Home |
Search |
Today's Posts |
#1
|
|||
|
|||
Designs for a single frequency high performance AM-MW receiver?
This is an unusual (and maybe fun) request to discuss the design
possibilities for a high-performance, single frequency (single channel) medium-wave receiver. That is, the receiver will be designed to receive only one MW frequency, such as, for example, 830 kHz. Of course, some would say this is silly, but I'm curious to see how the design would evolve when the need to tune across a significant portion of the medium-wave band is removed (any "tuning" included in the design will be very narrow, such as 1-2 kHz to compensate for slightly off-channel transmission.) To clarify, what I mean by high-performance is very good sensitivity, selectivity, dynamic range, etc. -- to be excellent for DXing as well as having excellent sound fidelity for strong stations -- to be an all things to all users receiver (albeit single frequency/channel.) Obviously, it can include several controls for fine-tuning, varying bandwidth, etc., etc. -- be innovative. Since I'm fond of tubes, the design should be open to using them (but not required) for achieving some performance goal, such as sensitivity. Of course, other unusual components can be considered, including the venerable crystal. It can be a hybrid design of the very new with the very old (although I'd also like to consider, for nostalgia-sake, designs employing only components used in 1930's radios.) The design need not be superhet, and in fact I prefer to explore simpler and innovative non-superhet circuits. It should have a pre-amplified line-out to drive a separate audio amplifier (so pure crystal sets are out, although a crystal set with a pre-amp is of possible discussion, I suppose.) Ok, that's the constraints. I hope several find this of sufficient interest to contribute to the discussion. Thanks. Gary Jensen |
#3
|
|||
|
|||
Though it most certainly wouldn't classify as an antique radio, one of
the easier paths for a single station receiver would be to use the MK484 chip. It is an entirely self contained TRF receiver needing only the input tank circuit (ferrite bar and compression trimmer, for example) and an audio amp. No external antenna is required, and performance is comparable to a simple superhet. This might be far off the mark in how you wnat to build your receiver, but the 484 is a fun chip to play with just the same. Joe, N6DGY www.kirtland.com for e-mail John Byrns wrote: Hi Gary, Your request is interesting, I have built a couple of single channel AM-MW receivers in the past, and have recently been contemplating another design. The first single channel design I built was a crystal controlled tube superhetrodyne, and the second was a transistorized TRF receiver. Your design requirements are a little confusing to me and I am hoping you can elaborate a little further? My concept of a single frequency receiver is directed to receiving one particular station, which would determine the required sensitivity, selectivity, and other performance parameters on an individual basis. You are asking for high sensitivity and selectivity, with large dynamic range, and variable bandwidth. This seems unusual to me as I don't see why all this would be necessary in a single channel receiver dedicated to receiving one station or frequency? Can you elaborate further on the reasons behind your requirements? Does this receiver have a specific use, or is it to be easily adaptable to any possible use requiring a single frequency receiver? A couple of other questions come to mind. What sort of antenna input circuit the receiver should have? This would be determined by the type of antennas the single channel receiver must be able to work with. Also once built, should it be relatively easy to retune the receiver for a different frequency, or can it be designed for a single fixed frequency from the start, with no allowance for easily changing to a different frequency later? An interesting question in my mind is should the receiver use an envelope detector, or a synchronous detector? Before laying out my ideas I would like to hear more about what you want this receiver to do. Regards, John Byrns In article , wrote: This is an unusual (and maybe fun) request to discuss the design possibilities for a high-performance, single frequency (single channel) medium-wave receiver. That is, the receiver will be designed to receive only one MW frequency, such as, for example, 830 kHz. Of course, some would say this is silly, but I'm curious to see how the design would evolve when the need to tune across a significant portion of the medium-wave band is removed (any "tuning" included in the design will be very narrow, such as 1-2 kHz to compensate for slightly off-channel transmission.) To clarify, what I mean by high-performance is very good sensitivity, selectivity, dynamic range, etc. -- to be excellent for DXing as well as having excellent sound fidelity for strong stations -- to be an all things to all users receiver (albeit single frequency/channel.) Obviously, it can include several controls for fine-tuning, varying bandwidth, etc., etc. -- be innovative. Since I'm fond of tubes, the design should be open to using them (but not required) for achieving some performance goal, such as sensitivity. Of course, other unusual components can be considered, including the venerable crystal. It can be a hybrid design of the very new with the very old (although I'd also like to consider, for nostalgia-sake, designs employing only components used in 1930's radios.) The design need not be superhet, and in fact I prefer to explore simpler and innovative non-superhet circuits. It should have a pre-amplified line-out to drive a separate audio amplifier (so pure crystal sets are out, although a crystal set with a pre-amp is of possible discussion, I suppose.) Ok, that's the constraints. I hope several find this of sufficient interest to contribute to the discussion. Thanks. Gary Jensen Surf my web pages at, http://users.rcn.com/jbyrns/ |
#4
|
|||
|
|||
In article , Joe Bento wrote:
Though it most certainly wouldn't classify as an antique radio, one of the easier paths for a single station receiver would be to use the MK484 chip. It is an entirely self contained TRF receiver needing only the input tank circuit (ferrite bar and compression trimmer, for example) and an audio amp. No external antenna is required, and performance is comparable to a simple superhet. How does the "MK484" achieve selectivity comparable to a simple superhet with only an input tank circuit providing selectivity? Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
#5
|
|||
|
|||
Though it only uses a single tank, the chip has an input impedance of
several megohms. The selectivity would come from the overall Q of your tank circuit. In some of the crystal set forums I participate in, there are people that build crystal sets with better than 10KHz selectivity. It seems to come from the quality and load put on the tank circuit. Joe, N6DGY www.kirtland.com for e-mail John Byrns wrote: In article , Joe Bento wrote: Though it most certainly wouldn't classify as an antique radio, one of the easier paths for a single station receiver would be to use the MK484 chip. It is an entirely self contained TRF receiver needing only the input tank circuit (ferrite bar and compression trimmer, for example) and an audio amp. No external antenna is required, and performance is comparable to a simple superhet. How does the "MK484" achieve selectivity comparable to a simple superhet with only an input tank circuit providing selectivity? Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
#6
|
|||
|
|||
"Joe Bento" Though it only uses a single tank, the chip has an input impedance of several megohms. The selectivity would come from the overall Q of your tank circuit. In some of the crystal set forums I participate in, there are people that build crystal sets with better than 10KHz selectivity. ** You are running fast and loose with the concept of "selectivity" there. To reject strong adjacent channel signals on the AM band one tuned circuit will never do - at least several are needed as with a TRF receiver. As has been pointed out here recently, a single tuned circuit 3dB down 10 kHz away from the centre of a transmission has at most 9 dB attenuation 20 kHz away and 15 dB 40 kHz away. That is pathetic selectivity. ............. Phil |
#7
|
|||
|
|||
There's a jazz FM station in Atlanta that is nearly impossible to receive
from my home. I've tried at least a dozen receivers or tuners, and only a digital Hafler SS FM tuner could nail it. I've got front ends and crystal IF strips to play with. I'd sure be interested if anyone has a good sources for building a single frequency FM stereo receiver . . . From: (John Byrns) Organization: KAOS Newsgroups: rec.antiques.radio+phono,rec.audio.tubes Date: Sun, 25 Jan 2004 13:54:12 -0600 Subject: Designs for a single frequency high performance AM-MW receiver? Hi Gary, Your request is interesting, I have built a couple of single channel AM-MW receivers in the past, and have recently been contemplating another design. The first single channel design I built was a crystal controlled tube superhetrodyne, and the second was a transistorized TRF receiver. Your design requirements are a little confusing to me and I am hoping you can elaborate a little further? My concept of a single frequency receiver is directed to receiving one particular station, which would determine the required sensitivity, selectivity, and other performance parameters on an individual basis. You are asking for high sensitivity and selectivity, with large dynamic range, and variable bandwidth. This seems unusual to me as I don't see why all this would be necessary in a single channel receiver dedicated to receiving one station or frequency? Can you elaborate further on the reasons behind your requirements? Does this receiver have a specific use, or is it to be easily adaptable to any possible use requiring a single frequency receiver? A couple of other questions come to mind. What sort of antenna input circuit the receiver should have? This would be determined by the type of antennas the single channel receiver must be able to work with. Also once built, should it be relatively easy to retune the receiver for a different frequency, or can it be designed for a single fixed frequency from the start, with no allowance for easily changing to a different frequency later? An interesting question in my mind is should the receiver use an envelope detector, or a synchronous detector? Before laying out my ideas I would like to hear more about what you want this receiver to do. Regards, John Byrns In article , wrote: This is an unusual (and maybe fun) request to discuss the design possibilities for a high-performance, single frequency (single channel) medium-wave receiver. That is, the receiver will be designed to receive only one MW frequency, such as, for example, 830 kHz. Of course, some would say this is silly, but I'm curious to see how the design would evolve when the need to tune across a significant portion of the medium-wave band is removed (any "tuning" included in the design will be very narrow, such as 1-2 kHz to compensate for slightly off-channel transmission.) To clarify, what I mean by high-performance is very good sensitivity, selectivity, dynamic range, etc. -- to be excellent for DXing as well as having excellent sound fidelity for strong stations -- to be an all things to all users receiver (albeit single frequency/channel.) Obviously, it can include several controls for fine-tuning, varying bandwidth, etc., etc. -- be innovative. Since I'm fond of tubes, the design should be open to using them (but not required) for achieving some performance goal, such as sensitivity. Of course, other unusual components can be considered, including the venerable crystal. It can be a hybrid design of the very new with the very old (although I'd also like to consider, for nostalgia-sake, designs employing only components used in 1930's radios.) The design need not be superhet, and in fact I prefer to explore simpler and innovative non-superhet circuits. It should have a pre-amplified line-out to drive a separate audio amplifier (so pure crystal sets are out, although a crystal set with a pre-amp is of possible discussion, I suppose.) Ok, that's the constraints. I hope several find this of sufficient interest to contribute to the discussion. Thanks. Gary Jensen Surf my web pages at, http://users.rcn.com/jbyrns/ |
#8
|
|||
|
|||
"Jon Yaeger" wrote in message ... There's a jazz FM station in Atlanta that is nearly impossible to receive from my home. I've tried at least a dozen receivers or tuners, and only a digital Hafler SS FM tuner could nail it. ** Tried a high gain Yagi aimed in the direction ? Effectively an RF amp and pre-selector in one. .......... Phil |
#9
|
|||
|
|||
I do not know exactly what Gary's objectives are. But here are my two
