Questioning Basics: FM Stereo Decoding Schemes
- Thread starter Trnsfmr
- Start date
KentTeffeteller
Gimpus Stereophilus!
A great thread on FM Stereo decoding and MPX, explains it well for those less technical. Agree, Eico had great designs, and often better grade parts where cheap was used, improves these classics many times over into something which amazes one in performance.
Tubologic
Well-Known Member
Thanks to the O.P for this very interesting thread. I feel some small corrections/additions are in order:
Regarding the circuit used in the EICO ST-97 decoder (and his stand-alone brother, the MX-99) it must be noted that it neither used a matrix nor a switching type, but a clever and unusual combination of both. A detailed description of this circuit can be found in "FM Multiplexing for Stereo" by Leonald Feldman (SAMS FMS-1, 1st ed, 1962, p.135), a book I would highly recommend to anyone interested by this topic. (If you're into tube circuits try to get the 1st ed.)
Also, in a historical perspective the switching (aka "time-division") mod/demod method was not pioneered by The FISHER and was already known (and used) in the 1950ies by the military for data/telemetry transmissions,etc... its first commercial use in a stereo decoder can be found in the HH SCOTT 335, the very 1st MPX decoder available on the market and designed by SCOTT's chief engineer Daniel Von Recklingausen (incidentally also member of the FCC stereo multiplex comitee in charge of the evaluation of the various proposed encoding methods). SCOTT heavily promoted his time-switchting method in their ads as being superior to other techniques, and it proved to be the case. (most high quality decoders used and still use this method)
And regarding the supposed "detrimental effect" of the signal being sampled at 38 KHz by the decoder, it has been already sampled (or "switched") at the encoding (transmitting) side : supressed subcarrier matrixing or (time-multiplex)switching are just two different mathematical ways to describe the same process but the transmitted composite signal remains the same. And whatever type of decoder you're using, in broadcasted FM the upper frequency limit of the recovered audio signal is about 15 KHz (slightly less than half the sampling frequency).
Regarding the circuit used in the EICO ST-97 decoder (and his stand-alone brother, the MX-99) it must be noted that it neither used a matrix nor a switching type, but a clever and unusual combination of both. A detailed description of this circuit can be found in "FM Multiplexing for Stereo" by Leonald Feldman (SAMS FMS-1, 1st ed, 1962, p.135), a book I would highly recommend to anyone interested by this topic. (If you're into tube circuits try to get the 1st ed.)
Also, in a historical perspective the switching (aka "time-division") mod/demod method was not pioneered by The FISHER and was already known (and used) in the 1950ies by the military for data/telemetry transmissions,etc... its first commercial use in a stereo decoder can be found in the HH SCOTT 335, the very 1st MPX decoder available on the market and designed by SCOTT's chief engineer Daniel Von Recklingausen (incidentally also member of the FCC stereo multiplex comitee in charge of the evaluation of the various proposed encoding methods). SCOTT heavily promoted his time-switchting method in their ads as being superior to other techniques, and it proved to be the case. (most high quality decoders used and still use this method)
And regarding the supposed "detrimental effect" of the signal being sampled at 38 KHz by the decoder, it has been already sampled (or "switched") at the encoding (transmitting) side : supressed subcarrier matrixing or (time-multiplex)switching are just two different mathematical ways to describe the same process but the transmitted composite signal remains the same. And whatever type of decoder you're using, in broadcasted FM the upper frequency limit of the recovered audio signal is about 15 KHz (slightly less than half the sampling frequency).
Last edited:
Trnsfmr
Well-Known Member
Hi Tubologic (great moniker by the way),
Sounds like a great book, I ordered one from Amazon. thanks. The ST97 came out in '63 after the book.
I am certainly willing to be corrected about Fisher inventing the "chopper", or sampling method of decoding, as I could find very little data about it. But, as discussed above in other postings, it seems the Scott 340 receiver along with the 350 series tuners used matrix decoding? at least that is what was presented earlier in the thread by another poster.
There is no ring demodulator in the Scott, but two bridge type rectifiers perhaps achieving the same thing?
