Need help with Mitsubishi DA-F20


Joe D
 

Looking for someone to repair/align my DA-F20. I brought it to one person who was recommended but the alignment results were less than stellar. I got the impression he did not have a lot of experience with the Mitsubishi although he had a lot of tuner experience. The unit appears to pull in stations ok BUT it does not stay in stereo at all and often mutes itself even on strong signals. So I am looking for someone who has at least some experience with the DA-F20 in particular.
Anyone?


newaag
 

Did the the tech who worked on it have the proper service manual? Always ask that question first before sending gear. If not, I can see why it could be odd. There are two detectors on that one.

Sounds like a simple detector stereo VCO alignment per the manual is needed. Unless it is sick...
If he had the manual and sent it back like that, tell us privately who did, if we recommended at TIC or in this group. I don't do this work for others...
Bob


Joe D
 

Well, he is recommended on TIC. I assumed he had a manual since I can get one myself on hifiengine. In fact I think I mentioned that. He even had another DA-F20 in for service so, at that point, I didn't ask if had worked on them before. I made a bad assumption at that point.
Wish I could figure out how to send a private message on this board.


newaag
 

Would anyone like to see more info on setting up the DA-F20 correctly? I have one open here, and can document a few of the more subtle, but very important settings, if anyone needs them. The service manual could be better...but it is correct as written.  
Bob


Rocker
 

Yes please! I have one here but I have not used it for a while as it seems to be a little too sibilant for my liking! I really need to either move on from it or, improve it! Decisions, decisions! Thanks Bob cheers… Rocker

Sent from my  Space gray  iPhone 11 Pro Max

On Mar 1, 2020, at 9:11 AM, newaag via Groups.Io <newaag@...> wrote:

Would anyone like to see more info on setting up the DA-F20 correctly? I have one open here, and can document a few of the more subtle, but very important settings, if anyone needs them. The service manual could be better...but it is correct as written.  
Bob


Robert Chambers
 

On Sun, 1 Mar 2020 09:51:42 -0700, Rocker <rocker@...> wrote:
Yes please! I have one here but I have not used it for a while as it
seems to be a little too sibilant for my liking! I really need to either
move on from it or, improve it! Decisions, decisions! Thanks Bob
cheers… Rocker
Hi Bob,

I would echo what Rocker said, as I have one languishing. When mine was
in-system, I found it took some time to stabilize tuning wise...I found it
easier to just leave the unit on.

So I would be interested in learning what you find re setup/refurb/mods.

Thanks and Regards,

RC in VT

Sent from my Space gray iPhone 11 Pro Max
On Mar 1, 2020, at 9:11 AM, newaag via Groups.Io wrote:

Would anyone like to see more info on setting up the DA-F20 correctly?
I have one open here, and can document a few of the more subtle, but
very
important settings, if anyone needs them. The service manual could be
better...but it is correct as written.
Bob _._,_._,_
-------------------------


Rick Price
 

I don't have one (yet), but sure...

Rick P.


newaag
 

I'll post some comments here on the DA-F20 in a few days, when I can take a few pictures and do it all at once.
It will be a new topic.
Bob


Rocker
 

Cheers and thanks Bob...Rocker


On 3/3/2020 1:25 PM, newaag via Groups.Io wrote:
I'll post some comments here on the DA-F20 in a few days, when I can take a few pictures and do it all at once.
It will be a new topic.
Bob


newaag
 

I've been sidetracked a bit on this one, but here is what I can report so far -
The unit I'm working on has a low IF frequency, about 30 kHz below 10.700. This is somewhat tolerable in wide, but not so good in narrow, where you only have 150 kHz total bandwidth to play with. Making things worse, the filters fitted to narrow were centered at 10.72 MHz. So yeah, it was not optimal...
So what determines the IF ? For this tuner in particular, it appears to be the digital tuning / PLL that pulls the local oscillator with a variable capacitor to "lock" onto the station. Testing shows the 10 MHz oscillator that drives the whole thing is running slightly low. So I'm now working on re-tuning that, one way or another, and it likely may involve new parts.

BTW, this sort of thing also affects most later tuners made after 1980 or so, that use an internal oscillator to "lock" onto the station. I have discussed this before, and so have others. One way to solve it - change the value of the small load capacitor on the oscillator crystal. But it depends on the error (in PPM), as a crystal can only can be adjusted so far...  In most cases, you can go +/- 100 PPM. In this case, the error is almost 300 PPM, so we'll see. You can also simply replace the crystal and capacitor with a new tightly specified parts, which I will likely do, if the existing one can't be adjusted. 

