Numbers and real world facts not just speculations please. How much do these cavity filters contribute to the distortion? What's the group delay introduced to the passband?
Describing views you disagree with as "speculation" is unhelpful. Demanding peer reviewed evidence yet offering none yourself is unhelpful.
90dB S/N in an FM stereo tuner is unlikely, except possibly in a lab - 80dB is about the best usually seen and many tuners can only manage 70dB. (Oops, sorry, more "speculation"). 70dB separation is unlikely; 40dB is more common and quite adequate. I assume the difference you heard was confirmed by careful double-blind tests and adequate statistics?
90dB S/N in an FM stereo tuner is unlikely, except possibly in a lab - 80dB is about the best usually seen and many tuners can only manage 70dB. (Oops, sorry, more "speculation"). 70dB separation is unlikely; 40dB is more common and quite adequate. I assume the difference you heard was confirmed by careful double-blind tests and adequate statistics?
Yes, been there. Owned, tried, tested and modified over 200 FM tuners.
There are many ways to get good performance from an FM tuner. It is never based on one circuit feature, i.e. the detector. It's easy to make good wide band low distortion detectors, and many tuners show great examples using pulse count, PLL, quad, and ratio detector technology. Sansui and Yamaha mostly used ratio detectors, for a good reason. They are simple, and work very well.
In general, the IF path filters contribute most to the distortion. If they screw it up, and they can, you can't undo it or fix it later. You need phase linear IF filters with flat group delay for good sound. And usually a 2nd option with narrow IF filters for distant and weak stations. Hence the better tuners all have wide / narrow IF filters.
To really appreciate a great FM tuner, you likely need an outdoor 8-10 element outdoor log periodic antenna on a rotor, though.
The multipath problem is greatly reduced, or, dare I say eliminated, by using a directional antenna with good front to back ratio.
For many many years top FM tuners were made that achieved 0.03% ballpark distortion, in stereo, at 100% modulation. This was when FM was taken seriously.
Today, most manufacturers use car stereo chips in home FM components, so yep, today .5% is typical for home theater receivers, for various reasons.
Mostly it's the filters and sloppy alignment, and also the use of "all in one" FM chips. It's always easier to do less, of course, and usually it costs less too.
But current designs using a digitized IF, and DSP for IF filters and stereo decoding can be pretty remarkable spec wise. A good example is the Sony XDR-F1HD. I modified one of these to give it a proper audio power supply (20 VDC) and discreet output stage, and was pretty impressed at the measured performance, i.e. S/N, distortion, separation - all in analog mode, not HD. Not sure if Sony was losing money selling these new for $100, but it's really too bad they are obsolete now. With proper DIY mods, they can be really nice machines.
Attachments
In a set of tuner reviews in 1987 (when FM was still taken seriously), most good tuners were in the 0.1-0.2% distortion region. Some were worse (but still sounded OK); few were much better. The very best was about 0.03%, so no 0.01%. Stereo S/N was typically around 60-65dB down. So my memory was not perfect, but not far out.
I agree that modern tuners are often worse. Those using SDR techniques seem to be particularly bad.
I agree that modern tuners are often worse. Those using SDR techniques seem to be particularly bad.
It doesn't mean something you haven't seen doesn't exist. If I divulge the tuner I aligned tested with these kind of numbers and the sensation it creates, I'll make it unfordable even to myself given how the market operates..
I know my FM stereo generator can go beyond 90dB S/N and THD <0.01% because my HP 8901A + HP 3585A shows FM residuals and fundamental harmonics are low at 75kHz deviation. I'm confident of the stereo separation is >70dB after confirming 19kHz vs 38kHz phase error is minimal, the crosstalk between left and right is over >80dB on the composite signal, again with an HP3585A.
Come and visit the Tuner Information Center - Vintage Stereo Tuners website and better yet, join the Yahoo FM Tuner group
I know my FM stereo generator can go beyond 90dB S/N and THD <0.01% because my HP 8901A + HP 3585A shows FM residuals and fundamental harmonics are low at 75kHz deviation. I'm confident of the stereo separation is >70dB after confirming 19kHz vs 38kHz phase error is minimal, the crosstalk between left and right is over >80dB on the composite signal, again with an HP3585A.
Come and visit the Tuner Information Center - Vintage Stereo Tuners website and better yet, join the Yahoo FM Tuner group
You are mixing up THD with THD+N, that's your problem right there. Most if not all tuner reviews use THD+N. My numbers are THD, I quote S/N separate and I don't use 'A weighting' which gives inflated S/N number by chopping off low and frequency noise, yes the ones where are our ears are most sensitive.