I have looked at the Scott schematic and since there is no accompanying block diagram or explanation of how the circuit works, it is difficult to analyze. I had looked at it before (350 tuner), and had determined that it was a switcher, but I wasn't certain.
Somebody else above in the thread said he thought the ST97 was a combination of switcher and matrix. I don't understand how you can come to that conclusion. Maybe you could explain that to me? I can follow each part of the circuit and understand (I think) exactly how each step works. I color coded the sections for that on the schematic.
Briefly, the green highlight is the composite signal in and buffered to the 38khz circuit and to the matrix combining circuit. The 38khz, the yellow highlight, is applied to the ring demodulator with the composite signal applied on the other two corners, done through the matched resistors, so that the resulting signal is L-R, the pink section. Because the 38khz is a balanced signal at this point, summing the two sides into the L-R amp, nulls out any residual 38khz signal (which could contain 38khz sampled components) leaving just the L-R. It is amplified and summed with the L+R composite signal in the blue sections which is a -(L+R) on the plate side and a +(L+R) on the cathode side of the input buffer. The resulting -L is then inverted to get the +L signal in the inverter, which is buffered with a cathode follower and drives the output, as is the R signal.
I don't see where there is any switching/sampling going on. The ring demodulator is normally where the switching occurs, but it is not being used in that way in this circuit. It is being used to produce the L-R signal. Any switching/sampled products are nulled out in summing the two signals going to the L-R amp.
The ring demodulator is being used in the same way a SSB signal would be demodulated. The missing carrier is re-inserted and then the AM detector (ring demodulator) turns it into audio. In this case the audio is the L-R signal.
As far as encoding goes, I have not studied various encoders, but I would be surprised if there were not a multitude of ways for it to be achieved. It is my understanding that the standard way to produce a suppressed carrier signal is with a balanced modulator (analog techniques). But in today's modern world, I would expect the most recent encoders to be building the signal with computer, and I would be very surprised if any sampling to build that signal was done at 38khz. I, of course am open to learning about that.
Feldman and Recklinghausen to the contrary, I do not subscribe to absolute statements from brilliant engineers from 50+ years ago. I've been designing in pro audio for 30 years, and over and over again, what was known and accepted by the best and brightest is found to be lacking in some way and can then be improved upon. It is how the state of the art constantly improves.
You make exactly my point about the 38khz sampling. Yes, 38khz sampling will provide a 15khz frequency response, or there abouts. But just as in amplifiers, it is found the best sound is produced when the actual frequency response goes way past the limits of hearing. For example an amp with extended response out to 50-60khz will make the most transparent audio. An amp with an upper cutoff at 15-20 khz will not sound near as good.
Likewise, in the world of sampling it has been found that the same effect applies. To get truly transparent audio, pro audio professionals have settled upon 96khz sampling which gets the frequency response up into the same range. The standard here is to compare the resulting sampled signal with the original analog signal. And guess what? Regardless of how the signal was encoded on the transmitting end it is basically an analog signal that is presented to the receiver as the L+R (frequency modulated) and the L-R (amplitude modulated) signals.
My point is Analog is better than sampling. Sampling is done in order to do digital recording or to transfer data over digital networks. Transparent sampling occurs at 96khz and above. 38khz is not going to sound as good as either 96khz or the original analog signal.
Saying that mathematically, two different circuits or processes are the same, to me, is simply non sequitur. It has been proved in audio, over and over, that what matters is what sounds better. Audio professionals and music listeners don't care about the math. If you are into tube gear, it would seem that you would be on board with this concept.
This all falls into the realm of advanced technology that engineers prefer, not necessarily sounding as good as simpler techniques. Unfortunately, most design engineers can't hear the difference between high quality audio and mediocre, and will state over and over again that there is no hearable difference, simply because that's what their engineering background says..
To take a processed, encoded signal, and then process it again with sampling, to me is ludicrous, especially when it is basically analog encoding to begin with that can be decoded with analog circuitry.
My opinion on the original chopper circuit was that it worked better than anything else at the time, and especially gave good specs. It also was easy to manufacture, as the only way to get good specs with matrix decoding was to hand select component values, which is a complete no no in manufacturing. The goal, always, especially with consumer items where price is the ultimate competition, is to produce the unit with as little skilled labor as possible.
