Koolertron For The Win, Or How My Cruddy Understanding Of Signal Generators And Oscilloscopes Led Me To Believe Mine Was Messed Up…
The mark of an expert is knowing not simply what you know, but knowing what you don’t know. There are things in life I am truly an expert in. Electricity is decidedly not one of those things. For every new tidbit I learn, a door opens to reveal a seemingly infinite number of topics I haven’t scratched the surface of or even realized existed. In essence, I really don’t know what I don’t know when it comes to electronics. My last blog post was evidence of this.
The nuts and bolts of the previous post was that my Koolertron 60MHz Function Generator was not outputting a stable voltage (amplitude) as I swept across the frequencies from 1MHz to 60MHz. I was trying to test a little breadboard bandpass filter that should have had a passband somewhere near 7 MHz. I’ll repost the video so you can see what I mean. Watch the way the height of the waveforms increase and decrease somewhat randomly seeming as the frequency of the waveforms increases (and then decreases) as it sweeps. What I wanted to see was the waves get closer together as the frequency increases (they do), and the height of the waves to remain the same throughout the test, except when they are around 7 to 8 MHz where I expected them to increase to near 5V peak-to-peak. This did not happen. In fact, the random odd voltage attenuation was present even when I connected the signal generator directly into the oscilloscope. Keep in mind I have a 300 MHz oscilloscope, so at least theoretically, it should be able to handle a 60 MHz input signal I believe.
I then tested the voltage output of the signal generator directly with an analog multimeter; the voltage was all over the place, including a 7.2Vrms (or 20.4Vpp) output at 24 MHz. Importantly, I clamped the leads of of the signal generator to the leads of the multimeter to obtain this result. I am re-posting the pic again below so you see what I mean. Keep this in mind, it is what I am going to discuss in this post! Ultimately I concluded that the problem was an unreliable signal generator. And while I am sure a budget signal generator does have limitations, I am reconsidering this conclusion.
Why the reconsideration then? Let’s start with the way that I was connecting the oscilloscope to the signal generator. I was simply using a piece of 50-ohm coaxial cable with BNC connectors at each end. Well really one end was a SMA connector to a BNC adapter, but I do not think that played a role here. I had started thinking about, and doing a little research on, the possibility of an impedance mismatch between the cable and the oscilloscope being the cause of the attenuation of the output signal of the generator. In other words, could it be that the cable itself was the reason behind the unstable and decreased voltage I was seeing on the oscilloscope waveform? My first order of business therefore was to test my coax jumper using the RigExpert AA-650 Zoom, which is an antenna analyzer, or essentially a 1-port vector network analyzer. (And for all you out there groaning at the thought of me not using the RigExpert to test my bandpass filter now that you know I have one, or even my 2-port NanoVNA for that matter, are missing the point! I am using this as an exercise to understand the signal generator and oscilloscope).
I terminated one end of the coax with a 50-ohm dummy load (shown above), and ran an SWR sweep from 0 MHz to 30 MHz. The SWR rose to a tiny high of only 1.05:1 at around 24 MHz (left asterisk below), dipped very slightly and rose back up at 50 MHz to 60 MHz to again 1.05:1. It did not seem like that SWR increase would be the cause of my voltage woes.
I swtiched over to the impedance (RX) plot, where the resistance is shown holding steady (yellow line) at around 48.29 ohms across the entire 0-60 MHz sweep. Reactance, the blue line, has two small dips, one maxing to a Xc of -j1.98 ohms (left-hand asterisk) centered around the 24 MHz mark, and then again at the end of the frequency (reaching -j3.21 ohms) after 50 MHz or so. Again, really quite negligible. This is definitely not a case of bad coax.
But in my research on why oscilloscope voltage readings can appear wrong, I encountered several factors that I have considered. For one, and perhaps the most important in this case, I ran across the idea that coaxial cable was a poor choice for oscilloscope leads unless terminated with 50 ohms. That led to the discovery that many oscilloscopes offer a low-Z input option so that coaxial cable can actually be used directly. Indeed, I did find a DC coupling 50 ohm setting on my oscilloscope as shown below. I had been (until now) conducting my testing using the 1M ohm coupling. An option I came across to correct this was to use instead of coax, an oscilloscope probe tip-to-BNC connector to avoid the coax low impedance problem by simply using the probe lead instead.
