So, I KNOW about impedance matching and all that, but bear with me while I tell a tale about staying up waaay too late to play with some junk on the workbench, and what that meant to my sleep schedule.
So technically yesterday, at around oh, 9:45pm, when most normal people are getting ready for bed or watching TV, I decided to start cleaning up my workbench. If you really want to split hairs, its my “main” workbench, as all 9 miles of benches in my garage are practically unusable at the moment. Yes I admit that, but its partly to do with a recent flood that found me frantically picking up stuff off the floor and tossing it onto whatever flat surface was nearby. I digress…
I didn’t just want to clean up the bench, part of the mess was several projects in various states that I just haven’t gotten back to. Since the floor is currently the cleanest part of the shop, I’m likely going to take advantage of that and re-organize the entire shop layout, but I can’t move a workbench that has altitude; it has to be cleaned itself, and I wanted to wrap up any projects laying on it. (I’m also halving the amount of workbenches as well; one project at a time!)
One of those projects was a WaveTek 178 Function Generator. A few months ago I went to use it, and to my surprise, it didn’t turn on; the fan spun up, but nothing on the display. I’m so used to test gear just working that I was honestly surprised, despite the fact that it’s probably 20+ years old if not more.
I cracked it open and eventually determined that one of the 15v rails off the power supply was shorted to ground. I suspected a bad cap, and after some time, found a shorted cap; C176, a 22uf/20v tantulum located near the low frequency mixer section on the upper board. Tantulums are starting to become a more failure-prone electronics component in older gear, and when they fail, they often short, and depending on the device they’re in, they often explode in a fiery way. In this case, the power supply was designed well and shut down.
I also learned that finding 22uf tantulums were not as common as you’d think, and many options on Digikey were not cheap! I bit the bullet and bought a handful to at least shotgun all the 22uf tantulums on that board; there wasn’t a ton of them. Since these are likely bypass caps I could have replaced them with a slightly higher capacitance (uf) rating, but since this device deals with somewhat precision frequency, I opted to try to stay to the original stats, with only a voltage increase change.
Anyhow…fast forward oh, 8-ish months later (don’t ask), and I finally placed the order and got the parts in. I shotgunned all of the 22uf tantulum caps on the upper board, then reconnected everything and gave it a test run. SUCCESS! The box fired right up and everything seems happy again. At this point it was about 11:30pm, so I put all 27,000 screws back into the board, gave it some cleaning and put the covers back on. Once I was done (around midnight) I figured I’d play with it a bit on the scope, check the outputs and all. I simply connected the scope probe to the function output BNC.
Everything seemed great at first, waveforms looked good, etc, until I noticed that the amplitude seemed off. In fact, it was double what it should have been! Crap…did I forget to connect a wire or plug inside? Once again opened up both covers, and looked all over for a loose wire or plug. Nothing! Next I started going through the amplitude troubleshooting steps and checking ICs and what not. Now its 1:30am.
After spending some more time fiddling and checking things, I got to the part of the manual that described the calibration process. I found the section on amplitude, set it up, and start going through it. After some confusion on the setup process (Freq: 1E3 = 1 and 3 zeros, or 1000hz aka 1khz), I noted the amplitude was still off. THEN I SAW IT: That little note at the top of the list: 50 ohm terminator required. HAH!
Yeah…I’m so used to fiddling with oscopes and multimeters that often have 1+ megaohm impedance, that I totally forgot this device had 50ohm output This means that it’s designed to connect to something with a 50ohm impedance (resistance) to deliver the most amount of signal to the test device. Connecting an oscilloscope with one MEGAohm imedance means the signal wasn’t loaded down at all, and was double what it should have been.
Connecting a 50 ohm resistor across the output fixed all; the amplitude was perfect, and late-night lesson learned! 2:30am I finally locked up the garage for the night, and I can *hear* the weight of my eyeballs today 😉
Highly recommend this video on signal termination, amongst many others out on the interwebs: