The TDS-420 is one of Tektronix's first oscilloscopes in the TDS line and while, at 150MHz, it is somewhat old and slow by today's standards it can now be had for relatively little money and it uses the modern TDS interface, making it popular with hobbyists and in education. We use them at MIT in one of the teaching labs and getting them all going again has been a pet project of mine on and off for a couple of years now. Since I am just now (finally) finishing the last of these beasts I thought I would share some of the common ailments and their cures with everyone so that others may get some benefit from my (many) long hours of troubleshooting. The service manual for this beast is available at the link below, however, it is all-but-useless since it does not contain schematics and the calibration instructions amount to "Install calibration software on computer. Connect oscilloscope to computer. Follow on-screen directions."

http://i3detroit.org/wi/images/0/0e/460-ServiceManual.pdf

Capacitors, Capacitors, Capacitors!
The first thing to know about these oscilloscopes is that they suffer from the capacitor plague of the early 1990's. For some reason, almost all of the SMD electrolytic capacitors from the early 90's are failing now, seemingly irrespective of use and storage conditions. I have heard tell that there was an incident with counterfeit capacitors made using a stolen electrolyte recipe that was missing a key corrosion inhibitor, but this may just be a story. The ones in this oscilloscope are especially bad as when they fail they leak corrosive electrolyte all over the boards. This electrolyte rapidly corrodes away traces and, due to electrolysis, this process occurs much more rapidly in the presence of a DC bias. Thus, if you have a TDS-420 DO NOT PLUG IT IN until you have replaced all the SMD electrolytic capacitors.

Here is a photo of the type of capacitor that needs replacing (note that the picture was actually taken after the parts were replaced):


I recommend the following part numbers (available from both Mouser and Digi-Key) for replacement capacitors:
25x 33uF@10V: Nichicon UUA1A330MCL1GS
18x 10uF@35V: Nichicon UUA1V100MCL1GS
2x 3.3uF@50V: Nichicon UUA1H3R3MCL1GS

Note that you will need to replace capacitors on the following boards:
-Attenuator board
-Acquisition board (this is the main board at the bottom of the oscilloscope)
-Video Board
-Serial / Parallel board
-Front panel board

My procedure for replacing the capacitors was as follows:
1) Remove capacitor
2) Use solder and wick to thoroughly clean pads
3) Check the continuity of all surrounding traces. The damage from corrosion can occur right where a trace meets a pad where it is almost invisible, so continuity checks are the only way to go.
4) Remove, clean, and test any surrounding resistors and capacitors that show significant corrosion. (Remove anything that looks black from corrosion - a bit of a mat finish on the pads is probably OK.)
5) Scrub the surrounding area with alcohol and acetone using q-tips.
6) Install the new capacitor and re-install any removed components.
If any of the little brown 1206 SMD ceramic capacitors are corroded beyond help, they are 0.1uF@50V and are replaceable with AVX part number 12065E104ZAT2A.

Take it slow here. If you miss a damaged trace, it will be HOURS of work to find again later.

tbiRampTest FAIL
All of the units which I have worked on so far failed the power on self test with an 'ACQ - FAIL' message before I replaced the bad capacitors. In the error log (look it up in the service manual if you can't find it) the errors have all been related to the 'tbiRampTest.' Replacing the capacitors did not fix the problem on this last one I was working on and the error log said "tbiRampTest FAIL - It appears the ramp didn't stop". In operation, the symptom was that when triggered the oscilloscope would rarely or never refresh the waveform shown on the screen unless the 'Force Trigger' button was manually pushed. However, if no signal was present (allowing the waveform to 'roll') the scope would seem to work normally. With no schematics it was a real bear to find the source of this problem. The only reason I was able to do it was because I had a working unit to compare with. First, I swapped boards between the units and determined that the problem was on the acquisition board. Not too surprising as this is the board with the most capacitor leakage damage on it. I then put the working unit and the broken unit side-by-side on my bench and compared the waveforms on each pin of each transistor and IC while marking the signals that were incorrect on the broken unit with red sharpie. You can see these marks in the picture below. After many, many hours of work on this I was able to determine the following information about this ramp test. Armed with this info, it should not be too hard to troubleshoot similar tbiRampTest problems.

I don't know exactly what the ramp is for, but here is (somewhat of a simplification of) how it works.

1) The ECL control logic generates a positive edge (first waveform picture) which, after going though some other transistors, appears on the base of the ramp generator transistor.
2) The ramp generator transistor, operating as a constant current source, charges a capacitor to form a ramp waveform (second waveform picture). This waveform appears on the collector of this transistor.
3) The ramp appears on the '+' input (pin 2) of the LM311 ramp comparator. The other input of the comparator (pin 3) is held at ground and so when the ramp reaches zero volts, the comparator triggers producing a positive edge.
4) This signal goes though another few transistors and ends up going back into the ECL logic at the point I am calling 'TP1'.
5) The ECL logic brings the base of the ramp generator transistor low again, ending the ramp.
Note that there is much jitter in the length of the ramp. This is (apparently) normal.



