{Collins} 75S-3B Band Switch - The Good, the Bad and the Ugly :)

Greetings to the List -

Last weekend, I put out a plea for help on cleaning the Band Switch on the 75S-3B Receiver that I am (hopefully) bringing back to life.

Special Thanks to Steve/VK4VN, Tom/K0EOA and Mike/K9LSH for excellent insights into that switch.

I have the Band Switch now under control and learned some interesting things.

(0) You cannot see the Band Switch wafers until you remove the three Shield Cans. And you cannot remove the Shield Cans until you remove the drive rod that connects all of the wafers to the Band knob on the front panel. Loosen the Bristols on the coupling and move the rod with your fingers toward the rear. Give a gentle tap on the rod and it will pop a bit out the small hole on the back of the chassis.

(1) After removing the Shield Cans (photos) http://frobenius.com/75S-3B-Band-Switch-Beginning-with-Three-Shields-On.jpg that surround the switch wafers, the drive rod - made of some sort of phenolic/fibre - is quite easy to remove at least in my case once you loosen the Bristoled coupling http://frobenius.com/75S-3B-Band-Switch-Bristoled-Shaft-Coupling.jpg I was afraid that the drive rod might be gummed up or age-welded to the wafer causing me to break a wafer. But, that was not a problem. The drive rod was quite loose and simply came out through the small hole on the back of the chassis. Hooray! By the way, be very careful to not over tighten the nuts when putting the Shield Cans back on - not much more than finger tight - it is very easy to bend the tabs on the Cans.

(2) Before removing the drive rod, make a small mark on the rod near the coupling so you can get things more or less straight when reassembling. However, this is not nearly accurate enough to make things align up correctly - more on that in a bit. Here http://frobenius.com/75S-3B-Band-Switch-Shaft-back-in-for-Wafers-Cleaning.jpg you can see the small dot that I made on the drive rod up close to the Bristoled coupling after I removed the rod; removed the Shield Cans to expose the switch wafers; re-inserted the rod through the wafers to enable me to rotate the Band Switch; and then tightened the Bristols back. Of course, be careful to not rotate a wafer with no rod inserted more than a tiny amount.

(3) The four exposed Band Switch wafers look like http://frobenius.com/75S-3B-Band-Switch-Shaft-and-Shield-Cans-Removed.jpg and, especially, http://frobenius.com/75S-3B-Band-Switch-Wafer-Supports-with-Center-Rod-Removed.jpg - What a work of Collins art! Now, just slide the drive rod back in through the wafers and into the Bristoled coupling and tighten that so you can rotate the wafers with the Band Knob after applying DeOxit to the switch contacts.

(4) "Fun Video" of the Band Switch rotating to clean: http://frobenius.com/75S-3B-Band-Switch-Wafers-being-Cleaned-after-DeOxit.mp4

(5)  Now comes the most important thing that I learned:

This is a bit difficult to explain verbally, but each of the five bands (80 - 10M) has its own set of tuned circuit elements (see schematic) and, therefore, there are five sets of "things" switched in/out for each of the 14 Band Switch positions. For example, on V3 Crystal Oscillator, the first wafer up close to the chassis front and not shielded by a can switches in the correct crystal - e.g., a 8.5775 Mhz xtal for 14.0Mhz band. The second wafer (shielded) switches in for the 14.0Mhz setting C74 68pf and a parallel C73 8-50pf trimmer, which with slug tuned T2 (PRESELECTOR) forms, I think, a tank circuit for the Crystal Oscillator. That same C74/C73 combination is used for all three 20M positions (14.0Mhz, 14.2Mhz, 14.8Mhz).

Point of all that is that the five sets of capacitors for the first shielded wafer are used in the 14 Band Switch band settings meaning that Set #1 of capacitors C70/C69 is used for the three 80M positions; Set #2 is used for the two 40M positions; Set #3 for the three 20M positions; Set #4 for the three 15M positions; and Set #5 for the three 10M positions.

Therefore, the switch contacts on the Band Switch are not evenly spaced and Collins implemented this by having, for example for the three 14Mhz positions, the tab touches one contact, touches the other contact and then sort of straddles between the first and second contact - and that is the design weakness. It would have been far better to have had 14 contacts rather than just nine, but wafer switches were/are quite expensive.

This means that the exact rotation angle of the drive rod and therefore the wafers is very sensitive. If off by even a tiny amount, the wiper tab will not touch the correct contact and you will end up with no C74/C73 capacitor for the 14.0Mhz tank circuit for the Crystal Oscillator.

The way to deal with this, I found, is to use an ohmmeter and the schematic to identify the correct position of the tab for each of the 14 band selections - fiddle with the rod angle until all 14 positions touch the correct C/C combination - it is quite tedious to get this correct - and the ohmmeter is the only way I found to make this work. You can tell which set of capacitors is which by reading the values of the Micas and comparing to the schematic (however, C120 has wrong value as per the schematic - schematic says 15pf but is actually 120pf).

