Parasitic Oscillation on the Eico 666 Tube Testers Mike at Fourwater.com 1 Mar 2024 http://www.Fourwater.com So what is parasitic oscillation, why should I care and what does it have to do with my Eico tube tester? The short answer is that parasitic oscillation can be a problem and it can potentially cause serious errors in the test results. Note: If you don't want to read the rest of this article and just want to know the bottom line, check the lever settings for the test setup on your tester. If the plate and screen grid pins are both being set to the plate selection, then you may experience parasitic oscillation in your tester which will disrupt the test results. Find a later chart that properly sets the screen grid to the screen selection. Warning: Some tubes required the selection of the grid and plate pins to be set to the screen selection. That is ok. These are normally low voltage tubes (space charge tubes) that require a lower voltage than the lowest plate selection can provide, so the lower screen grid voltage is used. These tubes are low power, the parasitic oscillation is not normally encountered when testing them. If you are still having problems even after using the latest setup chart, or just want to know what this is all about, read on... ---- So how do I know if it is happening on my tester? It can be difficult to check because the nature of parasitic oscilation is that any attempts to hang a probe on the electronics can be enough to disrupt (stop) the parasitic oscillation. Still, it is always worth checking. One way to check is to place an osciliscope probe on the plate pin of the tube under test. Another place is the control grid pin. In my test case, I couldn't put the probe on the plate pin as the o'scope couldn't handle the voltage on the plate, but I was able to see the oscillation by putting the probe on the grid pin. If you don't see a sinewave, you are likely looking at parasitic oscillation. See http://www.fourwater.com/files/parasitic666osc-before.jpg and http://www.fourwater.com/files/parasitic666osc-after.jpg If you don't have an o'scope, try wrapping your hand around the tube under test. That can sometimes be enough to stop the parasitic oscillation. ** WARNING! ** Be careful! If you are testing a tube where the plate connection is via a cap on top of the tube. You don't want to touch it and get shocked! If the needle on the meter changes position when you wrap your hand around the tube (usually causing the test result to get better), that is a strong indication that you are seeing the result of parasitic oscillation. The capacitive effects of your hand disrupts the existing parasitic oscillation. It won't normally stop it, but it will change the characteristics of the parasitics in the circuit enough to change things enough to see it as a change in the meter reading. (It's similar to the way a metal detector works.) Parasitic oscillation occurs when an amplified signal is coupled into an input of the amplifier. It is referred to as parasitic when it is not by design. Typically it is caused by layout issues and is usually uncontrolled. Under controlled conditions, it is just regular ocsillation and is the intended operation of an oscillator circuit. Parasitic oscillation usually occurs due to poor layout of circuitry where the output signal is inadvertently coupled into the input. There are often multiple nodes which can cause more than one frequency to be generated, and often easily disrupted by any minor change in the enviornment around the circuit (such as putting your hand on the tube under test). Even a minor change in the power line voltage can sometimes change it. So what does that have to do with the Eico 666 or 667? It is a tube tester, not an amplifier. The short answer is that it IS an amplifier. It is testing the amplification quality of the tube. The tube under test is the amplifier. Ok, so what? The problem that can occur is that the layout of the Eico testers (and a few other tube testers for that matter), doesn't take the potential of parasitic oscillation into consideration. To be more specific, if you look at the wiring inside, you will find that they run the socket wires together in a bundle between the switch assembly and the tube sockets. That means the plate wire is running next to the grid connection wire. This provides the potential coupling needed for parasitic oscillation because the plate wire is the output, and the grid wire is the input. Unfortunately, due to the nature of the requirements of the tube tester, any of the wires can be the plate and any of others can be the grid, depending on the tube being tested and pin lever switch selections for the tube. The coupling that allows for parasitic oscillation can be either capacitive or magnetic, or as often is the case, to some extent both. If the wire is connected to circuitry at both ends, it is a closed path. The closed paths of both wires allows the magnetic field transfer. (Although there will usually be some capacitive coupling as well.) The wiring essentially becomes a transformer. (A transformer doesn't require a metal core. The metal is just used to make the transformer more efficent by focusing the magnetic field on the coupled wires.) An example of only capacitive coupling is a device called a parasitic capacitor (sometimes referred to as a stub capacitor), which is just two pieces of wire twisted together (or simply near each other) with only one end of each wire connected to its associated circuitry. Without