Buying, Restoring and Calibrating the Eico 666/667 Tube Tester by mikeday@backwater.com 22 Feb 2008 The Eico 666 and 667 Tube Testers are both one of the best and the worst tube testers ever made. The design of the tester itself is very good and even has advantages over the Hickok testers. The Eico's do a better job of testing emission and leakage. For most tubes, the Eico's also do an adequate job of testing the gain of the tube. Given the price of a good functional Hickok tester these days, an Eico can be a good low cost alternative for those on a budget. To test Compactron tubes, you will need an Eico 667 or else get a model 610 adapter for the 666 tester. The down side of the Eico Tube testers is that the tube charts for them are loaded with errors. If you are not careful and cross check the settings with a tube manual, you can easily destroy the tube. The difference between the 666 tester and the 667 tester is that the 667 replaced the old preoctal sockets with Compactron and Nuvistor sockets so that the newer tubes can be tested. Also the fuse was moved from inside to a panel mount to make it more easily accessable. The only other real difference was they changed the knobs on the selector switches from round knobs to paddle knobs. Myself, I prefer the round knobs, but that is just me. Inside the testers, the electronics is the same, so the same test charts will work for both tester models. While the 666 and 667 testers are good testers, they do have a couple of design issues that could have been "fixed". The most serious is that if there is a grid to screen or plate short in the tube, it can burn out the grid bias control. Both the grid bias and the plate load control are not available anymore. The only way to get replacements is to cannibalize another 666 or 667 for the parts. Modifications. Given the difficulty of finding replacement controls, I have added a modification to my 666 and 667 testers to protect the grid bias control. (The plate load control is not as much a problem, I've never seen one burned out.) The grid bias control protection modification is very simple and easy to do. It uses a trick borrowed from the late model Hickok testers. A lamp is placed in series between the grid control and the tube. This is done by unsoldering the wire to the center connection of the Grid control, and soldering one leg of a minature bayonet lamp socket to the center connection of the grid control. The wire you disconnected from the grid control then goes to the other leg of the lamp socket. When done, insert a type 47 lamp (6.3V at 0.15A) in the socket. Make sure that the socket is well mounted so that it doesn't move around and short out to anything. A pair of back to back 1N4148 diodes across the meter can help reduce meter damage by limiting the current drive into the meter when the voltage is excessive. Connect the cathode of diode #1 and the anode of Diode #2 to the meter + terminal. Connect the anode of diode #1 and the cathode of diode #2 to the meter - terminal. A common problem is a bent meter needle caused by a shorted tube or incorrect test setup for the tube. This can cause the meter needle to slam into the right stop bending the needle. The diodes help to limit the current into the meter by limiting the applied voltage to less than 0.6volts. The normal maximum voltage across the meter is 0.1volts, well below the limit. Since the 1N4148 diodes don't turn on until about 0.4volts or so, they have no effect on the normal operation of the meter. I've also seen some people put a 1uf to 10uf 6volt capacitor across the meter to slow down the current surge and thus how quickly the needle moves. I generally don't do that because it requires an electrolytic capacitor which can get leaky over time resulting in incorrect meter readings. A potentially useful part to add though would be a 0.1uf to 0.5uf ceramic capacitor across the meter. This won't help much to control the current surge, but it will help to reduce parasistic RF oscillations or external RF from affecting the meter reading. However, it can also change the meter readings because it acts as a storage device which will supply more power to the meter than would occur without the capacitor. Also since the diodes and capacitor can act like a detector in a strong RF field, it can actually make the problem worse in certain situations (like operating next to a transmitter tower). As a general rule I don't add the capacitor unless there is a real need to do so. Buying the testers. When looking for a 666 or 667 tester to buy, there are a few things to watch out for. The most important is that the meter should work. Power up the tester, and press the "Line" button. Moving the "Line Adjust" control should cause the meter needle to move around the center position on the meter. Also make sure that the needle is not bent and doesn't stick. Note: If the meter cannot be centered, this can sometimes be fixed (see the section under restoration), but the tester will not work properly if the meter can not be properly centered. Also if the meter doesn't move at all, it is likely that the meter is broken and the tester will only be good for parts. The next important thing is that there are 14 screws around the edge of the tester that hold it in the case. Make sure that all the screws are present. One or two screws missing isn't a serious problem, it may mean that they were stripped