Fix for FM syntheized tuner slightly off channel

I have a Technics ST-Z780 AM and FM stereo tuner. Probably lower end of their product line. In any event, I was getting distortion on FM stereo reception, as if the highs were getting mangled. Mono sounded fine. Noticed that the tuner appeared to be not accurately tuning in stations. Tune a station 100KHz low, and signal was weak, tune 100KHz higher and signal was significantly stronger.

...The following assumes expert knowledge of radio electronics...

Which means that the synthesized local oscillator (uses a LM7001 PLL chip, with a crystal reference oscillator) was a little off, causing the station's signal to get clipped on one side of the IF's passband (a ceramic filter). This clipping will impact the higher supersonic audio frequencies (the stereo difference information is on a supersonic suppressed subcarrier). Thus why stereo sounded distorted, but mono was fine.

attachment.php


I modified the reference crystal oscillator circuit. Used a variable capacitor trimmer (5 to 20pF) to replace a fixed 15pF cap on the osc input pin of the LM7001. This allows me to slightly change the crystal oscillator frequency, which in turn will make the PLL chip's synthesized FM front end local oscillator be more accurate (10'7MHz above the desired FM radio station' frequency) As I twiddled teh trimmer, I'd tune 100KHz above and below a station, to get a condition where the signal strength was equal. And check 200KHz high and low too, both should be equally weaker as well. And of course the station should be strongest on its frequency. You might need to tweak the FM IF receiver chip's quadature detector coil (if there is one), as the IF frequency has moved a little, should now be in the middle of the IF passband.
 

Attachments

  • crystalosctrim.jpg
    crystalosctrim.jpg
    8.5 KB · Views: 789
Another approach could be to put a TCXO on a little board and use that to drive the OSC input directly. TCXOs compensate for temperature change and are available down to better than 1ppm, should you need it. Typically TXCOs come in 5V and 3.3V powered versions in a little screened 4-pin can. The FM synthesizer was probably using a cheap +/-100ppm xtal, leading to the frequency error.
 
Hmm. It is actually uncommon for reference xtals to drift enough to cause a problem. Discriminator coils, by contrast, need tweaking on an almost regular basis, and ceramic filters have a known habit of drifting as they age, too (and differently at that).

So I would be inclined to double-check the main xtal frequency, which I imagine shouldn't be that hard to do if you have a half-decent receiver or transceiver covering 40 meters and a piece of wire as an antenna. For things to be off even by 50 kHz at 100 MHz, we would be looking at a whopping 3.6 kHz offset at 7.2 MHz. That ought to be so plainly detectable it's not even funny. Now with the trimmer in place, zero-beating the thing in SSB should be trivial.

Then recheck how the tuner is doing. Once you know that the XO is in spec, you can try to realign discriminator coil center frequency as outlined in service docs (it should involve zeroing out a DC voltage). You don't need a signal generator for this, most any real radio station will do fine as they are generally required to keep their frequency very accurately.

If any asymmetry remains beyond this point, sweeping the IF or individually taking a filter response using a network analyzer would be required. That could get rather more involved.
 
I measured the reference crystal, and instead of 7.2MHz, it was at 7.30095MHz (I used my Icom 756Pro ham transceiver) Which would make a local oscillator frequency of 110.7MHz (for 100MHz reception with a 10.7MHz IF) be about 15KHz high. Which doesn't sound like much. But I think I had about 40KHz off before, which may have been enough to clip one side of the station's sidebands enough to impair the quality of FM stereo, making distortion. My IF ceramic filter must be off a bit high, as setting the reference crystal to exactly 7.2MHz made reception of FM stereo a bit distorted. I also tweaked the FM discriminator coil. But best reception was found with the crystal at 7.20095MHz and the discriminator reset to that as well..
 
davidsrsb is correct; it's doubtful you could pull a 7.2MHz xtal more than a couple of kHz, especially if you are only adding 5p or subtracting 10p.

A 7.3 MHz reference oscillator will cause the receiver to tune 1.5375 MHz higher than expected, not 15kHz. With a ÷R of 72 and a ÷N of 1107 for a 100kHz channel step and an LO freq of 110.7MHz, the LO will operate at 112.2375MHz for a receive freq of 101.5375 MHz.
 
me:

oops! meant to say 7.20095MHz. :)

That is only 132ppm which is within the limits of your setup and methodology. The best place to measure the crystal frequency is the LM7001’s 400kHz output pin which is derived from the 7.2MHz timebase.

