Most constructors have never have
heard of this type of receiver, unless you're into top end FM tuners, or
you have read British electronics magazines from the mid 50's to the 1970's.
I first saw the design in a mid 1960's Practical Wireless magazine and due to the simplicity I thought it wouldn't work. How wrong I was! This was in 1990 when I was getting frustrated with the limitations of super regenerative receivers, prior to the development of my 12AT7 circuit, and was looking for other approaches to the 'simple FM receiver' problem.
Concept of Operation.
The pulse counting FM receiver is a superhet receiver, but differs from conventional FM receivers in two ways. First, the VHF signal is converted to an IF of 200Kc/s instead of 10.7Mc/s. This is because instead of a ratio detector or quadrature detector, a pulse counting detector is used. Its operation requires a low IF. Because of the low IF, no tuned circuits are required, except one for the local oscillator/RF input. This dispenses with the necessity to perform any alignment. Beyond the converter, all stages are resistance capacitance coupled.
Commercially made Pulse Counting receivers are a bit more elaborate than this. They're often dual conversion, with a conventional 10.7Mc/s IF which is then converted down to the ~150Kc/s IF with a crystal locked converter. This is done to prevent image response. These days, they often use a digital type pulse counting detector which provides good noise immunity. By passing the squared waveform through a Schmitt trigger, most of the noise will be removed.
However, for the homemade type of receiver, we don't need to go to such lengths to get a good quality receiver.
The earliest mention of pulse counting techniques appears to be in the early 1940's where such a receiver was used to monitor one of the first FM transmitters on the Empire State building. It used 807's for the RC coupled IF stage (a bit of overkill) and a 6H6 detector.
The next mention appears in Wireless World in the late 1940's when Thomas Roddam asks, "Why align FM discriminators?" and puts forward a pulse counting detector design. Then in 1956, M.G. Scroggie answers the question with a complete and practical Pulse Counting receiver. It was this design that provided the basis for subsequent valve receiver circuits until solid state versions appeared.
Once I had familairised myself with the
operation of this method of FM reception, I settled on a basic design which
has been used in all my receivers with minor variations.
I have built five such receivers up to the present time and all worked first time, with nothing to align. Other constructors who have read this article have also duplicated the design with excellent results. The design is not 'weird' or a 'fluke'. Sensitivity is close to, and better than some, commercially made superhets. Yes, there is an extra control to be adjusted which is a tradeoff, but for the person who wants high quality mono FM, using valves, with simplicity, there isn't a better receiver. However, the "extra control" can be a set and forget affair unless you're wanting to get every last microvolt of sensitivity for low power or distant stations.
RF Amp & Frequency Converter.
I have found best performance, coincident with ease of construction, results from using a 6BL8/ECF80 at the front end. The triode is used as a grounded grid RF amp, feeding the pentode functioning as an autodyne converter. Having such a low IF means that the signal can be tuned in with the local oscillator either side of the carrier. This can be advantageous if there's an interfering station on a nearby frequency. Image response is a non issue as FM stations do not transmit on adjacent frequencies in any given area.
The original Wireless World article used a 12AT7/ECC81 local oscillator driving a passive germanium diode for frequency conversion, but the gain is obviously lower than an active first detector.
Grounded grid RF amplifier feeds and autodyne converter to produce the 200Kc/s IF.
The VHF signal is fed into the triode of
a 6BL8/ECF80. This operates as a grounded grid RF amplifier stage and thus
suits a low impedance input. RF chokes are used for plate and cathode loads.
As such, this stage is untuned and provides not much gain. However, its
purpose is to isolate the aerial from the following stage; which is the
autodyne frequency converter.
The aerial needs to be isolated, otherwise aerial loading effects will cause the oscillation level to vary excessively, to the point of causing the oscillator to stop in some instances.
It also reduces local oscillator radiation from the aerial.
The frequency converter is of the autodyne type; that is, the mixer is self oscillating. Only one pentode (or triode) is required. Here, the pentode operates as an electron coupled oscillator by virtue of the cathode choke and the stray capacities in the valve. The tuned circuit feeds the grid. It determines the frequency of oscillation. Conveniently, because the IF is only 200Kc/s, the received frequency is so close to the oscillator frequency, that we can use the same tuned circuit for both tuning the RF input as well as the oscillator. While the tuned circuit is not right on the received frequency, the selectivity of this circuit is such that a signal 200Kc/s away is not attenuated.