by someone who over-tightened the screws. The reason for checking the screws is that if there are no screws, or only a couple of screws, it is likely that there is something wrong with the tester, and the last person who worked on it didn't consider it worth even bothering to put all the screws back in (they might have kept it as a parts unit). Also check for rust on the screws. Light rust is not a serious problem, but heavy rusting is an indication that the tester was stored in a damp area and may have corrosion internally. For the same reason, carefully check any tester that has new screws, as it indicates that the original screws were not in the tester or were rusted at some point so someone put new screws in possibly to hide the problem. It may just be that someone did a restoration on the tester, so don't worry about it too much if the tester seems to look ok otherwise. The ability to adjust the meter with the line adjust control is probably the single most important test for the tester. Check that all the knobs are present. Missing knobs are an indication that the tester was a parts unit. Although they may have just gotten lost. The knob on the merit switch does have a tendancy to pop off or even crack on a tester that has seen some use. Also check that the two plastic covers over the roll guide are intact. It is normal for these to be yellowed with age, especially if the tester was exposed to sunlight for any length of time. If these windows are missing, it is an indication that the tester may have been used as a parts source for another tester. Again, don't worry too much about the missing windows, the tester will still work fine if that is the only problem. The roll chart should be present, if it is not, this is another indication that the tester may have been used for parts. The roll chart should not bind when you turn the knobs. There may be some "tightness" and a bumpy feeling as you move the chart. That is not a problem. They used a rather stiff spring and things tend to bind a bit if it hasn't been used in a long time. There should not be any rips or tears in the chart. You don't actually need the internal chart, as you can use a chart book instead. Myself, I normally just use a chart book (so that I can "fix" the chart errors as I discover them). A missing or damaged roll chart does not itself prevent the tester from working, but a missing roll chart is an indication that the tester was used for parts so there is likely to be other problems with the tester. Check that the meter face is not cracked. This is a common problem, and it does not prevent the meter from working, but it is a distraction, and an indicator of potential trouble with the tester. It is normal for there to be some minor surface scratches and/or scuffing on the meter face if the tester has been used. Use eyeglass or camera lens cleaner to clean the meter face. Don't just wipe the meter with your hand or a dry cloth as it will pick up a static charge that will affect the meter reading. Make sure that the cap connector is intact. The bakalite insulator on these sometimes breaks. It does not prevent the tester from working, but does make it more dangerous and more difficult to use the tester. Also, the attaching wire should be flexible and without cracks or missing insulation. This is not serious if you are intending to restore the tester. You can replace the wire with Beldon #8899-100 black 18AWG 5000Volt test prod wire (or equivalent). If you can't find the test prod wire, you can use a lamp power (or speaker) zip cord and strip it down the middle. Cut off a 15 to 16 inch (about 40cm for you metric folks) length of wire and use that for the connector wire. The power cord should be intact and flexible without any cracks or missing insulation. If the power cord is stiff or cracked, it can easily be replaced with a lamp power cord or a two wire extention cord with the socket end cut off. So this is not a serious problem, but it does mean that you will have to fix it. A stiff or cracked power cord means that it was not stored well and so may have other internal problems. Although keep in mind that these testers are 50 to 60 years old, so there is going to be some stiffness in the power cord even if it was perfectly stored. If you can, check that the tube sockets are all clean and that the tubes fit in them tightly. The Octal socket, 9 pin minature and 7 pin miniature sockets are the most important ones to check as they are the ones that get used the most. For the 667 tester, also check the 9 pin Novar socket. This socket gets used for two different tube types that have different pin sizes. Forcing the large sized pin tubes into the socket will distort the socket so that the more common thin pin tubes won't be properly seated in the socket. You may have to replace the socket (or live with holding the tube off-center to make a good connection while testing it). On the 666 tester, also check the preoctal sockets (four large sockets top left corner). This assumes you will be using the tester to test old preoctal tubes. (Presumably this is why you bought it instead of the 667 tester.) They should all be tight. Keep in mind that the four pin socket is going to feel looser than the seven pin socket because there are fewer pins for the socket to grab on to. The sockets are a pain to replace, so it is desirable to get a tester that doesn't need them