This is a typical TX filter, https://radiolab.no/datasheets/1kW_3Cav_FM_Filter.pdf
This would be used is series with a tuned combiner if the antenna is shared
Note the lack of information on group delay
This would be used is series with a tuned combiner if the antenna is shared
Note the lack of information on group delay
What exactly are your concerns here? The data sheet shows the filter is about 1 MHz wide. That bottom plot is not phase, it looks to be return loss.
The FM channel is 150 kHz wide. Normal tuners with many gangs in the RF front end have effectively about the same ballpark filtering before the mixer.
If aligned properly, it has no effect on distortion, being 8-10 times wider than the working channel band. When the bandpass filter is that wide in comparison to the channel, usually there is no problem with group delay.
I can probably dig up quotes from an active broadcast engineer who speaks to typical modern FM transmission quality, but you can too, if you join Yahoo FM Tuners and peek in the archives.
Not saying that's how it's done everywhere in the world, but can speak about what is typical in the US.
Well, I guess it all depends on your perspective, like what you call a "good tuner". I would say most "good tuners" from 60's to the early 70's had 0.1-0.2% distortion. And in some cases, they were specified that way due to limits on the equipment used to test them.
I've aligned late 60's H.H. Scott tuners like the LT-112B to 0.06% with modern test gear. Any tuner worth considering from the early 80's through the late 90's had at least 0.05% or better stereo distortion specs, and most exceed that after alignment.
Getting below .03% is not unusual, depending on your skills and experience with tuner alignment. Same with S/N, your numbers are for 60's -early 70's stuff.
Unless you are talking about cheap receivers, or bottom or mid line models.
I only deal with the top of the line, or one below it, class of FM tuners. They are cheap and plentiful these days, so if you are serious, why mess with crappy lesser units?
And that list of top or near top units includes many models that can be had for a song today. I usually run outboard MPX decoders, from a tap right after the detector on the vintage models. That's the best way to maximize the sound with DIY efforts, IMHO.
Here is a TX filter with group delay graph Shively Labs FM Antennas and Accessories
SDR radios can have excellent performance on the bench, due to lack of ceramic IF filters and tuned circuit detectors. Real world performance with the crowded FM band is another story
SDR radios can have excellent performance on the bench, due to lack of ceramic IF filters and tuned circuit detectors. Real world performance with the crowded FM band is another story
OK, according to you these tuners were all either unusual or were not measured in 1987. I find it interesting that you seem to know more about these reviews than I do, yet it is I who has the book in front of me. Or did manufacturers send all their rubbish tuners to the UK market?BFNY said:
I don't 'deal' with any FM tuners these days. I designed and built one some years ago and still use it. How many have you designed?
The CA3189 gave around 0.1% with a double tuned coil. This may have been bettered a little with fancier IF filtering and maybe some fortunate group delay cancellation. I have built PLL and pulse count detectors and both of these are hard to get better than quadrature due to subtle noise effects on the power rails and pulse edges causing small kinks in their transfer slope
The linearity of the transmitter depends on how much care taken and cost spent on group delay correction. Nobody bothers too much these days when the feed from the station studio is re-compressed digital from mp3 sources to save money on leased lines
The linearity of the transmitter depends on how much care taken and cost spent on group delay correction. Nobody bothers too much these days when the feed from the station studio is re-compressed digital from mp3 sources to save money on leased lines
I never designed a tuner, no. But designed and built many modules to go into modding and improving FM tuners. I have modified many units to improve them, and not just the audio stages and power supplies. I keep a stock of many thousands ceramic filters, which are swept and sorted for center frequency, insertion loss, etc. I designed / built aspects of the test gear to do this, using the AD8307 as the detector for greatest dynamic range. Some tuners screw up the impedance match from the post mixer IF transformer to first filter, so usually fix that when I see it.
Anyways, about your "book", (unknown title, unknown author). It's not relevant, for me, in any way to this discussion. Why would anyone trust a book of reviews, when they can test the tuners personally and draw from first hand experience?
I also have a library of over 200 tuner repair manuals, with specs and schematics, which is a better place to start if one wanted to read something about tuners, their design, and performance. Many are posted on the Tuner Information Center, or in the Files section of the Yahoo FM Tuners groups (+ 4 file archive groups listed on the home page).