The operating point of the converter is important. Best conversion efficiency (and thus sensitivity) occurs when the oscillator is operating at a low level; just above the point where oscillation starts. Because the oscillator level varies across the band, it is necessary to adjust for the optimum operating point, and so a control (the 50K pot) is provided to vary the screen voltage of the pentode. For strong signals, the operating point is not so critical and the control only needs to be set once.
Very strong signals can actually cause the converter to stop oscillating, and from the second receiver onwards, I have include AGC. The negative voltage present at the limiter grid is used to reduce the gain of the RF amplifier.
Due to the low IF, the plate load of the pentode is a 22K resistor, rather than a tuned circuit. This feeds the IF amplifier which subsequently provides the necessary selectivity.
The original Wireless World article used a two stage IF amplifier using 6BX6/EF80 valves with the response as shown:
Gain is about 4000, and input voltage (to
the IF) should be at least 1mV for optimum performance of the following
limiter and detector. However, the converter gain is high, so the receiver
works well at signal levels much less than this.
So how do we get this response with no coils? It's quite easy actually when you consider things like valve input and stray capacitances. Each of the 6BX6's has a 4.7K resistor in series with the grid. In conjunction with the grid capacitance there is a degree of low pass filtering. The high value of plate resistors (18K), adds to this and so we have the top frequency response set to around 250Kc/s.
The low frequency response is set by the 270pF grid coupling condensers in conjunction with the 100K grid resistors. It starts to fall off at about 20Kc/s.
In the receivers I have constructed, I have departed from the Wireless World IF amplifier design in some instances, but the results are the same.
Complete IF amplifier, limiter, and detector.
Limiter and Detector.
Limiting is done in the conventional way with a third 6BX6 operating with low bias and a low value plate load. Because of the low frequency used, it is quite easy to observe the waveform with a CRO throughout the IF, limiting and detection stages. Because the limiter is operated with a low bias, the signal fed into the grid causes grid current to flow, and thus produces a negative voltage which is consistent with signal strength. This negative voltage suitably filtered can be used for AGC.
The clipped waveform from the limiter plate is differentiated, and applied to a pulse counting, or tachometer, circuit. The higher the frequency, the closer the pulses are together, and the the higher the resulting DC from the detector, and vice versa as the frequency decreases. So, we have a frequency to voltage converter which is what we want for FM demodulation. The greatest advantage is there are no tuned circuits to get out of alignment, causing distortion.
Despite its appearance as a voltage doubling AM detector, the circuit around the 6AL5/EB91 is anything but. The low value input condenser (47pf) and the low value load (4.7K) ensure the signal from the limiter is differentiated. Filtering is done with a simple RC circuit which also provides de-emphasis.
Audio output is low at about 100mV, and the recommended load is 500K. While an ordinary triode pentode audio amp can be just fully driven, an extra stage of gain is worthwhile.
First Pulse Counting FM Receiver, Winter 1990. This is my Super DX model with 3 stages of IF amplification. Valves are, from back to front, 6V4, 6BM8,6AU6 x 4, 6BL8 and 6AL5.
Under the chassis of my first pulse counting receiver.
I had the basic receiver assembled in one
evening, and due to my scepticism, I'd added an extra IF stage in case
the gain turned out to be inadequate. I was in a state of amazement when
I first powered it up, and was getting Hi-Fi sound with excellent sensitivity.
The following night I'd added a grounded grid RF stage to eliminate the
problems of aerial coupling and absorption effects. With an indoor TV aerial
I had no problems in bringing in all local stations, and the low power
community broadcasters that caused so much difficulty on the super-regen
sets. And there was no SCA/stereo subcarrier beat, no hiss, and no distortion!!
Subsequent experiments proved that stereo reception was not practical. It appeared that the demodulated signal is not of sufficient bandwidth, and to increase it would mean reducing the output. However, with high frequency boost prior to the LM1310 stereo decoder, it was possible to get results of some sort but separation was just too poor.
I did some simple tests with my first receiver, with the three IF stages, and a Hewlett Packard 8654B signal generator. Even with no aerial connected there was some difficulty in finding a clear frequency in the middle of the band as stations just roll in from everywhere. My initial tests were done at about 90Mc/s, with a 1000c/s tone. Deviation was set to 70Kc/s. 1uV was discernible, but only just. Had it been voice or music you wouldn't be able to understand it. 3uV was better; just readable. At 10uV, the signals are very readable, but quite noisy. Noise free reception commences at about 30uV. Not bad for an FM receiver with only one tuned circuit! For the limiter to start clipping, signal strength required is much greater; around 500uV. However, this is not important as VHF is not prone to interference.