cents: 1. Even it is a single channel receiver, I would still suggest the use of one or two intermediate frequency (IF) stages. Because to achieve good sensitivity you need to have enough gain. Having too much gain in one single frequency will cause the receiver to be unstable. 2. However, very high gain alone will not do the trick. If the first RF amplifier (let's call it a LNA or Low Noise Amplifier) in a receiver has very high Noise Figure (NF), the high gain receiver will simply bring up the thermal noise floor along with the desired signal. It's typical to have a transistor LNA of 1.2 dB NF with today's technology in 900 MHz, 2.4 GHz and 5 GHz (I am more familiar with these frequencies). I have no data on tube NF in the AM-MW band. But I seem to recall that transistor LNA has lower NF. I also recall that triode cascode amplifier has lower NF than sharp cutoff pentode. 3. I would take advantage of "single channel" to its extreme and use a very narrow bandpass filter (BPF) between the antenna and the LNA. Neither audio nor RF amplifiers have unlimited dynamic range. Clipping in the audio amplifier will translate into "intermodulation" in RF. Intermodulation is an irreversible process (a non- linear process), i.e., once the desired signal is corrupted by the intermodulation, you will not be able to recover it by further flittering (a linear process), even with an ideal or "brick wall" IF filter. [For the same similar reason, most good audio amplifiers have a low pass filter between the input jack and the input stage.] 4. But an elaborate BPF may come at the cost of high insertion loss, which will cause the sensitivity to suffer. So we need to strike a balance on the order of the BPF. 5. To minimize the intermodulation problem, we can also use a LNA with higher "head room" or IP3 in the RF alphabet soup. -10 dBm is typical in the transistor LNA. I seem to recall +10 dBm for some tube receiver. 6. I lived in the Portland area a few years ago and my favorite station was AM910. They played songs from the 50's and 60's and my 1940 Philco console was a perfect match for the station. There was only one problem though, the second harmonics of the 455 kHz IP (910 kHz) leaked and beated with the incoming AM910. The beat was not very loud but noticeable. The only way to get rid of this is to provide very good shielding and isolation of the entire IF stage, especially a very good LPF physical path after the AM detection. Another way to eliminate this is to use high IF frequencies like 10.7 MHz. Unfortunately 10.7 MHz LC filter from the tube FM has too wide a bandwidth for AM. So we have to use ceramic 10.7 MHz filters, which are not too friendly to tubes (requiring matching circuits). Jerry "Gary Jensen" wrote in message ... This is an unusual (and maybe fun) request to discuss the design possibilities for a high-performance, single frequency (single channel) medium-wave receiver. That is, the receiver will be designed to receive only one MW frequency, such as, for example, 830 kHz. Of course, some would say this is silly, but I'm curious to see how the design would evolve when the need to tune across a significant portion of the medium-wave band is removed (any "tuning" included in the design will be very narrow, such as 1-2 kHz to compensate for slightly off-channel transmission.) To clarify, what I mean by high-performance is very good sensitivity, selectivity, dynamic range, etc. -- to be excellent for DXing as well as having excellent sound fidelity for strong stations -- to be an all things to all users receiver (albeit single frequency/channel.) Obviously, it can include several controls for fine-tuning, varying bandwidth, etc., etc. -- be innovative. Since I'm fond of tubes, the design should be open to using them (but not required) for achieving some performance goal, such as sensitivity. Of course, other unusual components can be considered, including the venerable crystal. It can be a hybrid design of the very new with the very old (although I'd also like to consider, for nostalgia-sake, designs employing only components used in 1930's radios.) The design need not be superhet, and in fact I prefer to explore simpler and innovative non-superhet circuits. It should have a pre-amplified line-out to drive a separate audio amplifier (so pure crystal sets are out, although a crystal set with a pre-amp is of possible discussion, I suppose.) Ok, that's the constraints. I hope several find this of sufficient interest to contribute to the discussion. Thanks. Gary Jensen |
#10
|
|||
|
|||
Gary Jensen wrote: This is an unusual (and maybe fun) request to discuss the design possibilities for a high-performance, single frequency (single channel) medium-wave receiver. That is, the receiver will be designed to receive only one MW frequency, such as, for example, 830 kHz. Of course, some would say this is silly, but I'm curious to see how the design would evolve when the need to tune across a significant portion of the medium-wave band is removed (any "tuning" included in the design will be very narrow, such as 1-2 kHz to compensate for slightly off-channel transmission.) To clarify, what I mean by high-performance is very good sensitivity, selectivity, dynamic range, etc. -- to be excellent for DXing as well as having excellent sound fidelity for strong stations -- to be an all things to all users receiver (albeit single frequency/channel.) Obviously, it can include several controls for fine-tuning, varying bandwidth, etc., etc. -- be innovative. Since I'm fond of tubes, the design should be open to using them (but not required) for achieving some performance goal, such as sensitivity. Of course, other unusual components can be considered, including the venerable crystal. It can be a hybrid design of the very new with the very old (although I'd also like to consider, for nostalgia-sake, designs employing only components used in 1930's radios.) The design need not be superhet, and in fact I prefer to explore simpler and innovative non-superhet circuits. It should have a pre-amplified line-out to drive a separate audio amplifier (so pure crystal sets are out, although a crystal set with a pre-amp is of possible discussion, I suppose.) Ok, that's the constraints. I hope several find this of sufficient interest to contribute to the discussion. Thanks. Gary Jensen If the wanted station you want to tune to is a local strong station with nothing strong and local within 50 kHz of your stn, then you could set up a couple of double tuned critically coupled RF transformers with one RF amp tube between the two, and a low distortion detector and audio CF buffer output, so just one 6J7 pentode, one 12AU7 double triode, and a couple of germanium diodes, is all you'd need. For the widest AF bw, you may ned to have a 2nd RFamp and have low Q RF response. Forget DXing! these will be noisy no matter what you do. Just because you try to single out just one F, it don't mean that a one channel TRF will detect such a station without noise from other nearby stations, or distant stations on the same F. Patrick Turner. |
#11
|
|||
|
|||
Jerry Wang wrote: I do not know exactly what Gary's objectives are. But here are my two cents: 1. Even it is a single channel receiver, I would still suggest the use of one or two intermediate frequency (IF) stages. Because to achieve good sensitivity you need to have enough gain. Since you only want one channel, there is no need for a frequency converter or any IFTs or IF amps, and a TRF with four tuned circuits in the form of two critically coupled RF trannies will do nicely. Having too much gain in one single frequency will cause the receiver to be unstable. Use low gain; a low Gm pentode is all you may need, and gain and Q could be reduced by strapping R across the LC circuits of the RFTs. Then use two RF amps, each with low gain, and your'e away. 2. However, very high gain alone will not do the trick. If the first RF amplifier (let's call it a LNA or Low Noise Amplifier) in a receiver has very high Noise Figure (NF), the high gain receiver will simply bring up the thermal noise floor along with the desired signal. Noise in AM receivers isn't a problem with local stations, since signal strength is hihg. Receiver noise is low compared to signal noise with DX, but atmospheric/manmade noise and other stations will ruin DX well before receiver noise. A cascoded triode can be used for the RF amp, and this will have far less noise than a the 6J7. It's typical to have a transistor LNA of 1.2 dB NF with today's technology in 900 MHz, 2.4 GHz and 5 GHz (I am more familiar with these frequencies). I have no data on tube NF in the AM-MW band. But I seem to recall that transistor LNA has lower NF. I also recall that triode cascode amplifier has lower NF than sharp cutoff pentode. I repeat, receiver noise with AM MW is negligible for locals. Many great communications recievers like ancient AR7 had two RF amps, a mixer, and two IF amps, and two stage audio amp, and they were quieter than the antenna recieved noise. The comm recievers had very low bandwidth, about 3 kHz was the std, but much less bw of only say 100 Hz was possible to tune out morse code from a very weak signal otherwise lost in the noise picked up. 3. I would take advantage of "single channel" to its extreme and use a very narrow bandpass filter (BPF) between the antenna and the LNA. Neither audio nor RF amplifiers have unlimited dynamic range. Clipping in the audio amplifier will translate into "intermodulation" in RF. Intermodulation is an irreversible process (a non- linear process), i.e., once the desired signal is corrupted by the intermodulation, you will not be able to recover it by further flittering (a linear process), even with an ideal or "brick wall" IF filter. [For the same similar reason, most good audio amplifiers have a low pass filter between the input jack and the input stage.] Examine basic communications analog receivers to learn more about... 4. But an elaborate BPF may come at the cost of high insertion loss, which will cause the sensitivity to suffer. So we need to strike a balance on the order of the BPF. Gain will over come insertion losses. 5. To minimize the intermodulation problem, we can also use a LNA with higher "head room" or IP3 in the RF alphabet soup. -10 dBm is typical in the transistor LNA. I seem to recall +10 dBm for some tube receiver. Tubes offer better dynamic range, and clipping and IMD can all be easily avoided. 6. I lived in the Portland area a few years ago and my favorite station was AM910. They played songs from the 50's and 60's and my 1940 Philco console was a perfect match for the station. There was only one problem though, the second harmonics of the 455 kHz IP (910 kHz) leaked and beated with the incoming AM910. The beat was not very loud but noticeable. The only way to get rid of this is to provide very good shielding and isolation of the entire IF stage, especially a very good LPF physical path after the AM detection. Another way to eliminate this is to use high IF frequencies like 10.7 MHz. Unfortunately 10.7 MHz LC filter from the tube FM has too wide a bandwidth for AM. Try 2Mhz IF and this has better results. So we have to use ceramic 10.7 MHz filters, which are not too friendly to tubes (requiring matching circuits). Why ceramic filters at all? But then they can be used with tubes if the tube is say a cathode follower, and voltage is small, or have an RF transformer... Patrick Turner. Jerry "Gary Jensen" wrote in message ... This is an unusual (and maybe fun) request to discuss the design possibilities for a high-performance, single frequency (single channel) medium-wave receiver. That is, the receiver will be designed to receive only one MW frequency, such as, for example, 830 kHz. Of course, some would say this is silly, but I'm curious to see how the design would evolve when the need to tune across a significant portion of the medium-wave band is removed (any "tuning" included in the design will be very narrow, such as 1-2 kHz to compensate for slightly off-channel transmission.) To clarify, what I mean by high-performance is very good sensitivity, selectivity, dynamic range, etc. -- to be excellent for DXing as well as having excellent sound fidelity for strong stations -- to be an all things to all users receiver (albeit single frequency/channel.) Obviously, it can include several controls for fine-tuning, varying bandwidth, etc., etc. -- be innovative. Since I'm fond of tubes, the design should be open to using them (but not required) for achieving some performance goal, such as sensitivity. Of course, other unusual components can be considered, including the venerable crystal. It can be a hybrid design of the very new with the very old (although I'd also like to consider, for nostalgia-sake, designs employing only components used in 1930's radios.) The design need not be superhet, and in fact I prefer to explore simpler and innovative non-superhet circuits. It should have a pre-amplified line-out to drive a separate audio amplifier (so pure crystal sets are out, although a crystal set with a pre-amp is of possible discussion, I suppose.) Ok, that's the constraints. I hope several find this of sufficient interest to contribute to the discussion. Thanks. Gary Jensen |
#12
|
|||
|
|||
Jerry Wang wrote:
I do not know exactly what Gary's objectives are. But here are my two cents: 1. Even it is a single channel receiver, I would still suggest the use of one or two intermediate frequency (IF) stages. Because to achieve good sensitivity you need to have enough gain. Having too much gain in one single frequency will cause the receiver to be unstable. 2. However, very high gain alone will not do the trick. If the first RF amplifier (let's call it a LNA or Low Noise Amplifier) in a receiver has very high Noise Figure (NF), the high gain receiver will simply bring up the thermal noise floor along with the desired signal. It's typical to have a transistor LNA of 1.2 dB NF with today's technology in 900 MHz, 2.4 GHz and 5 GHz (I am more familiar with these frequencies). I have no data on tube NF in the AM-MW band. But I seem to recall that transistor LNA has lower NF. I also recall that triode cascode amplifier has lower NF than sharp cutoff pentode. On the MW AM band, the natural noise from the outside world is usually much stronger than the front end noise figure of more "modern" tubes and solid state. So one doesn't worry about noise figure in an AM MW set. NF is an issue in TV sets, as the VHF tuners used triode cascode and such circuits. Also FM tuners. 3. I would take advantage of "single channel" to its extreme and use a very narrow bandpass filter (BPF) between the antenna and the LNA. Neither audio nor RF amplifiers have unlimited dynamic range. Clipping in the audio amplifier will translate into "intermodulation" in RF. Intermodulation is an irreversible process (a non- linear process), i.e., once the desired signal is corrupted by the intermodulation, you will not be able to recover it by further flittering (a linear process), even with an ideal or "brick wall" IF filter. [For the same similar reason, most good audio amplifiers have a low pass filter between the input jack and the input stage.] Filtering before amplifying is a good idea. Then you don't have to back off on the gain to listen to the weak but interesting station near the local flamethrower. I have a set that needs more work in this area. 4. But an elaborate BPF may come at the cost of high insertion loss, which will cause the sensitivity to suffer. So we need to strike a balance on the order of the BPF. As the NF becomes a problem. 5. To minimize the intermodulation problem, we can also use a LNA with higher "head room" or IP3 in the RF alphabet soup. -10 dBm is typical in the transistor LNA. I seem to recall +10 dBm for some tube receiver. 6. I lived in the Portland area a few years ago and my favorite station was AM910. They played songs from the 50's and 60's and my 1940 Philco console was a perfect match for the station. There was only one problem though, the second harmonics of the 455 kHz IP (910 kHz) leaked and beated with the incoming AM910. The beat was not very loud but noticeable. The only way to get rid of this is to provide very good shielding and isolation of the entire IF stage, especially a very good LPF physical path after the AM detection. Another way to eliminate this is to use high IF frequencies like 10.7 MHz. Unfortunately 10.7 MHz LC filter from the tube FM has too wide a bandwidth for AM. So we have to use ceramic 10.7 MHz filters, which are not too friendly to tubes (requiring matching circuits). One could retune the IF up or down to move the 2nd harmonic away from that station (at the cost of trashing any DX on 900 or 920). And the cost of poorer tracking across the dial and a mismatch of the dial calibrations. But for a single channel set this is not an issue. |
