If you are the proud owner of a Tektronix 2445 / 2465 oscilloscope, you are probably aware of the infamous U800 horizontal output IC that so often fails in these fine instruments. The cause of this failure, among other things, is insufficient cooling usually caused by a bad fan, so if you have a 2445/2465 with fan trouble you MUST get the fan fixed before continuing to use the oscilloscope. However, depending on exactly what model you have that may be easier said then done. If you are lucky enough to own a 'letter model' oscilloscope, that is a 2445A, 2445B, 2465A, or 2465B, then you are in luck - your oscilloscope uses a standard square computer fan and so it is trivial to get a replacement. However, if you have one of the original models (2445 or 2465, no suffix) then you are in for a project. You see, these models used a sophisticated hall-effect sensor controlled motor designed by Siemens (P/N 1ad3001-0a) which is now totally unavailable as is the custom Tektronix-designed fan assembly it is used in (P/N 670-7390-00 or 670-7390-01). Worse, due to the fact that this fan assembly uses a squirrel-cage type impeller, you can't just substitute a standard computer fan without somehow getting a hold of a rear plastic cover from a letter model oscilloscope.

That leads us to the project of rebuilding the existing motor. You would think this would be easy - just take it apart, replace the worn bearings, reassemble, done - and you would be right if the motor was any normal motor design. But it is not. You see, the only way to get at the bearings is to remove the windings! Yes. That is correct. You have to rewind the motor to replace the bearings. I have succeeded in rebuilding one of these and plan to offer a rebuilding service in the near future. I will describe the process in detail below, and you are welcome to attempt it, but be warned. This is not for the faint of heart!

---Here we go---

The first step is to remove the power supply from the oscilloscope and then remove the fan motor assembly from the power supply. This is detailed in the manual and so I won't go into it here. Just be very careful when removing the fan impeller. The collet used to hold it to the shaft is VERY easily broken and the part is not available (although I am working on a replacement - stay tuned.) Once you have the fan motor assembly removed from the oscilloscope, unsolder the fan motor from the board. The motor has thirteen pins (yes, that is 13!) soldered directly into the board, so the only way I know of to get it out is with a vacuum de-soldering station. YMMV.

Once you have the motor out it will look the picture below. To start disassembly, use a small screwdriver to un-crimp the aluminum from the top and bottom end caps. Then, using a pick, remove all the gray adhesive from the hole in the bottom of the motor housing. (Note that I will refer to the end of the motor where the shaft comes out as the top and the end with the pins as the bottom thought this discussion.)



With this done, pry off the top end cap and place it aside. Take a flat piece of material, thicker then the length of the motor shaft, and drill a hole in it just big enough for the shaft of the motor. Place this piece on a firm surface and put the motor shaft into the hole. Then, grip the body of the motor and press it down against the material. This should force the armature and bottom end cap part way out of motor casing. Once the process is started as described, it should be easy to complete by hand. However, DO NOT PULL ON THE BOTTOM END CAP. Rather push the armature out from the top end. With the outer casing thus removed, the motor will look this this:



Take this opportunity to label the winding supports as shown. This is essential for rewinding the motor later. Use permanent marker and touch up the labels if they start to wear off later in the process. After labeling the winding supports, remove the thin strip of tape holding the hall-effect sensors / bottom end cap to the armature. Do this slowly and carefully to avoid damaging the very delicate hall effect sensors. Save the tape for reuse later by sticking it to the inside of a Zip-Lok bag. With these steps done, you can carefully pull the bottom end cap off of the armature. If you removed all the gray adhesive, it should come off without breaking. The armature should now look like this:



Take a moment to compare your motor to the wiring diagram at the bottom of this document to verify that you understand how to rewire the motor, then unsolder or cut all the wires off of the terminals. Once you have done this, carefully remove the tape covering the armature, again saving it for re-use. The armature will now look this this:



Decide at this point if you want to save the winding wire for re-use. Doing so will take more care in unwinding the motor, but will save tracking down the correct replacement wire. This is what I did. If you do choose to replace the wire, you will need to source two rolls of #34 magnet wire of different colors. Originally they used red and clear. Once you have made your decision, remove and save the strip of tape holding the loose end of the winding in place and unwind it. You shouldn't need to count turns as I have already done that for you. There are 135 +/- 1 turns per winding. After removing the first two windings (they are wound together with two parallel continuous strands of wire) your armature should look like this:



Now remove and save the last piece of tape and remove the last two windings. Once you have done so, the two halves of the plastic housing around the permanent magnet rotor will separate and the rotor can be pulled out. The bearings are held in with a metal clip with four tabs. Each of these tabs must be depressed at once while rotating the clip counter-clockwise to remove it. I found that a nut-driver does a nice job of compressing all the tabs at once and that there is enough friction between it and the clip to rotate the clip for removal. See the picture below.