a return path on the wire, there is no magnetic field developed, so the coupling is capacitive. In a magnetic coupling, there is also capacitive coupling, but unless the signal is very small and the coupling distance very long, the dominant coupling method with a closed path is usually magnetic. In my test case, I was testing a 6L6GC in an Eico 666 tester. I was using my reference 6L6GC tube which I know what the test results should be. In this particular Eico 666, the measurement was coming out low. Instead of 100 on the meter, it was reading 40. My first thought was that there was something wrong with the tube tester. So even though I had recently gone through the calibration procedure, I checked it again. It checked ok. So I figured maybe something was wrong with the plate or grid supply voltages. I usually measure the voltages on the plate and grid control pots since it is an easy location to attach the meter leads. The voltages checked out ok. So my suspicion was parasitic oscillation. The first quick check was to place my hand on the tube glass to see if it changed anything. It did, the meter reading almost went to the normal 100 for the tube. A very strong indication it was parasitic oscilation. So time to bring out the big guns... I connected an O'scope to the grid connection. I used the pin on the loktal socket because it was easy to get to. I used the grid connection because the O'scope couldn't handle the voltage at the plate. Fired up the O'scope and the Eico 666, waited until the tube had warmed up, then pulled the Merit lever. Yep, there it is, it's oscillating. See http://www.fourwater.com/files/parasitic666osc-before.jpg you can see the effects of the parasitic oscillation. The meter reading is low, and there is an unstable oscillation in the peak of the AC wave form. The o'scope couldn't resolve the frequency of the oscillation, so I'm guessing it is well above 200MHz. I'm guessing around 500Mhz, maybe higher. Note: If you're o'scope can't handle the frequency of the parasitic oscillation, it might look like this picture instead: http://www.fourwater.com/files/bad-666grid-signal.jpg Having discovered it was parasitic oscillaton, the next step was to find out what was causing the oscillation, and what to do to fix it. In a normal circuit, the fixes are usually fairly straight forward. Check the layout to make sure there are no uncontrolled feedback paths, and if needed, install damping circuits to drain away any unwanted freqencies. Unfortunately, in a tube tester it isn't quite so easy, there are multiple tube sockets that need to be wired with the wire connections unknown until the switches have been set up to place the desired signal on the desired wire. Add all this to the cramped room inside a tube tester, and you have the makings for parasitic heaven. To be trueful, I've always known about the potential parasitic oscillation problems. All tube testers have the problem to some extent. Some worse than others. Even the Hickok testers can have the problem. In most cases, it is minor and has little or no effect on the test results even if a small amount of parasitic oscillation occurs. However, it is made more likely to happen when a high powered tube is being tested. The higher the voltage and current the tube is being driven at, the greater the risk of parasitic oscillation becoming a serious problem. The larger the oscillation, the more trouble is causes. I was probing around in the guts of the 666 trying to get an idea of what was causing the problem, and contemplating whether I would need to come up with some sort of filter network to deal with it, which I really didn't want to do as it would definately be a nightmare considering the layout issues involved with a tube tester. I had decided to try some simple filtering (A 0.01uF capacitor on the plate connection), I only half expected it to work, so I was not surprised when it didn't fix the problem, Nor did putting the capacitor on the grid pin. I really didn't want to put capacitors on everything anyway, as that would not really be feasable to put capacitors on each of the pins of each of the sockets. But I was hoping it would give me some other ideas that would be more practical to try. As I was probing, I happened to move the green and orange wires which carried the plate and grid signals, although at the time I didn't pay attention to that. (see the red circled area in the pic at http://www.fourwater.com/files/parasitic666osc-before.jpg) A short while later, I tried another merit test to try out another idea of placing the capacitors on the grid and plate pots, I had no real hope of it fixing it as it was most likely that the oscillation was caused by the socket wiring, and the grid and plate pots were at the wrong end of the wires for that. So I was surprised when I pulled the merit lever and there was no more oscillation and the meter now reading correctly. A quick check with the o'scope showed no oscillation. (see http://www.fourwater.com/files/parasitic666osc-after.jpg) Note: The waveform is not a perfect sinewave, which is to be expected. The 6L6 puts a heavy load on the power supply when it is conducting on the positive portion of the AC cycle, so you will see a bit of sag on the waveform. The tube doesn't conduct on the negative half of the AC cycle, so that side of the waveform is more normal. So I'm sitting there scratching my head wondering what happened. I still didn't believe that putting the capacitors on the grid and plate pots had anything