Eico probably could have had the best sounding tuner on the market at the time, but their use of substandard (how about "horrid") caps throughout their gear, probably nixed their opportunity to be the best.
Sounds like a great book, I ordered one from Amazon. thanks. The ST97 came out in '63 after the book.
I am certainly willing to be corrected about Fisher inventing the "chopper", or sampling method of decoding, as I could find very little data about it. But, as discussed above in other postings, it seems the Scott 340 receiver along with the 350 series tuners used matrix decoding? at least that is what was presented earlier in the thread by another poster.
There is no ring demodulator in the Scott, but two bridge type rectifiers perhaps achieving the same thing?
I have looked at the Scott schematic and since there is no accompanying block diagram or explanation of how the circuit works, it is difficult to analyze. I had looked at it before (350 tuner), and had determined that it was a switcher, but I wasn't certain.
Somebody else above in the thread said he thought the ST97 was a combination of switcher and matrix. I don't understand how you can come to that conclusion. Maybe you could explain that to me? I can follow each part of the circuit and understand (I think) exactly how each step works. I color coded the sections for that on the schematic.
Briefly, the green highlight is the composite signal in and buffered to the 38khz circuit and to the matrix combining circuit. The 38khz, the yellow highlight, is applied to the ring demodulator with the composite signal applied on the other two corners, done through the matched resistors, so that the resulting signal is L-R, the pink section. Because the 38khz is a balanced signal at this point, summing the two sides into the L-R amp, nulls out any residual 38khz signal (which could contain 38khz sampled components) leaving just the L-R. It is amplified and summed with the L+R composite signal in the blue sections which is a -(L+R) on the plate side and a +(L+R) on the cathode side of the input buffer. The resulting -L is then inverted to get the +L signal in the inverter, which is buffered with a cathode follower and drives the output, as is the R signal.
I don't see where there is any switching/sampling going on. The ring demodulator is normally where the switching occurs, but it is not being used in that way in this circuit. It is being used to produce the L-R signal. Any switching/sampled products are nulled out in summing the two signals going to the L-R amp.
The ring demodulator is being used in the same way a SSB signal would be demodulated. The missing carrier is re-inserted and then the AM detector (ring demodulator) turns it into audio. In this case the audio is the L-R signal.
As far as encoding goes, I have not studied various encoders, but I would be surprised if there were not a multitude of ways for it to be achieved. It is my understanding that the standard way to produce a suppressed carrier signal is with a balanced modulator (analog techniques). But in today's modern world, I would expect the most recent encoders to be building the signal with computer, and I would be very surprised if any sampling to build that signal was done at 38khz. I, of course am open to learning about that.
Feldman and Recklinghausen to the contrary, I do not subscribe to absolute statements from brilliant engineers from 50+ years ago. I've been designing in pro audio for 30 years, and over and over again, what was known and accepted by the best and brightest is found to be lacking in some way and can then be improved upon. It is how the state of the art constantly improves.
You make exactly my point about the 38khz sampling. Yes, 38khz sampling will provide a 15khz frequency response, or there abouts. But just as in amplifiers, it is found the best sound is produced when the actual frequency response goes way past the limits of hearing. For example an amp with extended response out to 50-60khz will make the most transparent audio. An amp with an upper cutoff at 15-20 khz will not sound near as good.
Likewise, in the world of sampling it has been found that the same effect applies. To get truly transparent audio, pro audio professionals have settled upon 96khz sampling which gets the frequency response up into the same range. The standard here is to compare the resulting sampled signal with the original analog signal. And guess what? Regardless of how the signal was encoded on the transmitting end it is basically an analog signal that is presented to the receiver as the L+R (frequency modulated) and the L-R (amplitude modulated) signals.
My point is Analog is better than sampling. Sampling is done in order to do digital recording or to transfer data over digital networks. Transparent sampling occurs at 96khz and above. 38khz is not going to sound as good as either 96khz or the original analog signal.
Saying that mathematically, two different circuits or processes are the same, to me, is simply non sequitur. It has been proved in audio, over and over, that what matters is what sounds better. Audio professionals and music listeners don't care about the math. If you are into tube gear, it would seem that you would be on board with this concept.