#13
|
|||
|
|||
[Following up on a thread dating back to January, similar to one I
started recently. Responding to Patrick Turner's comments.] Patrick Turner wrote in January 2004: Jerry Wang wrote: 1. Even it is a single channel [AM] receiver, I would still suggest the use of one or two intermediate frequency (IF) stages. Because to achieve good sensitivity you need to have enough gain. Since you only want one channel, there is no need for a frequency converter or any IFTs or IF amps, and a TRF with four tuned circuits in the form of two critically coupled RF trannies will do nicely. Interesting. As I noted in a recent message, it is very intriguing to build a modernized, high-performance AM tube tuner using the "channel" approach. This takes advantage of the fact that licensed broadcasters today must broadcast on specific frequencies, every 10 khz in North America and 9 khz in Europe and elsewhere. So, instead of trying to be able to continuously tune across the BCB spectrum, we can think outside the box for the moment and consider the alternative of building reasonably optimized tuning circuits for each listened-to frequency. There'd be a switch to select from a number of channels, each associated with a specific frequency the user wants to listen to (suggesting a plugin mini-board for each channel, but there are other possible configurations.) I infer from what Patrick said that it is unnecessary for a single frequency AM tuner to be a super-het design, and that (I assume) a much simpler two RF amp TRF design is sufficient for good to excellent audio quality and good to excellent sensitivity and selectivity. (John Byrns implies the same in his various comments on TRF AM tuners.) So, with respect to the channel approach, the next question to ask is if we can use the same two critically coupled RF transformers (as Patrick notes), and *independently* vary several of the other smaller components (e.g., capacitors, resistors, and even inductors) in the two or three tuning stages (if we include the antenna tuner) so as to maintain, from channel to channel in the BCB, reasonably optimal bandwidth and other desirable tuning characteristics? [With traditional continuous tuning, achieved with multiganged air capacitors, we do indeed vary a few capacitors in the tuning circuitry, but because all of them track each other, in reality we only have one degree of freedom, leading to circuit design constraints for continuous "single knob" tuning. Now imagine, for each channel frequency, to *independently* vary the value of several components at the same time -- we now have several degrees of freedom to play with and thereby hope to achieve reasonably constant (as a function of frequency) bandpass characteristics. Obviously, architecturally implementing this in a practical AM tuner design is not trivial (we do benefit by throwing away the multigang air capacitor.) However, several ideas suggest themselves. For example, we can imagine having multiple plugin slots, where we plug into each slot a PCB mini-board specific to a particular frequency. The board will contain the few components whose values *independently* change as a function of frequency. They probably will have trimmers for fine calibration of the center frequency and other bandpass filter characteristics. We may need multiple mini-boards for each channel (one for each tuning stage) if necessary for shielding purposes (to prevent oscillation by stage-to-stage interference if that is a problem.) And if higher frequency channel boards require some minor changes in the circuitry configuration, and not just component value changes, that can easily be done, too. In principle, this tuner might even be able to extend a little beyond (on both sides) the 500-1800 khz MW band -- just plugin the right mini-board circuitry for the frequency desired. Of course, others here will probably have much better ideas as to how to implement the channel approach. Thoughts? Comments? Criticisms? Jon Noring (It's interesting to think of doing the same "channel" approach for an FM tube tuner. Will that also confer several advantages in simplifying the circuit design for the same overall performance level?) |
#14
|
|||
|
|||
Jon Noring wrote: [Following up on a thread dating back to January, similar to one I started recently. Responding to Patrick Turner's comments.] Patrick Turner wrote in January 2004: Jerry Wang wrote: 1. Even it is a single channel [AM] receiver, I would still suggest the use of one or two intermediate frequency (IF) stages. Because to achieve good sensitivity you need to have enough gain. Since you only want one channel, there is no need for a frequency converter or any IFTs or IF amps, and a TRF with four tuned circuits in the form of two critically coupled RF trannies will do nicely. Interesting. As I noted in a recent message, it is very intriguing to build a modernized, high-performance AM tube tuner using the "channel" approach. This takes advantage of the fact that licensed broadcasters today must broadcast on specific frequencies, every 10 khz in North America and 9 khz in Europe and elsewhere. So, instead of trying to be able to continuously tune across the BCB spectrum, we can think outside the box for the moment and consider the alternative of building reasonably optimized tuning circuits for each listened-to frequency. There'd be a switch to select from a number of channels, each associated with a specific frequency the user wants to listen to (suggesting a plugin mini-board for each channel, but there are other possible configurations.) The problem is that if you want a channel at 9 kHz intervals to choose from across the band, you need around 12 perfectly set up tuning circuits all with multiple LC circuits. Then you need sitable switching. Far better is to forget all that BS and use a PC to decode the antenna signal. I infer from what Patrick said that it is unnecessary for a single frequency AM tuner to be a super-het design, and that (I assume) a much simpler two RF amp TRF design is sufficient for good to excellent audio quality and good to excellent sensitivity and selectivity. (John Byrns implies the same in his various comments on TRF AM tuners.) But you won't sell many kits set up optimally for just one F. As soon as the owner moves to another area, the radio becomes useless. So, with respect to the channel approach, the next question to ask is if we can use the same two critically coupled RF transformers (as Patrick notes), and *independently* vary several of the other smaller components (e.g., capacitors, resistors, and even inductors) in the two or three tuning stages (if we include the antenna tuner) so as to maintain, from channel to channel in the BCB, reasonably optimal bandwidth and other desirable tuning characteristics? This has al been investigated before, and the conclusions were about as simple as possible by about 1927. Try studying basic L,C, & R theory, and work all this out for yourself. I once fixed a 1932 TRF Radiola with only two single tuned circuits. It gave OK local reception with about 5k of audio BW where the stations were 100 kHz or more apart. It used the then high tech new fangled type 22 tetrode. [With traditional continuous tuning, achieved with multiganged air capacitors, we do indeed vary a few capacitors in the tuning circuitry, but because all of them track each other, in reality we only have one degree of freedom, leading to circuit design constraints for continuous "single knob" tuning. Now imagine, for each channel frequency, to *independently* vary the value of several components at the same time -- we now have several degrees of freedom to play with and thereby hope to achieve reasonably constant (as a function of frequency) bandpass characteristics. 1925 TRFs had 3 or 4 separate tuning gangs, each set to a certain numbered position for reception of a given station. Finding stations was exciting. Try studying the history of radio, and you won't need to ask such questions here. Obviously, architecturally implementing this in a practical AM tuner design is not trivial (we do benefit by throwing away the multigang air capacitor.) However, several ideas suggest themselves. The 1932 Radiola did have its two single capacitor gangs connected by cables, which had corroded, so I used builder's line. It worked OK. For example, we can imagine having multiple plugin slots, where we plug into each slot a PCB mini-board specific to a particular frequency. ? The board will contain the few components whose values *independently* change as a function of frequency. They probably will have trimmers for fine calibration of the center frequency and other bandpass filter characteristics. We may need multiple mini-boards for each channel (one for each tuning stage) if necessary for shielding purposes (to prevent oscillation by stage-to-stage interference if that is a problem.) And if higher frequency channel boards require some minor changes in the circuitry configuration, and not just component value changes, that can easily be done, too. In principle, this tuner might even be able to extend a little beyond (on both sides) the 500-1800 khz MW band -- just plugin the right mini-board circuitry for the frequency desired. This idea is totally impractical for 120 different stations, and plug ins get lost or broken, or worn out. Of course, others here will probably have much better ideas as to how to implement the channel approach. You bet there are, and only possible with chip technology, with press button station selection, and digital station F read out, with digitally generated oscillator frequency for the F converter of a superhet, with ceramic filter IF. Grundig have been multiband radios for about 20 years +. Not a tube in sight inh these lightweight plastic radios bought cheaply by the masses to allow connection to the world's AM, FM, and HF bands, and even amateur SSB stations. But how to improve such designs to make wider AF BW is unknown to me. Try examining the history of Yeasu. Thoughts? Comments? Criticisms? Jon Noring (It's interesting to think of doing the same "channel" approach for an FM tube tuner. Will that also confer several advantages in simplifying the circuit design for the same overall performance level?) Study the way most post 1980 AM/FM tuners are constructed. Tubes cannot be used with such methods. I reckon you got a pile of reading to do. Patrick Turner. |
#15
|
|||
|
|||
(assorted snips because these threads get sooo long!) see reply below
Jon Noring wrote: [Following up on a thread dating back to January, similar to one I started recently. Responding to Patrick Turner's comments.] Patrick Turner wrote in January 2004: Jerry Wang wrote: 1. Even it is a single channel [AM] receiver, I would still suggest the use of one or two intermediate frequency (IF) stages. Because to achieve good sensitivity you need to have enough gain. Since you only want one channel, there is no need for a frequency converter or any IFTs or IF amps, and a TRF with four tuned circuits in the form of two critically coupled RF trannies will do nicely. Interesting. As I noted in a recent message, it is very intriguing to build a modernized, high-performance AM tube tuner using the "channel" approach. T So, with respect to the channel approach, the next question to ask is if we can use the same two critically coupled RF transformers (as Patrick notes), and *independently* vary several of the other smaller components (e.g., capacitors, resistors, and even inductors) in the two or three tuning stages (if we include the antenna tuner) so as to maintain, from channel to channel in the BCB, reasonably optimal bandwidth and other desirable tuning characteristics? [With traditional continuous tuning, achieved with multiganged air capacitors, we do indeed vary a few capacitors in the tuning circuitry, but because all of them track each other, in reality we only have one degree of freedom, leading to circuit design constraints for continuous "single knob" tuning. Now imagine, for each channel frequency, to *independently* vary the value of several components at the same time -- we now have several degrees of freedom to play with and thereby hope to achieve reasonably constant (as a function of frequency) bandpass characteristics. Obviously, architecturally implementing this in a practical AM tuner design is not trivial (we do benefit by throwing away the multigang air capacitor.) However, several ideas suggest themselves. For example, we can imagine having multiple plugin slots, where we plug into each slot a PCB mini-board specific to a particular frequency. Of course, others here will probably have much better ideas as to how to implement the channel approach. Thoughts? Comments? Criticisms? Jon Noring For a one-channel receiver it makes perfect sense. Beyond that any advantage is lost. Why would I say that? You can create a perfectly acceptable single IF filter with not so much ado. Lets use 455kc as the example. Its considerably easier to build a single 'custom' IF filter at 455kc to do what you want to do than it is a bunch of modules at three or four times that frequency. Yes, you could do as you suggest but I see no advantage in doing so. It would be more critical, more expensive and probably not yield as good a result as a nice 455 filter. One thing I haven't heard mentioned, and admittedly I have only been grazing what has been a very windy thread, why not use a WIDE 455kc IF with tunable traps on either side? You can get a very steep skirt on a good hi-q trap...likely steeper than in a transformer configuration that is inherently q-disadvantaged. This would come in handy at night when dozens of adjacent channels stations will be struggling to find their way into your wide bandpass - and this scenario alone is a huge negative about any wideband scheme that needs to be addressed. A savvy person might be able to 'gang' the two adjacent channel traps for a single knob "bandwidth" control. My apologies if I'm missing the point. I'm unsure if the motive of the discussion is that of a wideband AM radio or a discussion of ways to reinvent the wheel. -Bill M |