to do with it, so I removed them to see if the oscillation came back. It did not come back...ok so what else changed since the last check where there was oscillation. I was staring at the before picture (the after picture hadn't been taken yet), I noticed that the orange wire in the tester was placed differently in the tester compared to the picture. The orange wire was now separated from the green wire by about 1/4 inch. Could that be it? Checking the wiring, the green wire is connected to the grid on the 6L6GC and the orange wire is connected to the plate. Oh ho! That makes it highly likely to be the culprit. So I moved the two wires back together to the way they had been and tried the merit test again. Bingo! The oscillation was back. I moved the wires apart again, and the oscillation went away. So that's the source of the problem. In the picture at http://www.fourwater.com/files/parasitic666osc-before.jpg Notice inside the red circle with the arrow pointing to it. The green and orange wires are running next to each other where they travel between the right group of sockets on the panel and the left group of sockets. That is the source of this particular parasitic oscillation. The orange wire has the plate signal on it, and the green wire has the grid signal on it. The black alligator clip to the right is picking off the grid signal for the display on the O'scope. In the picture at http://www.fourwater.com/files/parasitic666osc-after.jpg you can see what happened when I separated the green grid and orange plate wires going between the 8 pin Octal socket and the preoctal sockets on the left side of the meter. The parasitic oscillation is gone, and the meter is now correctly reading 100. While that fixes the test with the 6L6GC tube, it doesn't necessarily fix it for other tubes which have different pinouts as well as being at other sockets on the panel. So I'm sitting there staring at the back of the tester panel trying to decide what to do about the problem. Seeing the wiring harness with all the wires going to the sockets bundled together, my first thought is; Didn't anyone think about parasitic coupling? Of course the answer is no, or they wouldn't have done that. Since the tester uses the 60Hz power line for the tests, who would be thinking about anything oscillating at 200MHz or more? Running the wire connected to the grid next to the wire that is connected to the plate provides a feedback path to generate parasitic oscillation. It's even made worse in this case because there is nothing connected to the wires that go to the other group of sockets, so it forms a parasitic capacitor, and that makes the probability of parasitic oscillation occuring even greater. But the adventure doesn't end there... I had switched to staring at the lever switch settings, wondering if there was a better fix than just separating the wires, which seemed to me to not be a very good fix, as there can easily be other areas that can be a source of trouble. Looking at the levers, it dawned on me that the screen lever is being set to the same position as the plate lever (as the 666-05 setup chart specifies). That is a potential problem. A key purpose of the screen grid is to shield the grid from the plate to inhibit parasitic oscillation. Tying the screen to the plate defeats that purpose since it is no longer doing it's job of acting as a shield. It just becomes an addendum to the plate. In an amplifier, the screen is normally set to a voltage less than the plate and it is decoupled to the cathode with a capacitor. A common alternate connection is to tie the screen to the B+ side of the output transformer which will have it's reference to ground. That reduces the maxium undistorted output since the screen will end up being slightly higher voltage than the plate, but it makes for an easy configuration, and still serves to provide the shilding between the control grid and the plate I was referring to the 666-05 test setup chart for the settings I was using. Looking at the later chart put out by Coletronics (1 Jan 1978), the test setup was revised to set the screen pin lever to the screen connection. Ok, let's use that test setup. With the wires close together to cause the parasitic oscillation, I pull the merit lever, and the oscillation is gone! While the screen grid connection on the 666 and 667 testers doesn't have a capacitor to decouple it to the cathode, it at least goes directly to the power transformer, so that reduces the potential for parasitic oscillation even if it does go through a maze of wires and switches along the way. The original 667-01 chart still shows the 6L6 setup with the screen and plate tied together, but the 667-02 chart shows the screen grid selection moved to the screen selection. I don't know when exactly it was changed on the 666 charts, but likely happened the same time they changed the setup chart to test dual tiodes sparately instead of with a single setup that caused bad test results if one of the triodes had significantly different emission than the other triode. (see http://www.fourwater.com/files/eicotesting.txt) So if you are having trouble with your Eico tube tester telling you a tube is bad which you know for a fact is perfectly good and you have gone over your tester with a fine tooth comb, checking calibration and all the voltages, (see http://www.fourwater.com/files/eico666-667-repair.txt) yet it is still giving you bad results, it may just be the result of parasitic oscillation. The first thing to do is to