Sunday, March 23, 2014

1942 Philips GM 3155B Kathograph II Oscilloscope

     About 35 years ago, an older gentleman I worked with gave me this Philips oscilloscope.  It was given to him many years before, and that man told him that it came off a German U-Boat.  It is possible, but I can't prove it.  Apparently a GI brought it home as a war souvenir.  This sat among my collection of radios until today, when I decided to see if it worked.  Why did I wait 35 years?  Because I was happy to have it on my shelf, and did not want to risk blowing the rare Valvo tubes.  But  now I need to sell it and other things in my collection, and I felt that if it were operable, I could get a higher price.
     I first had to tackle a problem that I have known about for 35 years, and was one of the reasons I simply put it on the shelf. Someone in the past tried to repair it and made a mess of things.

The Y gain control potentiometer failed, and someone decided to "fix" it with a volume control from an old radio. Trouble is, the volume control had a SPST switch at the full CCW position, while the original had a SPDT switch to change the range of the gain control.  To solve that problem, the repairman put a toggle switch inside the cabinet.  I knew it couldn't be wired right because he had the the wire from the braided shield connected to the switch, when it must go to ground.
At that point, I did not have schematics, so I took my best guess. I also rooted through my collection of potentiometers and found one with a SPDT switch. It was a 2 gang pot, but I will just use one.  I later found out it is the exact value I need, 500k ohms.

I connected the wires, and was ready to try it out.  All I needed was a power cord.  I did not want to solder wires to the mains socket as was done before, so I went back to my supply of electrical hardware. It turns out that an American appliance connector is a close fit for this vintage European connector.

This oscilloscope is very versatile, and can be used nearly anywhere in the world, thanks to it's multi-tap transformer and voltage selector switch.Both sides of the mains are fused. 

I verified I had the right voltage, and plugged it in.
Nothing. No lights, no hum, not even any smoke.
I discovered that the power switch is bad.  Both sides of the mains are switched, and both switches were stuck open.  I soldered jumpers across the switch, for I want to leave the original switch intact.  This way everything is original except for the Y gain control.
I tried it again.  I should mention that I don't just plug it in, rather I raise the voltage slowly using a variable transformer.  This way I can catch a little problem before it becomes a big one.
 Amazingly, all the tubes lighted up, and after perhaps 50 years, the oscilloscope came to life again! Soon I had a trace on the CRT.
I did not know it, but I was soon in for a shock!  A 610 volt shock, to be exact.  See the 4 filter capacitors in a row? Unlike what is commonly found, where the metal cans are grounded to the chassis, these are on insulated bases, and the can voltage is very high!
From left to right, the first one is at chassis ground, no surprise there. However, the second one is at +205 volts. The third one is at -405 volts!  Between capacitors 2 and 3 is a potential of 610 volts!  Cap #4 is "only" -262 volts.  So, if you ever encounter one of these oscilloscopes, do not touch the capacitors!

The two clear glass tubes are 1876 rectifiers, and the gold tube is a 4673 pentode for the Y amplifier.

 Impressively, all of these high voltage capacitors appear to be good.

A view inside the cabinet.
That little rectangle of paper is very important!
That is all that is preventing the capacitors,
and their high voltages, from touching the metal cabinet.

Here is a view of the other side:  The clear tube is a 4890 triode oscillator, and the shielded tube is a 4673 pentode amplifier for the X axis.

Underneath the chassis, it looks like the day it was manufactured, 72 years ago.

 A side view of the cabinet.  The 72 year old decals are still in fair condition:

Finally, here is it in operation for the first time in 50 years.  Most of the controls work well, and the trigger is surprisingly stable.

The trigger is disabled by pulling this jumper out:

 This simple scope does not have centering controls, and I did not see any trimmers inside.

 I have to translate the German language instruction manual to see if there is a way to center the image.  Unfortunately, two of the knobs were missing, and I had to substitute the ones seen in the first picture.

The Kathograph has found a new home at Volker K's Oscilloscope Museum.
Should you have any questions about an old oscilloscope, I highly recommend his website.

Tuesday, March 18, 2014

Lorex Camera Repair

I have a Lorex security camera system that works very well.  However, after a year in operation, two cameras developed blurry areas in the bottom 1/3 of the image, and the colors were not as good as they used to be.  At first I suspected moisture, since these are outside cameras.  But when I disassembled them, I found the day/night filter had come loose and fell out of place. This caused the blur at the bottom.
Here is how to disassemble and repair the camera. First, pry out the three white plugs in the lens, and then remove the screws.  One of the white plugs can be seen under the screw on the left. Then gently pry out the lens. I put a screwdriver in the holes and rock the lens loose, a little at a time. I change holes and rock some more. Do not exert too much pressure or you will crack the lens.

