Once the receiver is determined to be working properly, alignment is required.
To properly align the discriminator coil, two pieces of test equipment are essential - an accurate frequency counter with resolution to 1 KHz and a high impedance voltmeter, either a digital VOM, or a VTVM.
To align the discriminator:
1) place the voltmeter across the 5K6 resistor between pins 7 and 10 of the LM3089;
2) Using the frequency counter to measure local oscillator frequency, tune the radio to exactly the center frequency of a strong broadcast signal;
3) With the receiver tuned to exact center frequency, adjust the discriminator coil so that the meter reads exactly zero volts. The AFC will fight this process, so it will be necessary to repeatedly retune to center frequency and readjust the discriminator to zero volts until zero volts is realized at exactly center frequency of the broadcast channel.
Unfortunately, I could not properly adjust the discriminator coil by ear. I tried it several times by adjusting the discriminator coil for what I perceived to be the correct response, and it was off when I checked with the voltmeter. The IF transformer I am using for a coil has more than one response peak, and I picked the wrong one every time by ear. The receiver will work with an alignment by ear, but it won't work as well as it is capable of.
The commercial front end is pre aligned, and the coils should not be touched. The IF transformer might benefit from adjustment to the load presented to it, but it probably won’t make much difference. Align for maximum received signal at exactly the center of the IF channel with a scope or spectrum analyzer. It might be possible to align by ear by tuning to center channel of a noisy broadcast station and tuning for maximum quieting.
When properly aligned, the radio should be very sensitive, tune cleanly, and have very clear and undistorted recovered audio.
I printed an adapter to allow installation of the new module in the existing chassis. This avoids the inconvenience and expense of having to immediately print a new chassis, The adapter has mounting points for the 50 x 80mm RF board, and a second set of 50 x 70mm mounts for a perfboard to contain the stereo decoder if, or when, I get around to that. At present, I do not have a need for stereo reception. The expense of printing a new chassis is insignificant, but the inconvenience (time involved in printing) is not. It only took a few minutes to design the adapter and an hour or so to print it out.
I did not bother with installing connectors on the RF board, I simply soldered the existing wire directly to the relevant pads on the board to speed re assembly. I did reconnect the AFC with a 1M resistor on pin 7 of the LM3089 and the other end of the resistor to the tune pin on the tuner.
I ran into one bit of difficulty - on re assembly, there was a low-level audio tone on the audio output even with the volume control set to full attenuation. When volume was at a soft to normal listening level, the tone was masked, but I still considered this unacceptable. I first thought this to be a ground loop between the RF board and the LM386 audio board, but pursuing that theory resulted only in a reduction of the amplitude of the tone, not elimination. Ultimately, the tone was found to be coming in on the DC supply line to the frequency counter. The fix turned out to be simple - a 15 ohm 0805 chip resistor in the supply line to the counter, but it took me a while to solve the issue. It is now silent at any frequency I can hear with the volume control at max attenuation.
Interestingly, as part of that trouble shooting, I found that the new board had markedly more audio output than the prior board, enough to allow me to remove the gain set capacitor on the LM386, and turn its gain down to 20 from 200. On inspection of the removed board, I found that I had carelessly installed a 27K resistor in the RC de emphasis network instead of the correct 2K7 value.
This board works well; at first impression I believe the combination of better quality / smaller parts, and a true RF ground plane has made a worthwhile improvement. It seems to have better selectivity with one filter than the dual filter arrangement of the previous board, where there was potentially considerable blow by around the filters.
The KEC tuner is apparently a Far East market tuner as it tunes from about 75 MHz to 110 MHz. I don't need that kind of range, so the end point resistors on the tuning potentiometer will need to be adjusted. Although the data sheets specify that the KEC tuner is typically about 5 dB less sensitive than the KSE tuner, I cannot tell any difference in apparent sensitivity, The new tuner may be more sensitive, as it does not have the loss of a second filter. The KEC is very stable, it seems to easily hold + - 1 KHz tuning and has good capture when tuning to a station. The tuner module is secured to the ground plane of the pcb at four points - two at the center rear of the tuner, and one at each front corner. It is rigidly mounted and well grounded.