#16
|
|||
|
|||
Patrick Turner wrote:
Jon Noring wrote: As I noted in a recent message, it is very intriguing to build a modernized, high-performance AM tube tuner using the "channel" approach. This takes advantage of the fact that licensed broadcasters today must broadcast on specific frequencies, every 10 khz in North America and 9 khz in Europe and elsewhere. So, instead of trying to be able to continuously tune across the BCB spectrum, we can think outside the box for the moment and consider the alternative of building reasonably optimized tuning circuits for each listened-to frequency. There'd be a switch to select from a number of channels, each associated with a specific frequency the user wants to listen to (suggesting a plugin mini-board for each channel, but there are other possible configurations.) The problem is that if you want a channel at 9 kHz intervals to choose from across the band, you need around 12 perfectly set up tuning circuits all with multiple LC circuits. Then you need sitable switching. Far better is to forget all that BS and use a PC to decode the antenna signal. Well, if you recall, I did agree with you that the ultimate AM tuner will be all PC-based DSP as close to the antenna feed as possible, along with a true Class D digital amp for final output to the speakers. Everything inbetween will be only real-time digital signal processing. No need to sell me on that! So why are we even bothering talking about tube-based equipment? smile/ Because there is definitely an interest in tube-based equipment, for various reasons: nostalgia, the challenge, the aesthetics, and in some cases (such as high end audiophile amplifiers), The Sound (tm). When true Class D amplifiers mature, they will supplant tube amps for pure sonic quality. But that's still a few years off until PWM switching improves.) And even then, tube equipment is definitely of interest for aesthetic and nostalgic reasons. Regarding the channel TRF receiver being "BS", well that's in the eye of the beholder. smile/ I infer from what Patrick said that it is unnecessary for a single frequency AM tuner to be a super-het design, and that (I assume) a much simpler two RF amp TRF design is sufficient for good to excellent audio quality and good to excellent sensitivity and selectivity. (John Byrns implies the same in his various comments on TRF AM tuners.) But you won't sell many kits set up optimally for just one F. As soon as the owner moves to another area, the radio becomes useless. If the owner installed a number of "mini-boards" (or whatever) to receive stations (both local and DX), then moves, he simply either swaps mini-boards with new ones, or keeps the ones he has and adds new ones, so now he has even more channels to "surf". The boards don't become useless at all, especially if they're interested in casual DX. The mini-boards can be sold either as kit boards (just add the components of the right value, calibrate and plug-in) or buy them already made and calibrated from the kit supplier. For simpler bandpass tuning filters (not the complex ones like nine order Chebychev, as an extreme example), the mini-board may only have a few simple components to add. For example, for a given center frequency (check the chart) just add a capacitor here of a certain value, a resistor there of a certain value, an inductor over there of a certain value, etc. Not a big deal. I envision the mini-boards to maybe be as small as 1" x 2" in size, more like a stick, with terminals on the narrow end to plug into a slot connected to the main circuitry of the tuner (hopefully none of the components will be very large -- thus the question I asked you about making the critically coupled RF transformers common to all channels -- we don't want to have any more than two or three of them!) So, with respect to the channel approach, the next question to ask is if we can use the same two critically coupled RF transformers (as Patrick notes), and *independently* vary several of the other smaller components (e.g., capacitors, resistors, and even inductors) in the two or three tuning stages (if we include the antenna tuner) so as to maintain, from channel to channel in the BCB, reasonably optimal bandwidth and other desirable tuning characteristics? This has al been investigated before, and the conclusions were about as simple as possible by about 1927. Try studying basic L,C, & R theory, and work all this out for yourself. I have. This channel concept has nothing to do with "basics". It is a twist to TRF tuner architecture taking advantage of the fact that AM BCB is done in specific assigned frequencies, just like FM, like TV, like the CB band, etc. It will not be practical for general shortwave listening since that is a huge band (from 1.8 mhz to 30.0 mhz) and amateurs in particular pick their own frequencies (and over time even commercial SW broadcasts move around a lot, for those only interested in listening to the majors like Radio Australia, as I do many evenings on 15.515 mhz. It comes in loud and clear here in Salt Lake City.) Back in the late 20's and early 30's, on MW there was clearly a need for continuous tuning since broadcasts could be anywhere on the band. (And tubes then had poor gain, among other problems.) Today, a lot of the issues of building TRF circuitry is trying to overcome the limitations of one-dimensional tuning using, for example, a multigang air capacitor -- John Byrns is going through agony trying to find the magic formula to get what he wants with a multigang air capacitor. But with the channel TRF concept, the sky's the limit as to how many components in the bandpass tuning filter can be independently selected and hardwired for any given frequency. So one can optimally tune the bandpass characteristics for each and every frequency in the TRF without worrying how that affects other frequencies, since each channel frequency tuning circuit is now effectively decoupled (made independent) from the other channel frequencies. [With traditional continuous tuning, achieved with multiganged air capacitors, we do indeed vary a few capacitors in the tuning circuitry, but because all of them track each other, in reality we only have one degree of freedom, leading to circuit design constraints for continuous "single knob" tuning. Now imagine, for each channel frequency, to *independently* vary the value of several components at the same time -- we now have several degrees of freedom to play with and thereby hope to achieve reasonably constant (as a function of frequency) bandpass characteristics. 1925 TRFs had 3 or 4 separate tuning gangs, each set to a certain numbered position for reception of a given station. Finding stations was exciting. Try studying the history of radio, and you won't need to ask such questions here. With the channel TRF concept, the component values of the bandpass filter (or parts of the filter circuit) are hardwired on the channel plug-in board (and trimmed during calibration), so all the person has to do in listening to the tuner is switch to the channel, and the radio will be in tune to the desired frequency, with the optimal bandpass characteristics for that frequency. (There is likely to be a need for a very fine tuning control, maybe +/- 1 khz, to handle slight drift, both for tuner warmup, and for the inevitable long-term drifting of component values.) I suppose back in 1925 radio stations where in all sorts of weird locations on the dial, and constantly moving around, so hardwiring all the tuning components for a particular frequency, and likewise for other frequencies, was not even an option. For example, we can imagine having multiple plugin slots, where we plug into each slot a PCB mini-board specific to a particular frequency. ? You probably understand the channel TRF concept, but did not understand what I wrote the above, so let me restate with an example: I want my channel TRF tuner to tune in 830 khz (WCCO in Minneapolis), so I get the mini-board for that frequency already hardwired with the optimum configuration of the various tuner components, plug it in, and then listen to that frequency whenever I switch to whatever channel slot I placed that mini-board in (I am reminded of how components are plugged into PCs, such as via PCI slots.) Or, I buy the blank mini-board, check the kit-supplied chart for 830 khz, and then solder in a 50 ohm resistor in this spot, a 200pF capacitor in that spot, etc. -- probably will take me all of five minutes. Then calibrate it by tweaking the trimmers. If I instead want the mini-board to tune 1160 khz (KSL in Salt Lake City), I check the chart, put in a 75 ohm resistor in this spot, a 150pF capacitor in that spot, etc. (whatever values are called for.) Then calibrate it. Plug it in, listen to 1160 khz, knowing that the TRF bandpass tuning circuitry is now optimized for that frequency, and much better optimized than could ever be done with the one dimensional limits of a multigang tuning capacitor. For those who build tube kits, this will border on the trivial. And some hobbyists may find the channel TRF AM tube tuner architecture of real interest, since now they can more easily experiment with new higher-order bandpass filters of various mathematical functions to see how they affect TRF performance. This could lead to a revised mini-board to be issued at some future time based on all this research, and the channel tuner owner can, if they so choose, simply buy or build updated boards for the broadcast stations of interest, and instantly get better performance. It's possible to mix bandpass filters for different stations: a third order Butterworth for 1160 khz, and a fifth order Chebychev for 830 khz. The possibilities are endless. The board will contain the few components whose values *independently* change as a function of frequency. They probably will have trimmers for fine calibration of the center frequency and other bandpass filter characteristics. We may need multiple mini-boards for each channel (one for each tuning stage) if necessary for shielding purposes (to prevent oscillation by stage-to-stage interference if that is a problem.) And if higher frequency channel boards require some minor changes in the circuitry configuration, and not just component value changes, that can easily be done, too. In principle, this tuner might even be able to extend a little beyond (on both sides) the 500-1800 khz MW band -- just plugin the right mini-board circuitry for the frequency desired. This idea is totally impractical for 120 different stations, and plug ins get lost or broken, or worn out. I don't believe it is impractical for 120 different stations, for two reasons: 1) Those tube-o-philes who only want to listen to stronger local stations, or to particular distant ones, are likely only to want to have 10-20 stations (with the ability to add more if they want.) One purpose of picking TRF is its legendary high-fidelity audio capability which will appeal to audiophiles -- most won't want to listen to a very weak station 1000 miles away that can only be picked up some evenings. And I believe it is easier to sell tube-o-philes on the Channel TRF concept once it is explained how it maximizes audio performance for each and every broadcast frequency that cannot be done with a continuously tuned TRF. 2) Those who would use this for casual DXing (and note the hardcore MW DXers will use something like a Drake R8B or ICOM R75, or some digital receiver) will certainly be motivated to add more mini-boards, and can do so over time. The tuner will work with 1 channel board, or with all 130+ (if enough slots are provided. For the moment I am imagining the mini-board approach, but the sky's the limit for other ideas to implement the channel TRF AM tube tuner.) 