Skip to main content

Beaverton, we have a trace!

·6 mins

Buying #

Got my hands on a vintage Tektronix 464 storage oscilloscope. These were made between 1975 and 1985, so this unit was probably manufactured before I was even born.

It’s an analog storage scope - a thing of beauty and a true marvel of engineering.

This one is defunct (no trace), “sold as-is” for 100 ILS (~$30). From the seller photos it looks like the tube and HV supply are probably healthy, which is a good start. Some broken knobs are also visible, but that’s it.

Seller photos. Sold “as-is”.
Seller photos. Sold "as-is".

I’ll start with the usual LV supply checks (seller thinks 15V supply is bad, but who knows).

Unfortunately I can’t fully test CRT and HV supply until LV supply is good. Most components on these scopes can be replaced, but problems in specific parts (tube, HV transformer, HV multiplier, tunnel diodes, attenuators, and custom ICs) may bring this project to an abrupt end.

Fixing the low-voltage rails #

Taking it from the top, I checked the LV rails. All checked out, except the +5V which was reading 0.3V, and the -15V which was all over the place.

Looking at the schematics, I noticed the -15V rail relies the +5V rail, so it might just work if I fix that one first.

-15V needs +5V
-15V needs +5V

Since the +5V rail was reading low, I had two theories:

  1. Control circuit isn’t regulating; or
  2. Shorted downstream consumer that’s dragging down the rail.
Original board state. Note the botches if the left and right most transistors.
Original board state. Note the botches if the left and right most transistors.

The second theory is very annoying to test, as I had to chase down every consumer in the scope and/or start disconnecting boards to rule them out. So - being lazy, I started checking regulation, as the circuit seemed pretty simple.

Sure enough - after a bit of probing, I found a 2N2222 that’s decided it wanted to be diode.

I didn’t even have to solder anything - it was socketed. Replacing it brought back the +5V, and the -15V rail along with it.

A trace shortly followed, and the scope is even triggering - an encouraging sign that the tunnel diodes, tube, and HV supply are functioning!

After the trace appeared I started by testing in XY mode as it doesn’t rely on
triggering
After the trace appeared I started by testing in XY mode as it doesn't rely on triggering
Channel 1 is triggering!
Channel 1 is triggering!

While hanging around in the LV supply I replaced two TIP31 botches and replaced their crumbling isolation plastic with new mica pads. I also re-applied new thermal compound.

While examining the boards I found a disconnected ground wire between the main board and the vertical amplifier, which solved a minor noise issue.

Next steps are to clean with IPA all the mechanical switches, and start running through the calibration procedure to get it in-spec and find other problems.

I am not going to re-cap the thing, or replace the notorious carbon resistors/tantalum capacitors just yet. I will reconsider if calibration fails.

There are multiple botches on the board, but comparing them to the schematics
they all seem correct. I might touch-up some solder joints though.
There are multiple botches on the board, but comparing them to the schematics they all seem correct. I might touch-up some solder joints though.

All-in-all I am pretty optimistic about this project.

50 years old crumbling plastic #

After getting a trace on the Tek 464, I went ahead with cleaning up the vertical knobs and buttons.

Note to self - when cleaning up 50 years old plastics, go easy, or they crumble.

This time is was the overly-elaborate mechanism of the TRIG VIEW/BW LIMIT button. The plastic holder mounting got broken. Access to this part is horrible, and I had to disassemble the vertical amplifier board (yikes!). Twice.

Horrible access
Horrible access

In the first time I tried to super-glue the part back to its mounting pins that were still mounted on the PCB. This was challenging as this part holds another sliding plastic part, so I had to apply pressure with one hand and keep moving the slider with the other so it doesn’t stick. After 20 mins or so I though I was done and reassembled the board - only to discover the next day that pressing the button breaks the part again, as the super-glue didn’t catch. Eye roll.

For the second time I took no chances. Metal holds better than plastic. I filed down the plastic so it clears all the mounting holes and drilled out the plastic studs from the PCB.

I then fabricated an aluminium piece with a slot to go over the plastic part and hold it down. I used nylon screws as the holes, for some reason, were tinned.

I had to enlarge the PCB holes as they were just under 3mm, which is the screw size I used.

Luckily I had enough clearance from any trace near the holes

It isn’t beatiful, but it holds
It isn't beatiful, but it holds

With that I proceeded to complete the vertical calibration successfully (after cleaning up the contacts of the CH2 X10 attenuator). The scope did fail the BW limit test by ~6MHz, but I suspect my function generator as I failed the same test by the same margin on a 465 I previously calibrated with it.

Soapy water woes #

To wash out all the gunk I submerged the knobs and buttons in some soapy water.

Little did I know that in addition to the gunk it will also remove the label ink…

Now what? Do we print new labels? A short search in TekScopes@groups.io reveals this is probably Univers Condensed, so at least that’s an option.

But in the mean time I scvanged buttons from a Tek 465 carcass with a dim tube.

Maybe a future project could be printing new labels on photo paper with toner and heat-transfer them to the plastic.

50 years old crumbling plastic - part 2 #

Horizonal calibration quickly went sideways. The scope’s TIME/DIV knob controls both the A and B sweeps. In the regular state when you turn the knob both timebase are locked together, but if you pull the knob out a bit you can further adjust just the B TIME/DELAY.

But that mechanism was broken. I mean, literally broken.

Timebase board
Timebase board

The B delay is adjusted by an internal rod going through the main rod. When the knob is pulled out a small shim attached to the inner rod engages with the last barrel and allows you to rotate it individually.

That little shim was crumbling, and was slipping over the inner rod, preventing rotation of the barrel and chage of B timebase.

Disassembling these boards was scary. Everything is so delicate, and if broken replacements are impossible to find.

As if I didn’t learn from last time, I attempted to super-glue the crumbling plastic back to the rod.

After waiting a day and reassembling the scope, I continued the horizontal calibration. It wasn’t long before the super-glue snapped, and I had to disassemble the whole thing. Again.

Super-glued the shim unto the rod. You can guess how that turned out.
Super-glued the shim unto the rod. You can guess how that turned out.

I ended up 3D-printing a new shim. For some reason I can’t find pictures of that. It was a really tiny piece and the 3D-printer’s accuracy was no where near what’s needed - so I oversized the print and spend a good amount of time filing it to size.

With that, I was able to complete horizontal calibration.

A and B sweeps work in different timing using the new 3D-printed shim
A and B sweeps work in different timing using the new 3D-printed shim