If there is any variance between the most recent schematic sheet and the prior sheets, the most recent should control.
Here is the simple FM broadcast band tuner rebuilt unto a 50 x 80 mm prototype board by BusBoard Systems. I anticipate the pc board format with a true ground plane and smaller components / physical size will give improved RF performance over the original build.
Besides the obvious SMD construction, a few differences from the original: LM3089 rather than CA3089; a KEC (Korea Electronics Corp.) KCF201H tuner instead of the Kwang Sung tuner; one ceramic filter rather than two in series (for now).
Generally, I think sockets for RF IC’s are a bad idea because of the increased lead length, but I used a socket here to make it easier to swap out for an LM3189 at some point in the future. The socket legs are bent to a 90 degree angle and then soldered directly to the bus board.
The tuner pictured is fully functional, although AFC is not installed - I’m still studying how it is implemented on the LM3189. The delayed AGC of the LM3189 is also wired slightly different from the LM3089, although the chips are pin compatible.
The parts are mostly 0805 SMD, although there are a few 0603 10nF bypass caps. The resistors with flying leads are 1/8 watt. I put 10 nF disc ceramic bypass caps on the tuning and AGC line at the tuner, which is likely unneeded as the schematic shows internal caps, but does not list a value. I buy caps in bags of 1000 pieces, so I have decided not to be stingy with them after the power supply debacle. I used SMD electrolytics, although leaded caps are easier to install on these boards and actually have a significantly smaller footprint.
The output resistance of the LM3089 (and CA3089) is 5K. A 2K7 resistor and 10nF capacitor set the proper de emphasis for US broadcast standard and then the audio is taken from a 10 uF electrolytic.
The other picture shows a simple tuner built from a Phillips TBA120U (the suffix is important on these - there are a lot of variants). This chip is intended for television sound systems but is rated to 12 MHz. I did not pursue use of this chip because the preliminary circuit demonstrated that at 10.7 MHz, it is clearly inferior in terms of sensitivity and audio output relative to the CA/LM3089. To compete on equal terms with the ‘3089, it needs one, possibly two, IF stages, and an audio preamplifier. It does not have an AGC output, nor any provision for AFC.
Although CA3089 is approaching fifty years of age now, it still appears to me to be the chip of choice for homebrew FM receivers.
I think I am at a pause point with this radio, after some incremental improvements.
After using it the last week, just as when I first built it, I became undecided as to whether or not the extra gain of the IF amplifier ahead of the CA3089 was needed. RCA and National Semiconductor say that it typically is not needed. After using the device in two separate locations, I concur.
The CA3089 features a delayed AGC for the tuner front end, the voltage of which corresponds to signal amplitude at the input to the CA3089, and which is graphed out in the data sheets. I measured the delayed AGC voltages at pin 15 of the CA3089, and was seeing some in the 3.5 vdc range and many in the 4 volt range with only a few feet of wire for an antenna. After thinking on this, and temporarily bypassing the IF amp for real world comparison/analysis, I decided that the consequent gain reduction from AGC in the front end outweighed whatever benefit (if any) I was getting from the IF amplifier, and that actual sensitivity might be degraded by the excess IF gain, when the increase in front end noise figure from excess delayed AGC is considered. More gain is not always better. Sometimes it is just more gain.
I then permanently bypassed the IF amplifier by connecting the output of the first ceramic filter to the input of the second ceramic filter, and since 10.7 MHz IF transformers are now scarce, I removed it and put it back in my parts inventory. Selectivity has improved noticeably, and apparent sensitivity is not affected. Additionally, some intermod artifacts are no longer present, and overall the radio seems better behaved. I put a note on the board indicating the circuit modification.
With regard to local oscillator stability (which operates at received frequency plus 10.7 MHz), the KSE tuner I used does not provide a specification, but a similar varactor diode tuned device, the KEC KCF201, does.