3) And as noted above, hobbyists may find the "plugin" bandpass filter capability of particular interest. Of course, others here will probably have much better ideas as to how to implement the channel approach. You bet there are, and only possible with chip technology, with press button station selection, and digital station F read out, with digitally generated oscillator frequency for the F converter of a superhet, with ceramic filter IF. Grundig have been multiband radios for about 20 years +. I have a Radio Shack DX-399 (the Sangean 606A) which is a very good performer for casual MW (with the Radio Shack MW loop) and shortwave DXing. So I am very familiar with that hobby, and with the benefits digital systems bring to tuners. You need not sell me on that! See my previous note above on "why tubes then?" Not a tube in sight inh these lightweight plastic radios bought cheaply by the masses to allow connection to the world's AM, FM, and HF bands, and even amateur SSB stations. See my previous note above on "why tubes then?" (It's interesting to think of doing the same "channel" approach for an FM tube tuner. Will that also confer several advantages in simplifying the circuit design for the same overall performance level?) Study the way most post 1980 AM/FM tuners are constructed. Tubes cannot be used with such methods. O.k. But are you referring to tube-based tuners? Again, if all I wanted was an audiophile grade AM/FM tuner, and did not care about what was under the hood, I'd be open to solid state designs, but I'm specifically looking at tube-based tuners. I still assume that the channel approach to tube-based FM tuner design may confer some benefits, but maybe less since the frequency ratio to tune from the lower to the upper ends of the band (about 1.25) is much less than that for the AM BCB (a whopping 3.5 or so.) And there are probably other factors as well specific to frequency modulation. It's trying to tune the AM band with only one degree of freedom (e.g., air tuning capacitor) which is causing all the hassle in tube-based TRF AM tuner circuit design. One would want to pick a bandpass filter which is optimally tuned to the specific frequency we want to listen to, and this involves optimally selecting *several* component values, not just one as we are limited to by continuous tuning with a multigang tuning capacitor. The channel TRF approach appears to free up the TRF designer from the tyranny of having to compromise the bandpass characteristics over the entire tuning range which only one degree of freedom allows. Of course, superheterodyne is one solution to the TRF problem, and allows for continuous tuning. Note that super-het works because it uses "one channel" (the IF). So in a sense, superheterodyne supports the channel TRF approach for those who don't want to build a super-het, but rather want a pure TRF receiver (e.g., for sound quality reasons, or whatever.) ***** Now, I've made the call several times for classic and proven AM tube tuner designs of the past which have excellent audio quality (and wide bandwidth capability), are good for casual DX use, and can easily be "modernized" for a kit. There are no doubt many excellent super-het designs out there, but I've had very few recommendations. Patrick, since you appear to much prefer super-het over TRF for AM tube tuners, which classic super-het tube AM radio designs of the past would you suggest as candidates to consider? Anybody? I reckon you got a pile of reading to do. Yes, I have been reading. That one-year equivalent of EE training back in 1974 at the University of Minnesota is slowly coming back to me. Back then we spent a few weeks on tubes, and only a couple days with transistors. Things have changed a lot since then. And it was interesting reading about Chebychev bandpass filters today since I wrote a lot of Fortran code years ago to do various types of numerical analytic processing including integration using quadrature with orthogonal polynomials (mostly Legendre polynomials.) It was especially cool to see how the higher order Chebychev polynomials U(x) plot out in the desired shape (well approximately) for a bandpass filter (but with that slight ripple within the bandwidth.) I'm not saying all this to brag, but to give a better idea of my background. Definitely I have a lot to learn, of course, and your posts are helping me to better understand things. I still believe the channel TRF concept is viable for those who want to build the best possible TRF tube tuners where for each frequency the absolute best bandpass characteristics can be chosen without worrying about how it impacts the other frequencies since each channel is now largely independent. The obvious downsides with the tube-based channel TRF concept a 1) The practical, real-world implementation of it (I believe it is doable, I suggest one approach), 2) Losing the ability to continuously tune, which for BCB is not an issue as I've noted several times, and 3) Calibration of each mini-board if done by the kit-builder (I think this is solvable, but it is an issue to consider.) The upsides are several, as previously noted. Thanks for your helpful comments! Jon Noring |
#17
|
|||
|
|||
Bill (exray) wrote:
Jon Noring wrote: Nice to hear from you again, Bill! I'm still in the process of restoring the Philco 37-670 console, and will need your advice on a couple of issues, such as how to replace the rubber insulators on the RF chassis and on the back end of the tuning capacitor, which are all disintegrating due to the radio being exposed to the LA smog for decades. For a one-channel receiver it makes perfect sense. Beyond that any advantage is lost. Why would I say that? You can create a perfectly acceptable single IF filter with not so much ado. Lets use 455kc as the example. It's considerably easier to build a single 'custom' IF filter at 455kc to do what you want to do than it is a bunch of modules at three or four times that frequency. Yes, you could do as you suggest but I see no advantage in doing so. It would be more critical, more expensive and probably not yield as good a result as a nice 455 filter. I think the ultimate explanation is the desire for the tube tuner to remain a pure TRF design, for audio quality purposes -- John Byrns has discussed this as well (yes, we've hammered to death the poor quality of most AM broadcasts, but that's been covered elsewhere.) As soon as one decides the tube tuner is to be a pure TRF, then one is instantly confronted with the very difficult problem in how to get optimal bandpass characteristics for all the frequencies from 500khz to 1800khz. As I read the many messages on this from the Google archive, it clearly borders on a nightmare to overcome when the only degree of freedom the TRF designer has to work with is a variable air capacitor. John Byrns is wrestling with this issue even as I write, trying to find the magic formula. When confronted with an intractable problem in design, it is time to think outside the box. It is obvious we need to have more degrees of freedom in tuning, but for continuous tuning all this does is add more knobs to tweak, not unlike the TRF designs of the 1920's. Do we want to go in that direction? But since we observe the stations on the BCB are restricted to specific frequencies, this means we don't *need* to have continuous tuning, and from this paradigm shift the channel TRF idea springs forth. As I noted in a parallel message I just sent out, the channel TRF has its problems for practical implementation, and it goes against the almost 100 year paradigm of continuous tuning that is so ingrained in BCB radio tuner design, but I think it solves that otherwise intractable problem with TRF tube tuner design. But, if John Byrns or someone else can discover the magic way to allow one degree of freedom to give optimal enough bandpass design for a TRF tube tuner, then that's the direction I'd recommend going, and not the channel TRF approach, interesting as it is. (Of course, understandably many still recommend super-het.) Jon |
#18
|
|||
|
|||
Jon Noring wrote:
Bill (exray) wrote: Jon Noring wrote: Nice to hear from you again, Bill! Indeed, hi again. I think the ultimate explanation is the desire for the tube tuner to remain a pure TRF design, for audio quality purposes -- John Byrns has discussed this as well Firstly, I'm not getting the WHY this (TRF idea) is of such great import. Conceptually its a nice idea to not add 'unnecessary' stages but if one harkens back to why this was (and still is) the panacea to overcome the TRF ills then maybe they shouldn't be categorically discarded as bad things. As soon as one decides the tube tuner is to be a pure TRF, then one is instantly confronted with the very difficult problem in how to get optimal bandpass characteristics for all the frequencies from 500khz to 1800khz. As I read the many messages on this from the Google archive, it clearly borders on a nightmare to overcome when the only degree of freedom the TRF designer has to work with is a variable air capacitor. John Byrns is wrestling with this issue even as I write, trying to find the magic formula. I'm of the mind that going pure TRF is not necessarily the answer to your original request. But we can run with that for the sake of discussion. There may well be some magic combination of ganging inductors and caps but upon finding that we'll still have to weigh in the cost, complexity, repeatability, performance, etc compared to a superhet. Radio folk haven't reached that point yet in 80 odd years so there's no disagreement to be found :-) And don't assume that radio minds are in a 'box'. The crystal radio fanatics beat this issue to death on a daily at a very sophisticated level. When confronted with an intractable problem in design, it is time to think outside the box. It is obvious we need to have more degrees of freedom in tuning, but for continuous tuning all this does is add more knobs to tweak, not unlike the TRF designs of the 1920's. Do we want to go in that direction? I don't...at least not for the purpose of hooking up something to my home stereo for e-z audiophile listening. But since we observe the stations on the BCB are restricted to specific frequencies, this means we don't *need* to have continuous tuning, and from this paradigm shift the channel TRF idea springs forth. I disagree 180 degrees. If BCB channels could be counted on as equivalent building blocks maybe this would apply but we are talking three octaves of frequency range. As I noted in a parallel message I just sent out, the channel TRF has its problems for practical implementation, and it goes against the almost 100 year paradigm of continuous tuning that is so ingrained in BCB radio tuner design, but I think it solves that otherwise intractable problem with TRF tube tuner design. But, if John Byrns or someone else can discover the magic way to allow one degree of freedom to give optimal enough bandpass design for a TRF tube tuner, then that's the direction I'd recommend going, and not the channel TRF approach, interesting as it is. (Of course, understandably many still recommend super-het.) I fully understand what you are suggesting and all I can say is that we've been there and done that. When I stated that you could build a nice hi-q BCB circuit that would yield 3kc bandwidth at 550 and 25 kc width at 1600 I wasn't exaggerating. Intuitively one might think that hey, I'll twist the LC combo somehow and come back to the same Q across the band simply doesn't work...either in numbers or worse still in practice. I'd like to say you can't obtain a sharp 3kc bandwidth at 1600 with a simple LC circuit but thats too open-ended. Suffice it to say that it ain't easy. One can visualize some scenarios of mechanical (or electrical) ganging of components that might approach this goal but that visualization typically falls in the ditch once one tries to transfer the idea from the brain to an actual breadboarded version of the concept. Going back to some of the earlier filter flatness discussion, well toss that idea into the mix when you think in terms of TRF. Not only do you want to achieve a specific width but you want it to be flat. My 3/25 kc TRF scenario isn't flat at all. Its a big peak that just broadens out. When we say a 'bandwidth'number we are relating to something specific like 3 or 6 db down from the peak. Its still a peak in this context. So whats happening at 20 db down? You guessed it, that 25kc number is 150 kc wide. I dunno how you could control the width AND the flatness AND the skirts. I'm a fairly recent convert to crystal radios. For the sake of discussion there's little difference in xtal technology vs trf technology in that both are non-superhet. I get absolutely glorious quality audio from my xtal set when fed thru an amp. With 6 or 8 knobs on the front panel and top notch components I can find a dead spot between semi-locals on 680 and 690. With a local station on 1370 it takes traps and VERY hi-q stuff to ferret out semi-locals on 1240, 1290 and 1480. Its as if it were a totally different radio from one end of the band to the other and this has been the plague of TRF circuits since day one. If I didn't have the local 1370 I could safely say, hey Jon, this is the ticket, but there's scarce few of us who don't have a strong undesired local station to bollox up the works. Go superhet, my man. -Bill M |
#19
|
|||
|
|||
Jon Noring wrote: Patrick Turner wrote: Jon Noring wrote: As I noted in a recent message, it is very intriguing to build a modernized, high-performance AM tube tuner using the "channel" approach. This takes advantage of the fact that licensed broadcasters today must broadcast on specific frequencies, every 10 khz in North America and 9 khz in Europe and elsewhere. So, instead of trying to be able to continuously tune across the BCB spectrum, we can think outside the box for the moment and consider the alternative of building reasonably optimized tuning circuits for each listened-to frequency. There'd be a switch to select from a number of channels, each associated with a specific frequency the user wants to listen to (suggesting a plugin mini-board for each channel, but there are other possible configurations.) The problem is that if you want a channel at 9 kHz intervals to choose from across the band, you need around 12 perfectly set up tuning circuits all with multiple LC circuits. Then you need sitable switching. Far better is to forget all that BS and use a PC to decode the antenna signal. Well, if you recall, I did agree with you that the ultimate AM tuner will be all PC-based DSP as close to the antenna feed as possible, along with a true Class D digital amp for final output to the speakers. Everything inbetween will be only real-time digital signal processing. No need to sell me on that! So why are we even bothering talking about tube-based equipment? smile/ Well you are the one wanting an avaliable kit which had everything, including an ability to glow in the dark ;-) I already got my answer in my kitchen. WTF are all you other keen dudes gonna do about getting good AM to listen to? If I can do it, so can you. Because there is definitely an interest in tube-based equipment, for various reasons: nostalgia, the challenge, the aesthetics, and in some cases (such as high end audiophile amplifiers), The Sound (tm). Tubes do sound the best when they are good, imho. They are lousy devices for computers. When true Class D amplifiers mature, they will supplant tube amps for pure sonic quality. But that's still a few years off until PWM switching improves.) I will believe it when I see it. And even then, tube equipment is definitely of interest for aesthetic and nostalgic reasons. Regarding the channel TRF receiver being "BS", well that's in the eye of the beholder. smile/ Bold Scheme, perhaps, maybe even Naughty Electronics Endeavours, or NEL. I infer from what Patrick said that it is unnecessary for a single frequency AM tuner to be a super-het design, and that (I assume) a much simpler two RF amp TRF design is sufficient for good to excellent audio quality and good to excellent sensitivity and selectivity. (John Byrns implies the same in his various comments on TRF AM tuners.) But you won't sell many kits set up optimally for just one F. As soon as the owner moves to another area, the radio becomes useless. If the owner installed a number of "mini-boards" (or whatever) to receive stations (both local and DX), then moves, he simply either swaps mini-boards with new ones, or keeps the ones he has and adds new ones, so now he has even more channels to "surf". The boards don't become useless at all, especially if they're interested in casual DX. I doubt your idea would ever catch on........ The mini-boards can be sold either as kit boards (just add the components of the right value, calibrate and plug-in) or buy them already made and calibrated from the kit supplier. For simpler bandpass tuning filters (not the complex ones like nine order Chebychev, as an extreme example), the mini-board may only have a few simple components to add. For example, for a given center frequency (check the chart) just add a capacitor here of a certain value, a resistor there of a certain value, an inductor over there of a certain value, etc. Not a big deal. I envision the mini-boards to maybe be as small as 1" x 2" in size, more like a stick, with terminals on the narrow end to plug into a slot connected to the main circuitry of the tuner (hopefully none of the components will be very large -- thus the question I asked you about making the critically coupled RF transformers common to all channels -- we don't want to have any more than two or three of them!) Your'e dreamin..... So, with respect to the channel approach, the next question to ask is if we can use the same two critically coupled RF transformers (as Patrick notes), and *independently* vary several of the other smaller components (e.g., capacitors, resistors, and even inductors) in the two or three tuning stages (if we include the antenna tuner) so as to maintain, from channel to channel in the BCB, reasonably optimal bandwidth and other desirable tuning characteristics? This has al been investigated before, and the conclusions were about as simple as possible by about 1927. Try studying basic L,C, & R theory, and work all this out for yourself. I have. This channel concept has nothing to do with "basics". It is a twist to TRF tuner architecture taking advantage of the fact that AM BCB is done in specific assigned frequencies, just like FM, like TV, like the CB band, etc. It will not be practical for general shortwave listening since that is a huge band (from 1.8 mhz to 30.0 mhz) and amateurs in particular pick their own frequencies (and over time even commercial SW broadcasts move around a lot, for those only interested in listening to the majors like Radio Australia, as I do many evenings on 15.515 mhz. It comes in loud and clear here in Salt Lake City.) Ah well, long ago I gave up backing ideas in which basics meant SFA. Back in the late 20's and early 30's, on MW there was clearly a need for continuous tuning since broadcasts could be anywhere on the band. (And tubes then had poor gain, among other problems.) The no 22 had plenty of gain, gm was around 1 mA/v at least, and plenty by 1930. Continous tuning kept radios affordable. Implementing your scheme, whatever it may be, would have never caught on in 1935. Today, a lot of the issues of building TRF circuitry is trying to overcome the limitations of one-dimensional tuning using, for example, a multigang air capacitor -- John Byrns is going through agony trying to find the magic formula to get what he wants with a multigang air capacitor. But with the channel TRF concept, the sky's the limit as to how many components in the bandpass tuning filter can be independently selected and hardwired for any given frequency. So one can optimally tune the bandpass characteristics for each and every frequency in the TRF without worrying how that affects other frequencies, since each channel frequency tuning circuit is now effectively decoupled (made independent) from the other channel frequencies. You just need 120 optimised sets of tuning circuits... An electronic 120 position switch should be a doddle. You won't get anyone to finance your endeavour, or pay the patent fees. [With traditional continuous tuning, achieved with multiganged air capacitors, we do indeed vary a few capacitors in the tuning circuitry, but because all of them track each other, in reality we only have one degree of freedom, leading to circuit design constraints for continuous "single knob" tuning. Now imagine, for each channel frequency, to *independently* vary the value of several components at the same time -- we now have several degrees of freedom to play with and thereby hope to achieve reasonably constant (as a function of frequency) bandpass characteristics. 1925 TRFs had 3 or 4 separate tuning gangs, each set to a certain numbered position for reception of a given station. Finding stations was exciting. Try studying the history of radio, and you won't need to ask such questions here. With the channel TRF concept, the component values of the bandpass filter (or parts of the filter circuit) are hardwired on the channel plug-in board (and trimmed during calibration), so all the person has to do in listening to the tuner is switch to the channel, and the radio will be in tune to the desired frequency, with the optimal bandpass characteristics for that frequency. (There is likely to be a need for a very fine tuning control, maybe +/- 1 khz, to handle slight drift, both for tuner warmup, and for the inevitable long-term drifting of component values.) I suppose back in 1925 radio stations where in all sorts of weird locations on the dial, and constantly moving around, so hardwiring all the tuning components for a particular frequency, and likewise for other frequencies, was not even an option. That was an age where nations and states on the same continent built railways mainly with different guages. It was a natural for man to fight man, and millions were slaughtered in 20th century wars, and having selectable and agreed radio station Fs wasn't ever going to prevent all that stupidity. Billions were wasted keeping lawyers fabulously wealthy. Many arguments were over radio ideas and patents. But having channels spaced at 9 or 10 kHz hasn't revolutionised receivers. press auto tune on many, and they just go searching for what's there, and lock ono it, and no drift, and no tubes, just rotten fidelity. For example, we can imagine having multiple plugin slots, where we plug into each slot a PCB mini-board specific to a particular frequency. ? You probably understand the channel TRF concept, but did not understand what I wrote the above, so let me restate with an example: I want my channel TRF tuner to tune in 830 khz (WCCO in Minneapolis), so I get the mini-board for that frequency already hardwired with the optimum configuration of the various tuner components, plug it in, and then listen to that frequency whenever I switch to whatever channel slot I placed that mini-board in (I am reminded of how components are plugged into PCs, such as via PCI slots.) It would peave me if I had to buy seperate plug ins for each station, and peave me greatly if i had to find the darn plug in after the dog or child ran off with it, or buy another after treading on one. The plug in wears out. Its ok for plug in coils like in a HRO, for a full band, but not for one station F. Or, I buy the blank mini-board, check the kit-supplied chart for 830 khz, and then solder in a 50 ohm resistor in this spot, a 200pF capacitor in that spot, etc. -- probably will take me all of five minutes. Ye are hopeful; such farnarcling around, such skyborne dreams..... Then calibrate it by tweaking the trimmers. AHHHHHHH..... If I instead want the mini-board to tune 1160 khz (KSL in Salt Lake City), I check the chart, put in a 75 ohm resistor in this spot, a 150pF capacitor in that spot, etc. (whatever values are called for.) Then calibrate it. Plug it in, listen to 1160 khz, knowing that the TRF bandpass tuning circuitry is now optimized for that frequency, and much better optimized than could ever be done with the one dimensional limits of a multigang tuning capacitor. For those who build tube kits, this will border on the trivial. Where is my Smith and Western? I need to put a chronic dreamer out of his misery.... And some hobbyists may find the channel TRF AM tube tuner architecture of real interest, since now they can more easily experiment with new higher-order bandpass filters of various mathematical functions to see how they affect TRF performance. This could lead to a revised mini-board to be issued at some future time based on all this research, and the channel tuner owner can, if they so choose, simply buy or build updated boards for the broadcast stations of interest, and instantly get better performance. It's possible to mix bandpass filters for different stations: a third order Butterworth for 1160 khz, and a fifth order Chebychev for 830 khz. The possibilities are endless. The possibilities will end. The board will contain the few components whose values *independently* change as a function of frequency. They probably will have trimmers for fine calibration of the center frequency and other bandpass filter characteristics. We may need multiple mini-boards for each channel (one for each tuning stage) if necessary for shielding purposes (to prevent oscillation by stage-to-stage interference if that is a problem.) And if higher frequency channel boards require some minor changes in the circuitry configuration, and not just component value changes, that can easily be done, too. In principle, this tuner might even be able to extend a little beyond (on both sides) the 500-1800 khz MW band -- just plugin the right mini-board circuitry for the frequency desired. This idea is totally impractical for 120 different stations, and plug ins get lost or broken, or worn out. I don't believe it is impractical for 120 different stations, for two reasons: 1) Those tube-o-philes who only want to listen to stronger local stations, or to particular distant ones, are likely only to want to have 10-20 stations (with the ability to add more if they want.) One purpose of picking TRF is its legendary high-fidelity audio capability which will appeal to audiophiles -- most won't want to listen to a very weak station 1000 miles away that can only be picked up some evenings. And I believe it is easier to sell tube-o-philes on the Channel TRF concept once it is explained how it maximizes audio performance for each and every broadcast frequency that cannot be done with a continuously tuned TRF. 2) Those who would use this for casual DXing (and note the hardcore MW DXers will use something like a Drake R8B or ICOM R75, or some digital receiver) will certainly be motivated to add more mini-boards, and can do so over time. The tuner will work with 1 channel board, or with all 130+ (if enough slots are provided. For the moment I am imagining the mini-board approach, but the sky's the limit for other ideas to implement the channel TRF AM tube tuner.) 3) And as noted above, hobbyists may find the "plugin" bandpass filter capability of particular interest. I leave answering points 1 thru 3 for others more patient than myself... Of course, others here will probably have much better ideas as to how to implement the channel approach. You bet there are, and only possible with chip technology, with press button station selection, and digital station F read out, with digitally generated oscillator frequency for the F converter of a superhet, with ceramic filter IF. Grundig have been multiband radios for about 20 years +. I have a Radio Shack DX-399 (the Sangean 606A) which is a very good performer for casual MW (with the Radio Shack MW loop) and shortwave DXing. So I am very familiar with that hobby, and with the benefits digital systems bring to tuners. You need not sell me on that! See my previous note above on "why tubes then?" Not a tube in sight inh these lightweight plastic radios bought cheaply by the masses to allow connection to the world's AM, FM, and HF bands, and even amateur SSB stations. See my previous note above on "why tubes then?" (It's interesting to think of doing the same "channel" approach for an FM tube tuner. Will that also confer several advantages in simplifying the circuit design for the same overall performance level?) Study the way most post 1980 AM/FM tuners are constructed. Tubes cannot be used with such methods. O.k. But are you referring to tube-based tuners? Again, if all I wanted was an audiophile grade AM/FM tuner, and did not care about what was under the hood, I'd be open to solid state designs, but I'm specifically looking at tube-based tuners. The tubes are nice to use, but any scheme of discrete gain devices limits your own channel approach, which is so far free of any details, and probably impossible as it is impractical unless you care to prove otherwise with a fully made prototype. I still assume that the channel approach to tube-based FM tuner design may confer some benefits, but maybe less since the frequency ratio to tune from the lower to the upper ends of the band (about 1.25) is much less than that for the AM BCB (a whopping 3.5 or so.) And there are probably other factors as well specific to frequency modulation. true. It's trying to tune the AM band with only one degree of freedom (e.g., air tuning capacitor) which is causing all the hassle in tube-based TRF AM tuner circuit design. No, its not just the tuning cap. Its the cost and effectiveness, low drift, and serviceablity and selectivity of the superhet which makes the TRF look like a dinasoar. One would want to pick a bandpass filter which is optimally tuned to the specific frequency we want to listen to, and this involves optimally selecting *several* component values, not just one as we are limited to by continuous tuning with a multigang tuning capacitor. The channel TRF approach appears to free up the TRF designer from the tyranny of having to compromise the bandpass characteristics over the entire tuning range which only one degree of freedom allows. Of course, superheterodyne is one solution to the TRF problem, and allows for continuous tuning. Note that super-het works because it uses "one channel" (the IF). So in a sense, superheterodyne supports the channel TRF approach for those who don't want to build a super-het, but rather want a pure TRF receiver (e.g., for sound quality reasons, or whatever.) Sound quality don't have to suffer