Your original problem is likely the quadrature detector not zeroed when tuned to a strong signal which was already pointed out earlier. The quadrature adjustment is SO sensitive that it is almost always never gets aligned properly off the factory. I have never seen an FM tuner that uses a quadrature detector and measure close to 0V +/- a with few mV. Time is money and factories are time sensitive, they won’t squeeze the last mV when aligning a mass market item. Tuners drift over time but the factory is the most likely culprit more than anything else. An engineer with all the time in the world working on an FM tuner is a different scenario. I fall in the latter category.

You need a decent FM stereo generator and a plethora of tuner related test gear if you want to correctly diagnose a problem and publish your results with a high degree of confidence.
 
How many ceramic filters of what type does this tuner even have? Your average North American market low-end tuner would have been shipping with two 280 kHz jobs (typically Murata SFE10.7MA). These should be so wide that a mere 13 kHz should not matter much at all... they're generally good enough for some half-decent 200 kHz separation but certainly not DX grade. Something may be rather out of whack there.

Possible candidates?
1. There's generally an LC circuit after the mixer, before things enter the IF. This may appreciate being peaked up.
2. Someone may have "hot-rodded" the unit with some narrower filters which may or may not have been decently matched (NWAs and filter test jigs aren't exactly growing on trees). This can get critical once the likes of 110 kHz types are involved.
3. One or more of the filters may have drifted. Theoretically it should be possible to "reset" them using heat (they are high-K ceramic), but I don't know whether they wouldn't require some pre-aging after that. In any case I would consider installing some sockets and getting a bag full of 180 kHz jobs (10.7MS3) to merrily mix and match until you find a good combo.
4. While unlikely, a dead semiconductor in the IF chain could be throwing things off as well, or anything else that substantially changes termination impedances.

For properly sweeping the IF you do in fact want some of the right test gear. However, peaking things up by ear with some ingenuity may already get you close enough.
 
...Your original problem is likely the quadrature detector not zeroed when tuned to a strong signal which was already pointed out earlier. The quadrature adjustment is SO sensitive that it is almost always never gets aligned properly off the factory. I have never seen an FM tuner that uses a quadrature detector and measure close to 0V +/- a with few mV. Time is money and factories are time sensitive, they won’t squeeze the last mV when aligning a mass market item. Tuners drift over time but the factory is the most likely culprit more than anything else. An engineer with all the time in the world working on an FM tuner is a different scenario. I fall in the latter category.
...
I find that the tuning slug in the main coil of my Sony tuner physically moves in the thread over time, due to low level vibration and temperature cycling
 

Tom911

Member
2020-10-21 7:23 pm
wonderfull thread;)


I have a Kenwood KT-2010L Tuner from back 1990.


The device is a LM7001, LA1265, AN7470 based design.



I do have the exact same problem.


When seeking an FM station, it always stops above the specified freqency.
fe. auto-seeking to 103.7Mhz will stop at 103.75 with audio being distorted.
Then manually stepping down to 103.7Mhz will give good FM stereo reception.


As the thread creator stated, it could be the 7.2Mhz crystal for the LM7001 being off.


I do not have total insight of the LM7001(PLL), ...it divides the crystal-F by a factor and outputs a DC voltage to an VCO. The VCO returns a feedback to the LM7001 which it uses to re-adjust its output voltage.


I am wondering if the crystal frequency is such critical....and being the reason (root cause) for the off-drift..????
 
Bright red or pink nail polish has served well in many circumstances, but doesn't always adhere well enough if one of the surfaces is waxed cardboard.

Machinists use a product called Thread Locker to keep set screws in place. It comes in a wide range of permanence, different formulations suitable for use on different materials, and handy little squeeze tubes.

Once you get it dialed in, try it.

Cheers
 

Tom911

Member
2020-10-21 7:23 pm
I set up a "functional overview of the lm7001" :
ieb2I2nUKC4AAAAASUVORK5CYII=



Just to get an understanding of the working.


I manually tuned to an FM stations specified frequency (here 99.2Mhz)
stereo reception was ok.


Then i adjusted the discriminator coil (LA1265) to minimum/0-volt at testpoint with voltmeter.


eh voila...the auto-tune function now stops at the stations specified frequency.


However..then the audio quality is again distorted as not being in center tune.
 

Tom911

Member
2020-10-21 7:23 pm
LM7001 functional overview.

Success;)
This discriminator coil is so sensitive to adjust.
I fine tuned it now veery slowly and everything works.

auto-tune works, stereo is excellent. lovely old piece of equipment;)
 
Last edited by a moderator:
  • Like
Reactions: 1 user