Two specifications are provided: ordinary oscillator stability, and frequency drift against strong signal input, which I take to mean frequency drift when AFC is active. The former spec is typically +- 200 KHz, and the latter is typically +- 5 KHz.
Ordinary oscillator stability is observed by just removing the antenna input. My tuner module will wander around over about a +- 50 KHz span under these conditions. After careful adjustment of the quadrature coil, and when tuned to a signal sufficient to quiet the receiver, my receiver exhibits stability under AFC control consistently +- 3 KHz, and on a strong station it will hold +- 1 KHz. On cold startup, the oscillator may be as much as 40 KHz from center channel as last tuned, but within a minute the AFC will begin moving the oscillator to center channel and won’t stop until it gets there. It’s rather fascinating to watch, in the first minute or so the frequency change can be dramatic, and as it approaches center channel, the rate of change slows. Most FM tuners only display to the nearest 100 KHz, so AFC action on received frequency is never seen. AFC also aids manual tuning, as it makes the radio try to lock unto a station frequency.
In short, I think the oscillator stability is satisfactory, and AFC action is also satisfactory, and working properly.
I removed the cheap generic LM386 from the premade board and replaced it with a genuine 1980's vintage LM386-4. This made a noticeable improvement in audio quality.
I removed the capacitor to set the gain to 20 and decided this was inadequate. So, I reinstalled the capacitor and reset the gain to 200. For now, this provides a satisfactory level of audio output.
I installed a phono socket for an antenna plug (I anticipate changing this to BNC) and a barrel connector for DC input. I printed a big hole for a snap in on / off switch identical to the switch used in the power supply, but right now I just unplug the radio to turn it off. Or turn off the power supply.
Many of the holes in the rear panel are not presently used and were printed in anticipation of possible future needs
Cosmetically, I printed a fairly large tuning knob for the radio.
I printed a bezel to go around the panel display to give it a little more pleasing appearance.
I printed a “lens” to cover the LED display from translucent blue plastic. The trick to doing this is to use a translucent or semi translucent plastic, make the lens thin, and use minimal infill. This lens is 1mm thick with 10% infill. The difference between an unfiltered display and filtered can be easily seen in the side-by-side pictures with the power supply. The lens makes the display more attractive, at least in my opinion, and when powered on the display has a bit of the same appearance as the old Noritake or Futaba vacuum fluorescent displays on imported late 70's, early 80's HF ham rigs.
That’s it for now. The radio is satisfactory for its intended purpose. Future improvements (if any) are anticipated to be using the top deck of the chassis for better / different audio amplifiers. That will likely be the subject of a separate thread.
I built the tuner back in August 2017, and haven’t powered it up since, so it seemed prudent to make a functional test before proceeding further. When originally constructed, it was designed for and planned that this receiver would stay in my ham shack, but five years later I am spending the majority of my time at our second house up in the mountains at a local lake, so it will stay here, at least temporarily.
I considered replacement of the terminal blocks with plug in connectors similar to what I used in my general coverage receiver and SSB transceiver. After thinking about it, other than aesthetics (this board will never be a beauty queen) I decided that as almost all of the terminal blocks are just carrying DC or audio, there was no material improvement to be had in replacing them, and there is always the risk of breaking something in the process of fixing something that was not broken, not to mention the time involved. So I just left them alone.
The radio powered up and performs satisfactorily from an RF perspective. For testing, I moved the CA3089 audio output from the LM1310 multiplex chip and connected it to a de-emphasis network for mono reception. For quick testing purposes, a store-bought module using an LM386 audio chip with integral volume control drives the same 4 inch homebrew speaker used with my other gear up here. Scraps of leftover wire are being used as an antenna.
Interestingly, one of the reasons it was built in the first place was DX reception of sports broadcasts. Here at the lake, my Yamaha AV receiver with a pair of rabbit ears (behind a big screen TV) cannot quiet on the sports station I need to receive, and portable FM radios are very position sensitive and hand capacity sensitive. The ugly homebrew fully quiets on the target
Alignment
Once the receiver is determined to be working properly, alignment is required.