Waveform at the base of the ramp generator transistor:


Waveform taken at the collector of the ramp generator transistor (the ramp):


Waveform taken at TP1:


In the end, the problem with my unit was a bad ramp generator transistor. This transistor is marked 'M3J' and is a Motorola 'MMBTH69' for which a datasheet is available here:
http://www.datasheetarchive.com/dlmain/Datasheets-21/DSA-408130.pdf
The only place I could find which carries this part is American Micro Semiconductor:
http://www.americanmicrosemi.com/

CMOS Batteries
Another weird problem with this oscilloscope was that, even when turned off, the fan would 'twitch' - every second or so the blades would jump a bit. After much board swapping, I determined that this problem is caused by a low, but not completely dead, CMOS battery on the CPU board. I replaced all the CMOS batteries in all of the TDS-420 oscilloscopes I have worked on, so here is some info on how to do that.

First, there are two CMOS batteries in these oscilloscopes. Both are held in clips on the chassis and connected with wires to the boards. One is connected to the CPU board and the other to the DSP board. The CPU board is identifiable because it in in the first slot from the top of the unit and has the GPIB connector on it, and the DAQ board is identifiable because it is shorter then the CPU board, is in the bottom most slot and has no external connectors on it.

The battery on the CPU board may be disconnected for replacement with no ill effects. However, I recommend that before replacing the battery on the DSP board, you connect a 3V power supply in series with a 1N34A diode to the battery terminals on the board. This will assure that the contents of the memory are preserved while the battery is being replaced. If the contents are lost, it is not a major problem (the calibration data is safely in EEPROM) but it will cause your error log to be cleared and replaced with a single battery low error. The error log can't be cleared without special software (which I have never been able to find) so it is nice to keep it 'clean'.

Each of these batteries is replaceable by Panasonic BR-2/3AT2SPN. However, to prevent the sharp clips from cutting through the thin insulating material on the batteries and shorting them to the case, each battery must be covered, like the originals, in another layer of insulating material. The correct material is 31mm (flat) by 4mil thick clear vinyl heat-shrink ('battery-wrap'). I bought mine from eBay. Use a 1-5/8" long strip and shrink it around the battery with a heat-gun, then transfer the battery leads from the old battery and insulate them with heat-shrink like the originals.

Adjusting the display
Sometimes the displays get dim or raster lines start showing up. This is usually a simple adjustment. Follow the procedure in the service manual. Note that it seems some scopes, even when properly adjusted, will show raster lines for a few minutes until fully warmed-up. I think this is just aging CRTs.

Signal Path Compensation
If nothing else works, or you are having some really weird problem, try running the signal path compensation. This is available in the 'Cal' menu and has worked magic for me a couple times. First I discovered that when one of these appears to be 'missing samples' when running in equivalent time (ET) sampling mode running signal path compensation will fix it right up. The picture below shows the problem. Very odd.



Second, I tried running signal path compensation before fixing the tbiRampTest FAIL problem. This put the scope into a weird mode where even after the original problem was solved, it still wouldn't refresh the screen when triggered. Running SPC again solved it for good.

Finally, if the DC balance is off (the trace moves up and down when changing the attenuation setting, SPC should fix it.

Attenuator repairs
If someone manages to blow the 50-Ohm resistor in the attenuator(s) of one of these beasts, the higher voltage ranges (10V and 20V I believe) won't work. This can be fixed as I described in a previous thread:

viewtopic.php?f=8&t=216792

Hope this help someone!

-Matthew

Wow! Matthew, you are quite an inspiration for me. You've shown that even this modern surface mount gear is fixable.

If you haven't done so yet, you need to post this on the Yahoo! Tekscopes site.

Thanks guys! Upon Dean's suggestion I have posted this on Tekscopes and Tekscopes2.

Tekscopes:
https://groups.yahoo.com/neo/groups/TekScopes/conversations/messages/104826

Tekscopes2:
https://groups.yahoo.com/neo/groups/Tek ... ages/10801

-Matthew

Yah, many thanks,

Its amazing work..

I may even decide to keep my eyes open for one of these, inspite of the lack of service data.....

David

Yes, you have to do that SMD stuff when you are young. Hit 40 and you'll be needing quadruple the magnification just to see any of it, I squint even thinking about it When it came to an all digital scope I decided to limit my “modern-ness” to late 80’s technology with a HP 54502A 400 MHz 400 Ms/sec model. These will get the job done for most hobby needs while having the convenience of on screen measurement, and they are noted as being fairly reliable. At least reliable in the sense you won’t have to go in and replace every single one of any given component as a precaution. The fun pretty much ends for me with surface mount crap, but it's great to see people can actually troublehoot and repair it.

-Mark-

Thanks Mark! I have an HP 54502A as well. It seems like a very well-made scope and aside from having to replace the CMOS backup battery and clean the attenuator contacts when I got it, it has worked flawlessly. I especially like the self-calibration feature - super easy and no special gear needed.

-Matthew

I don't know exactly how those surface mount caps are made, but most if not all manufacturers had major problems with them, usually within only 10 years. I worked on broadcast equipment for about 11 years and replacing SMD components was an everyday chore. It was quite common to have boards with 80 to 100% failure of electrolytics; if the board was no longer available it meant hours of work to get it going again.