If people are interested, I can post here pdfs of the tuners available in 1987 with specs, from the Audio Annual Equipment listings. The units from companies that really invested in FM Tuners, like Accuphase, Carver, Denon, Harman Kardon, Hitachi, JVC, Kenwood, Luxman, Onkyo, Pioneer, Revox, Sansui, Sony, Sumo, Tandberg, and Yamaha show the state of art for tuners in 1987. I would not call the list "small" or "exceptions".
Most of the top models had 1kHz stereo distortion well below 0.1%, and many were listed at 0.02%. Most had S/N (stereo) way better than 75 dB, and some listed numbers in the upper 80s and low 90s. Most had stereo separation way better than 50dB, many showing mid 60's and even 70dB.
Yes, there were worse spec'ed models, many from boutique brands that had expensive cases, high price tags, not much tuner inside. Or bottom line units for 199. list or cheaper. So, turns out 1987 was a pretty good year for tuners.
Last edited:
Most of the tuners used on reviews are not tweaked for best performance and come from assembly lines where the quad coil is set for 0V and the front end VCO voltage is set for maximum range. I seriously doubt the manufacturers employ FFT for adjusting THD and stereo separation.
I got into this discussion because somebody said that there is no need for tuners having <0.01% THD. I don't believe that notion for a moment. It's like saying who needs to design power amplifiers with THD <0.003% with slew rate of 300V/usec when the speaker THD is already several percent.
I'm not here to post my CV/resume and brag about it. I know my stuff when it comes to FM tuners which is not my only interest or experience.
I got into this discussion because somebody said that there is no need for tuners having <0.01% THD. I don't believe that notion for a moment. It's like saying who needs to design power amplifiers with THD <0.003% with slew rate of 300V/usec when the speaker THD is already several percent.
I'm not here to post my CV/resume and brag about it. I know my stuff when it comes to FM tuners which is not my only interest or experience.
With the LM/CA3189 having 0.1%, is this THD or THD+N, at what bandwidth and your S/N? What make and model is the FM stereo that you used? These all play in the numbers you are giving and how relevant they are with the discussion. I use a Panasonic's VP-812x/813x FM stereo generators.
The cavity filter link you sent shows insignificant group delay on the passband, it is esentially flat which barely affects the modulation characteristics. We are talking less than 50ns for a 300kHz bandwidth. Look up the spec for low group delay MX series ceramic filters from Murata that are now discontinued, they range from 0.25u-1us. This number is a lot worse than your dreaded cavity filter. Tuner designers incorporate group delay equalizer to minimize group delay and preserve the demodulated audio's shape, which is your THD.
The cavity filter link you sent shows insignificant group delay on the passband, it is esentially flat which barely affects the modulation characteristics. We are talking less than 50ns for a 300kHz bandwidth. Look up the spec for low group delay MX series ceramic filters from Murata that are now discontinued, they range from 0.25u-1us. This number is a lot worse than your dreaded cavity filter. Tuner designers incorporate group delay equalizer to minimize group delay and preserve the demodulated audio's shape, which is your THD.
From the CA3189 data sheet
"Distortion in a CA3189E FM-lF System is primarily a function of the
phase linearity characteristic of the outboard detector coil."
The better the quad detector coil, the lower the distortion.
There are many FM tuner designs using Quad detectors that produce very low distortion, .02% in stereo. And they can sound pretty good. Ever listen to a Rotel RHT-10?
The engineering teams with tons of money and experience at say Accuphase, Pioneer, Sansui, or Kenwood probably achieved a lot more performance with all the detector designs you mention.
"Distortion in a CA3189E FM-lF System is primarily a function of the
phase linearity characteristic of the outboard detector coil."
The better the quad detector coil, the lower the distortion.
There are many FM tuner designs using Quad detectors that produce very low distortion, .02% in stereo. And they can sound pretty good. Ever listen to a Rotel RHT-10?
The engineering teams with tons of money and experience at say Accuphase, Pioneer, Sansui, or Kenwood probably achieved a lot more performance with all the detector designs you mention.
I was part of a team in GEC McMichael developing these detectors (part of Marconi), for professional applications, so there was plenty of RF skill, top end test equipment and not much budget constraint
The hard part with PLLs is making a low noise and linear VCO, something JLH failed at badly with his CA3046 based version.
Pulse counters depend on a monostable period being absolutely constant, whatever the input frequency, they are horribly supply sensitive.
I feel that a well implemented quadrature is as good as anything. I don't see much point in going beyond a correctly tuned dual tank and the wideband Murata MX filters
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.