I also examined the performance if the oscillation level control, which sets the frequency converter's operating point was simply used as a preset control. Much to my surprise, the change in sensitivity was not as great as first thought as I tuned across the band, maybe a loss of 10uV or so. Interestingly, this seemed to be around the centre of the band.
Second Pulse Counting Receiver, October 1992.
This receiver uses 6SN7's for IF amplification, and also uses a 6SN7 fed by a 12AX7 for the audio amplifier. The resulting 'triode sound' is the nicest sounding of all my FM receivers. A series heater circuit is used with modern low voltage transformers for the power supply. This receiver was operated daily for several years at my place of work and never failed.
Under the chassis of the 2nd receiver.
The third pulse counting receiver I built is for 12V operation. It uses a vibrator power supply. Tuning is by a ten turn pot and varicap diodes. Its valves are 2x 6U8, 6BX6, 6DX8, 6AL5 and 6BQ5. It provides the highest audio output of all my pulse counting receivers. While it has good sensitivity and sound quality, the varicap diodes cause drift, as the set warms up over about 20 minutes. An AFC circuit will have to be provided, or a return to variable condenser tuning, to eliminate this.
4th Pulse Counting Receiver
This set was constructed in the Winter
As previously, I stuck to my tried and trusted 6BL8 RF/converter stage, as I've found from experimentation that other valves or circuit configurations don't produce as good results. If you've seen the 12AT7 receiver article, you'll notice my RF amp design there. It's a grounded grid circuit with input and output untuned to save winding coils and their alignment hassles. There isn't really any gain but it does have excellent isolation, which is the main purpose. As with the super-regen sets, direct aerial connections can cause oscillation to become unreliable at certain parts of the band, or with the aerial in certain positions, but the RF amplifier eliminates this problem completely.
The frequency converter uses a 6BL8 pentode operating as an electron coupled oscillator, just like in the 12AT7 receiver. I have found this type of oscillator to be the simplest and easiest to get going of all the VHF oscillators I've tried, so I use it in all my valve VHF work. The screen voltage is made adjustable by the 50K pot. This is necessary to set the optimum operating conditions for the converter, so it oscillates reliably and provides maximum gain. Of course, the optimum setting varies from one end of the band to the other. So, despite being a superhet, there's still an extra control.
Maximum sensitivity occurs just after the converter has gone into oscillation. However, if you are content with only the higher power stations, just set the screen grid voltage at the 108Mc/s end of the band and leave it. Sensitivity will drop off as you tune to the 88Mc/s end, but the performance will still be better than any super-regen set. A vernier dial is highly recommended for the tuning control, or other geared down arrangement, although a large knob is adequate. As usual, the tuned circuit consists of a 15pF variable condenser and 4 turns of 18 gauge tinned copper wire on an air cored 10mm former.
The IF appears across the 22K plate resistor and thence is amplified by the two 6BX6's. Limiting is achieved by a third 6BX6, whereupon the clipped output is fed to the pulse counting detector, a 6AL5. The IF, limiter, and detector is exactly as per the Wireless World design, except I've used a valve detector instead of germanium diodes. Other valve diodes can of course be used, such as 6H6 or EB41.
My last three Pulse Counting Receivers have incorporated AGC. Without it, strong signals can actually overload the front end to the point where nothing is heard on that station (i.e. the receiver front end becomes blocked). While placing an attenuator in the aerial lead solves the problem, overall sensitivity is then reduced for all stations. An ideal source of AGC voltage is obtainable from the limiter grid. Due to grid rectification, the voltage here becomes increasingly negative with an increase in signal strength. This is fed to the RF amp grid via a suitable voltage divider to reduce the gain sufficiently.
Although I have not done much work with it, the limiter grid appears to be also suitable for providing an AFC voltage. With the receiver correctly tuned, the voltage is at its maximum negative. Early experiments on my first receiver, with a varicap diode connected to the oscillator coil via a low value condenser, seemed to indicate the idea works.
One can also add a magic eye at this point to indicate correct tuning, as several constructors have done.
Circuit of my 4th Pulse Counting FM receiver. For full size circuit, click here.
Audio amplifier has been revised around the pentode: 180R now 330R. 180V supply, now 200V @10mA. 2.2K bridged out.
RC network across transformer primary replaced with .0047uF. 2.2M in series with 1000pF connected between triode and pentode plates. Speaker transformer primary changed to 15K.
Turning now to the audio amp, I've used
another 6BL8. This is probably the most common TV valve in Australia. Europeans
know it better as the ECF80 or its series heater version, the PCF80 or
9A8. Initially, I used a 6DX8/ECL84, but the severe heat problems I discovered
when I completed the receiver meant I had to do everything I could to reduce
heat, and decreasing the power of the audio amp was one of the steps I
had to take. Output power is about 250mW.