check that the screen grid pin selection is tied to the screen selection and the plate is tied to the plate pin lever connection. If they are both tied to the plate or screen connection, then get a later chart that has the test configuration setup properly. Note: Certain low voltage tubes (space charge tubes) require that the plate and screen both be set to the screen lever position as the lowest available plate voltage is too high to test those tubes. This is ok as those tubes generally don't use enough power to cause the parastic oscillation. If you are still seeing problems, try a different tube. Particularly one that is not a power tube to see if it checks ok. If you have an o'scope, you can check it by measuring the signal on the grid pin on the tube under test while the merit lever is pulled, but be VERY CAREFUL! If you short the pin to an adjacent pin or to ground, you may damage the Grid control if you haven't yet installed the grid protection lamp. (In my test case, the grid lamp had no impact on the parasitic oscillation, neither initiating it, nor making it worse.) If you're o'scope can handle the voltage, you could alternately check for the oscillation on the plate pin of the tube under test. I did later build a resistor divider so I could look at the plate pin with the o'scope, and the signal was there. I also saw it on the center pin of the plate contol pot. I tried putting a capacitor there to ground to see if it would dampen the signal, but it didn't have any real impact on the problem. If you are still seeing the parasitic oscillation, even with the correct setup chart and the wires separated (not readily possible on the 667), at this point, I have no real solution. Putting a 0.1uF 400V ceramic disc capacitor on the plate pin of the tube under test helped, but it had to be on the socket pin and going to a nearby ground (ie the frontpanel). But since any pin can be the plate, depending on the tube under test, that's not a viable solution to be putting those big capacitors on every socket pin. Nor did I try it, for all I know it might even make things worse. Note: It needs to be a ceramic disc capacitor because they have very little inductance Other types of capacitors have more inductance and can potentially contribute to the parasitic oscillation because of the inductance rather than inhibit it. The only solution (other than making sure the screen and plate lever selections are separate) I've found is to make sure the wires going between the two groups of sockets (between the left and right side of the meter) are separated by at least 1/4 inch is a solution that worked. How that might affect the the 667 tester, I don't know I didn't see the oscillation on the one I checked. That doesn't mean it won't happen. With the wires being bundled together on the 667, it is a potential problem waiting to happen. Another solution that might work, but I didn't try, is to replace the wires between the two socket groups with shielded wires and tie the shield to ground. But that is rather drastic and a lot of work as well as creating the potential of making a mistake and damaging the tester. Nor is doing that mod providing any assurance of it resolving the problem (which is why I didn't try it). So the best suggestion at this point is to make sure you are using the lastest test setup chart, and make sure the wires between the two socket groups on the 666 are separated. (I wouldn't try that on the 667 unless you actually have the parasitic oscillation problem, as there is a greater likelihood of damaging the tester trying to do that. If it ain't broke, don't fix it.) It's the long wire runs with the gird and plate wires close to each other that are important, they have the greatest potential to cause the parasitic oscillation. The short runs between the sockets are generally not as serious a problem as they are generally too short to generate much of a capacitive or magnetic coupling. Besides which, there is not much you can do about them. Note: Another thing to take note of is that the diode protection circuit on the meter tends too make the parasitic oscillation worse if it happens. The reason why is because the peaks of the oscillation can exceed the 0.4 Volt point where the diodes begin to conduct to protect the meter. The meter itself goes full scale at 0.2 Volts rms, or 0.288 volts peak. So under normal test conditions, the voltage across the meter never exceeds the protection diode's conduction point. When the diodes do start to conduct, they can cause additional parasitic effects by causing the impeadance of the meter circuit to shift as the signal goes over the 0.4 Volt conduction point. (The diode goes into full conduction at 0.6 volts and can rise to 0.7 volts at maximum current.) There is no fix for this (other than to remove the protection diodes, which is a bad idea. But it is only an issue if the parasitic oscillation is already occuring, which means the measurement is being disrupted anyway. Also see the companion files: http://www.fourwater.com/files/restr666.txt http://www.fourwater.com/files/666-667-mod.png http://www.fourwater.com/files/eicotesting.txt http://www.fourwater.com/files/eico666-667-repair.txt http://www.fourwater.com/files/mutualconductance.txt http://www.fourwater.com/files/eico666meter-power-notes.txt http://www.fourwater.com/files/eico666tester-meter-check.txt http://www.fourwater.com/files/eico666parasitics.txt http://www.fourwater.com/files/testertypes.txt http://www.fourwater.com/tubeinfo.htm