The other way to solve this problem - fit new ceramic filters that are tested / sorted, and match the low IF frequency. In this case it is ~10.67 MHz (I'll show how to measure it later). The problem with this approach is a very practical one - you need a supply of filters, and means to test them. But my experience has also shown - even if you can do that, most filters, if they deviate from 10.70, test high, not low. Finding ones that test low, especially minus 30 kHz, is very difficult. You may get 1 out of 100, making it really hard to find 3 or 4 filters....and what to do with all the rest of those filters ? So adjusting the IF / digital circuit makes more sense.

As you can see, there is some thought that needs to be applied to solving this problem, it's not just a case of "re-capping" the unit...but I'll share all the details later, once the job is done.
Bob 
   


newaag
 

One more point on possibly why the sorted 10.72 MHz filters were fitted to this DA-F20. It is possibly due to confusion about something that can happen in a mixer, called "spectral inversion". This effect take place when the LO is higher than the RF tuning frequency (also called high side injection). To recall, a mixer will produce IF outputs that include both sum and difference frequencies. It is the job of the IF filters to get rid of the unwanted signals from the mixer output, and only use one frequency. Most FM radios use a higher LO than RF, such that when tuning LO - tuned RF = 10.700. More on this here, they explain it way better than I can.
http://www.rfcafe.com/references/electrical/spectral-inv.htm

It can be confusing, depending on how you measure the IF error, to know how to apply the fix, i.e. select the right ceramic filters. The best way to do this, to avoid the confusion, is to measure the IF frequency directly, by probing the last IF stage (10:1 high impedance scope probe), using a strong RF signal, with no modulation applied.  In this case, you should see a pure sine wave, which can hopefully be measured directly on a frequency counter.        
Otherwise, you may end up thinking you need 10.73 MHz filters, when you really needed 10.67 ones, of course, making the problem even worse if you fit the 10.73 ones...
Bob


newaag
 

Quick update for those following along. Briefly, I removed the load cap on the crystal oscillator, and replaced it with an adjustable one, 3-22 pf. This cap looks like a very mini tuner gang with interleaved plates.
I adjusted it until the frequency at pin 56 of the digital board was exactly 1.25000 MHz. This moved the measured (at R215) IF frequency from 10.673 to 10.700. Nice!
more on what is going on -
The signal at pin 56 is the output of a divide by 8 counter that is fed by the 10.00 MHz crystal oscillator. It was originally 1.24967 MHz, or an error of 330 Hz. This 1.25000 Mhz clock is further divided down to 50 kHz, and then used in a phase locked loop (PLL) to sample the output from the local oscillator, to create an error signal that than feeds/tunes the local oscillator voltage tuned capacitor "gang".

This PLL feedback circuit, in essence, is supposed to lock the station on the tuned frequency, but also keep the IF at exactly 10.7000 MHz. Any error here is multiplied by the ratio of the local oscillator over this 1.25000 MHz clock. I was working at 98.00 MHz, so the LO was at 98.00 + 10.7 = 108.7 MHz. And (108.7 / 1.250) X 330 Hz error = ~ 28 kHz, which is almost exactly the IF error I measured. So by making this error zero, I was able to bring the IF back to 10.7000. 

Bob 


rick.savas@...
 

Nice, thanks for sharing

On Wed, Mar 11, 2020 at 6:34 PM newaag via Groups.Io <newaag=yahoo.com@groups.io> wrote:
Quick update for those following along. Briefly, I removed the load cap on the crystal oscillator, and replaced it with an adjustable one, 3-22 pf. This cap looks like a very mini tuner gang with interleaved plates.
I adjusted it until the frequency at pin 56 of the digital board was exactly 1.25000 MHz. This moved the measured (at R215) IF frequency from 10.673 to 10.700. Nice!
more on what is going on -
The signal at pin 56 is the output of a divide by 8 counter that is fed by the 10.00 MHz crystal oscillator. It was originally 1.24967 MHz, or an error of 330 Hz. This 1.25000 Mhz clock is further divided down to 50 kHz, and then used in a phase locked loop (PLL) to sample the output from the local oscillator, to create an error signal that than feeds/tunes the local oscillator voltage tuned capacitor "gang".

This PLL feedback circuit, in essence, is supposed to lock the station on the tuned frequency, but also keep the IF at exactly 10.7000 MHz. Any error here is multiplied by the ratio of the local oscillator over this 1.25000 MHz clock. I was working at 98.00 MHz, so the LO was at 98.00 + 10.7 = 108.7 MHz. And (108.7 / 1.250) X 330 Hz error = ~ 28 kHz, which is almost exactly the IF error I measured. So by making this error zero, I was able to bring the IF back to 10.7000. 