replaced. Restoring the tester. As long as the tester is in otherwise good condition, there is little work that needs to be done to an Eico 666 or 667 tester. Use mild soap and water to clean the front panel. For the meter face, use eyeglass or camera lens cleaner. You want to use something that is non-abrasive, non-corrosive, does not leave a film, and is anti-static. Since the meter cover is made of plastic, it will easily pick up a static charge even just by touching the meter face. This can cause the meter to read incorrectly. If the meter was working but then stopped returning to zero, try spraying some eyeglass or camera cleaner on the meter face. If the needle drops back to zero, it was being affected by a static charge on the meter cover. You can use contact cleaner to clean the switches in the testers. I use Deoxit DP5. Be careful that you don't accidently spray the paper roll chart. Holding a piece of paper or cardboard in front of the chart can help prevent that problem. Also be careful of not spraying too liberally or you will end up with a dripping mess all over the place. The plate control can sometimes develope contact corrosion on the resistor coils inside. You can pop off the back cover with a screwdriver and spray it with contact cleaner as well. After you are done and it has dried out, just snap the cover back on the control. The grid control is more of problem because of the switch. But, you can pop the back cover off of it too. First move the control to the center position (around 50 or so). Then pop off the back cover with a screwdriver. Be careful not to damage the locating tab located above the center connection of the control. Spray the control with contact cleaner and move the control to work it in. Once the control has dried out, make sure the control is positioned in the center (around 50) and carefully put the cover back on. The locating tab will fit through the hole in the back cover (be careful not to damage it). Be very careful not to spray the two calibration controls, R7 and R18. These are snap in potentiometers mounted on plates. One (R7) near the filament switch, and one (R18) near the line calibration control. These controls are not sealed, so over time they tend to collect dirt and grime. As a result, they become "leaky". This is the single biggest cause of calibration failure in the 666 and 667 testers. Luckily it is easy to fix. You can either replace the controls with new ones (100K ohm 1/2 watt linear trim pot), or you can clean the existing pots. If you decide to replace the trimmer pots, use sealed ones so that they don't collect dirt like the original parts do. To clean the existing pot, you will need to first remove it. To remove the pot, first carefully mark down the connections to the trimmer pot, then unsolder and remove the connections. It is possible to remove the pot with the connections intact, but I don't recommend it for a novice. Use a pair of needle nose pliers or a screw driver to gently compress the two snaps on the control shaft and push the control out of the mount plate. There are four bent tabs that hold the back shell on the pot. Unbend these tabs enough to remove the back shell. I use a pocket knife to pry them loose, then a pair of needle nose pliers to open them up. The shaft and front snap-in plate will also come out at the same time since the back shell holds it all in place. Now that the resistor surface of the potentiometer is exposed, take a soft cotton rag (such as an old tee-shirt) and soak a portion of the rag in isopropal alchohol. Now carfully wipe the surface and surrounding area of the resistor area of the control. *DO NOT* scrub it. If you scrub it, you will damage the resistor. Just very lightly wipe it clean. Once you have cleaned the control, put it back together. Place the shaft back in with the stop "ridge" on the back of the control shaft 180 degrees away from the middle connection of the pot. Place the snap in plate over the shaft on the front. Now place the back shell on the back of the pot. Bend the tabs back down with a pair of pliers. Make sure that the cut out on the backshell is over the pot connections, or you will short them out. (Yes, I've done that a couple of times.) Luckily it doesn't hurt the tester other than it won't work until you fix the mistake. Reinstall the pot back in the tester. (Hopefully you remembered to note down how it was connected.) Once you are done cleaning (or replacing) the calibration pots, you will have to recalibrate the tester (see the calibration section below). There is a 10mfd 150V electrolytic capacitor located near the filament switch. This is connected between the chassis ground [positive] and the power rectifier [negative] (which is also connected to a 270K resistor). There is also a 10mfd 25volt electrolytic capacitor located on the transistor test switch (this is only used when testing transistors). It is usually good to replace electrolytics in old equipment when doing a restoration. The electrolytics dry out and can get leaky over time and don't work as well (if at all). The precision carbon film resistors will usually not have any problems. There are four carbon composition resistors as well (round brown resistors). These can sometimes go bad, although it is rare. The two wire wound resistors are usually trouble free. There are two diodes in the tester, one selenium rectifier for the tube circuits, and