I have recently revised the audio amplifier to improve the tonal qualities, and to bring the pentode operating conditions closer to what they should be. The changes are described in the note below the circuit diagram. As I am using a 240V to 6.3V power transformer as the speaker transformer (for mains isolation), other kinds of output transformer may warrant changes to the 1000pF condenser in the feedback network. This will have to be determined experimentally to obtain the desired tone. Undistorted output is now about 160mW (similar to small valve portable or pocket transistor radio). Emails I get from constructors indicate that most are using their own amplifier circuits fed via a cathode follower stage.
The power supply is unusual as far as Australian design goes, using a live chassis with a transformer for the heaters only. This was done for space reasons. There was no way any valve type power transformer capable of powering this circuit could fit in the box with everything else. For the B+, the 240V mains is rectified by a 1N4007 and smoothed with RC filtering. No hum is evident even with sensitive headphones.
Because the receiver has a live chassis, I have not included the power supply circuit.
Very compact inside with 6 valves! On the left are the front end valves. The bottom row is the 6BX6's, and the 6BL8 converter and 6AL5 are on the top. Over at the right is the 6BL8 audio valve. This receiver has a live chassis and has suitable precautions to prevent the user coming into contact with the mains.
Powering up the new receiver
As with previous pulse counting receivers, this one brought in stations straight away. I did have to do a minor adjustment of the tuning circuit but nothing that required test instruments. The audio amp required a few minor component alterations to get the correct voltages, and likewise the converter screen grid or 'oscillation' control needed to have its voltage divider resistors optimised. Sensitivity was good as usual, with one FM station on 96.1Mc/s at 5KW, receivable from 100km away with no aerial (Wentworth Falls > Gore Hill).
However, it was becoming apparent that the heat build up inside the box with the lid on was a real problem. With the lid too hot to touch and the internal components subjected to such heat there was a risk of failure, as well as the chassis coming loose from the plastic studs securing it to the box.
To try and cool the inside of the cabinet as much as possible, I put ventilation holes at the rear and underneath the case. Furthermore, the heater and B+ dropper resistors were moved to outside the box inside a perforated metal shield mounted on the back. This was connected to mains earth of course.
This helped, but I was able to improve things even more. I replaced the 6DX8 audio amp with the 6BL8. This lessened the heater current nearly 300mA, so the transformer ran cooler. Also, the lower plate current of the 6BL8 lessened the heat generation.
Testing at home in the Blue Mountains gave the expected results. I was able to receive a tourist information station in the Southern Highlands with my 5 element Exastereo outdoor FM aerial. These stations only transmit with a couple of watts, so it was a good indication of the excellent sensitivity.
One thing that was becoming obvious, and annoying, was the dropping out of the oscillator at the high end of the band. I eventually discovered it was due to the original 1.5mH cathode choke. For some reason the one I'd used wouldn't allow proper oscillation. Changing to a different type (axial 1mH) fixed that completely.
Ideas which I haven't actually implemented but are worth experimenting with. Note that these in no way imply this is an unstable receiver that suffers from drift. They are merely good practice to include in any VHF receiver.
1)AFC: The DC present at the detector output or perhaps the DC on the AGC line could be used to provide AFC as these voltages peak up on correct tuning. Varicap diodes or using the miller effect of a triode could be used to control the oscillator frequency. I did experiment with a varicap diode fed from the detector output with my first receiver and the idea seemed to hold promise.
2)Regulating the B+: When the mains voltage changes and the receiver is adjusted to just past oscillating for the most sensitive point, it may drop out when the mains voltage decreases. Regulating the B+ to the converter stage at least would overcome this. Also, the frequency at which the 6BL8 pentode oscillates is affected by plate and screen voltages (the screen control can actually be used for fine tuning within limits). Regulation would therefore improve frequency stability.
3)Regulating the 6BL8 heater: I have noticed that after a large drop in mains voltage that within a few seconds the receiver may drift off frequency. The time delay suggests the heater temperature of the 6BL8 pentode has an effect on oscillation frequency. So, it would be worthwhile to provide a regulated supply for this as well. Easiest way is with a three terminal regulator set to 6.3V. As the regulator has to be fed with DC, consideration has to be given to the other heaters. Two options are: a) separate heater windings (or separate heater transformers), one feeding the other valves with 6.3VAC and the other feeding the bridge rectifier and regulator (it will need to be about 9V to allow for regulator headroom and rectifier losses), b) one winding feeding a rectifier and regulator to provide DC for all the heaters. The reason for separate heaters is of course with bridge rectification one cannot earth both the input and output of the rectifier, and it is essential that one side of the heater line is earthed.