This all falls into the realm of advanced technology that engineers prefer, not necessarily sounding as good as simpler techniques. Unfortunately, most design engineers can't hear the difference between high quality audio and mediocre, and will state over and over again that there is no hearable difference, simply because that's what their engineering background says..
To take a processed, encoded signal, and then process it again with sampling, to me is ludicrous, especially when it is basically analog encoding to begin with that can be decoded with analog circuitry.
My opinion on the original chopper circuit was that it worked better than anything else at the time, and especially gave good specs. It also was easy to manufacture, as the only way to get good specs with matrix decoding was to hand select component values, which is a complete no no in manufacturing. The goal, always, especially with consumer items where price is the ultimate competition, is to produce the unit with as little skilled labor as possible.
Eico probably could have had the best sounding tuner on the market at the time, but their use of substandard (how about "horrid") caps throughout their gear, probably nixed their opportunity to be the best.
Last edited:
Jim; of all of the FISHER gear I have in house (9 consoles before 1965 and like 5 late 60's Solid State and a 400 and 2 800C's), the only one I have that I can identify with a chopper in it is the 700-T from 1967. The 550-T (from the same year) has a mpx board with 3 IF can's on it with no chopper that I can identify. The 700-T's also has 3 on a board that is better than 1/2 smaller. IIRC all of the console MPX, the 400 and the 800c were all matrix decoders. The model #'s for the decoders are mpx-65 and mpx-125 (consoles) with the 65's on the 400 and 800C. As a matter of fact FISHER's chopper went out of manufacture and FISHER sent out a bulletin on how to make one from available parts. (see below). PM me your email and I'll cull out the mpx schematics for the FISHER consoles, receivers, and tuners from the manuals I can get hold of.
Chopper bulletin.
700-T MPX board.
550-T MPX Board. the 550-T was directly below the 700-T in the Lineup.
Chopper bulletin.
700-T MPX board.
550-T MPX Board. the 550-T was directly below the 700-T in the Lineup.
Last edited:
FISHER forum has a FISHER MPX ADAPTER's thread in the Sticky section. A bit light on the technical stuff, but plenty of pictures of the tops of the chassis's and a few underskirts. http://audiokarma.org/forums/index.php?threads/fisher-mpx-adapters.412418/
Larry -- The terms "chopper", "switching", and "time division" are all synonymous with one another.
Tube -- The book reference you mention is in deed excellent. I would also concur with you that "sampling" already takes place at the transmission end, so any degradation due to it has already occurred before the composite signal hit the modulator.
Trnsfmr -- A fascinating thread you've started! FM Stereo MPX is also a topic I have been very interested in, and have studied intently for may decades now. One thing I like to offer as a correction however, is your comment that (at the time) the FCC was unconcerned as to what the reception format was (paraphrasing your words). I'm not sure if you made that comment in this thread or your Eico MPX thread over on the Tube forum. As well, maybe I misunderstood your comment, but in reality, they were ultimately made to understand just how important the reception end was, as all the existing FM stations that were transmitting SCA information were very concerned that the new FM Stereo system would be compatible with existing reception equipment for that medium. In fact, that is what won over the FCC to approve the AM based sub-carrier of the GE/Zenith system, over that of the FM based sub-carrier Crosby system.
I think the time division system won out, because it proved to be more stable and much less problematic over time -- as you have indicated. BTW, to add to your list, the early Fisher MPX-100 units were a matrix based design, so you might check that out as well.
Dave
Tube -- The book reference you mention is in deed excellent. I would also concur with you that "sampling" already takes place at the transmission end, so any degradation due to it has already occurred before the composite signal hit the modulator.
Trnsfmr -- A fascinating thread you've started! FM Stereo MPX is also a topic I have been very interested in, and have studied intently for may decades now. One thing I like to offer as a correction however, is your comment that (at the time) the FCC was unconcerned as to what the reception format was (paraphrasing your words). I'm not sure if you made that comment in this thread or your Eico MPX thread over on the Tube forum. As well, maybe I misunderstood your comment, but in reality, they were ultimately made to understand just how important the reception end was, as all the existing FM stations that were transmitting SCA information were very concerned that the new FM Stereo system would be compatible with existing reception equipment for that medium. In fact, that is what won over the FCC to approve the AM based sub-carrier of the GE/Zenith system, over that of the FM based sub-carrier Crosby system.
