Crack noise, or pop noise with electron tubes

Jac van de Walle

ISO9000 is a worldwide used quality system, widely abused by companies that only qualify for it, in order to write ISO9000 on their products. So there is a lot of inflation today, on what is called ISO9000. However, the basic idea is very good. It says quality is not absolute, it is relative. What does that mean? Lets say you produce nails, to hammer into wood. You say they are 30mm long +/- 5mm and you write that on the box. 100% quality means none of them is too long or too short, and the average is 30mm indeed. You competitor produces nails that he claims are 30mm +/-2mm. However, when you check those out, it appears 10% are 27mm. Which means too short. So he has only 90% quality though his nails are more precisely build then yours. However, you have 100% quality.

Here is how ISO9000 defines a basic Quality System. It must be on paper. In there is defined: Say what you will do. Then you go to work and must do what you said you would. Then control yourself. Document the results, and take corrective action if needed. Then if corrective action was needed indeed, the way to go is again: Say what you will do, then do what you say, control the resukts AGAIN etc. The only way to end a issue is either change the product requirements, or solve the issue. If both can't be done, you must remove the ISO9000 from the product. This, and nothing else is quality.

1) Bad Job with socket glue.

Let's begin with a silly error mechanism, seen often enough though. If the glue inside tube sockets forms a bridge from the anode to one of the other electrodes, the result is a crackeling or pop noise sound. Not a very loud sound, but it is there. It's a kind of random sound, like .ptpt...ptrt...rtrt, and it can come and go, with another air humidity

2) Bad circuit design

Please don't take this personal, but this article is about a commonly made mistake. Probably "the mistake", alsways seen with Cascade amplifiers It is such a very common mistake that there is some probability your products suffer from this as well.

The mistakes are often this:

  1. The voltage between cathode and heater is chosen close to or at the maximum specification of the tubes. (wrong approach to high quality designs). With power devices (all things that get warm), going to maximum limits means reducing lifetime to a minimum, and increase problems to a maximum. This is per definition, be it a car engine, a human being, a light bulb, and also an electron tube.
  2. "Positive and Negative" of the datasheet is mixed up with each other.When we read Uh-k is +50V ... -150V, that means for Uh-k=-150V you connect + to the heater and - of your meter to the Cathode, and then read -150V. (Like this, and no other way!) May seem a silly advice, but it is a popular mistake, since the positive voltage limit and the negative voltage limit is another one.

    We see these mistakes just as well with expensive brands, as with DIY products. For some reason However, ECC88 designs are subject to this often, but... particularly this tube is not forgiving mistakes. Problems come after some use period, and not in the beginning. Let me point out here, that the difference between professionals and DIY, is that DIY are interested to talk about it, and professionals not. I have seen this so often, that I say this is a rule with almost no exceptions. Almost any professional we see making a design mistakes, hides the mistake, hides the schematics, ignores the mistake, or even try to tell you that using a maximum voltage limit is something normal. All on the end where they are proven wrong, they say their products are very delicate and will only work on "so and so" brand of tubes, that never developed a problem with that since 25 years. So what they are doing is work AT the break down limits, and they find that only Telefunken tubes will survive that. However, my definition of a good design is, it works excellent with any normal tube, and my dear friends, THAT is real designer craftsmanship.

    What you read in this small article is not a guideline to professional design, but I try to create awareness, that a professional design needs more than being a "tube doctor". Best tube circuit designers are those with knowledge what quality really is. What defines a quality circuit design, what defines product product, and what defines quality components.

    The aim of this article is to explain this particular mechanism of crack noise and pop noise, and hopefully help some designers to get their feet back on the ground.


Crack Noise

 

Tubes like ECC88 and 5687 are nice for Audio, because they offer good gain and low output impedance at the same time. They even can do so at low voltage, which is convenient for the power supply. The nice combination of good gain and low output impedance is achieved by relatively good heated cathodes, and the isolation layer between cathode and heater is always a bit more stressed as with smaller cathodes.

 



This schematic shows the typical operating point of ECC88

Now, with two of those triodes in one bulb, many designers have the luminous idea to build what is called a cascade stage, or a mu stage (these are not the same).