Mark very carefully the rod to record the exact position with a very thin marker.

The real hat-trick is that to reassemble, you must get everything correctly positioned; pull the rod out; replace the shield cans; put the rod back in; and then tighten the Bristols - but, with the Shields on you cannot see the wafers! :)

Check with the ohmmeter after the re-assembly that all 14 band positions cause the correct C/C combination to be switched in - it is "tedious" :)

Once all that works, then put a very small shim (VK4VN says use a small piece of razor blade) along the edge of the rod in the Bristoled coupling so that when you tighten the coupling the Bristols are at 90-degree angles with the rod - otherwise, the rod will shift over time.

The mistake that I made originally was that I assumed that the Bristoled coupling is machined to exactly fit the non-round shape of the rod so that alignment of the wafers would be forced - it is not! The coupling has simply a nice round hole in which the more or less rectangular rod happily rotates. The rod is sort of rectangular in shape so as to drive the wafers.

I see no other way to do this mechanical alignment - I am all ears for a better less tedious way.

(6) What prompted all of the above - to "simply" clean the Band Switch - was that output of the Crystal Oscillator to input into the First Mixer V3A was intermittent. Many bands were totally dead - I could slightly wiggle the switch and a band would come alive - or not.

In addition, some of the bands that were alive showed only about 0.2V RF from the Crystal Oscillator and not the 2.0V in the manual - and so I descended into this morass.

The wrong 0.2V came from, I think, the fact that the Crystal Oscillator in those sort of alive bands had no C/C combination switched in and, therefore, no tank circuit.

On the Spectrum Analyzer, I typically saw 0.2V (on the 'scope) as White Noise running up to 100Mhz+.

After cleaning the Band Switch contact and aligning the wafers mechanically so the correct C/Cs were switched in, I now see the correct 2V RF from the Crystal Oscillator in all positions... ...Life is good!

(7) ...except in the 14.0Mhz position. There, I would usually see a very low frequency signal that after a second would change to something that looked reasonable but with considerable jitter; or I would see some sort of White Noise gibberish; or I would see nothing - all that randomly.

I could visually see that the wiper tab was flexing the contact and the ohmmeter showed that it was switching in the correct C/C - C74/C73 in the 14.0Mhz case - but, weird stuff was coming out of the Crystal Oscillator.

I thought, of course, a dirty switch contact - I re-cleaned - no.

How could this be - the three contacts 14.0, 14.2, 14.8 are all wired together - I then tried jumpering around the 14.0 contact and got exactly the same thing - how is that possible?

Then, I remembered that there is a variable - the different crystals being switched in by the front wafer.

Bad crystal perhaps - I checked and to my amazement, the 14.0Mhz crystal 8.5775Mhz is simply not there - just two holes! Sort of ironic as 14,005/CW - the only dead band - is my usual hangout in all of the bands :)

I did not know that a crystal controlled oscillator with the crystal removed will, in fact, oscillate - I suppose, frequently, at or roughly at the frequency of the tank circuit - or generate White Noise - or nothing at all randomly - an interesting phenomenon.

So, after a lot of tedious stuff, the Band Switch and the Crystal Oscillator now work correctly. I ordered an 8.5775 Mhz Collins XTAL today from eBay.

But, what about the other two shielded wafers on the Band Switch - front-end tuned circuit with T1 on the antenna input and the L2 output of the RF Amplifier?

At least on my Receiver, when I got the first shielded wafer aligned, the other two were also aligned. Hooray! The other two also have only one wiper tab and the exact same contact configuration with five sets of "things" switched in an identical fashion complete with required tolerance among the 14 positions.

Lesson Learned: When you clean the Band Switch, do not depend on just a rough marking on the drive rod. That will not suffice. Use an ohmmeter to adjust the tweaked rotation of the drive rod until all 14 contacts are correctly connected to the correct "thing" set. Shim between the rod and the Bristoled coupling hole to fix the rod so that the Bristol screws tighten down onto the rod at a 90-degree angle so the rod does not drift over time.

I do wonder if the center holes in the wafers have slowly worn and slightly widened with increased play over the past sixty years...

I learned a lot - and that counts - a lot of fun - a couple of Rum & Cokes helped :)

Best to the List -

Jack, WØYJ
Evergreen, Colorado USA

Jack Harper, President
Secure Outcomes Inc
2942 Evergreen Parkway, Suite 300
Evergreen, Colorado 80439 USA

303.670.3750 (fax)

http://www.secureoutcomes.net for Product Info.

This archive was generated by a fusion of Pipermail (Mailman edition) and MHonArc.