With the clip removed, the bearings can be removed for replacement. The bearings themselves are oil-impregnated spherical bronze bushings. I have not found anyone who stocks these parts, but I was able to get a small run of replacements made by LM-Tarbell (lm-tarbell.com). They are a really helpful and friendly company which I highly recommend. PM me if you want to buy a pair. If you want to have some made for yourself, the dimensions are 4.45mm diameter, 3mm length, 2mm ID. Below is a picture of the new bearings next to the old:



Before installing the new bearings, be sure to clean all the parts carefully, including the felt oil washers. Here I am soaking them in isopropanol to remove residual dried oil.



Install the new bearings the same way as the old ones were removed. For reference, here is a photo of the bearing, washer, and clip together ready for the clip to be installed with the nutdriver.



That completes the replacement of the main bearings, but the thrust bearing probably also needs attention. The thrust bearing is a modified set-screw installed under all the gray adhesive in the hole in the bottom end-cap. With the adhesive gone, it can simply be unscrewed with a small flat-blade screwdriver.



Once it is removed, check it for wear which will show up as a worn dimple in the middle. Mine was pretty bad so I decided to lap it flat again. Here is the bearing before lapping:



To lap the bearing, I made a small jig out of Delrin plastic. The jig just consists of a 'puck' of the plastic with a hole drilled in it which is just barely smaller then the OD of the setscrew. This way the setscrew can be threaded into the hole and will bind enough to stay put. To lap the part, the setscrew is advanced such that its end just barely protrudes through the bottom of the puck. The bottom of the puck is then polished against a piece of sandpaper placed on glass (to assure a flat surface) until the setscrew is lapped down to the level of the bottom of the puck at which point the setscrew is advanced a little bit and the process is repeated. See the picture below.



Once there is no more evidence of the dimple, the sandpaper is replaced with a fine polishing stone and the end of the setscrew is polished to an almost mirror finish:





In a perfect world, the end of the shaft which meets the thrust bearing would also be resurfaced, but I don't know of an easy way to do this which preserves the correct hemispherical shape. Thus, this is as far as we can go. Reinstall the thrust bearing. Before reinstalling the rotor, check for and remove any rust or dings on the shaft which might damage the new bearings. Also, be sure the bearing surfaces on the shaft are clean and have an almost mirror-finish to them. Then wet the oil washers, bearings, and shaft with just a drop of motor oil. I think 50W is about right for this application. See the photo below. Reinstall the rotor (make sure no filings are stuck to it - it is a magnet) and reassemble the two halves of the plastic rotor housing. You are now ready to rewind the motor!



Rewind the motor according to the diagram below. Reinstall the tape in the same places you took it off of - one piece to hold the loose ends of the first two windings and another to hold the loose ends of the second two windings.



Due to stretching, if you reuse the old wire you will have a bit of extra. Put on a couple extra turns or cut it off - it doesn't really matter. When finishing a winding, make sure the loose ends end up toward the outside of the end of the armature as shown in the picture below. They should not end up coming out near the shaft end in the center.



Before reinstalling the large piece of tape that covers the entire armature, use your fingers to work each winding such that none of the wires stick out past the winding supports. If you don't do this, the armature won't fit back into the casing! You can carefully test-fit the armature once you have the tape on, but don't do it before then as you will damage the windings. Getting it to fit will take some iteration. With the winding done and the armature taped, carefully reconnect all the wires to their corresponding terminals per the diagram below. Note that when I refer to the inner winding I mean the two windings you made first, while the outer winding refers to the two windings you made second.



With the connections made, carefully snap the armature back into the bottom end cap, paying attention to the fact that the hole it fits into is 'D' shaped and thus it can only be (correctly) assembled one way. Then press the housing back over the armature, snap on the top end cap, and re-do all the crimps with a small screwdriver. Before soldering the motor into the board, adjust the thrust bearing (accessible though the hole in the bottom end cap) such that the shaft can move in and out of the motor by about 0.3mm. A dial caliper with a depth gauge works well for measuring the play. Then secure the thrust bearing with a bit of lacquer or similar removable adhesive. Solder the motor back into the board, reinstall the assembly into the oscilloscope and watch it work! If done correctly, the motor should be completely silent with only the very slightest air noise when the fan is attached. You can try heating up the thermistor on the board with a soldering iron to see the motor accelerate. In case you run into trouble, you can use an ohm-meter to verify that each winding is about 25 ohms and that they are connected correctly per the diagram above.