To properly align the discriminator coil, two pieces of test equipment are essential - an accurate frequency counter with resolution to 1 KHz and a high impedance voltmeter, either a digital VOM, or a VTVM.
To align the discriminator:
1) place the voltmeter across the 5K6 resistor between pins 7 and 10 of the LM3089;
2) Using the frequency counter to measure local oscillator frequency, tune the radio to exactly the center frequency of a strong broadcast signal;
3) With the receiver tuned to exact center frequency, adjust the discriminator coil so that the meter reads exactly zero volts. The AFC will fight this process, so it will be necessary to repeatedly retune to center frequency and readjust the discriminator to zero volts until zero volts is realized at exactly center frequency of the broadcast channel.
Unfortunately, I could not properly adjust the discriminator coil by ear. I tried it several times by adjusting the discriminator coil for what I perceived to be the correct response, and it was off when I checked with the voltmeter. The IF transformer I am using for a coil has more than one response peak, and I picked the wrong one every time by ear. The receiver will work with an alignment by ear, but it won't work as well as it is capable of.
The commercial front end is pre aligned, and the coils should not be touched. The IF transformer might benefit from adjustment to the load presented to it, but it probably won’t make much difference. Align for maximum received signal at exactly the center of the IF channel with a scope or spectrum analyzer. It might be possible to align by ear by tuning to center channel of a noisy broadcast station and tuning for maximum quieting.
When properly aligned, the radio should be very sensitive, tune cleanly, and have very clear and undistorted recovered audio.
73,
Win W5JAG
I printed an adapter to allow installation of the new module in the existing chassis. This avoids the inconvenience and expense of having to immediately print a new chassis, The adapter has mounting points for the 50 x 80mm RF board, and a second set of 50 x 70mm mounts for a perfboard to contain the stereo decoder if, or when, I get around to that. At present, I do not have a need for stereo reception. The expense of printing a new chassis is insignificant, but the inconvenience (time involved in printing) is not. It only took a few minutes to design the adapter and an hour or so to print it out.
I did not bother with installing connectors on the RF board, I simply soldered the existing wire directly to the relevant pads on the board to speed re assembly. I did reconnect the AFC with a 1M resistor on pin 7 of the LM3089 and the other end of the resistor to the tune pin on the tuner.
I ran into one bit of difficulty - on re assembly, there was a low-level audio tone on the audio output even with the volume control set to full attenuation. When volume was at a soft to normal listening level, the tone was masked, but I still considered this unacceptable. I first thought this to be a ground loop between the RF board and the LM386 audio board, but pursuing that theory resulted only in a reduction of the amplitude of the tone, not elimination. Ultimately, the tone was found to be coming in on the DC supply line to the frequency counter. The fix turned out to be simple - a 15 ohm 0805 chip resistor in the supply line to the counter, but it took me a while to solve the issue. It is now silent at any frequency I can hear with the volume control at max attenuation.
Interestingly, as part of that trouble shooting, I found that the new board had markedly more audio output than the prior board, enough to allow me to remove the gain set capacitor on the LM386, and turn its gain down to 20 from 200. On inspection of the removed board, I found that I had carelessly installed a 27K resistor in the RC de emphasis network instead of the correct 2K7 value.
This board works well; at first impression I believe the combination of better quality / smaller parts, and a true RF ground plane has made a worthwhile improvement. It seems to have better selectivity with one filter than the dual filter arrangement of the previous board, where there was potentially considerable blow by around the filters.
The KEC tuner is apparently a Far East market tuner as it tunes from about 75 MHz to 110 MHz. I don't need that kind of range, so the end point resistors on the tuning potentiometer will need to be adjusted. Although the data sheets specify that the KEC tuner is typically about 5 dB less sensitive than the KSE tuner, I cannot tell any difference in apparent sensitivity, The new tuner may be more sensitive, as it does not have the loss of a second filter. The KEC is very stable, it seems to easily hold + - 1 KHz tuning and has good capture when tuning to a station. The tuner module is secured to the ground plane of the pcb at four points - two at the center rear of the tuner, and one at each front corner. It is rigidly mounted and well grounded.
