FM Follies

FM Follies  if you’re new to this, and really interested, you might want to scroll down and read this from the beginning, starting in 2014.

April 23, 2016 This page has really become like a blog.  So, I’ve sorta changed the format and will put new updates at the top.  The earliest stuff will be toward the end.

I recently posted a piece on Audiokarma, the Tuner group, or Tuna as they smilingly call it.  Here’s a link if you want to read it.

http://audiokarma.org/forums/index.php?threads/questioning-basics-fm-stereo-decoding-schemes.712259/

The reception has been pretty good, a lot of knowledgeable folks on that group, and they rightly question what I have to say.  According to one of the posters, I was wrong about Fisher coming up with the Chopper (sampling) FM tuner.  It was H.H. Scott who made the first one.  Interesting too, because an earlier poster told me than an early Scott receiver, the 340, was Matrix.  It doesn’t seem to be to me, but there is no explanation with the schematic, no block diagram, so I am not sure.

Turns out there were several technical mistakes in my earlier writings on this page.  I went back and corrected them, so as to make it the least confusing as possible.

I recently bought another ST97 and just last week, two  MX-99’s.  I’ve been trying to buy one for years and in one week I was able to get 2 of the MX’s at a good price.  So, I now have 5 Eico tuners and 2 mx decoders.  Excessive, I know, but I paid less than $25 ea for most of them.  Lest you think I’m nuts, we have a customer or two who have 4 or 5  9001’s.

I’m working on a piece, which I will put up here, and at Audiokarma about restoring an ST97 or an MX99.  I think we are going to offer parts for doing it, if not all the parts, at least matched sets of the matrix resistors and a matched set of schottky diodes.  Matching the parts to very close tolerances is what makes it work, and, what makes that luscious and so beautiful stereo sound.

I really want people to experience to the fullest the quality audio that is possible with this 55 year old design.

People on AK questioned how well it works as a tuner.  How well does it pull in stations, crowded band performance, all that?  Well, we in our shop here in Crazytown are in a very bad location for FM.  We are about a mile from “antenna hill” where most of the transmitters and antennas are located that serve this metropolitan area of  250K people, both TV and radio.

So I can’t tell you about fringe area performance, but here at the shop we rate tuners on how many listenable stations can be tuned in.  About the best number has been 17 or 18 stations, a level of performance turned in by several good tuners and receivers.  The ST97 will tune in 13 listenable, which is pretty good.  About as good as most of the receivers we see.  You can hop in your car and drive a few miles away and do better than 17.  It’s a torturous environment for an FM tuner.

Hopefully the restoration article will be next.  Stay tuned.

March 2016  I haven’t written anything on this page for a while, but have continued to study FM decoding off and on.  I’ve probably studied the schematics for close to 20 different Fm Tuners, and even looked at the schematic and theory of operation of some Integrated Circuit based decoders.

As far as I can tell, they all use a variation of the original Fisher “chopper” decoding scheme.  There might be others, but even top of the line vintage tuners like the very well thought of Marantz, Fisher, and McIntosh units use the “chopper”.

Here’s a brief description of how the “chopper” works.

The 19khz signal is derived from the original FM signal.  The 19khz signal is doubled to 38 Khz and it and the composite waveform are sent to a circuit called a ring demodulator.  The left and right channels then are produced at the L and R outputs.  How does it work?

That’s pretty complicated… but basically…

When the 38 khz signal peaks positive the demodulator will put out the sampled L channel at that point in time. When the 38khz waveform peaks negative, the sampled R signal is put out.  And so on… back and forth.  With good channel separation.

The 38 khz signal acts as a control gate, alternately opening a window to pass through the L ch signal, and then a window for the R ch signal.  You can see why they called it “the chopper”.

The comparison to digital sampling is immediately obvious.   There are differences though, the main difference being that the channels are sampled in turn, not at the same time.  The peaks are then smoothed by a small capacitor.  That’s the basics of it.

I don’t know about you, but there is a sound, the sound of FM to me, that I especially hear with pop music, rock n roll, etc.  where there is a lot of high frequency energy like guitar or perhaps drum cymbals.  It sounds like noise, and it all sounds the same.  If you find other sources for the same exact music, it doesn’t have “that” sound.  I think this is the result of the “chopper”, but that is just really my opinion.

Now, after reading all this description, imagine a completely analog signal… not cut up in little pieces, recombined and smoothed out… and that is what the “Matrix Decoder” provides, like is in the ST97.  I’m sure there are other tuners that use the Matrix scheme, I just haven’t found them.

One reader sent me info on an early Heathkit tuner, the AJ41, and it obviously used a version of this Matrix decoding, with some adjusting required each time you tune a station, to make it work.  The design obviously hadn’t been fully developed yet.   That’s the only other one I have studied so far.

Please email me with anything you might think about this.  I assure you, no one is talking about FM stereo decoding schemes.  LOL

Aug 14, 2015

Well, I have remained very interested in this.  Turns out the Sansui 5000 has a transformer (tuned coil) IF strip as I said above, but it has the standard Chopper type decoder.  It seems everything has this.  The Eico seems to be the only one with what is called a Matrix Decoder.  I’m searching for another analog matrix stereo tuner, but haven’t found one yet.

Aug 3, 2015.  I have gotten a few letters regarding the above, and so here is more information for the technically inclined.  First of all, the above is really simplified, but none the less, I stand by it and I hope if you read some of these links you will understand why.

I have found a wonderful book, published in 1974 that covers pretty much the state of the art at that time for FM stereo.  It’s called “Stereo FM Radio Handbook”, and you can find it on line  Here.

It’s a good book, but like all engineering texts of that time, and even today, sound quality is not a big topic.  The only place I can find any reference to it is in the very first sentence of the Author’s preface, where he says, “The desire to improve the quality of sound in recorded music led to the introduction of stereophony.”  Then, the whole rest of the book never mentions it again, or even the idea that some of the many methodologies discussed might sound better than others.

It’s part of what I call, “Engineering Arrogance”, which is taught, pretty much unknowingly, to all engineers.  “Everything can be explained by theory, and math.  Opinions don’t matter.”  The basic flaw in this viewpoint, is that there is always new theory, and new designs, new ways of doing things that weren’t even thought of before.

We happen to have a Sansui 5000A in the shop being restored at the moment, and I spent some time with it and with the schematic.   Turns out the 5000 is a pretty simple tuner, with no tricks, just good solid circuit design, very much like the Eico.  Without any particular attention paid to the multiplex decoder on my part, it has a very full and rich sound.  Vocals pop out at you, and  music sounds dynamic.

So here’s a bit more explanation of why the modern tuners don’t sound so good.

An FM signal comes into the tuner and is converted in the front end to a 10,7 Mhz signal.  This is called the IF, or intermediate frequency.  Then there are a series of amplifiers, usually at least 2, but 3 or 4 is better, one after the other, each tuned to 10.7 Mhz.   The effect is to get selectivity, so that stations on either side are greatly reduced in amplitude.

The first problem with FM tuners to appear was the removal of the tuned coils between each of these IF stages and replacement with ceramic filters.  In the 50’s and 60’s various types of crystal/ceramic filters had been developed that give very good selectivity.  They were originally developed for communications equipment, and early top of the line SSB ham receivers had these filters.

Problem is, those filters don’t sound good.  They are very selective and pass just a small slice of the frequency spectrum, and when cascaded in 3 or more sections, make for a very selective receiver.  That’s not the problem, the problem is what happens within that passband.

The frequency response of that filter, in the passband, is rough, with sharp ups and downs, and lots of phase problems.  Engineers would say the filter has poor group delay response.  They vary from unit to unit right off the manufacturing line.  Ceramic filters with the best and more uniform passband are selected by testing and these are used in high end tuners.  Still, they cannot begin to compare to a good IF transformer which is totally smooth in response by its very nature.

In the Sansui line of FM tuner development, the 5000 was the last one to get the old fashioned IF strip, with tuned coils.  The Eight had ceramic filters, and thus continuing the march into mediocrity.

If you look at the pictures of the naked Fisher and Eico, above, you will see squarish metal cans mounted between vacuum tubes.  These cans contain the IF coils.  I don’t think any electronics manufacture even makes them anymore.