Let’s talk about my oscilloscope probes then to rule out other possibilities for voltage misreadings. My oscilloscope probes are passive with high impedance, and have a 10:1 attenuation factor setting. But before you think that their attenuation feature may be the source of the low voltage, they also have a sense pin for the 10x setting, and my oscilloscope indeed has a sense ring, thus automatically adjusting the voltage to account for the 10x attenuation (which I do tend to use). So this is not the source of my poor voltage readings! Along those lines of user error mistakes, I double-checked that the volts/div vernier control (var knob) is locked into the calibrated position – again not the source of my misreadings. Finally, the calibration the of the probes is also in check; apparently poorly calibrated probes can also effect voltage readings particularly at higher frequencies. Circulating back, I think it is indeed the transmission line effects of the coaxial cable jumper that may be responsible for the voltage irregularities seen on the oscilloscope, and NOT a bad signal generator after all (sorry for my doubt Koolertron!).
To test out the cable theory, I decided to rerun a 1 MHz to 60 MHz sweep with an amplitude set to 1 volt peak-to-peak. I used the coaxial cable with BNC connectors directly from the signal generator to the oscilloscope, but this time I switched the oscilloscope from its normal high impedance setting to the low Z (50 Ohm) mode. And this video shows what I got… near perfectly stable voltage at approximately half of the signal generator set output voltage throughout the entire sweep. Apparently the voltage divider effect of using the low Z oscilloscope mode halves the detected voltage. So the fact that the signal is seen as around 500 mVpp instead of 1Vpp is the actual expected result.
The last test I wanted to run was to set up the probes in the same way that I had for the original scoping of the band pass filter circuit I had tested in the previous blog post. Again, I was wondering if impedance variations in the cabling could be the cause of the voltage fluctuations seen at the oscilloscope. For this test, I alligator clipped the oscilloscope probes to the signal generator leads. I used a 50-ohm dummy load to terminate the oscilloscope probe’s BNC connection, whereas I attached the signal generator BNC connector to the RigExpert (with some adapters in between.)
I ran an SWR sweep from 1 MHz to 60MHz, and sure enough, there was two areas showing significant decreases.
The first dip was is shown below, where it reached its nadir of 2.1:1 SWR at 29 MHz.
And then again decreasing from around 35 MHz all the way until 53 MHz or so, with its lowest SWR reading 1.32:1 SWR in this frequency region. At low SWR, I am expecting the impedance to be mostly purely resistive at around 50 ohms. Unfortunately, I did not bother checking the RX plot. Regardless, at low SWRs, my guess is that the signal outputted from the generator is going to have more of its voltage “show up” at the oscilloscope, versus being reflected back toward the generator, and as a result the oscilloscope will show a higher voltage at frequencies with low SWR than those with high SWR. I am now strongly considering that it is indeed this variation in impedance through the test leads that is causing the voltage fluctuations seen on the oscilloscope during a frequency sweep.
What have I learned from all this? Who knows?! Okay, a bit more generously, I have learned that electronics are not particularly straightforward. I am partial to testing. And electronic testing is full of various challenges, nuances, and esotericisms that I have been barely able to peel the first layer of. The pragmatic nature of the electrical engineer will most certainly bristle at this thought, but to me the divide between the present day electronics enthusiast and the medieval alchemist is not entirely clear. Even as I learn the various formal numerical models that are used to represent electrical phenomenon, I am touched with a sensation that what we do not know about this thing called electricity is still so much impressively larger than what we do know. Maybe it really is with a twist of sly humor that reactance is designated “imaginary”. I know, I know, to the dear reader who is chuckling in amusement right now at my novice naivete, thinking to himself, this all makes wonderful sense if you just were to learn it properly…I would simply challenge you to look deep within yourself and answer this very simple question: Does it? Does it really?
Forever yours.
KM1NDY