The KT88 tubes is a curious tube. On the one hand, we always told not so mistreat a tube. So heat it up slowly, pack it carefully, do not drive the amplifier in full distortion, just for the fun of it. Circuit designers should not run the tubes excessively hot, use not extremely high voltage, and avoid high plate current. Power tubes are expensive items, and we want them to last long time of course. You must have seen (some?) of those requirements at already before. Well it is so curious, KT88 is the strangest exception of all. KT88 is low cost, and is mistreated in all possible ways, by unknowing users or "so called" design experts. It is amazing it goes wrong only in so few cases. What can be the reason for this? Frankly, I do not know the full answer, but a few things I can say here.
Below here, are some of the items that can cause quality problems with KT88
COSTS. Lets just go through the complete cycle of a tube. It begins with materials and manufacturing. KT88 is made from relatively low cost materials. The most expensive is probably the cathode and the first grid. So just by using those few words here, we already can say where the savings will be, and we can try to understand the effects of cost saving at this point. Things that can go wrong here, will be loss of emission, too low initial emission, emission depends too much in heater voltage, grid current, grid rattle, grid to heater leakage (hum), or a cathode to grid short. Each of these things can be tested nicely. It is however also a quality issue, meaning problems wit few tubes in the beginning, will most likely spread to more tubes when they get older.
Apart from new made Russian KT88, vacuum is normally good when tubes are factory packed. Under normal use conditions, the vacuum begins only to lower when the tube's out gassing can no longer be compensated by the getter flash any more. When the vacuum gets totally lost, the tube stops working of course, but there will be a moment where the vacuum begins to get less good, and the tube is still capable to work. Of course this is the moment to replace the tube, but it is very difficult to say exactly when that will be. At first the the tube will begin to show some blue gas cloud inside the anode. You can see this when looking inside the inspection holes of the anode. This is a light-blue haze, evenly distributed, inside the whole anode, and the anode only. It needs a trained eye. Do not confuse this with the more common blue light effects. which is a dark blue fluorescent lighting of glass, some outside parts of the anode, or edges of the inspection holes. This shows the tube has good vacuum, and even so very old tubes don't do this any more. Though this is no requirement, and two fully good, new tubes of a pair, may have one with fluorescent lighting, and the other doesn't have that. So keep apart dark blue fluorescent lighting spots from light blue overall "cloud" lighting. Only the last stands for a gassy tube.
THE AUTO BIAS TEST CARD . (Order nr: 380-080-67)
Final inspection in tube factories is really a weakness. A KT88 production line of a large factory is always automated. Meaning it has high overall costs, which have to be spread over all tubes. So when a batch is low quality, there comes this factory manager with a white coat on, and his function is rather to try to pass as many tubes as possible, instead of throw away anything which is suspicious. It just means tubes with known issues, will leave the factory still, and there is this excuse thinking. "we replace them" if there is a problem. However in real life, the tubes can be on stock with dealers, and/or pass several persons before YOU get them, and in the end the last person selling them, or the buyer, is going to be the looser. It is also for this reason, unserious, inferior brands, refuse to give good incoming inspection criteria. They just say the tubes are "good" and "well-tested" and that's all you get: Words. No numbers. You may find data on the KT88 box, saying: IP=44mA, Gm=10.000 and that's it. This is however matching data only. Though matching and quality are two fully different things. For this reason, we use the old G.E.C. datasheet for the KT88 which has such data called "Characteristic". This is not the normal use data, because Characteristic is at 250V. Nobody uses KT88 at 250V. This is just a relatively low voltage, because this is for incoming and outgoing inspection. The advantage of 250V is also, you can now use the KT88 at higher plate current as normal, and yet not reach maximum plate dissipation. We all know that at lower voltage, it is hard to pull a lot of current. (And vice versa, low current at high voltage is easy to do). For this reason, KT88 should be tested at 250V (Plate + Screen) like in the G.E.C. Datasheet, and then you should be able to pull 143mA as it says there, at a control voltage of app 15V. And Transconductance should be 11.5 at 143mA.
To make it easier, you can use the tube under test in an autobias circuit, so you don't have to worry about the control gird voltage any more. This is our test card Nr: 80-080-67 for the Russian L3 tester. The idea here is, when a USED tube can not supply 140mA in Auto Bias any more, it has probably low emission. Of course, it MAY also be such a tube which by default has a strange bias setting and not low emission at al. (So a problem with internal geometry - See the next point below here, under packaging). However such a tube can be easily detected, because an internal geometry (drop damage) problem leading to a lower bias value will also lead to a too higher Gm value at the same time. So this is really easy to see. If it is like this, you rather have a manufacturing issue. Would it be reversed, so too high bias, and too low Gm, it is either a shipment drop problem, or a gas problem. Wheras a gas problem can be seen from the gas cloud, and also too much grid current. If both are too low, it's rather a low emission tube. If both are too high, is actually unlikely.