Bob 


newaag
 

I will go ahead and order new oscillator parts and install, as I don't consider this job done. Why? 

I removed the adjustable cap for the 10 MHz oscillator, and it measured 10 pf. It is supposed to be 20 pF. 
This is probably right on the edge of even working as an oscillator. It is pulling the crystal about 300 PM higher.

Normally, you can adjust the load cap for a crystal +/- 100 PPM, at best, that is all designers recommend. At some point, it just stops working. At the edge of oscillation / not working, stability and jitter get much worse.

So in this case, it makes sense to simply replace the crystal, and load cap, bring them back to stock values with new parts. For the load cap, it need a 20 pf C0G spec part. For the oscillator crystal, a decent +/- 30 ppm crystal (or better). They are so cheap, it's best to order 10 of each, plus caps +/- 2 pf form 20, to fit the best combination.
The other option is to install a tightly specified TCXO oscillator, and eliminate the on-board one. This is more work, but for those with the chops, not too hard. Most of these are now CMOS running at 3.3V, which means it needs to interface with old tech TTL digital parts. Not sure if this would be a problem or not. So to keep it simple,going with parts replacment as stock. But you can also buy TTL 5V units, but they are either really expensive ($50+) or not TCXO, just can oscillators, and not tightly spec'ed, with no adjustment.

Some may wonder why this problem happened? Was it like this since new?
Probably not. Just like people, these crystals, depending on how they are made, "slowly age" and drift over time. Most data sheets say 5 PPM / year. Well this one is 40 years old!
If it was off +/- 100 PPM to begin with (not unusual), it could be normal aging, and alas, in 2020, off by almost 300 PPM.
Which is why I am documenting all this stuff. There may be other decent tuner units out there (digital tuned) that could use this cheap fix. This sort of instruction (like cleaning gangs) will NOT be in the service manual. :-)
Bob 
To be continued...

 


newaag
 
Edited

I wanted to update this topic with more information. Here's a quick recap of the problem. The tuner, a DA-F20, did not play in stereo, and reception was poor.
A quick alignment showed the both detector zero adjustments were way off. The narrow detector on this unit is used to enable both muting, and also stereo reception. If that detector is not at zero, stereo will not engage, regardless of any other settings (muting off, etc). So this was a quick fix.
But there was more wrong - the reception sensitivity was poor, and distortion measured high.
It turned out the crystal oscillator was off specification, (as per previous messages), tuning the IF to 10.67 instead of 10.70. It appears the previous tech who worked on this unit also saw this, and tried to correct it, but installed the wrong ceramic filter set in narrow (4 filters). They were marked, in handwriting, 10.72 on each filter, so they were obviously tested and sorted before installation. But the correction went the wrong way - 10.67 filters should have been used.  
It was likely an easy mistake to make, though. I frequently do a very low RF signal test, in IF narrow. Then tune the RF signal generator slightly above and below the station frequency (say 98.10 MHz), to see if noise goes lower and sensitivity improves slightly off frequency. It is a good test for crystal oscillator based tuners. 
For this tuner, initially the sensitivity improved, with lower noise, with the stereo generator tuned about 30 kHz above the actual target station - in this example, it was best at 98.13 MHz. 
This could give one the impression that the IF filters need to be moved "up" in frequency from 10.70. In fact, if the actual IF is measured, it would be where the oscillator told it to be, at 10.67. This is due to frequency inversion, as noted in previous messages. But...
The "fix" may not be using new narrow filters selected for 10.67 MHz. Because one needs to measure the other IF (wide) IF path, and see where it is "centered". In this case it was measured, and found to be correct, at 10.70 MHz. When going to all this trouble, you want the IF path centers to "match" each other, AND also match the oscillator. And in this tuner, the wide LC filter is a closed box and non-adjustable. So the wide IF drives which filters you select for narrow, and where you adjust the crystal oscillator.
Long story short - a new 10.000 MHz crystal was installed, and the 20 pf cap was selected so it ran exactly on frequency. Then 4 new narrow filters were fitted, all selected for 10.700 MHz. 
Then the tuner was completely re-aligned - this is a MUST after moving the IF frequency (from 10.67 to 10.70) and fitting new filters.
These changes completely transformed the tuner - it now is both sensitive, selective, and tests very low for distortion in both wide and narrow, with excellent stereo separation (over 50 dB in both wide and narrow).
That is basically the end of the story. I will document the alignment "how to" specifics elsewhere at a later time. The service manual could be a lot better for this unit.
Bob