation. Reversing the phase of one of the loudspeakers tends to spread out the sound between the two loudspeakers. I have tried this with a pair of Realistic Minimus-7 die cast body speakers. The effect is exactly as described. The sound previously (mono) originated from the center of the two speakers and at the front of the operator. But with the phase reversal the sound originates from the sides, close to the operator and near his ears. It is difficult to explain the effect, but there is a huge difference. In contrast to true stereo or pseudo stereo that is based on frequency separation between channels, all the signals/frequencies are present on both channels. It is just the source point of the sound that been altered, from a central point at the front, to two points, each close in to the operator's ears.

In general, the filter does an overall good job, offering lots of settings to play around, but at the same time without being so difficult to tune.


UPDATE 19-3-2020

Here is an updated version of the filter/amplifier. Another amplifier chip is used and a special mute circuit has been combined so that the mute relay is not needed anymore. A LPF at the speaker output composed of a 1uF non-polar electrolytic capacitor and a 4R 2W resistor, takes care of the hiss of the filter without affecting the AF tone much

Values of the filter for minimal effect on the wanted audio:

Speaker    R          C
4 ohms     5 ohms     10 uf
8 ohms     10 ohms    4.7 uf
16 ohms    15 ohms    2.2 uf


The resistor R should be as close to the speaker impedance as possible. I had two speakers connected in parallel so the 4 ohms was the ones to use. However, I only had a 2.2uF non-polar electrolytic capacitor and a 10R 2W resistor available, and these worked satisfactorily too.

The 50k peak frequency control potentiometer must be connected like shown in the diagram (with the wiper tied to the lower end), but in such a way where when the potentiometer is rotated counter clockwise, then the peak frequency is at maximum. Connecting it otherwise did not achieve an even spread throughout the whole potentiometer rotation. This is really depended on your logarithmic potentiometer curve, so a bit of testing is needed.

The 5k notch frequency control potentiometer must be connected like shown in the diagram (with the wiper tied towards the electrolytic), but in such a way where when the potentiometer is rotated counter clockwise, then the notch frequency is at maximum. Connecting it otherwise did not achieve an even spread throughout the whole potentiometer rotation. This is really depended on your logarithmic potentiometer curve, so a bit of testing is needed.



Back to main site