An experienced tester will also observe the speed at which a cold tube begins to develop plate current. KT88 has a massive cathode. So when you plug a cold tube in the L3 tester, you will see nothing happen for a time which is much longer as with smaller tubes. This unusual long delay means nothing. It can take some time before the tube is only at 20% of maximum. However when plate current BEGINS to rise above 20%, it can be said, the faster it rises from now on, the better quality the tube is. This applies until you reach 80% of normal. The range from 80% to 100% takes longer with all tubes. On the other hand, it is also good, when 100% is reached faster. Particularly such tubes with not extremely strong emission, benefit from the higher heat development inside the anode, which back-radiates also on the cathode, so emission improves by greater heat. This takes several minutes. Whereas tubes with extremely strong emission don't need this, and reach 100% level a lot faster. This difference can be amazing, but you need to work yourself into this, and learn to understand it from testing many tubes. I can not compress this in a few lines easily. Very short: The additional heating up of the cathode, by back radiation of the anode, can be seen on the L3-3 when the KT88 is DC heated. The higher temperature will raise the resistance of the heater wire, and you will see the 6.3 Heater voltage rise a little bit, because of this. Of course you need to re-adjust the 6.3V before final measurement, but it is just this rise which indicates the heat development.
So you see here, the Auto Bias card tells you right away very much about the tube, when looking at the right things. Particularly the analog effect, so the way you see the bias change, you can not replace that with any digital testing.
While story telling here.... I remember when I was 15 years old, I took my old Radio Tubes to the local radio shop, and the man had a tube tester at the counter. It cost 25 cents per tube, which was the price of portion of French Fries. So you would think twice before testing 20 tubes. He could see at the speed at which the needle went up, if the tubes were weak or strong. Of course we waited until fully hot, but when he said: Oh, that's a good one, just by looking at he speed at which the needle moved, he was always right. So I adopted that method from this experienced old man. There is no absolute number for this, it is also tube type and brand dependant. So just compare some strong and weak KT88 of the same brand, and you will see what I mean.
THE ADJUSTABLE TEST CARD . (Order nr: 380-072-55)
This card is not so much for tube quality testing, but rather for matching. If the tube will auto bias or not, this is of course also very important, but the proof of that is given with the auto bias card. However the auto bias card will bias just somewhere, the higher the better. With the adjustable bias card, you can precisely adjust the bias current, and now the question is, do two tubes have the same Transconductance at the same plate current.
If two tubes have different transconductance at different plate current, this is normal. However the auto bias card may come up with this result, as it rather tests for emission first.
Normally the KT88 is used around 350...500 Volts, and few tube testers can do that. Besides there is no use to test at higher voltage, because changes this meets the amplifier voltage AND amplifier plate current at the same tube is virtually zero anyway. Unless somebody can say the tubes have to be tested at 350 Volts, 85mA. This is not real life. In reality somebody says I have a "Balanced Audio Tubes" BAT-7 amplifier and I need 4 matched KT88. So we have to match tubes for an unknown working point anyway. Can this be done at all? The answer is. Yes!
First, you need to make sure the tubes have good emission. (Use the auto bias card). If not, the following does not apply. So provided the tubes have good emission, you test them for plate current at working point "A". This working point means: fixed plate voltage and fixed Screen (Ug2) voltage. You can test the for instance at 300V 120mA. Then at this working point you find a transconductance (Gm) and a control grid (Ug1) voltage needed. Now provided you have two tubes with the same transconductance, later on these will be used at (say) 390 Volts 90mA in the BAT-7 amplifier. So how will you be sure of matching? First question is: Will the plate current be the same? For this, the transconductance at 300V plays a role. If it is an identical 11mA/V for both tubes, the auto bias circuit of the BAT-7 will provide the same grid (Ug1) volt for both tubes, and so the tubes will bias the same. This is of course a very important thing, and it is good to know this. Next question is, the transconductance at 350V 93mA will it be the same for both tubes. To answer this question, you need to investigate how much the tubes change their transconductance when going from one plate current to another. If that is the same too for both tubes, confidence these tubes will bias perfectly in the BAT-7 is very high.
Conclusion: We rarely know the real operating point in the amplifier. So the requirement is actually: The tubes must show good matching at a RANDOM operating point. This is a very hard requirement. To ensure this, the tubes should show good match at TWO different operating points, and you are usually safe. Suppose we only test for operating point "A". In that case, we can assume good matching in Operating point "B" as well, provided the grid voltage (Ug1) at Point A was the same for both tubes, as in case 1. If you find differences in Ug1, you should do a second operating point "B", and you can get CASE 2 or CASE 3 then. When matching a greater number of tubes, the tubes 3 + 4 should not be paired. For a small number of tubes, you may match them when it appears like in CASE 3.
PACKAGING. What packaging? KT88 is packed with almost no boxing at all. It's crazy. Though grid to cathode short is a known problem with KT88, it is a miracle this doesn't happen even too often. However there is another thing to mention. You should try this with a KT88 that you want to throw away. If a KT88 is dropped, the plate current will go up and the transconductance will go down. This is cause by deformation of the internal structure. Also, the distance between grid and cathode is small, and the grid is very fine, so small particles can get in between and cause a short there.