With a cascade stages and Mu stages, many advantages are achieved, like stability, more gain, lower distortion,. etc, but it is not the scope of this article to explain this. The audiophile advantages are clearly there.

However, is there a disadvantage? Yes, there is! The problem is, the voltage between filament and cathode must always be as low as possible. Though this voltage is theoretically allowed as high as possible, positioning yourself here, is THE classical cause for spurious crack noise (very loud, speaker damaging kind of sounds), or some softer background noise, of the kind you cannot localize well.

 

Here is a typical schematic I found. I am not sure if it is works, so better not re-build it.

What you see here is a cascade stage with 5687. The cathode of the tube half "U2" is at 125V, since Cathode-to-Grid voltage if 5687 (also ECC88) is just a few Volts. However, the cathode of tube half U1 is at 5Volts. So we just put the results of that in the table below

 

Line #
Filament Ground reference
Cathode of U1 to filament
Cathode of U2 to filament
1
0V
5V
125V
2
30V
25V
95V
3
60V
55V
65V
4
50V
45V
75V
5
100V
95V
25V

So, as you can see from this table, whatever you try, either tube half U1 or tube half U2 will be seeing a relatively high heater to cathode voltage. These are DC voltages. With most tubes datasheets, there is given the peak voltage. This means, they do not want to know how you bias the tube, you just are supposed not to exceed that peak voltage. (The result is a CRAACCCKKK noise). Let's look at the RCA datasheet of 5687 here. As a designer of cascade circuits, of course you will immediately look where you always look first (ahem....) and you find the right number immediately on the second page. It is 100V peak value.

Now the mistakes possible with the heater supply are various. We have to list it up a little bit in the text mode. Sorry, I have not much time to draw pictures. Still the text also describes what I see go wrong to often.

Possible mistakes:

1) All 6.3V filaments of the whole amplifier are connected together, and the ground reference is.... (none). So the heater winding is connected only to the tubes and that's it. VERY hard to explain to some people why that is wrong. The error is here the undefined DC potential of the heater wires. So in the above table you are somewhere on line 1...5. Just somewhere, and nobody knows. Most likely some leakage effects inside the tubes and transformer will create some potential. Note that the 6.3V AC winding of a bad transformer has a relatively high common mode voltage on it, like 300V AC is possible. This voltage (300V?) has a very high internal resistance. So the moment you load it with something, it drops. However, HERE is is loaded with the very high resistance of the isolation layer between cathode and heater. You feel the problem coming up? That layer is now stresses exactly with that voltage it doesn't like, and limits the leakage voltage of the transformer that way. Needless to say that damages the tubes over time. That could produce some undefined crackling noise, or a noise like ptt...ptt. Sure not with all tubes, and you might end up tube picking, and blaming good tubes for problems. (AND YES, professional companies LOVE this mistake). Just for your interest, de solder all connections from the 6.3Vols transformer. So you have two ends, with 6.3V in between, and they aren't connected to anything. Now use a vera good, very high impedance voltmeter, set it to the highest voltage range, and measure between one end of the 6.3Volt and ground. (transformer must be "on"). What do you read? I bet it is not zero volts. So latest after seeing this, you will believe it.

2) The 6.3V filaments are grounded at one end. So you are at line #1 of the table. Please look there, and you will have no further questions.

3) You have DC heating. The electronics is floating. Same error as 1)

4) many more errors are possible!

 

SOLUTION

In the table above you see that the best DC level for the filaments is +55V.

1) They may be some low impedance points in the amplifier where you can connect it to. With AC heating, the best is first make a center tap on the 6.3V by means of two 1K resistors. This center tap is connected to 55Volts. Do not use grid circuits of other tubes for this, since the parasitic capacitor's current of the 6.3V winding will flow through this. However, the cathode voltage of the end tube is a good one. With tube like 2A3 you have usually 45...60V. Tubes like 300B have around 80V, and you can make a small voltage divider on it. That can be with 0.25 Watt resistors, since any current is not drawn from it.

2) Another way to get the +55V needed, is make a voltage divider on the mains voltage with 0.25 Watt resistors.

3) many more solutions are possible!

Subscribe or unsubscribe our Mailing List
© COPYRIGHT NOTICE  All Rights Reserved