a germanium diode for the transistor test circuit. I've never seen these go bad. If you do need to replace them, the germanium diode can be replaced with pretty much any other good quality germanium diode. The selenium rectifier can be replaced with a silicon rectifier such as a 1N4004. However, there is no need to replace them (and I do not recommend it) unless there is a problem with the original parts. The two replacable items in the testers are the fuse and the pilot lamp. As long as you are doing a restoration, you may want replace the pilot lamp. The pilot lamp is a type 47 (6.3volt 0.15A) located just above the meter. This lamp tends to burn out rather quickly. You can replace it with any small bayonet 6.3V lamp, although I wouldn't use anything more than 0.25A as they will burn out even faster, and will place an added load on the power transformer. You can alternately use a higher voltage lamp such as 12.6volts, which will make the lamp last much longer (as much as 10x), but it will also reduce the brightness of the lamp as well. There are also some led replacement lamps starting to show up (or, you can make your own using a dead small bayonet lamp). Using an led would mean never having to replace the lamp again in most situations. The lamp is not required for the tester to operate. It's only real purpose is to let you known that the tester is turned on. The fuse for the 667 is accessable on the front panel, so you don't need to pull the tester from its cases to access it. On the 666 the fuse in inside, mounted next to the power transformer. I have never had the fuse in the 666 tester blow on me, so having it located inside is not a big problem that I can see. Charts. The last chart Eico made for the 666 tester was 666-07. It corrected a few errors and changed the way most dual triodes were tested (making the test more cumbersome, but more accurate as well). The most common chart found in the 666 tester is the 666-05 chart. Care must be used with this chart as it has a number of errors. In some cases it is serious enough to destroy the tube. The type 34 tube is listed as having a heater voltage of 8.0V, however this tube has a 2.0V heater, setting it to 8.0V will burn out the heater in short order. The type 35/51 tube is listed as having a 25V heater. However, this tube actually has a 2.5V heater. That dropped decimal point will cause the heater to instantly burn out if you set the filament switch to the specified setting. The same problem occurs with the type 57 tube. It is listed as being 25V, but the actual filament in the tube is 2.5V. The type 816 tube is another one with a dropped decimal point. The chart says 25V, but it is really 2.5V. Always check a tube manual for the correct voltage setting rather than relying on what the Eico chart says. They had good electrical engineering, but their document quality control was bad. The last Eico 666 chart I've seen that was published was produced by Coletronics in 1978. It appears to mostly be a copy of the 666-07 chart. It does not have the old preoctal tubes listed, so you will need to find a copy of an old Eico suppliment chart to get the setting for the older tubes. However keep in mind that the settings have some errors. The filament voltage errors listed above are the most serious ones that I know about. If you note those on your chart, at least you won't blow up those tubes if you try to test them using the original stated chart values. Over the years I've found many of the settings on the Eico to be less than optimal for the tube being tested. In some cases the setting will show the tube to be much better than it really is, in other cases it will cause a perfectly good tube to be shown as being bad. I have my own "tweaked" chart that I use and have been periodically threatening to make an electronic copy of it and providing it to the rest of the world out of the frustration with the charts supplied by Eico. Testing Unlisted Tubes. Eico also has in its manual a procedure on how to develop test settings for tubes not listed in their charts. While these settings are a good place to start, don't rely on them to give you a valid test result. The only valid way to determine the test settings is to get a known good tube that has been tested on a known good calibrated tube tester. Myself, I use a calibrated Hickok 533 tester as my reference. I then average the readings from ten known good new tubes (if I can find them) to get my reference setting. Another important thing to know is that tubes have a wide range of varience in the readings +-10% is normal, and some tubes have even wider range of readings (such as pentagrid converters). They can also have significantly different results depending on the tube tester they are being tested on (pentagrid converters are particularly a problem). Most manufacturers target a tube at 110% emission when new, with a tolerance of +-10%. The good manufacturers would sell the tubes that fell out of the range to the second tier sellers who would relabel the tube and sell them under their own brand name. The important thing to know is that even a new tube can have a wide difference in the test result from the declared average perfect tube. There is nothing wrong with the tube and it will last just as long. However since people don't like to see a tube that shows a test result 'below' the perfect average setting, the tube manufacturers