Since 1993 Copyright Notice

We have to begin with some words about matching.

Some limited use for testing with fixed grid voltage is with matching. That is because depending on transconductance of the tube, any differences between two tubes become artificially "magnified". And even that doesn't always take place. It is completely normal when NEW tubes at fixed grid voltage show lower than average plate current, but higher than average transconductance, or vice versa. Reason for this is to be found in anode distance tolerance, becoming visible to the user as grid voltage tolerance. (more about this later).

At fixed grid voltage testing and matching, what people do: Plate current could be the SAME, and they feel great about it. They call it a very well matched pair. But then, verify these so called pairs at another grid voltage, and you will see, they totally off balance. Sometimes just a little bit, sometimes very much, which is all completely normal. Specially with tubes like EL34 or KT88 this causes great problems, because sellers match them at low voltage like 250V, and the amplifier uses them at 450 or higher. That can give differences of 35% easily. We can be lucky, most auto bias circuit have no problem with that, and the amplifier works fine. Just trying so say here, such matching is for the birds.

For those, who do have a tube tester with controlable settings, I can only advise to TRY IT OUT. This is so totally easy to do, and you will fall off your chair if you see the enormous difference, always thinking you did it right. Just take any of your matched pairs EL34 or KT88, which you matched so perfectly at 250V, using fixed grid voltage. Now take this pair and test it at another fixed grid voltage, and you will see now, there some pairs stay pairs, and some others will drift apart. Still talking about 250V. In the amplfier they are used anywhere from 350 t0 480V. How about verifying that? Is the pair also a good pair at anode the voltage of this amplifier, using 480V and fixed grid voltage? I can tell you: Forget it.

Some very few tubes have always the same working point, such as 2A3, or the 45. That makes it easier. For most other tubes, the working point depends on the amplifier design.

It is possible at all, to match tubes in such a way that they are a good pair at any random working voltage or plate current? Well of course it is. Just not with fixed grid voltage. The method for this is, set the tube for the SAME Plate current, at 250V, take use any grid voltage this needs. Then, match them for transconductance. Which value should be IDENTICAL, and not further below average than 15% or so. These will be a good pair, regardless test voltage and current. (Including fixed grid voltage). You will see a few things now: 1) This is a lot more work to do. 2) You need a larger quantity of tubes to make good pairs. 3) Some tubes can't be matched at all.

What is the use of fixed grid voltage testing?

Differences look at least 10x larger as they are in a real circuit, which is GOOD, because it t makes work easier, and you can let 10% difference pass, and get fine selected tubes. Now there are better ways for matching, which I use myself, but the fixed grid voltage method is not a bad one. That is because for instance with 10% difference, you can be sure, there is no difference at all in an auto bias amplifier circuit. But matching is only matching, and testing for tube condition is something else.

The major problem with fixed grid voltage "testing" (which is not real testing at all...) is that you still have to say which are good and which are bad tubes. So in some effort to get a result, it is unavoidable to say that Bogey tubes (which test as average as possible) are 100% good tubes, and those with more plate current are "better" and those with less current are "weak". An that is where stupidity begins. So people even prefer those with 130% plate current, because they are so "strong". And they kick aside the weak ones with 70%. And no, people are not surprized at all when even such (weak?) 70% tubes play fine. They will just say these won't last long.

Hammering on pentodes

What a title for this paragraph:) This demonstrates how silly fixed grid voltage matching is. I know some Ebay sellers use this hammering method. It works only on pentodes, because they react on this with a smalll deformation of the screen grid. The screen grid has very large windings, which are stretched very little, so this deforms easily. You repeat this entirely on you own risk! But it works like this: Take a used up EL34, which is to be thrown away. So a tube, where the test result with fixed grid voltage is too low. Now hit the tube on the palm of your hand. You don't even have to do this very hard. Rather repeat it a few times. Now re test the tube, and there you are: The plate current has increased. The intention of this is only to show how little value fixed grid voltage testing has. Since the tube tests BETTER now, by this method, while for sure you know it is worse as ever before. Even so you can do this on a Hickok tester, you will see the same.

The RS1003 data sheet. Clear words by Siemens

That is because if you do so, you get the functionality from the tube which you can judge if you accept it not.

So you see, at fixed gird voltage, a 70% plate current tube (called weak) tube, just take the correct method instead. Set it for 100% plate current, do as Siemens tells you. In THAT CONDITION when the gain is good, and Gm is good, and all the things are good, then the tube is good. Why does is to show 70% with the wrong test method? Well the only thing you had here, was grid voltage tolerance, which is completely normal, as long as all other parameters are good. Even so, the higher Gm, the higher the grid voltage tolerance. But the thing is, high Gm tubes will have high gain as well. Particularly such, despite the higher tolerance of grid voltage, will bias nicely in an auto bias circuit, because the feedback on the auto bias DC current works very good.

Now you may still ask, how come it is at 70% with fixed grid voltage then? I will explain later in this article how this is possible. But asking this question is the whole cause of the problem, because it suggest there must a reason for such a problem. Yet fixed grid voltage testing by itself is the problem, and it is written on NO DATA SHEET at all you need to do so.

An exeption is a hand full of tubes with applications where specified grid voltage is important. Such as battery radio tubes, where the bias voltage is generated by connecting the heater to a negative 4V battery. So the heater is at -4V, and the grid voltage is also -4V. These tubes take the bias voltage straigt from the heater. Even so, when the heater battery gets low, and heater voltage would drop, also grid voltage would drop, keeping up the plate current to some degree. What an excellent idea that is! But it never was a break through. These stayed very rare tubes.