     The lens and electronics come out as a single unit.  There is an O ring around the lens and each screw.  These are critical, even in indoor use, to prevent moisture from fogging the camera.
     Remove the two screws in the black plastic plate, and the lens and infrared LEDs will come loose.  Unplug the LEDs and set them aside.
     Remove the center screw in the white plastic, and the two black screws close to the motor, which is centered above the white plastic. 

     Turn the camera over and lift off the motor and filter assembly. Be careful not to lose the filter!  it may fall out as soon as the screws are loosened.  Notice how the filter has one half that is red when viewed under certain lighting angles.

On my camera, the red half is on the right in the disassembled picture. It is important that it is oriented properly. The filter has a front and back. The small side fits into the plastic frame.  It appears that it was intended to be a snap fit, since I do not see evidence of glue. However I have to now glue mine to secure them in place.  Use very small amounts of glue and do not get any on the lens or the filter.  If you do, the camera will be ruined.

Stretch the O ring over the lens. Try to get it on without it twisting.  It will seal better if it is back on the way it was originally.
Press the lens back into the body, and see that it is uniformly bottomed out. Do not use the screws to pull it in place, or the lens may crack. Tighten each screw a little at a time.  Your camera should now be as good as new.
Here are some tips on identifying the camera models. There are 50 viewing angle and 90 degree viewing angle cameras that look nearly the same.  You can tell them apart by looking at the lens.
      The 50 degree camera is on the left, and has a small lens in the rear, behind the day/night filter. In these cameras, if the filter is loose, you can see it.
  The 90 degree camera is on the right.  It's wide angle lens is close to the front, and the day/night filter is behind it.  You cannot see the day/night filter without disassembling the camera.
The electronics are different, too.

Here is the back of the 50 degree camera.  Prominently missing is the white plastic piece, which is a heat sink.  That heat sink on the 90 degree cameras makes a huge difference in night vision performance.

 Here are pictures taken at night from each camera. They are both looking at my driveway from the same distance. The driveway is barely visible in the picture taken with the 50 degree camera, while it is easily visible in the picture from the 90 degree camera.

Daytime performance of the two cameras is similar, but I prefer the wider viewing angle of the 90 degree camera.

The 50 degree cameras had IR LED boards that could hold 30 LED's, but only had 18.  The 90 degree boards have 18 LED's and no extra holes. I decided to experiment, and I took one of the 50 degree boards and added 12 more LED's. These LED's I mounted at an angle to reduce the bright spot in the center of the night view.  Notice in the picture below how the LED's are angled out.

I installed this board in one of the 90 degree cameras, and it made a huge difference.

Compare this picture to the other night pictures. It is unfortunate that Lorex did not use 30 LED's for they would have a much better product. 

Thursday, March 13, 2014

My Fiber Optic Polishing Machine

I do research with lasers and fiber optics, and I needed a polishing machine customized to meet my needs.  I started with an old Fibertek machine, which could only polish standard fiber connectors.

 I need to be able to keep my parts perfectly vertical, and also have control over how hard the parts pressed on the polishing pad.  I made a guide using 4 hardened stainless posts with linear bearings. A dial indicator measures the polish depth, and counterweights in the rear control the pressure.

This circular cut was made on the mill using a rotary table. 

The accuracy of the rotary table cut approaches that of a lathe.  Here a lathe cut part fits inside with about 0.002" clearance.
 Here is a partially completed fixture that will hold 10 items in the polisher.   The pie shaped pieces will each hold a single part, and they can be removed without disturbing the other parts.  They are held by a single screw at the small end, and a dovetail at the big end.  All of the holes were done with the bolt hole function on my Shumatech DRO.  Here I did not use the rotary table at all, for it was much easier to program the Shumatech.

The outside diameters and the dovetails were cut on the lathe.  Then I cut the upper disc into the pie shaped pieces using a bandsaw, and then I milled the saw cut edges smooth.
 My high powered lathe has no trouble turning these diameters, even in steel.  However, it puts about 300% more stress on the drive belt pulley as does a 7x10 or 7x12 lathe.  The pulley finally failed, so I ordered a replacement from Little Machine Shop.   Here is a picture of how the belt connects to my gear reduction unit.   Someday I will make a metal replacement pulley.

I needed a way to hold an optical fiber connector, known as an MT connector, very accurately, and with no free play.  The MT's measure 3mm high x 6 mm wide.  Here is one in a rectangular hole I made with tolerances of 0.1mm

I began by milling a slot slightly smaller than the MT's dimensions.  Then I drove a rectangular punch I made from a hardened steel bolt into the slot.  Even though my method was crude - I drove the punch in with a ball pein hammer - I had good repeatability from one part to the next.