The first digital tester I saw, was the SOFIA tester by Audiomatica Italy, but I have to say, this tester can test the tube in pulse mode, for making tube charts, as well as test the tube in an analog way, by using fixed DC voltages to warm it up fully, and do quality tests and precision testing of parameters. Besides this was by the company Audiomatica in Italy, who never intended it for DIY users. I use it as a reference still today.
Methods I have seen so far:
- Fully programmable power supplies. The Sofia tester by Audiomatica now is not the only tester of it's kind. (End of production in 1995). Also the German ROETEST works like this. Such power supplies are quite complex. The Sofia features four power supplies, which are analog controlled, so they have no digital inputs. The micro computer has D/A converters with 16Bit, 20kHz. These can control the power supplies, and they give any kind of pulse, voltage ramp up, or steady voltage needed. The power supplies are heavy, because the can do 2x 700V 250mA continuously. I have never seen anything like that before. Amazing is also the small size and the elegance of the PCB. The later AT1000 works in a similar way, but has only 500Volt, which is enough for parameter testing of most tubes, but it is a bit poor for tube curves. The ROETEST you can see as a much upgraded AT1000, and it is a kit version.
- Pulse testers using the mains voltage as power supply. I have first time seen those from a French enthusiast. He still sells them on Ebay for 65 Euro only, but it's without transformer which you need to get yourself. Around 2008 I bought his kit, but I had no time yet to build it. This gives really the simplest electronics, and the circuit board, all I can say, has almost no parts on it! If you look at it. Crazy and amazing. And yet it is a complete curve tracer. Too bad the operation is in French, so you have "Tension de grille" instead in grid voltage.
- Pulse testers generating it's own voltages. The first is the u-trace, by the genius circuit designer Roland Dekker from Holland. Perhaps the oldtimers remember the 'videoscope' from the 1970's which was a big DIY circuit board, which was a digital oscilloscope, using a normal Television (via antenna signal) as a screen. As a project, Professor Ronald Dekkers re-build this into an 8-pin single chip u-processor, many years later, and the called it u-scope. Look here for his board. This is what the u-scope looked like. His next, similar project was the u-trace. The hardware is one of the mist genius pieces of electronics I have ever seen. I am not just saying so! Roland published all schematics, and it is just amazing to see how he got all of that functionality with so little low power electronics, on such a small board. I know a little bit about electronics, and this board is to my opinion a master piece. The next one of this kind is the E-tracer by Chris Chang from Taiwan. We are in email contact about this, and he made this tester as he wanted originally to use the u-trace, but he could not convince Roland Dekker to add the features he wanted to have. So he made his own.
Full power testing vs. Pulse testing
This is am important section, and we need to speak about some of the disadvantages of a pure 'pulse only' tester. This comes later. We look first at the advantages.
What is good about a pulse tester:
- Reliability. If you ever build an electronic power supply yourself, which combined high power, with high voltage and high current, you will know about the many smoke clouds this produced during the design and test phase. And then, using that to test unknown, and sometimes shorted devices, and is really a major challenge. What is good about 'pulse only' devices, they do have high voltage and high current, but not high power. So things don't get hot, and this combination of HEAT with already stressed electronics is simply avoided. So reliability is a lot higher that way. And with a lot, I mean A LOT! This advantage you will learn to value when the equipment gets older.
- Speed. Good software can be really fast. My Sofia is perhaps 10x faster than the AT1000.
- Cost and Size. Cost is obvious, but also size is important. I sold my Neuberger RPG375 analog tester because of it's totally crazy weight, and my favorite Hickock (750) is so large, it take the whole workspace. Compared with this, the Sofia is already small, but the E-tracer and u-trace are a dream.
What is the weakness of a pulse tester:
- Cold testing. There is little awareness for this amongst designers of test equipment. Everybody who has a fully analog tube tester, so one that can warm up the tube anode, will know that just warming up the heater will not give the final reading. After the tube is initially working, it begins to warm up the anode, until you get thermal balance. So it is only natural that you wait until nothing changes anymore, and THIS and nothing else is a correct test result. Unfortunately, a pure 'pulse only' tester cannot do this. Let's just reverse this, and suppose I offer tested tubes, and I tell the buyer I warm up only the heater, and don't bother about the anode warming up. I just say this is unimportant. Period. Would YOU buy such tubes? Now, look at the AT1000, it has all the electronics on board, but when I click 'automatic test' it only warms up the heater and not the anode. This is so painfully stupid, I contacted Chris Terraneau from Amplitrex about it, and ask for an update. Probably the most important this machine needs. Chris considered it not important, because there is a way to work around for that, by hand. Yes there is, I know that myself, but you have to do it by hand indeed. So re-start the tests, which becomes really annoying for double triodes, as it misses also the timer function, and you need to look at it all of the time. Like this it costs me 2x 3 minutes per tube. Which is a pain when I want to test 50 tubes, and also it makes no sense to stand up and do something else 50 times for 3 minutes. This is a typical case of bad software making good hardware not work as good as it can, The opposite is with pure pulse testers, read the next points.