Good luck! (and I hope this guide actually helps someone)

(In the spirit of full disclosure, I must admit that the photos in this post were taken during reassembly of the motor, not disassembly as they are presented, but for all intents and purposes they are the same.)

-Matthew D'Asaro

Matthew, an EXCELLENT presentation. All I can say is, I'm glad my 2465 is an "A".

Yeah, nice Matthew, but my gosh ... the work involved. It looks like the same motor used in the 465, 465B, 475, 475A, 464, 466, etc. If so, I think I'd scrap out a beat-up 465 for the motor rather than going through all that! Lazy? No. Old? Yeah. But that is a nice presentation. Did you up-load it to the Yahoo! Tek site? If not, you should. It's a lot more important than the 4,562 photos of 2465 waveforms that are posted there.

Thanks for letting me know the motor is also used elsewhere. Good to know. I did link this post to the TekScopes2 Yahoo group, but since you can't do formatting of messages in that group I tend to like to post things here and link them in from the various relevant Yahoo groups.

http://groups.yahoo.com/neo/groups/TekS ... ons/topics

-Matthew

Safe to say I doubt I am going to try this complex of a project any time soon, but it is very interesting to see how Siemens made that motor. They must have had an engineer on the staff who was really bored, so they asked him to design the most complex of small DC motor he could come up with, their intent to sell it to Tektronix in revenge for something in the past.

-Mark-

Excellent job, Matthew! There are several threads for repair of this motor on the TekScopes group, but none as well presented as yours.

Bill

I don't think the complexity was revenge. The overall circuit (as used in the 465, for instance) is elegant. The motor includes Hall-effect sensors, so it's brushless. I've only had one motor go bad in my years on the Tek bench, and then it was cheaper for the customer to replace the entire module rather than charging for the cost of the motor and all the labor to suck the old one out (risking board damage) and soldering the new one in.

The driver "chip" for that circuit is actually a four-transistor array and it's in a symmetrical package. One symptom was a motor with a dead spot. Bad motor or array? With the motor sitting on the dead spot, simply reverse the array in its socket and turn the scope back on. If motor goes, bad array. If motor is still stalled, bad motor. The circuit included a thermistor for speed control. Turned slow in a cold scope, fast in a hot one. Spray that thermistor with freeze spray and you'll bring the motor to a halt!

I recall looking at the fan setup in my 7854 and thinking to myself wow, why all this garbage for just a fan. I've been careful about preventive maintenance in regards to it, as I never want to fix this fan setup.

-Mark-

Just because something is over-engineered doesn't necessarily make it better. My 2¢

I would suspect the design thinking went something like this:

Brush-less to avoid any noise issues that might find their way to the display.
Variable speed to maintain as constant internal temperature as possible to avoid thermal drift.

Paying attention to the small things is what made Tek the gold standard in scopes, just like Fluke and meters.

I made a few measurements of a known-good low-hours fan motor in a 2445. Hopefully these numbers will help if you are trying to see if your motor is working correctly after rebuild.

All measurements made at room temperature with 15V supply voltage in free air.

No blade: 70mA drive, unable to measure RPM
With blade: 105mA, 20 RPM

Note also that the driver chip on this fan got quite warm to the touch after running a few minutes so this is (apparently) normal.

-Matthew

I recently acquired a 2465 on eBay. At $200+shipping & duties I got a scope described as "in fair condition, with a very loud fan". When the scope showed up, the power switch turned out to be wonky, it would only latch on every now and then. After replacing the switch, the fan was the next problem - it squeeled like it was begging to be put out of its misery, poor thing.
I tried lubricating it with a silicone-based lubricant. I managed to get the lubricant at the front bearing through the axle. My fan control board has the solder-filled hole, through which I could get some lube at the back bearing - I think. After this the fan stopped squeeling, but now it seriously rattled. The rattling was bad enough that every now and then it would drop into resonance with my whole workbench.
After reading through these (awesome) instructions and peeking inside the motor, I decided that this particular restoration is beyond me and my hobby budget ATM.
Instead I replaced the fan with a 60x60x20mm 27cfm case fan, driven by the temperature-controlled fan control circuit from the 2465B (an LM317 and an NTC). This runs the fan at ~8V for 25C ambient, and ramps up to ~13V at 65C ambient.
I now have a scope that no longer rattles with perfectly decent ventilation, but boy is the fan loud compared to the original when it wasn't rattling .

My next couple of hundred hobby dollar budget is going on a decent DMM, but after that I think I'll be asking Matthew to restore my original fan motor. There really is a world of difference in the noise level.