I think the time division system won out, because it proved to be more stable and much less problematic over time -- as you have indicated. BTW, to add to your list, the early Fisher MPX-100 units were a matrix based design, so you might check that out as well.
Dave
Trnsfmr
Well-Known Member
Hi Larry,
Well, now this is getting interesting. I spent most of my spare time yesterday going over schematics again. I think I truly am obsessed with this whole Matrix tuner thing.
thanks for the "Chopper Bulletin". That exact circuit appears in the 800C, 500C, MPX100 and many others and most Fishers I have looked at. I have three 800c's, unfortunately I haven't restored any of them yet. The interesting one to me was the 550T, which you included a schematic for. I found it on HIFI engine, so I could print it out and mark it.
The 550T is definitely not a Chopper. It seems to be a rather poor attempt at a matrix decoder. It seems the detection for the 38khz subcarrier L-R signal is attempted in a transformer coil (Z-405) when there should be two diodes at least, and then the mixing, or adding the L+R with the L-R is being done in Q407 and Q408. It is done by applying the composite signal to the collector and the L-R to the base. Odd, using simple resistors would seem to be so much better and easier for this. Perhaps the designers were determined to not have to use hand picked, super tight tolerance parts.
The concept of adding/subtracting signals, which is really just mixing, is very well known to pro audio designers, who design mixing consoles. Perhaps these concepts were just not understood by HiFi and FMstereo designers in the early 60's..
In the pro audio world signals are sometimes combined together in a transformer, but it is a very different situation, and certainly not done to achieve AM signal detection.
In any case, another Matrix decoder to add to my list, or at least a poor attempt at one. I wonder how good the decoding was with the 550T? I would imagine not very good, with poor separation, especially at high frequencies.
Do you have any experience with the 550T?
If other manufactures made similar matrix decoders, I'm sure the poor performance led to the development and acceptance of the chopper. The Eico ST97 was probably late on the chain, and of course Eico and Heathkit were not industry leaders that any of the majors would want to follow. The Heathkit implementation of matrix decoding was unwieldly, with the additional step of phase adjustment required whenever a station was tuned in.
Well, now this is getting interesting. I spent most of my spare time yesterday going over schematics again. I think I truly am obsessed with this whole Matrix tuner thing.
thanks for the "Chopper Bulletin". That exact circuit appears in the 800C, 500C, MPX100 and many others and most Fishers I have looked at. I have three 800c's, unfortunately I haven't restored any of them yet. The interesting one to me was the 550T, which you included a schematic for. I found it on HIFI engine, so I could print it out and mark it.The 550T is definitely not a Chopper. It seems to be a rather poor attempt at a matrix decoder. It seems the detection for the 38khz subcarrier L-R signal is attempted in a transformer coil (Z-405) when there should be two diodes at least, and then the mixing, or adding the L+R with the L-R is being done in Q407 and Q408. It is done by applying the composite signal to the collector and the L-R to the base. Odd, using simple resistors would seem to be so much better and easier for this. Perhaps the designers were determined to not have to use hand picked, super tight tolerance parts.
The concept of adding/subtracting signals, which is really just mixing, is very well known to pro audio designers, who design mixing consoles. Perhaps these concepts were just not understood by HiFi and FMstereo designers in the early 60's..
In the pro audio world signals are sometimes combined together in a transformer, but it is a very different situation, and certainly not done to achieve AM signal detection.
In any case, another Matrix decoder to add to my list, or at least a poor attempt at one. I wonder how good the decoding was with the 550T? I would imagine not very good, with poor separation, especially at high frequencies.
Do you have any experience with the 550T?
If other manufactures made similar matrix decoders, I'm sure the poor performance led to the development and acceptance of the chopper. The Eico ST97 was probably late on the chain, and of course Eico and Heathkit were not industry leaders that any of the majors would want to follow. The Heathkit implementation of matrix decoding was unwieldly, with the additional step of phase adjustment required whenever a station was tuned in.