- There is no heat development in the device under test. An example: On the Sofia or AT1000 you can make a 230V, 15 Watt light bulb burn if you want so. On a pulse tester you can try, and even make a curve chart, but then you will see, even though a 230V bulb is tested at 230Volt, it will still not burn. Moreover, the curve chart will be totally wrong, as the bulb is not burning, even not at places where it says 230 Volt. Meaning the filament is still cold, and the current tested is a multiple factor too high. This is not a practical example for real tube testing of course, but it does show, that pulse testing is not normal testing.
- Cold testing with pulse testers using AC supply voltage. I still have that kit from France, but I had no time to build it yet. Hopefully it does warm up the tube with it's AC voltage at least a little bit. Technically it is possible. Same as some of the Hickok testers have a 'lock' function for the TEST button, in order to heat up the anode, and do more accurate hot testing. .
- Cold testing with pulse testers using it internal DC voltage. We have to be fair, such testers have no option to heat up the anode, the concept doesn't allow it. So we cannot call it is missing feature. We rather need to praise the amazing results from such simple hardware. We do have to say however, this is how it is, and the user much decide for himself, if he requires the comfort of such a pulse tester, or rather the higher precision of such testers, like the Sofia, ROETEST or AT1000 (and a few more) which heat up the anode at working temperature.
Difference between cold anode and hot anode testing
Like said before, when we would sell matched tubes, and we say we match them right after heater warm up, and we do not wait for thermal balance, because it is 'not needed', this would be hard to accept.
Many users of tubes will not have access to a tube tester which can fully warm up the anode, and not be aware how large a difference this can make. The heater itself has must less power than the anode. So most of the tube heat is generated in the anode, and not by the heater. The warmer the tube, the better it works, but also leakage (if any) is only present in hot condition.
Reasons why hot anode testing is better are:
- The anode expands after warming up. It's simply a metal box, and plate distance gets larger. If you know, plate distance tolerance in the tube factory, is the #1 factor for parameter tolerance, then it should be clear, the tube needs to be tested with a hot anode. These little holes you see in the tube anodes, at the sides of big tubes like KT88, are intended to adjust the anode distance by hand, before the bulb is closed. With poem other tubes, the adjustment is done by pressing the plates together, with a distance holder in between.
- The heat of the anode is really very much. A tube like EL84 can be touched with only the heater on (and all other voltages off). Yet when fully warmed up, even isolated wire would begin to smoke when it touches the glass.
- Most forms of leakage and similar errors, like grid emission, only appear at full anode dissipation, and not even at 70% dissipation. Heater to cathode leakage needs often even 100% dissipation before it becomes a problem, and begin to decrease within 2 seconds after the anode current is switched off (and heater still on).
- Normal cathode working temperature needs a hot anode. Reason is simply, the heat of the anode radiates in two directions: 50% the outside world, and 50% back into the tube where is the cathode and the grids. This will significantly contribute to the cathode temperature, and the whole tube is of course designed in such a way that the cathode temperature is ideal when the tube is used at it's intended anode dissipation. Some switching tubes are designed such, that the cathode is already at ideal temperature when the anode is cold. A normal amplifier tube is not such a tube.
What difference does it make?
This depends on a lot of factors. Here are some numbers I collected myself:
2A3 RCA NOS. Tested at 250V -45V
- Only heater warmed up: 56mA
- Anode warmed up 5 minutes: 60mA
6922 RCA NOS. Tested at 90V -1.3V. Note, this test method is not even allowed by the datasheet, as fixed grid voltage may only be applied up to 5mA. We do it anyway:
- Anode warmed up 5 minutes:13mA
- Only heater warmed up: 11mA (18% error)
- Anode warmed up 5 minutes:14mA
- Only heater warmed up: 13mA (7% error)
Tube2 (system 1+ 2 tested identical )
- Anode warmed up 5 minutes:15mA
- Only heater warmed up: 14mA (7% error)
6922 RCA NOS. Tested at 90V in auto bias.
- Tube #1 reached 15mA on both anodes, after 10 minutes.
- Tube #2 reached 15mA on both anodes, after 1 minutes
Not all kind of tubes are listed here, just read the conclusion:
Sometimes it is hard to estimate how much a tube will benefit from reaching it's normal working temperature, but it seems needed with almost any tube. Tubes that need it most are: Used tubes, and tubes that have seen many (or to many...) decades of unused storage. Difference of 10% is nothing special. Particularly weak used tubes benefit more. So a tube may seem initially vera much used, almost too much. Then, after a good warming up, the tube appears not so bad at all. Or, a tube may test "bad" initially, and after good good warming up, test just good. Tubes which show no difference at all, exist just as well.
One reason why weak tubes benefit more, is also that during the heat up phase of the anode, also the current increases slowly, and from that it gets even warmer. So after 3...10 minutes, the tube draws almost normal current. Of course one may further speculate for the reasons of this, and you may call it good it bad. Just one thing we need to see clearly, it is the function of the tube tester, to show this to us. Then, judging it, we can do ourself.
Knowing this, I often use the cold testing method when I am short of time, and when I only need to know if the tube is good. In case a tube reaches 90% or more with only the heater on, I already know the tube is fine. Yet for matching that is not the way to go of course, this needs thermal balance.