set the "average" test results to actually be the low end of acceptable test results for a new tube. What that means is that the vast majority of the tubes tested will fall in the 110% test result area. People feel much better about test results between 100% and 120% rather than 90% to 110%. There is no real difference, it is just playing games with the numbers to make people feel good about what they are buying. This varience even shows up in "good" testers like the Hickok tube testers. Early test settings were developed for the average result of a new tube. However poeple complained when their new tube measured below the average result. "Your trying to sell me a used tube as new!" As a result, later test settings were developed to be at low end of acceptable for a new tube. The importance of knowing the variablity and the target they try to build towards is knowing how that reflects in determining the settings to test the tube. To transfer the measurement made on a calibrated Hickok tester (which provides a reading in micromhos) to an Eico, divide the measured micromho reading by the average reading from the Hickok chart. This is the percentage you will setup on the Eico. An alternate approach requires more skill and knowledge of how to set up the test. Get a known good new tube that is in the center of "new" readings. Then setup the Eico to show the reading at 110% (or offset the reading by how much the tube differs from the ideal average tube). Getting the readings will require setting up a test jig or using a curve tracer to determine the characteristics of the tube in question, and understanding how the results relate to the appropriate documents for the "ideal" tube. Since most people don't have the equipment, knowledge or skills to do the later, it is much easier to do the former and transfer the readings from a known good calibrated tester. Assuming of course that the settings on the tester being used as the reference are correct as well. You should keep in mind that some tubes are particularly difficult tubes to test. Pentagrid converters are in this class of tubes. Tube testers do not normally have the circuits to properly test these tubes. Pentagrid tubes are normally operated at radio frequencies with high impedance circuits. Further an important aspect is the interaction between grids of the tube which is how it performs its work of converting the high frequency RF down to the intermediate IF frequency. In addition, these tubes are normally not running very hard, so they can operate at lower emission levels than other tubes and they can deal with a much wider varience in tube gain. The result is that the readings for these tubes are more variable, and the acceptable operating range is much wider. Thus it is not uncommon to see pentagrid tubes as much as 150% gain above the average declaration. Test settings for the pentagrid tubes can vary. Some of the early settings specified for the Eico applied filament voltage to the signal grid in an attempt to measure the grid interaction. I have not seen this to be of much help, and it tended to result in irratic test results. In my revised test settings, I revise these to more "normal" test setups. Power tubes can also be difficult to test. They can require high filament current and high plate current to properly test these tubes. Most tube testers will not properly test these tubes as they can not provide the voltage and current levels that the tube normally operates at. As such any test result is really only going to tell you that the tube is not yet dead, but it may not tell you much about how much life is left in the tube. Calibrating the tester This is very easy to do, all you need is a voltmeter. With the tester unplugged and in normal operating position, adjust the meter needle to zero position using the mechanical adjustment on the face of the meter. Remove the tester from its case. ***WARNING*** Caution! Be careful, there are lethal voltages present and exposed with the tester out of it's case. If you don't know what you are doing or feel uncomfortable doing the calibration, get a qualified technician to do it for you. ***WARNING*** Connect the voltmeter across the 130Vac connections of the power transformer. Normally these connections are at the power fuse for one side, and the outside connection of the line adjust control for the other connection. When the control is centered, you will see around 130Volts +-10volts, depending on the current line voltage. Press the reset button to release all the push switches. Set all the levers to position #1. Set the grid and plate controls to 0. Set the filament switch to 6.3. Set the Transistor Test switch to "Tube". Start with the Line Adjust control in center position (120V). Adjust the line adjust control until the attached voltmeter reads 130 volts. Press the "C" push button so that it latches. Touch the cap lead to the chassis. Adjust R7 so that the needle is on "0" of the inter-element leakage scale of the meter. Press Reset to release the "C" button. Press down and hold the "Line" button. Adjust R18 so that the meter needle is centered over the Line Adj. mark at the center of the meter. You are now done calibrating the tester. Unplug the tester and remove the voltmeter connections from the tester. Place the tester back in it's case. That's it, you are done. You now have a fully calibrated and functional tube tester.