Hey Matthew,

this is an awesome write up, as usual.
I'm about finished fixing up an 2430 that I acquired as an awful mess. The poor thing had suffered battery corrosion, and had apparently been operated wet.

In the course of fixing it, I happened to take a close look at it's "jitter correction ramps", which are described quite thoroughly in the SM, and the ramp(s) you talk about here sound like exactly the same thing.


In the 2430, the jitter correction ramps are used to measure the time between a trigger and the sampling clock edge. Once a trigger hits, a relatively high current (20mA?) is blasted into a fairly small (100pF?) integration capacitor. On the following sampling clock edge, the current source is turned off, and the capacitor is bled at a much lower current (1250/1 or so), while keeping time on how long it takes (there are counters off a 5MHz clock IIRC to time the bleeding).
In the 2430 there are two of those, one measures trigger to rising sampling clock edge, the other to the falling edge. According to the SM, the firmware will then choose the more reasonable of the two measured time (words to that effect).

This allows the 2430 to place the aqcuired samples with a time resolution of something like 100ps, against a a max 100MHz sampling clock. Since the TDS-420 has an equivalent time mode, it stands to reason it's supported by similar machinery.

Siggi

Thanks for the information! Have you run across a problem where the scope a TDS420 or TDS460a, would only boot if the ON/STBY button is held down? The problem I'm experiencing is after the scope's main power switch has been off for a period of time I have to hold the ON/STBY switch for it to boot. Once I release the button the scope with power off. However, I left the main power (the switch on the back) for a few days and press the ON/STBY once and the scope booted fine, however, I couldn't power it back off without using the main power switch. I replaced the power supply and the main board as well as trying different front panels. The only thing I didn't try was the CPU board since I didn't have another one to test with.

Thanks!

Sounds like a bad CMOS battery on the CPU board. There should be two of them. Check them both. If replacing one that is still good, I recommend using a diode and an external power supply to maintain voltage on the board during replacement and prevent loss of stored data, although if the data is lost it is not catastrophic, only the contents of the error log and some other life data will be lost.

-Matthew

Thanks for the fast response Matthew! I should have mentioned I replaced both batteries as well with new ones and verified the voltage. I was actually hoping that was the problem when I first experienced it but unfortunately replacing the batteries didn't help at all. BTW, the old ones were completely dead with zero voltage.

Intriguing... OK. Here is another possibility. It sounds to me like it could be high resistance leakage between the silver ink conductors that make up the membrane switches around the CRT bezel (including the power button). I saw similar odd behavior on one that I worked on, although the switch that was misbehaving was one of the other buttons (not the power button). To check for this, remove the connector where this membrane attaches to the front panel. Then, using a DMM check for high resistance leakage (>1M can still be a problem) between each and every one of the conductors on that connector. The leakage need not be between adjacent conductors, so you need to try all possibilities. If you find leakage, blow it clear with a 10uF electrolytic capacitor charged to about 100V. Leakage forms when moisture dissolves and oxidizes the silver ink that makes up the conductors in the membrane and spreads that silver oxide until that it brides connections. Since silver oxide is highly conductive this causes leakage. This is the same process that causes silver mica disease in the IF cans of AA5 radios.

-Matthew

Thank Matthew! I'll give that a try. Will it need to be a 100v? I think the largest cap I have laying around is about 65v.

Nothing magic about those numbers. Just try higher voltages until the leakage goes away. There are no active components in that membrane so it is hard to hurt anything. Obviously, if you find traces that are connected by only a few ohms they probably are supposed to be connected so don't try to blow those!

-Matthew

I acquired (quite cheaply) a TDS220 (I think) that worked fine except for the vertical position control - move it and the trace disappears. Can use it again by pressing "autoset". Have been using it that way for a while, but it may have other problems lately. Thanks for the leads!

Hey 7jp4-guy, You should really start making video's. That's a great explanation.

Paul

Thanks for the compliment and the suggestion, but I am honestly not really a fan of videos. They take a lot of time to make and watch, they are very hard to skim quickly for useful information, hard to search for, hard to cite or quote, and hard to embed useful information like links, schematics, ROM images, etc. I far prefer to read and write information in a more classic text and image format.

-Matthew

Actually, It's pretty easy now a days. Once you make one, it's easy from that point on. I do agree about the video's taking time though. When I get into an in-depth video about a radio resto or something of that nature, 1 hour of video is usually 1 week of work. Explanations and schematics are live, so the guess work is minimal in my video's. But, to each their own, right.?.

Paul

I picked up a TDS460 in a local sale for 50.

Had the typical "aquisition" failure. Replaced all the surface mount caps and had to run a jumper wire on the front panel board due to corrosion of a trace from the leaked electrolyte.

I used radials instead of surface mount.

I also had to get buy some silver epoxy and reworks all the pads for the soft buttons around the crt.

Have had no issues since those repairs were made.

Lucky for me, this scope has the FFT and external display options. I got lucky a few days ago and found a Belkin EGA to VGA adapter for a whole 99 cents and now I can use a spare 20" lcd monitor with it.