NB: this is my first scope, and I didn't know about the A/B sweep features until I used it in anger the first time - WOW! These instruments must have been indisinguishable from magic in their day.

I got lucky enough to score a brand new fan motor on eBay for a very reasonable price. De-soldered the old rattler, and installed the new one without much drama. God what a relief to rip out that case fan, and restore the scope to humane sound levels - well worth the trouble and expense.

Matthew,

Thanks so much for the write up! I just finished my fan motor rebuild, and to my utter amazement it worked!

I turned new bushings out of some oilite stock I had laying around. I didn't bother restoring the thrust bearing, and it runs just fine. Other than that I followed your instructions to the tee. Yes, it was a royal pain getting the windings squished down to size. I reused the old wire, and came up a few turns short on inner and outer windings. I'm amazed that you actually came out with extra wire. You must have rewound a few motors in your time.

-- Trey

Holy Slipstream, Batman! What a project. Excellent job and writeup. If my 2445, or 465, ever needs this procedure, I am simply going to cut a hole in the case and run a duct into my HVAC system If that cools it down too much, I'll install a thermostatically controlled damper door. Sheesh!!

Or, "someone" may be getting a fan motor to rebuild for hopefully less than the cost of another scope

That motor also looks like ones I've seen in some old HP test equipment. They came out with a retro-fit replacement kit with a simpler motor at one point in time.

Thanks Matthew for writing this article. It has given me a heads-up for rebuilding my Siemen's fan motor in my Tek 2465A DV. My scope has the Siemens motor because it has the DMM option. Since I had your article to go on, I didn't have to go into this project blind. Thanks!

I have my fan motor disassembled right now. I am waiting for my Oilite bushings to get here that I ordered from Patrick at LM-Tarbell. I will be using brand new wire because I don't trust used wire, and that's just the way I do things.

I will also be writing detailed articles about the rebuild in two parts. My hope is to show people that it really isn't very difficult to rebuild the Siemens motor and they should not be afraid of the job. I am in the process of writing Part 1 right now.

Robert

Hi Matthew. Your post was back in 2013 and it looks like the last comment was in 2018. I'ma wondering:
(1) did you ever make this into a Service Business? If so, please let me know how to become a customer.

(2) Did you ever come up with something to replace the plastic collet? If so, how can I obtain a couple?
Thanks
Bob

I am with Mike up above. I sure am glad my scopes are "B" models.

John

The problem with U800 relates to the product of 2 events:

These are the magnitude of the temperature rise above ambient temperature, and the number of times that it is cycled. This in conjunction with the weakness within the IC allowing the output transistors to delaminate from their connections.

It is not the absolute magnitude of temperature rise being the main cause, the temperature rise above ambient for the IC tab, without a heat sink is fairly modest for the IC at about 32 degrees. In a 30 deg C environment though the tab can get to 62 degrees.But that is a fairly warm workshop.

If there was no temperature cycling of any significance, and U800 was kept at room temperature as a spare part, no measurable degradation occurs. I have verified this and tested NOS aged U800's and there are no detectable early failures.It is possible to detect early failure in these as a small DC offset at their outputs. It increases with time.

I have recordings of a U800 gradually failing with power cycling of a 2465B over 12 years. I did not know I was making these at the time, but inspecting screen photos I had been making over the years for various articles showed the horizontal shift to the left (typically) of the data display, occurring very slowly, year by year, until one of the output stage transistors finally lost complete contact and the beam deflected off to the left. Sometimes it can go the other way depending on which of the 4 output transistors fails.

To save U800's and get the best out of them, since you cannot avoid temperature cycling them if you use your scope especially if you use it daily then the max temperature rise of the the IC's metal tab must be decreased to the minimal possible magnitude as likely it is temperature increase x thermal cycling events which unmasks the defect in U800. And obviously it helps if the fan is in good order.

There are solutions attaching heat sinks to U800, I found though that the common ones were minimally effective and only lowered the tab temperature by about 4 to 5 degrees C.

So I designed some large copper heatsinks which are properly thermally bonded to the tab with a machined copper washer.

One system/design is for a U800 that is in a socket, it has two copper flanges, attached to both the upper and lower metal surface with thermal compound. The other single flange system is for an original U800 soldered into the pcb as they often are.

Both have a similar surface area and lowers the tab temperature on U800 by 18 to 19 degrees C.

U800 then, for each use of the scope only gets thermally cycled by about 13 degrees above ambient, less than half of "normal" this will substantially increase the life of U800, and delay the time before its internal defect occurs.

(I'm not sure why the geometry markings didn't show up on the .png of the two flange heat sink I can post them if anybody wants)