Next came multiplex decoder chips.  These use all kinds of schemes to get good separation, low noise, and low distortion.  If you can get through the above two links, you’ll see how digital techniques, with up to 8 times oversampling can be applied to get good specs.  If I wanted digital, though, I wouldn’t be on this site.  :o)

So, I am going to get myself a 5000A, and see just what I can do with its FM tuner.  Wouldn’t it be cool if I could get it to sound as good as the Eico?  It already sounds very good.

Stay tuned.

Sometime in 2014

I’ve been wanting to write something about FM stereo, as I feel rather strongly about it.  This is likely to be something different than you have ever read before about FM stereo.  It’s the story of technology making things worse, but measure better.  Anyway…

FM stereo started in the early 60’s and went downhill from there.  What do I mean?  The tuners got worse.  The engineers wanted lower manufacturing costs, and better and better specs, and they got them, but the sound got worse.

Here’s what actually happened.  Stereo records came out and broadcasting executives thought it might be an opportunity to create a whole new broadcast medium, FM stereo.  The FM band had been in existence for many years, but essentially no one listened, as AM was adequate for existing broadcasting.   Engineers came up with some ideas, different ways to do it, and these different schemes were evaluated by the FCC and they eventually picked the scheme we still use today.

It’s not necessary to get technical and explain how it worked.  It was a pretty good system, the only drawback being that you needed a strong signal for the stereo to work properly.

On the transmit side, since cost is not that much of an issue, equipment was designed that produced a very good stereo signal with low distortion and good separation.  On the receive side, mass produced FM stereo receivers did not fair so well.  There was no cheap way to decode the signal.  Expensive components and precise alignment were required to get low distortion and good separation.

It mostly came down to good separation.  The stock scheme required a very well aligned receiver using some fairly expensive components to get good separation.  Expensive parts and high labor costs were not something manufactures wanted.

Engineers saw it as a problem of getting better specs.  Receivers using the original decoding scheme only got 8-10db of separation,  15 db was considered excellent.  Never mind the fact that most recorded music didn’t have much more separation than that to begin with.

So the engineers went to work.  Back in the early 60’s Fisher was the industry leader.  There also was HH Scott, and a couple others.  Scott came up with a chop scheme, where the signal was chopped up about 38,000 times per second, with every other bit sent to each channel in turn.  It produced much better specs and kind of became the standard of the day.  Engineers seized on this design and maximized the design over many years.

In the 70’s, all kinds of component advances followed for a cheaper and easier to build tuner.  First were Ceramic Filters, and then Integrated Circuits.  Today it takes just a few components to make an FM tuner, with amazingly good specs.  Only problem is, well… it sounds like…  FM.  You know, dull and lifeless.

One poster on a HiFi stereo list said he wasn’t much of a fan of FM. I replied that none of us critical listeners are, or at least very few of us.

The reality is, that Sansui did just like everyone else.  As time progressed so did the sophistication of the parts in the Sansui tuners, and, as a result, the older the tuner, the better it sounds.

Probably the best FM tuner in the SS Sansui  line is in the model 5000 receiver.  The 5000, came out in 1968, and the 5000A, and 5000X are pretty much the same tuner.   It was pretty much the last one to have an IF strip with transformers, and the FM decoding is old school, no integrated circuits.

You might wonder what I think is the best tuner ever made?  You would probably think the Fisher 1000 or that McIntosh one, I can’t remember the model number.  They are pretty good, regardless of what I said above.  You might think it’s because they are tube tuners, but that is not it.  It’s because of the circuit design.

They used transformer based IF filters, and there were no integrated circuits.  The same could be done with solid state.

But no, I would say the best tuner ever made is the lowly Eico ST97. Sold in the 60’s and a cosmetic match to their ST40 and ST70 amplifiers, which were ok, but not stellar amps.  The amps can be modified to be very good, though.  Recently a design engineer has redesigned the whole output section of those amps to get about 10 db less noise, and lower distortion, but that’s another story.

Why was the ST97 so good?  Because, for some reason, the design uses a different decoding scheme. They were the cheapest makers of component stereo, especially if you bought their kits, and they just designed a straight up FM tuner using the completely stock decoding scheme as in the original FCC description.  They didn’t cut corners, they made it as good as they could.

Perhaps in original configuration, with the very cheap electronic parts that Eico always used,  it did not sound as good as the Fisher and the Mac, and many others.  These things are so old though, that the only way we can even hear one today is to restore it.  AND, if you put a little extra effort into the restoration, AND upgrade a lot of the electronic parts, well… you know where I’m going with this… in my opinion, it’s the best one.

I have one, that I restored, kind of on a lark.  The real reason is that when I was a kid in high school, I built, from kits, an Eico ST40 and then an ST70.  I also had two Acoustic Research AR2a speakers, and an AR turntable, all paid for from my own hard earned cash.  I had the best stereo of any kid in my high school.  Every one of them would agree once they came over and listened to it.

I wanted the ST97 tuner that matched up to my amp, but could never afford it. Actually there was a competing expense to my stereo hobby in my life, and you can probably guess what that was, and it took more and more precedence.  So I bought one on ebay for $25 a few years ago, and restored it.  While doing the restore, I saw a whole bunch of places to make it better and did so.  Just like I would restore any old amp.

You know what I’m talking about, hi quality and bigger caps, precisely matched 1% resistors, newer immensely better, matched Schottky detector diodes.  Everything I could see to make better I did.

Eico was a funny company, and I have done a bit of research about them.  There were several engineers there, none of them the big names in the hi fi world, but young guys with lots of enthusiasm and ideas.  They would design the best equipment they could to fit the products, and then they would have to put it together with the cheapest parts available, to keep their products down in price.

In their hi fi gear, Eico had a nasty habit of using cheap ceramic disc capacitors for almost everything.  I mean the really bad ones, called Z5U.  Back then, the idea that a better component might actually sound better, was a novel idea.  If it made spec, that was fine.

The most interesting story about the Eico team was about a loudness switch.  It seems a couple of the designers didn’t like the idea, and refused to put them in their ST40/70 design because it wasn’t accurate to the original sound, even though everyone else was doing it.  They were real purists.  There was evidently a lot of infighting about it.  They eventually ended up putting it in, but put the wrong value parts on it, so it didn’t work.  LOL

Perhaps the engineers decided their decoding scheme was the right way to go, as it was the pure implementation of the FCC specification.

When I was done and hooked the ST97 up in my system.  I was stunned.  I had only just hoped that it would work, and I could sign off on that unrequited desire from those teen years.  I didn’t expect it to be so good that I would write articles about it, or rant and rave about it like I have done on the Yahoo Eico list.

I actually stayed up all night when I first got it working.  I just kept listening and did not want to go to bed.  We have two high quality stations in this university town where I live, and I was convinced that they really care about quality, by the time the sun came up.  FM stereo was performing as designed.  It was like a pair of stereo cables from the station to my amp.

That is the idea, you know.  The signal comes out of the studio/audio part of the FM station, goes to the transmitter, gets encoded and sent out over the airwaves, where your tuner grabs it and turns it back into audio again.  If everything is done right, the whole transmit receive process should be transparent.  And if the audio part of the FM station is all high quality audiophile grade stuff, well…

A guy on the Yahoo Eico list recently followed my advice and restored one like I did.  He was the 2nd or 3rd one to do so.  He was amazed at the sound.  He’s the one who said he was not that big of a fan of FM.

I mean, you gotta get the concept here, of FM.  High quality Stereo-Quad is expensive.  But you can spend amazing amounts of money on source material: records, CDs, whatever.  Suppose an FM station would kind of specialize in music that you liked?  They’re on the air 24 hours a day.  They get a copy of every record or CD made, the record companies send them promotional copies for free.  They have high quality equipment and you can listen for free.

That was the idea, tailor made for us audiophiles.  Only a bunch of engineers got in the way and dumbed down the sound in search of better specifications, and lower manufacturing costs.  Today there are many variations of decoding schemes for FM Stereo, and none of them follow the original spec as outlined by the FCC.  All they had to do was make a high quality tuner that followed the original scheme.

I didn’t know all this history stuff when I restored my ST97, I looked all that up after I listened to it for a while, searching for an explanation. All the while, I was noticing what FM stereo sounds like in the Sansui receivers, and all the other brands too.  Please no offense, I love my Sansuis as much as you do, but then you’re probably not a big fan of FM anyway.  snicker, snicker…

 

 

 

 

 

 

As a side note, I always liked that 1964 song Iko Iko, by the Dixie Cups.  Here