If there is any variance between the most recent schematic sheet and the prior sheets, the most recent should control.
73 and good DIY,
Win W5JAG
Apparently, the pause was short lived.
Here is the simple FM broadcast band tuner rebuilt unto a 50 x 80 mm prototype board by BusBoard Systems. I anticipate the pc board format with a true ground plane and smaller components / physical size will give improved RF performance over the original build.
Besides the obvious SMD construction, a few differences from the original: LM3089 rather than CA3089; a KEC (Korea Electronics Corp.) KCF201H tuner instead of the Kwang Sung tuner; one ceramic filter rather than two in series (for now).
Generally, I think sockets for RF IC’s are a bad idea because of the increased lead length, but I used a socket here to make it easier to swap out for an LM3189 at some point in the future. The socket legs are bent to a 90 degree angle and then soldered directly to the bus board.
The tuner pictured is fully functional, although AFC is not installed - I’m still studying how it is implemented on the LM3189. The delayed AGC of the LM3189 is also wired slightly different from the LM3089, although the chips are pin compatible.
The parts are mostly 0805 SMD, although there are a few 0603 10nF bypass caps. The resistors with flying leads are 1/8 watt. I put 10 nF disc ceramic bypass caps on the tuning and AGC line at the tuner, which is likely unneeded as the schematic shows internal caps, but does not list a value. I buy caps in bags of 1000 pieces, so I have decided not to be stingy with them after the power supply debacle. I used SMD electrolytics, although leaded caps are easier to install on these boards and actually have a significantly smaller footprint.
The output resistance of the LM3089 (and CA3089) is 5K. A 2K7 resistor and 10nF capacitor set the proper de emphasis for US broadcast standard and then the audio is taken from a 10 uF electrolytic.
The other picture shows a simple tuner built from a Phillips TBA120U (the suffix is important on these - there are a lot of variants). This chip is intended for television sound systems but is rated to 12 MHz. I did not pursue use of this chip because the preliminary circuit demonstrated that at 10.7 MHz, it is clearly inferior in terms of sensitivity and audio output relative to the CA/LM3089. To compete on equal terms with the ‘3089, it needs one, possibly two, IF stages, and an audio preamplifier. It does not have an AGC output, nor any provision for AFC.
Although CA3089 is approaching fifty years of age now, it still appears to me to be the chip of choice for homebrew FM receivers.
73,
Win W5JAG
Epilogue August 2022
I think I am at a pause point with this radio, after some incremental improvements.
After using it the last week, just as when I first built it, I became undecided as to whether or not the extra gain of the IF amplifier ahead of the CA3089 was needed. RCA and National Semiconductor say that it typically is not needed. After using the device in two separate locations, I concur.
The CA3089 features a delayed AGC for the tuner front end, the voltage of which corresponds to signal amplitude at the input to the CA3089, and which is graphed out in the data sheets. I measured the delayed AGC voltages at pin 15 of the CA3089, and was seeing some in the 3.5 vdc range and many in the 4 volt range with only a few feet of wire for an antenna. After thinking on this, and temporarily bypassing the IF amp for real world comparison/analysis, I decided that the consequent gain reduction from AGC in the front end outweighed whatever benefit (if any) I was getting from the IF amplifier, and that actual sensitivity might be degraded by the excess IF gain, when the increase in front end noise figure from excess delayed AGC is considered. More gain is not always better. Sometimes it is just more gain.
I then permanently bypassed the IF amplifier by connecting the output of the first ceramic filter to the input of the second ceramic filter, and since 10.7 MHz IF transformers are now scarce, I removed it and put it back in my parts inventory. Selectivity has improved noticeably, and apparent sensitivity is not affected. Additionally, some intermod artifacts are no longer present, and overall the radio seems better behaved. I put a note on the board indicating the circuit modification.
With regard to local oscillator stability (which operates at received frequency plus 10.7 MHz), the KSE tuner I used does not provide a specification, but a similar varactor diode tuned device, the KEC KCF201, does.