Last edited:
Trnsfmr
Well-Known Member
Dave,
As far as I can determine, FM mpx encoding is done at the transmitting station by many different methods depending on what devices each station has. I would expect Direct Digital Synthesis to be the order of the day in the most recently designed encoders. If this is the case, then the original stereo signal would be sampled, I'm sure, but certainly not at a 38khz rate. Probably more like a 96khz or 192khz rate with a 24 bit depth. At those sampling rates a well implemented A/D converter will achieve complete transparency with no degradation of the signal.
Sampling per se is not inherently degrading to audio quality. Just has to be done well.
I found an interesting paper on an encoding device that uses both DDS and analog techniques.
http://www.inase.org/library/2014/santorini/bypaper/ROBCIRC/ROBCIRC-29.pdf
FCC process: Most interesting. Didn't know about the process the FCC went through to decide on the scheme. My statement was more about that the FCC didn't care so much about what us listeners would use for equipment. The SCA inclusion was very important, and I'm sure the commercial users and stations filed their opinions with the FCC. HiFi listeners at home, not so much. It would be interesting to find out if there was any opinion filed regarding home listener FM stereo decoding. That would probably have been done by the HiFi manufactures, who were probably small companies when compared to Zenith, Crosby, and Fisher. Maybe Fisher had something to say about it.
As I mentioned in the previous post, I think the MPX 100 was a chopper decoder as the same chopper circuit shared by Larry is part of the schematic.
I was looking through a Heathkit manual I have. They were really good at explaining stuff. Here's the page where they describe matrix decoding. I thought some of the readers of this thread might be interested.
Last edited:
Ah! Pg 50 out of Heath's model IG-37 FM Stereo Generator manual. Great information. That model, the H.H. Scott model 830, and the Fisher model 300 all use diode ring modulators to develop the DSB supressed carrier 38 kHz sub-carrier -- the latter two using dual balanced ring modulators. This thread has piqued my interest to refresh my mind how that signal is generated in the ST-1000A (just checked, it's a switcher) -- that generator (iirc) claiming an excellent separation figure of 50 db @ 1 kHz. Of all four of these generators, I find the Fisher and the ST-1000A to be the most stable and practical for everyday use.
It was the very earliest version of the Fisher MPX-100 that was a matrix design, but the vast majority of that model was indeed produced as a chopper design.
Dave
It was the very earliest version of the Fisher MPX-100 that was a matrix design, but the vast majority of that model was indeed produced as a chopper design.
Dave
Trnsfmr
Well-Known Member
That's impressive, you must know your way around the Heathkit universe! There's a very good matrix description in the AJ-41 manual, which was their matrix tuner, replaced by the switching AJ-43 when they went to solid state. The AJ-41 is a beautiful piece of early 60's styling. I linked to an AJ-41 manual somewhere early in this thread.
l My IG-37 was beautifully built and must have been used an hour or two before I got it. I tried to save the multi cap can, but it went quickly.
I checked the early MPX schematic and completely different. Maybe I'll analyze it another time, that's tough brain work to do every day.
Not to derail your thread, but since you have a Heath generator, then check R62 in the power supply section. It is an 820 ohm resistor spec'd as a .5 watt unit, and is what was installed. But based on the current draw through it, it should be properly rated as a 2 watt device with everything operating properly. The one in my generator was notably burned and doubled in value. Any added current draw from a deteriorating can cap will only add insult to injury. One of Heath's "oops" moments.........
Dave
Dave
arts
Addicted Member
I've got an IG-37,the later,sexier IG-5237,and a cute little FMO-1.They all work very well (after rebuilding) especially considering what they sold for back in the day.
For most work I'm using an RCA WR-51A (hey,if it was good enough for McIntosh,it's good enough for me!).
I had the ST-1000A,but sold it years ago as there just wasn't enough work to justify keeping it. That and I spent more time working on it than the tuners I was supposed to be working on. Glad to hear you're having better luck with your's, Dave
For most work I'm using an RCA WR-51A (hey,if it was good enough for McIntosh,it's good enough for me!).
