An exasperated engineer struggles to troubleshoot a scope with a funky frequency response
Several years ago I had a calibration lab perform
cal on a 60 MHz Tektronix scope that I had bought at the TRW swap meet. Shortly after calibration, I noticed that one
of the two channels had a low frequency rolloff at a much higher
the other channel. This anomaly was only
noticeable in AC coupling mode, of course. Further investigation determined that the rolloff was
identical on both channels except when the vertical scale on the
channel was set to 0.1, 0.2, or 0.5 volts per division, a major clue.
So I went back to the swapmeet and bought a manual
scope. In the schematic, I found that a
certain attenuator circuit was switched in only for to 0.1, 0.2, or 0.5 volts
per division. I
studied the attenuator circuit and noticed
that there were some capacitors in parallel with carbon composition or
wirewound resistors in voltage divider circuits. Also
in the circuit was a shunt trimmer
capacitor right next to the AC coupling capacitor.
Well, I tried to determine the transfer function
attenuator as a function of frequency, but I quickly got lost in the
algebra! Too many j?C’s and whatnot. I thought, “That trimmer cap must be
interacting with that AC coupling cap.” So I gritted my teeth and gingerly tweaked its capacitance with
trimmer tool while sweeping the frequency going into the scope with a
generator. I just barely touched it, and
like magic, the frequency response of the scope dropped down to where
The moral of the story is that with some decent
troubleshooting skills (and some spare time that it seems like I no
have), some intestinal fortitude, and a bit of luck, some pretty
problems (alliteration, anyone?) can be solved. Oh, and I never did use that calibration lab again!
Jim Ford is officially Senior Staff Engineer,
Broadcom Corporation. In reality, his
work involves design and layout of test circuit boards for wireless
Bluetooth, WLAN, GPS, cellular, and/or FM radios in them. His hobbies are electronics and playing bass,
keyboards, and guitar (not simultaneously!).
Thanks for commenting, everyone! Yes, the problem could have been a flaky trimmer cap. It's been close to 20 years ago now, so I don't recall all that much. Anyway, I sold the scope to a fellow EE in 2002. Yes, there are lots of cheap scopes on ebay these days. The last one I bought was a Tek 7904 with 7A26 and 7B85 plugins for $100 in perfect working condition! This was after selling my Iwatsu 250 MHz scope with a broken trigger mode switch (stuck in single-shot mode, the least often used mode!) for $150. I've seen the Tek 7000 series plugins for as low as $10, and a cherried-out 7A26 will only set you back about $70. Add to that all the USB RF/microwave test equipment out there for less than $1000, and it's a great time to be a garage entrepreneur. Assuming you have the time, unlike me.
Granted that a good calibration procedure would include cleaning all switches and verifying operation in all modes. However it seems the author was fooled by a symptom disappearing as he made a adjustment on his oscilloscope. It could happen to anyone not familiar with the design of the input circuits, but I just wanted to point this out. I am intimately familiar with the input circuits on those oscilloscopes and the HF frequency compensation trimmers for the stepped attenuator would affect high frequency response only. The AC coupling on the input sets a single-pole high pass filter at a subsonic frequency, apparantly this roll-off was occurring at a higher frequency than it should. Based on my experience the issue was more likely a dirty contact on the input coupling mode switch or a bad connection somewhere.
This is a great experience, but the aging effect was more pronounce on Analog Devices, but due to digital electronics the thing are getting changed and it is a tough time of survival for calibration labs.
ttt3 wrote: "I've always wondered exactly what goes on at these calibration labs."
No kidding I was having a problem with my Tek TDS420A scope. The probe sensor circuitry that automatically detected what kind of scope is connected was flaky. I dropped the scope off with the calibration people with a note explaining the problem. They gave it back with a clean bill of health. It immediately failed when I tried it. When I challenged the lab people (they hadn't left yet) they admitted that they didn't use the scope probes when they tested the scope. They just plugged a BNC cable between the scope and the signal generator. That also meant that all of their testing was being done using the 50 ohm input impedence mode on the scope.
What a rediculous idea, the probes can have a huge impact on the accuracy of the scope and virtually 100% of the time I am using a 10:1 probe with my scope. I got them to agree to at least function test the probes on my scope during annual calibration.
Turned out I needed to replace the BCN connector PCB on my TDS420A.
What we learn here is that it is not always a requirement to have the whole manual, provided that a copy of the circuit is available and you understand what is supposed to be happening, After that, when you find out what the difference is between what is happening and what is supposed to be happening, often the exact failure becomes obvious. At least, that is the second best approach to fault diagnosis. The first best is to consider what usually fails on an item and check that part, but it only is a useful approach when the failure symptom matches the previous failure symptom. As an example, there is one model of Sony cassette recorder that would develop a hum on playback. No amount of part checking in the power supply area could turn up an out-of-spec part. It seems that for that particular model, the cure for the hum is replacementof the volume control potentiometer. That was it every time, for all of that model that came in for repair, for over a year, about 26 different machines.
My experience with the manuals is that the older and LESS integrated the circuitry, the more adjustments were necessary/possible, and the more likely the adjustment procedure was described. Yes, in the pre-microprocessor instruments from Tek and HP particularly, there is a checklist set up to copy and perform step by step. And a cal lab needs to do more than cal, it needs to retroactively inform if an instrument is found to be out of cal, so that notification is made that measurements made with that instrument may be suspect. Not just in-cal for the future, but affirming measurements made in the past.
It's always good when you can get a manual. Some time ago someone gave me a 30-year old Fluke bench DMM. Fluke still had the service manual on their web site. I also managed to get (not frum Fluke, although they were very helpful) an LCD display for it. Result, a seriously useful DMM for amost nothing. I'm a big fan of Fluke and others who do their best to support their older products.
Who's to say it was the lab? I've had intermittent issues like this crop up from time to time. Just the fact the unit was moved and the vibration induced in the process can often wreck havoc. I'm not defending a lab if it was poor work on their part but to paint them as the source is a bit of a stretch. Jim's own comment "I just barely touched it, and like magic, the frequency response of the scope dropped down to where it should be!" I'd of suspected a connector, oxide in the trimmer, etc... and the mere fact you touched it was enough to overcome the issue.
I've always wondered exactly what goes on at these calibration labs. I think most of the time they just run a few tests, the instrument passes, they slap a new sticker on and send the equipment back out the door. Exactly how those tests are written (and who deems them acceptable) has always been a bit of a mystery to me. Does Tek have a "calibration manual" for their scopes, or does each lab come up with their own test procedure?
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.