Two specifications are provided: ordinary oscillator stability, and frequency drift against strong signal input, which I take to mean frequency drift when AFC is active. The former spec is typically +- 200 KHz, and the latter is typically +- 5 KHz.
Ordinary oscillator stability is observed by just removing the antenna input. My tuner module will wander around over about a +- 50 KHz span under these conditions. After careful adjustment of the quadrature coil, and when tuned to a signal sufficient to quiet the receiver, my receiver exhibits stability under AFC control consistently +- 3 KHz, and on a strong station it will hold +- 1 KHz. On cold startup, the oscillator may be as much as 40 KHz from center channel as last tuned, but within a minute the AFC will begin moving the oscillator to center channel and won’t stop until it gets there. It’s rather fascinating to watch, in the first minute or so the frequency change can be dramatic, and as it approaches center channel, the rate of change slows. Most FM tuners only display to the nearest 100 KHz, so AFC action on received frequency is never seen. AFC also aids manual tuning, as it makes the radio try to lock unto a station frequency.
In short, I think the oscillator stability is satisfactory, and AFC action is also satisfactory, and working properly.
I removed the cheap generic LM386 from the premade board and replaced it with a genuine 1980's vintage LM386-4. This made a noticeable improvement in audio quality.
I removed the capacitor to set the gain to 20 and decided this was inadequate. So, I reinstalled the capacitor and reset the gain to 200. For now, this provides a satisfactory level of audio output.
I installed a phono socket for an antenna plug (I anticipate changing this to BNC) and a barrel connector for DC input. I printed a big hole for a snap in on / off switch identical to the switch used in the power supply, but right now I just unplug the radio to turn it off. Or turn off the power supply.
Many of the holes in the rear panel are not presently used and were printed in anticipation of possible future needs
Cosmetically, I printed a fairly large tuning knob for the radio.
I printed a bezel to go around the panel display to give it a little more pleasing appearance.
I printed a “lens” to cover the LED display from translucent blue plastic. The trick to doing this is to use a translucent or semi translucent plastic, make the lens thin, and use minimal infill. This lens is 1mm thick with 10% infill. The difference between an unfiltered display and filtered can be easily seen in the side-by-side pictures with the power supply. The lens makes the display more attractive, at least in my opinion, and when powered on the display has a bit of the same appearance as the old Noritake or Futaba vacuum fluorescent displays on imported late 70's, early 80's HF ham rigs.
That’s it for now. The radio is satisfactory for its intended purpose. Future improvements (if any) are anticipated to be using the top deck of the chassis for better / different audio amplifiers. That will likely be the subject of a separate thread.
73,
Win W5JAG
I built the tuner back in August 2017, and haven’t powered it up since, so it seemed prudent to make a functional test before proceeding further. When originally constructed, it was designed for and planned that this receiver would stay in my ham shack, but five years later I am spending the majority of my time at our second house up in the mountains at a local lake, so it will stay here, at least temporarily.
I considered replacement of the terminal blocks with plug in connectors similar to what I used in my general coverage receiver and SSB transceiver. After thinking about it, other than aesthetics (this board will never be a beauty queen) I decided that as almost all of the terminal blocks are just carrying DC or audio, there was no material improvement to be had in replacing them, and there is always the risk of breaking something in the process of fixing something that was not broken, not to mention the time involved. So I just left them alone.
The radio powered up and performs satisfactorily from an RF perspective. For testing, I moved the CA3089 audio output from the LM1310 multiplex chip and connected it to a de-emphasis network for mono reception. For quick testing purposes, a store-bought module using an LM386 audio chip with integral volume control drives the same 4 inch homebrew speaker used with my other gear up here. Scraps of leftover wire are being used as an antenna.
Interestingly, one of the reasons it was built in the first place was DX reception of sports broadcasts. Here at the lake, my Yamaha AV receiver with a pair of rabbit ears (behind a big screen TV) cannot quiet on the sports station I need to receive, and portable FM radios are very position sensitive and hand capacity sensitive. The ugly homebrew fully quiets on the target