I had the ST-1000A,but sold it years ago as there just wasn't enough work to justify keeping it. That and I spent more time working on it than the tuners I was supposed to be working on. Glad to hear you're having better luck with your's, Dave
Continuing the derail, I have an IG-37 and it's a handy piece, but I've never understood the alignment of it. They say to minimize the signal at the composite output by adjusting T2, T3 and the balance control. When I do that, the transformer slugs are in wildly different positions and the signal isn't minimized at the junction of D2 and D3. When I minimize it between D2 and D3, everything works as I'd expect, with equal slug positions, but signal at the composite output. If there's something wrong, I've never been able to figure out what.
Trnsfmr
Well-Known Member
"Not to derail your thread, but since you have a Heath generator, then check R62 in the power supply section. It is an 820 ohm resistor spec'd as a .5 watt unit, and is what was installed."
I found it before I turned it on. It was the 330 ohm one and it was blistering. I just happened to have a 3w one, so I put it in. The 820 was a 1000 ohm in mine and it seemed to be ok so I left it as is.
I found it before I turned it on. It was the 330 ohm one and it was blistering. I just happened to have a 3w one, so I put it in. The 820 was a 1000 ohm in mine and it seemed to be ok so I left it as is.
KentTeffeteller
Gimpus Stereophilus!
Crosby Stereo failed due to the inability to run SCA services, in that era very critical to FM station revenue as most FM stations then were barely staying afloat. HH Scott was a major company involved in the FCC testing of FM Stereo, and their engineers built Stereo generation equipment as well as prototype decoders for the various systems under FCC test. In 1961 the Zenith and GE system was approved for use by the FCC. That said, I find in real life, HH Scott's decoders work best for me on less signal strength in the rural area than most others, Fisher being great also. HH Scott also built FM Multiplex Stereo generators through the early 1970's through their instrument division.
Oddly enough looking at the bottom of all your shoes with a telescope and grasping a very small percentage of what's being discussed I am enjoying the thread. I'll fade back into the background for now.
Trnsfmr
Well-Known Member
What a great thread. I really appreciate everyone's input. I have learned a lot...
Practically everything I thought about the early history of FM stereo decoding was wrong. HH Scott made the first chopper. Fisher started off making Matrix decoders, the early MPX100 and their 550T, were both matrix. The early MPX100 schematic on HFE, is not a complete schematic but it is obvious that it is a matrix. Connections are missing on the schematic to a crucial cathode where the signals get combined, so I can't analyze it and put it up here.
So... there were at least 5 matrix decoders, 2 from fisher, 2 from Eico, and 1 from Heathkit. Most of them seem (from today's viewpoint) to be lacking in common sense design, but that is often how it is, with each new design building on the mistakes and weaknesses of the previous one.
Just as each discipline has its strength and weaknesses, for example, comparing SS and tubes... Tube amps can never match the power levels possible with SS designs (thousands of wpc in modern amps), and SS amps have a hard time matching the detail and musicality of the best tube designs.
Here we have two topologies for decoding the rather complex analog encoded FM stereo signal.
In the beginning of stereo there was a huge rush to be first with true decoding. As described by the FCC, the process would seem to be best decoded by a matrix analog decoder as that is essentially how the encoded signal was constructed to begin with. However engineers obviously had a very hard time getting there. Some very sharp Scott engineers figured out how to do it with sampling, 30 years before digital sampling would take over the marketplace, giving quite good specs and very passable audio quality.
However, do to the nature of the encoded FM signal this sampling was limited to what today would be a substandard sampling rate, and by today's standards, FM stereo quality decoded by the chopper is sorely lacking. You only have to read a few postings on most any forum where people talk about FM to find people expressing their opinions about this.
In fact it is so bad that the industry has come up with a new FM scheme, digital FM, which is adding more and more processing to what was basically a very low distortion analog medium to begin with, that just needed to be decoded properly (at least, that is my opinion).
Eico seems to have been late on the chain with matrix decoder design. They gave up on trying to do the matrix mixing functions in transformers, tubes, and transistors, and instead accomplished the mixing in precision resistors, and balanced detectors, as anyone familiar with pro mixing console design would do today. With precision resistors it is possible to achieve very good channel separation, without that "hard to describe poor audio quality" that comes from sampling a signal at low sampling rates.