with frequency conversion. This truth knocks the life out of TRF fanatics. ***** Now, I've made the call several times for classic and proven AM tube tuner designs of the past which have excellent audio quality (and wide bandwidth capability), are good for casual DX use, and can easily be "modernized" for a kit. There are no doubt many excellent super-het designs out there, but I've had very few recommendations. Patrick, since you appear to much prefer super-het over TRF for AM tube tuners, which classic super-het tube AM radio designs of the past would you suggest as candidates to consider? Read all my other recent AM radio posts again and you will see my preferances repeated. Anybody? I reckon you got a pile of reading to do. Yes, I have been reading. That one-year equivalent of EE training back in 1974 at the University of Minnesota is slowly coming back to me. Back then we spent a few weeks on tubes, and only a couple days with transistors. Things have changed a lot since then. I doubt they spend more than a single sentence on receiver tubes in courses today, and all revolves around chips, in which the inner workings are never to be fully understood, and only the uses are known. And it was interesting reading about Chebychev bandpass filters today since I wrote a lot of Fortran code years ago to do various types of numerical analytic processing including integration using quadrature with orthogonal polynomials (mostly Legendre polynomials.) It was especially cool to see how the higher order Chebychev polynomials U(x) plot out in the desired shape (well approximately) for a bandpass filter (but with that slight ripple within the bandwidth.) I'm not saying all this to brag, but to give a better idea of my background. Definitely I have a lot to learn, of course, and your posts are helping me to better understand things. I still believe the channel TRF concept is viable for those who want to build the best possible TRF tube tuners where for each frequency the absolute best bandpass characteristics can be chosen without worrying about how it impacts the other frequencies since each channel is now largely independent. The obvious downsides with the tube-based channel TRF concept a 1) The practical, real-world implementation of it (I believe it is doable, I suggest one approach), 2) Losing the ability to continuously tune, which for BCB is not an issue as I've noted several times, and 3) Calibration of each mini-board if done by the kit-builder (I think this is solvable, but it is an issue to consider.) The upsides are several, as previously noted. Thanks for your helpful comments! Jon Noring I'll delegate you to chief honary prototype developer, and let you spend the next 20 years building something for AM that nobody else has. During my wait, I'll live a bit, then I'll die. Patrick Turner. |
#20
|
|||
|
|||
Snip, Going back to some of the earlier filter flatness discussion, well toss that idea into the mix when you think in terms of TRF. Not only do you want to achieve a specific width but you want it to be flat. My 3/25 kc TRF scenario isn't flat at all. Its a big peak that just broadens out. When we say a 'bandwidth'number we are relating to something specific like 3 or 6 db down from the peak. Its still a peak in this context. So whats happening at 20 db down? You guessed it, that 25kc number is 150 kc wide. I dunno how you could control the width AND the flatness AND the skirts. I'm a fairly recent convert to crystal radios. For the sake of discussion there's little difference in xtal technology vs trf technology in that both are non-superhet. I get absolutely glorious quality audio from my xtal set when fed thru an amp. With 6 or 8 knobs on the front panel and top notch components I can find a dead spot between semi-locals on 680 and 690. With a local station on 1370 it takes traps and VERY hi-q stuff to ferret out semi-locals on 1240, 1290 and 1480. Its as if it were a totally different radio from one end of the band to the other and this has been the plague of TRF circuits since day one. If I didn't have the local 1370 I could safely say, hey Jon, this is the ticket, but there's scarce few of us who don't have a strong undesired local station to bollox up the works. Go superhet, my man. -Bill M The only way to gain enough RF bw at any F on the BCB to ensure there is no sideband cutting which would restrict the AF bw, you have to use two LC circuits and couple one to the other, and I used a 39k resistor. At the low end of the BCB, the coils are tuned about 10kHz apart, and at the top end, they are tuned to the same F. If you only have two LC circuits, and the bw is 25 kHz for each at 1,500 kHz, then the Q is 60 only . Using two LCs with a Q like that in cascade, the bw will be reduced to 19 kHz, or thereabouts, and the selectivity away from the pass band will be twice that of a single circuit. But another powerful station at 50 kHz away will be heard, although it won't be loud. Once you are 50 kHz away from say 1,000 kHz, the rate of attenuation is at 6 dB /octave only for the one tuned circuit. So a station at 500 kHz of equal strength is only -6 dB below the 1,000 kHz station. You need multiple tuned circuits to give decent selectivity, and here the superhet is king. But it is possible to series three double LC twin gang stagger tuned stages. This gives 6 tuned circuits. The final Q has to be 60 to allow full audio bw, The initial Q therefore has to be much much lower, maybe 15 only at 1,500 kHz, allowing 100 kHz of bw. Such a tuned circuit has a blunt nose, and no advantage can be had as with with the double tuned IF transformer's flat topped steep sided bandpass characteristic The only reason for RF input selectivity with relatively low Q tuned front ends in AM BCB sets is to make sure the mixer does not get overloaded by too much signal from a poweful unwanted station which would then try to cross modulate the mixer tube. The purpose of the RF front end is spelled out in RDH4. For those struggling on Q issues, go find out, I'm sick of repeating text books. Patrick Turner. |
#21
|
|||
|
|||
The fixed-step (10 kHz) tuner is what killed music on AM radio. You
can get a much more pleasant sound by detuning a few 100 Hz. This PLL crap sounds like ****. On Sun, 13 Jun 2004 00:36:21 GMT, Jon Noring wrote: [Following up on a thread dating back to January, similar to one I started recently. Responding to Patrick Turner's comments.] Patrick Turner wrote in January 2004: Jerry Wang wrote: 1. Even it is a single channel [AM] receiver, I would still suggest the use of one or two intermediate frequency (IF) stages. Because to achieve good sensitivity you need to have enough gain. Since you only want one channel, there is no need for a frequency converter or any IFTs or IF amps, and a TRF with four tuned circuits in the form of two critically coupled RF trannies will do nicely. Interesting. As I noted in a recent message, it is very intriguing to build a modernized, high-performance AM tube tuner using the "channel" approach. This takes advantage of the fact that licensed broadcasters today must broadcast on specific frequencies, every 10 khz in North America and 9 khz in Europe and elsewhere. So, instead of trying to be able to continuously tune across the BCB spectrum, we can think outside the box for the moment and consider the alternative of building reasonably optimized tuning circuits for each listened-to frequency. |
#22
|
|||
|
|||
Hi Fi AM--
This was "almost" popular as a design project in the 60s. Even RCA in their RC-19 Tube Manual have a circuit labelled "TRF AM Tuner-- for High- Fidelity Local Broadcast Reception." (Circuit 19-8, p. 357) Fidelity on an AM signal requires that most common circuits used in radios be eliminated: 1: No AVC. This distorts the low frequency frequency response 2: No cathode bias bypass. 3: No diode detectors, unless the signal feeding them is greater than 10Vrms. 4: No AC coupling if diode detector is used (the "AC-loading" distortion described in Terman, et al). 5: Speaker resonance 30Hz. Assumes the line out goes to a real "Hi Fi" system. All these "don't do" can be found in Terman, the Radiotron Designer's Handbook and others. If you are willing to live with about 5-10% THD, then you can use more common circuits. However, there are dozens of "Hi Fi" AM circuits published by the hobby magazines, tube vendors and kit makers. Have a look at them. The RC-19 circuit uses a 6BA6 as an RF amp, followed by a 12AU7 used as a detector and audio amplifier. Good luck. Steve. -- Steven D. Swift, , http://www.novatech-instr.com NOVATECH INSTRUMENTS, INC. P.O. Box 55997 206.301.8986, fax 206.363.4367 Seattle, Washington 98155 USA |
#23
|
|||
|
|||
David wrote: The fixed-step (10 kHz) tuner is what killed music on AM radio. You can get a much more pleasant sound by detuning a few 100 Hz. This PLL crap sounds like ****. Howcome? Patrick Turner On Sun, 13 Jun 2004 00:36:21 GMT, Jon Noring wrote: [Following up on a thread dating back to January, similar to one I started recently. Responding to Patrick Turner's comments.] Patrick Turner wrote in January 2004: Jerry Wang wrote: 1. Even it is a single channel [AM] receiver, I would still suggest the use of one or two intermediate frequency (IF) stages. Because to achieve good sensitivity you need to have enough gain. Since you only want one channel, there is no need for a frequency converter or any IFTs or IF amps, and a TRF with four tuned circuits in the form of two critically coupled RF trannies will do nicely. Interesting. As I noted in a recent message, it is very intriguing to build a modernized, high-performance AM tube tuner using the "channel" approach. This takes advantage of the fact that licensed broadcasters today must broadcast on specific frequencies, every 10 khz in North America and 9 khz in Europe and elsewhere. So, instead of trying to be able to continuously tune across the BCB spectrum, we can think outside the box for the moment and consider the alternative of building reasonably optimized tuning circuits for each listened-to frequency. |
#24
|
|||
|
|||
Steven Swift wrote: Hi Fi AM-- This was "almost" popular as a design project in the 60s. Even RCA in their RC-19 Tube Manual have a circuit labelled "TRF AM Tuner-- for High- Fidelity Local Broadcast Reception." (Circuit 19-8, p. 357) Fidelity on an AM signal requires that most common circuits used in radios be eliminated: 1: No AVC. This distorts the low frequency frequency response 2: No cathode bias bypass. 3: No diode detectors, unless the signal feeding them is greater than 10Vrms. 4: No AC coupling if diode detector is used (the "AC-loading" distortion described in Terman, et al). 5: Speaker resonance 30Hz. Assumes the line out goes to a real "Hi Fi" system. All these "don't do" can be found in Terman, the Radiotron Designer's Handbook and others. Let me say a few words. 1. Many pages of RDH4 are devoted to AVC. The time constant for AVC application is very long, comprising of 1M and 0.047 uF, and measurement of bass distortions resulting from well applied AVC is low enough to be negligible. 2. Nothing wrong with cathode bias, especially nowdays when cheap large value elcaps are plentiful, and we have better plastic caps. RDH4 speds a lot of time on cathode bias. 3. Diode detectors are quite low distortion detectors even with very low voltages of 100 mV if there is a constant current trickeled through the crystal diode to keep them turned on with their forward conducting voltage. I gave details yesterday in another post of a detector which will change your views about diode detectors. Diodes can be used with DC shunt feedback around an RF opamp, and thd is negligible. 4. AC coupling is fine from an RC load fed by a diode. The impedance fed by the audio + RF ripple voltage should be high, like a cathode follower grid. 5. I have tried my radio with various speakers, and no trouble making full amplitude signals at 20 Hz. The LF pole is determined by the audio amp in the radio, but at the detector, the pole is much lower. On which pages are RDH4 and Terman "dont's" spelled out? If you are willing to live with about 5-10% THD, then you can use more common circuits. True, but onje doesn't have to live with 5-10%. linearize the IF amp and detector, and thd plummets. However, there are dozens of "Hi Fi" AM circuits published by the hobby magazines, tube vendors and kit makers. Have a look at them. The RC-19 circuit uses a 6BA6 as an RF amp, followed by a 12AU7 used as a detector and audio amplifier. RDH4 has the circuit for the Selsted and Smith "infinite impedance " detector, where a 12AU7 performs as credible detector, and as a diode, but I think I'll stick with a germanium diode fet by a 12AU7 CF. But does the RC-19 have enough tuned circuits to give over 70 dB rejection of signals which are 50 kHz away from the wanted station at any place on the band? The 6BA6 is a variable U tube, with a non linear transfer curve. If a large voltage is detected, there is quite a bit of distortion of the detected wave form. Preferable is a 6AU6, a sharp cut off RF amp, with a more linear transfer curve, although too much gain could be a problem, if so, use a lower Gm pentode like a 6J7. But having a 6AU6, or perhaps a 6BX6 is OK if biased with a cathode resistance, and this R is left unbypassed to further reduce thd in the envelope amplification. With a load on the R amp typically of 25k, and Gm = 3 mA/V, gain is 75, and if the Rk = 500 ohms, gain is 30, and thd reduced by 6 dB. In a superhet, the voltage levels of the IF envelope coming from a mixer tube are usually so low that the mixer don't maul the linearity of the IF signal, even when the amplification is backed off with AVC. Patrick Turner. Good luck. Steve. -- Steven D. Swift, , http://www.novatech-instr.com NOVATECH INSTRUMENTS, INC. P.O. Box 55997 206.301.8986, fax 206.363.4367 Seattle, Washington 98155 USA |
#25
|
|||
|
|||
Steven Swift wrote:
Hi Fi AM-- This was "almost" popular as a design project in the 60s. Even RCA in their RC-19 Tube Manual have a circuit labelled "TRF AM Tuner-- for High- Fidelity Local Broadcast Reception." (Circuit 19-8, p. 357) John Byrns web site has the circuit diagram for the RCA design you mention (or a related one if there's more than one of them): http://users.rcn.com/jbyrns/pics/RC-17-8.jpg Fidelity on an AM signal requires that most common circuits used in radios be eliminated: [snip of good list] All these "don't do" can be found in Terman, the Radiotron Designer's Handbook and others. [snip] The RC-19 circuit uses a 6BA6 as an RF amp, followed by a 12AU7 used as a detector and audio amplifier. As the diagram at John Byrns site shows. What's intriguing is how simple this design is -- it has one RF stage, which indicates that a one RF stage TRF for local, high power stations makes sense when audio fidelity is the overriding criterion. Now, I wonder how much improvement in the audio quality is possible if the channel TRF approach is used (which optimizes the bandpass for each broadcast frequency)? Or does it not make sense by the law of diminishing returns? Being able to use a higher order plug-in bandpass filter (such as a constant delay/linear phase one), optimized for each frequency, is intriguing. Jon Noring |
#26
|
|||
|
|||
David wrote:
The fixed-step (10 kHz) tuner is what killed music on AM radio. You can get a much more pleasant sound by detuning a few 100 Hz. This PLL crap sounds like ****. Assuming that indeed a more pleasant sound is had by detuning off the center frequency by a few hundred Hz (Patrick was skeptical), this is not a problem for the Channel TRF tube tuner design since I believe a very fine tuning control will be necessary, due to both tuner warmup, and long-term drift as bandpass component values slowly change over time between "calibrations". Only guessing, I think the control should vary the center frequency for a channel by about +/- 1 khz, enough to cover the several hundred Hz of deviation for those who think this makes the station sound better. Jon Noring |
#27
|
|||
|
|||
Nothing wrong with a channel approach. Simplifies most of the design tradeoffs. BTW, I measured all the AM stations at my house (Seattle). Using a typical AM radio (a Sony SW7600GR with internal antenna), I tuned all the stations that were easily received without too much QRN/QRM (day time). I then checked the field strength of each using a spectrum analyzer and a calibrated antenna with 1 meter effective electrical length. I received 15 (reasonably clear) stations. The strongest station, only 2 miles away, gave -55dBuV (about 1.8mV/m), while the weakest of the 15, gave -85dBuV (about 56uV/m). QRN, due to skywave, increased substantially at night. This is "same channel" interference, which we get from the fact that there are no "clear channels" anymore. Better antennas actually make the QRN worse on some channels-- even while helping local daytime reception. I would design the radio to only work with strong local signals, if fidelity is the goal. Again, I say "good luck." Steve. -- Steven D. Swift, , http://www.novatech-instr.com NOVATECH INSTRUMENTS, INC. P.O. Box 55997 206.301.8986, fax 206.363.4367 Seattle, Washington 98155 USA |
#28
|
|||
|
|||
Patrick Turner writes:
Let me say a few words. 1. Many pages of RDH4 are devoted to AVC. The time constant for AVC application is very long, comprising of 1M and 0.047 uF, and measurement of bass distortions resulting from well applied AVC is low enough to be negligible. To work properly for fading, the AVC needs to be about 100ms. This causes significant distortion at bass audio. If a longer time constant is used since only local (non-fading) stations will be tuned, then you are right. 2. Nothing wrong with cathode bias, especially nowdays when cheap large value elcaps are plentiful, and we have better plastic caps. RDH4 speds a lot of time on cathode bias. Cathode bias is great, but do not bypass the resistor. The degenerative feedback will improve audio performance, but you lose gain. 3. Diode detectors are quite low distortion detectors even with very low voltages of 100 mV if there is a constant current trickeled through the crystal diode to keep them turned on with their forward conducting voltage. I gave details yesterday in another post of a detector which will change your views about diode detectors. Diodes can be used with DC shunt feedback around an RF opamp, and thd is negligible. I agree that this can be made mostly true using active filters and such, but a perfect diode, with perfect modulation has lots of distortion. I am willing to take a look at your analysis, but if you use Volterra series expansion, you simply can't prove that you'll get better than a few percent distortion. Somewhere in my old grad school notes, I have a derivation done by Prof. Meyer (of Gray and Meyer, UC Berkeley) which shows the limits. I'll look for your other post. Better than a few percent is NOT possible with just an RC load (diagonal clipping) except for low modulation percentages. 4. AC coupling is fine from an RC load fed by a diode. The impedance fed by the audio + RF ripple voltage should be high, like a cathode follower grid. 5. I have tried my radio with various speakers, and no trouble making full amplitude signals at 20 Hz. The LF pole is determined by the audio amp in the radio, but at the detector, the There is a discussion of speaker/cabinet resonances in RDH4 somewhere. Lots of distortion when you approach resonance. On which pages are RDH4 and Terman "dont's" spelled out? Each chapter has "crumbs" of knowledge in it. I have yet to find a nice do/don't do list anywhere. I may also be "integrating" Terman and Henney. I think the guy who started the thread should build his idea and try them on a few "beta" testers. He'll be able to sell enough to pay off his costs. Maybe pick a few channels in a few big markets (LA, New York, Chicago) to keep the work load down. Steve. -- Steven D. Swift, , http://www.novatech-instr.com NOVATECH INSTRUMENTS, INC. P.O. Box 55997 206.301.8986, fax 206.363.4367 Seattle, Washington 98155 USA |
#29
|
|||
|
|||
Jon Noring writes:
David wrote: The fixed-step (10 kHz) tuner is what killed music on AM radio. You can get a much more pleasant sound by detuning a few 100 Hz. This PLL crap sounds like ****. Hmm, and I thought it was a phase noise problem. A lot of these radios used a 10kHz reference frequency, multiplied up to the LO, giving lots of close in phase noise. Due to FM/AM conversion, this then showed up in baseband. Jon-- are you the originator of this thread? Steve. -- Steven D. Swift, , http://www.novatech-instr.com NOVATECH INSTRUMENTS, INC. P.O. Box 55997 206.301.8986, fax 206.363.4367 Seattle, Washington 98155 USA |
#30
|
|||
|
|||
In article ,
Patrick Turner wrote: Snip For those struggling on Q issues, go find out, I'm sick of repeating text books. These long winded arguments are not enlightening and I consider them off topic for rec.radio.shortwave. Please delete them from newsgroup header. There are several of these threads now cross posted into the RRS. Thanks in advance for your cooperation. -- Telamon Ventura, California |
#31
|
|||
|
|||
|
#33
|
|||
|
|||
Telamon wrote:
Please don't cross post to rec.radio.shortwave. Perhaps you would be happier if you just learned to use your delete key instead of expecting everyone else to conform to your demands. Jeff -- "They that can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety." Benjamin Franklin "A life lived in fear is a life half lived." Tara Morice as Fran, from the movie "Strictly Ballroom" |
#34
|
|||
|
|||
In article ,
Jeffrey D Angus wrote: Telamon wrote: Please don't cross post to rec.radio.shortwave. Perhaps you would be happier if you just learned to use your delete key instead of expecting everyone else to conform to your demands. Oh I would be much happier if people did not cross post. Thanks for your consideration. -- Telamon Ventura, California |
#35
|
|||
|
|||
In article , Patrick Turner
wrote: Steven Swift wrote: If you are willing to live with about 5-10% THD, then you can use more common circuits. True, but onje doesn't have to live with 5-10%. linearize the IF amp and detector, and thd plummets. However, there are dozens of "Hi Fi" AM circuits published by the hobby magazines, tube vendors and kit makers. Have a look at them. The RC-19 circuit uses a 6BA6 as an RF amp, followed by a 12AU7 used as a detector and audio amplifier. RDH4 has the circuit for the Selsted and Smith "infinite impedance " detector, where a 12AU7 performs as credible detector, and as a diode, but I think I'll stick with a germanium diode fet by a 12AU7 CF. The "Selsted and Smith" detector is not the same thing as the so called "infinite impedance" detector. The "infinite impedance" or "reflex" detector was designed by RCA, while the "Selsted and Smith" detector was designed by, well "Selsted and Smith", or at least "Selsted" who is still around, or was a year or two ago. The "Selsted and Smith" detector differs from the "infinite impedance" detector in that it has a diode in series with the grid, and also a diode load resistor. There is no peak detection capacitor across the diode load, so the diode does not act as an ordinary diode peak detector, and the triode doesn't act as a cathode follower. The triode is the actual detector operating in a fashion similar to the "infinite impedance" detector, with the diode apparently serving to linearize the "infinite impedance" detector. The input impedance of the "Selsted and Smith" detector is not infinite due to the presence of the diode load resistor. For that matter the input impedance of the so called "infinite impedance" detector is also not infinite, and can even have a negative resistance component which can cause stability problems. The negative resistance effect can occur when circuit conditions are right, similar to the conditions that can cause oscillation in cathode and emitter follower circuits if you aren't careful. But does the RC-19 have enough tuned circuits to give over 70 dB rejection of signals which are 50 kHz away from the wanted station at any place on the band? The 6BA6 is a variable U tube, with a non linear transfer curve. There is nothing wrong with the 6BA6, it was specifically designed for this service and has very low odd order distortion which is all that matters since the even order distortion products can't get through the IFT. I hope I got that the right way around, if not it is explained in some detail in some of the old texts, I think "Radio Receiver Design" by Sturley is one that explains it. You only get in trouble if you try to run the tube at a very high signal level, simultaneously with a high AGC voltage applied for a large gain reduction. This is mainly a problem in the stage driving the diode detector, so it is best to avoid AGC on that stage, but in a minimal radio that is of course problematic. This is one of the many topics that the RDH4 gives short shrift. The 6BA6 is even usable as a gain control element in audio circuits where even order distortion does matter. IIRC the peak limiter at a radio station where I once worked used four 6BA6s in the audio path, where they were connected in push pull, presumably to cancel the even order nonlinearities which are inherent in the design of the tube. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
#36
|
|||
|
|||
In article , wrote:
Steven Swift wrote: Hi Fi AM-- This was "almost" popular as a design project in the 60s. Even RCA in their RC-19 Tube Manual have a circuit labelled "TRF AM Tuner-- for High- Fidelity Local Broadcast Reception." (Circuit 19-8, p. 357) John Byrns web site has the circuit diagram for the RCA design you mention (or a related one if there's more than one of them): The circuit is the same, with one exception, in all the editions of the RCA tube manual in which it was contained. The one change that occurred in later editions was a change from a 5Y3GT rectifier in the power supply, to a rectifier with an indirectly heated cathode, I think it was a 6X4. This circuit was only published during the 1950's, it did not appear in editions before about 1950, and was dropped from later editions after about 1960. I have often wondered if it was inspired by Williamson's late 1940's design for a radio feeder unit. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
#37
|
|||
|
|||
In article , (Steven
Swift) wrote: Patrick Turner writes: 3. Diode detectors are quite low distortion detectors even with very low voltages of 100 mV if there is a constant current trickeled through the crystal diode to keep them turned on with their forward conducting voltage. I gave details yesterday in another post of a detector which will change your views about diode detectors. Diodes can be used with DC shunt feedback around an RF opamp, and thd is negligible. I agree that this can be made mostly true using active filters and such, but a perfect diode, with perfect modulation has lots of distortion. I am willing to take a look at your analysis, but if you use Volterra series expansion, you simply can't prove that you'll get better than a few percent distortion. Somewhere in my old grad school notes, I have a derivation done by Prof. Meyer (of Gray and Meyer, UC Berkeley) which shows the limits. I'll look for your other post. Better than a few percent is NOT possible with just an RC load (diagonal clipping) except for low modulation percentages. It would be interesting to see the derivation you speak of! It was my impression that if we had a "perfect diode" it could be used make a perfect envelope detector, with the exception of the "tangential clipping" problem that you mentioned. "Tangential clipping" is not just a function of the modulation level, but is also a function of the modulating frequency. As Patrick mentioned using a higher IF frequency will allow using a smaller peak hold capacitor which will reduce "tangential clipping". Also doubling the IF frequency by using a full wave detector will reduce the "tangential clipping". It is hard to believe that the distortion of a reasonably designed diode detector is anywhere near "a few percent", as simple diode detectors were used in the modulation monitors used by AM broadcast stations in times gone by, and they had to have distortion low enough to measure the system distortion at modulation percentages up to 100%. I will have to look up the specifications of a few. Of course there is the issue that negative peak clipping that is visible on a scope may represent only a small fraction of a percent distortion on an RMS basis. Regards, John Byrns Surf my web pages at, http://users.rcn.com/jbyrns/ |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
on topic: we need a rec.audio.pro.ot newsgroup! | Pro Audio | |||
Artists cut out the record biz | Pro Audio |