4)Automatic level control for the oscillation amplitude of the frequency converter. This would eliminate the "extra control". Other kinds of frequency converter could also be investigated; e.g.. pentagrid valves, or separate oscillator and modulator valves.
Thoughts on Stereo:
An attempt was made to connect an LM1310 type stereo decoder to the original receiver. If you wish to experiment, connect the decoder input to the junction of the two 47K resistors in the detector output filter. Change the 1000pF to 100pF. Although results of a sort were obtained, separation was poor, and stereo reception was unreliable. Later research revealed the problem is lack of bandwidth with the pulse counting detector as presented. Don't forget the design was optimised in an era when multiplex transmissions had not been thought of. Don't forget that the stereo signals are centred about 38Kc/s, hence the need for greater detected bandwidth. Detected output voltage, and bandwidth of the detector is dependent on the value of coupling condenser from the limiter to the detector diodes. The larger the condenser, the wider the pulse, the higher the output, and the the narrower the bandwidth. So, if you have ideas about trying the design for stereo reception, a good start would be to reduce the 47pF, as well as the filtering on the detector output. However, the detected output will now be of lower amplitude. It would appear therefore, to compensate for this, the B+ for the limiting stage will need to be increased. This may mean the 6BX6 will be outside its ratings, and if so will be necessary to change the limiter valve to something higher powered. Having said all that, the Zenith stereo multiplex system used for FM stereo is a limited performer to start with, and requires a signal far better than that required for good mono reception.
Version 5 Pulse Counting Receiver.
This receiver was the first where I'd included
a magic eye tuning indicator. It is based on my original version, but with
one less IF stage. To enable the 6BM8 to be driven with gain to spare,
an extra audio stage has been added. It is simply a 6AU6 wired as a triode,
and the cathode resistor is unbypassed, as only a low gain is needed.
I have also taken the opportunity to try my homemade VHF chokes as the originals are almost impossible to obtain. Constructors of my VHF receivers have difficulties in getting the original commercially made chokes I used, and often the substitutes perform poorly or stop the receiver working altogether. So, now this problem has been solved. I now recommend the homemade chokes for all the valve pulse counting designs in all three positions. The same goes for the super regenerative sets. Because the power transformer only had a 2A 6.3V winding, it was necessary to include another filament transformer for the 6BM8/ECL82 and 6FG6/EM84
The magic eye is very useful for tuning, and as a relative signal strength indicator. It also makes it very clear when the 5K oscillation level pot is adjusted optimally.
Note that the screen grid voltage for the frequency converter is fed from the same B+ point as the RF amp which is AGC controlled. The common 2.7K resistor means that the screen voltage fluctuates with signal strength, and this reduces the necessity to adjust the 5K pot. For local stations it is possible to leave the 5K pot preset.
Note also the extra components in the screen circuit; the 10K and .15uF and .082uF. These were only included because of the wiring distance to the 5K pot in this particular receiver - they are not normally needed as the previous circuits show.
Circuit of the 5th Pulse Counting Receiver. This incorporates an extra audio stage needed to drive the 6BM8 to full power. A magic eye simplifies tuning.
|INPUT (uV)||AGC (V)||QUALITY|
|10000||-19.8||magic eye closed|
This table shows ACG voltage versus signal input. Frequency was 100Mc/s, deviation 50Kc/s, modulation 1Kc/s. 5K pot was adjusted for maximum sensitivity at the start of the test. Signal generator was a R&S SMS.
This is the best performer so far. It also includes the often asked for magic eye circuit. It follows the original circuit in that 6AU6's are used for the IF and limiter, but there is one less IF stage. This is because in the original receiver this contributed more noise than gain. If you think you're seeing 43K resistors in the circuit, you're right - they were to hand so were used. 47K is suitable instead.
This Pulse Counting Receiver was made by
Josef from the Czech Republic.
It uses the EF80 IF amplifier stage and includes the EM84 magic eye. The power supply is external and connects via an octal socket at the back.
I thought this was a particularly good example of correct construction technique. As can be seen, wiring is kept short and direct. The aluminium diecast box makes an excellent chassis for RF projects such as this. The particular box used was this http://www.tme.eu/cz/details/hm-1550we/univerzalni-krabicky/hammond/1550we/#.
You can see the YouTube video here https://www.youtube.com/watch?v=hZjDmox4QzA. It gives a good idea as to how the receiver operates.