In my opinion the Eico decoder when restored with modern high quality parts, finally shows the true capabilities of analog encoded FM stereo, where the quality only seems to be limited by the quality of the encoders and audio processes at the originating FM station.
I am still trying to get an early Heathkit reciver (AJ41), which, as they describe in their explanation of matrix decoding, uses resistors for achieving the matrix mixing, just as Eico did. I'd love to get a chance to hear the early Fishers also, but I think those pieces are pretty rare, and not available on ebay for less than $100, like the Heathkit.
Again thanks for everyone's input on this complex subject, and please add more information if you can. Incidentally, today I finally found on line a rundown of the history of the FCC stereo approval process...
http://files.eric.ed.gov/fulltext/ED296411.pdf
Practically everything I thought about the early history of FM stereo decoding was wrong. HH Scott made the first chopper. Fisher started off making Matrix decoders, the early MPX100 and their 550T, were both matrix. The early MPX100 schematic on HFE, is not a complete schematic but it is obvious that it is a matrix. Connections are missing on the schematic to a crucial cathode where the signals get combined, so I can't analyze it and put it up here.
So... there were at least 5 matrix decoders, 2 from fisher, 2 from Eico, and 1 from Heathkit. Most of them seem (from today's viewpoint) to be lacking in common sense design, but that is often how it is, with each new design building on the mistakes and weaknesses of the previous one.
Just as each discipline has its strength and weaknesses, for example, comparing SS and tubes... Tube amps can never match the power levels possible with SS designs (thousands of wpc in modern amps), and SS amps have a hard time matching the detail and musicality of the best tube designs.
Here we have two topologies for decoding the rather complex analog encoded FM stereo signal.
In the beginning of stereo there was a huge rush to be first with true decoding. As described by the FCC, the process would seem to be best decoded by a matrix analog decoder as that is essentially how the encoded signal was constructed to begin with. However engineers obviously had a very hard time getting there. Some very sharp Scott engineers figured out how to do it with sampling, 30 years before digital sampling would take over the marketplace, giving quite good specs and very passable audio quality.
However, do to the nature of the encoded FM signal this sampling was limited to what today would be a substandard sampling rate, and by today's standards, FM stereo quality decoded by the chopper is sorely lacking. You only have to read a few postings on most any forum where people talk about FM to find people expressing their opinions about this.
In fact it is so bad that the industry has come up with a new FM scheme, digital FM, which is adding more and more processing to what was basically a very low distortion analog medium to begin with, that just needed to be decoded properly (at least, that is my opinion).
Eico seems to have been late on the chain with matrix decoder design. They gave up on trying to do the matrix mixing functions in transformers, tubes, and transistors, and instead accomplished the mixing in precision resistors, and balanced detectors, as anyone familiar with pro mixing console design would do today. With precision resistors it is possible to achieve very good channel separation, without that "hard to describe poor audio quality" that comes from sampling a signal at low sampling rates.
In my opinion the Eico decoder when restored with modern high quality parts, finally shows the true capabilities of analog encoded FM stereo, where the quality only seems to be limited by the quality of the encoders and audio processes at the originating FM station.
I am still trying to get an early Heathkit reciver (AJ41), which, as they describe in their explanation of matrix decoding, uses resistors for achieving the matrix mixing, just as Eico did. I'd love to get a chance to hear the early Fishers also, but I think those pieces are pretty rare, and not available on ebay for less than $100, like the Heathkit.
Again thanks for everyone's input on this complex subject, and please add more information if you can. Incidentally, today I finally found on line a rundown of the history of the FCC stereo approval process...
http://files.eric.ed.gov/fulltext/ED296411.pdf
Last edited:
Trnsfmr -- On the early Fisher MPX 100 schematic, the schematic is in fact complete, but hardly indicated so very clearly: The "missing" cathode connection for V4-B is due to the fact that this cathode is internally electrically common with the cathode for V4-A, with both cathodes then appearing at pin 3 of the ECH84 tube.
I hope this helps!
Dave
I hope this helps!
Dave
