• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Uneven Filament Wear of DC heated DHT’s, Truth or Myth

Using my Utracer, I measured a type 45 that has seen some use in a DC heated amp. I don’t know how many hours of use it has had and I don’t have results from when the tube was new to compare. The book valve states 34ma at 250vdc, -50v bias, but she only gives around 25ma. That said the curves still look okay and with reduced heater voltage they’re still quite steep so I guess there is still some life left in her.

Anyway, this got me wondering about uneven heater wear, so I retested it switching the heater terminals (still measuring w.r.t the 0v lead), and the results were exactly the same which suggests to me that uneven filament wear is a myth. The low heater/ high bias of the type 45, perhaps wouldn’t make it a good example for demonstrating uneven filament wear. Perhaps the critical point is to maintain the “space charge” by not exceeding the current limit, and if that is accomplished then no uneven filament wear occurs.

A lower filament temperature would presumably reduce the rate of barium evaporation at the expense of available current to maintain the space charge. Testing the 45 at reduced heater voltages of 2.2v and 1.9v (30v-bias), the curve/gradient for 2.2v was similar to 2.5v but plotted a small step beneath. At 1.9v it followed the same principle until about 30ma, above which the gradient levelled off, which makes me think at that point the current is starting to be drawn directly from the filament rather than from the space charge. Perhaps an optimum filament voltage can be established from this approach for a given tube.

I’ve not got much past experience to work on and was wondering if others have measured differences on a tester or perhaps found different bias points when switching the filament terminals.

I would think that if uneven filament wear was truth then it would be well proven/documented, but I haven’t spotted anything.
45 tests2.png
 
Administrator
Joined 2004
Paid Member
It's not a myth, but I have not exactly had a problem with it either. I suspect the emission has dropped due to lots of hours. I have run JJ300B for up to 10K hours without a filament failure despite the dire predictions.

Reversing the filament connections should not have any effect on DHT emission, but you might want to try an external DC source for heating DHTs if you have an older uTracer as I do, the internal heater PSU is not very accurate at low voltages. I had low emissions and reduced transconductance in particular with 1.5V, 2.0V, and 2.5V filaments.
 
  • Like
Reactions: 1 user
I believe the logic is that the filament supply voltage drops across the filament, so with DC one end is consistently at a higher potential than the other, and since that would make it less negative wrt the grid, there is more emission from the end with the higher potential.
 
  • Like
Reactions: 6 users
...you might want to try an external DC source for heating DHTs if you have an older uTracer as I do, the internal heater PSU is not very accurate at low voltages. I had low emissions and reduced transconductance in particular with 1.5V, 2.0V, and 2.5V filaments.
After building the Utracer a year or two ago, I soon gave up on the internal DC supply and predominently use an external DC supply. I then realised that for directly heated types that the cathode measurement needed to be connected to the DC -ve lead, as reading from the +ve lead would skew the results.
 
  • Like
Reactions: 1 user
It's not a myth, but I have not exactly had a problem with it either. I suspect the emission has dropped due to lots of hours. I have run JJ300B for up to 10K hours without a filament failure despite the dire predictions.

Reversing the filament connections should not have any effect on DHT emission
So with 10K hours on the 300B, is that without switching the heater connections. That seems good for new production.

If it's not a myth, would you expect there to come a point after say 20k hours where it would become measurable by reversing the filament... that said, the 300B like the 45 is low heater voltage / high bias so during operation, the difference in emission between either end of the filament wouldn't be that great. Perhaps it's only worth bothering with switching the connections periodically for tubes like the ux201a, 5v heater ~9v bias
 
  • Like
Reactions: 1 user
I've just noted something from the 300B datasheet. The book values are with AC heaters, so measuring on my Utracer the comparable curve is that which is halfway between the two curves I'd get w.r.t the +ve and -ve terminal. So I guess my 45 tube is in slighly better condition. i.e to match the book value of 34ma at 250vdc, -50v bias, I should use a bias of -48.75 on the Utracer.

300B Datasheet.PNG
 
I believe the logic is that the filament supply voltage drops across the filament, so with DC one end is consistently at a higher potential than the other, and since that would make it less negative wrt the grid, there is more emission from the end with the higher potential.

Maybe is harder to imagine but the supply for filament is floating , it doesn't matter where is plus and minus
Floating supply means that the current is flowing just in the filament and is not in series , parallel or how do you want with the B+ ,so can't change the bias
If not , any direct heated tube with DC filament would be highly problematic , the two sides with different bias don't cancel each other or make an "average current" .
 
Last edited:
  • Like
Reactions: 1 users
A heater/filament is quite literary just a wire that heats up.
There isn't much more to it.

So a lot of stuff I read doesn't make any sense.
Just draw an norton/thevenin equivalent circuit.
It's just a matter of P = I² * R and following Kirchhoff's rules.
Meaning that the current trough this wire MUST be equal at the beginning or the end.

It's the current that causes heat = emission, not the voltage.

The only thing I can imagine, is that locally the resistance at the pins will be slightly higher.
Or the local temperature will be slightly higher in a bend or so.
Or when the diameter of the wire has quite big tolerances (which seems very unlikely)
Which can have an effect where the wire breaks first/wears down first.
 
  • Like
Reactions: 1 users
It is the potential difference (bias voltage) between the grid G1 and the cathode that controls the emission between the elements inside the tube.
Perhaps I was wrong when I said the above. The other posts have made me rethink.

It is the anode/plate that draw the current from the space charge primarilly and only directly from the cathode at excessive currents. The anode/plate voltage is uniform* across it's surface.

The grid 1 acts as a control between the anode/plate and the space charge. The voltage is again uniform* across it's surface.

As b_force says, it is the heat which causes emission. This would lead me to conclude that the space charge is uniform*.

*Perhaps not exactly uniform, due to geometry.

So I'm now thinking that it is the potential difference between the grid G1 and anode that matters, and cathode to G1 is not relevant. But I could be just getting confused.
 
It is true that the local emissive current is skewed to a higher value across the length of the filament, because the filament→grid voltage is lower (though this is somewhat offset by the lower filament→anode voltage). But so long as the total anode current (which adds together in the substrate wire) AND the anode power dissipation are lower than the data-sheet's Maximum Value by a good margin, the effect on the filament lifetime will be small, and usually near zero.

The lifetime of the emissive surface of an oxide coated DHT is limited by the availablility of the emissive atoms throughout the coating. The atoms (e.g. Barium) are drawn to the surface at a speed determined by the temperature of the cathode. Once all the sub-surface atoms are depleted, the emission fails.
It's difficult to meaure the filament temperature directly, so the filament voltage is used as a proxy for temperature. Too-High temperature (voltage) speeds the evolution of Barium, and wastes lifetime. But undervoltage is not helpful - the emissive surface runs out of Barium and the emission begins to fail.

Having the anode-cathode current skewed toward one end of the filament is unlikely to make much difference to the filament temperature (e.g.10mA in a 2Ω section of the filament is 20mW; but a 300B has around 6.5W of heating power) OTOH, even a small increase in heating-voltage will increase heating current at the same time, and combine to raise the temperature unhelpfully.

Rather than spend time thinking about skewed cathode current, spend time ensuring that the filament temperature is as near as possible to the value it was designed to operate at. Setting the voltage to be stable and to 5.0V - as nearly as possible (for 300B).

With AC-heating, this is much more difficult, since the heating voltage wanders around with the mains Line voltage.

If you are still worried by the idea of the skew, wire the DHT's heating pins of your stereo amp in opposite polarity, and swap the tubes over every 6 or 12 months, remembering to recheck the anode current at the same time.
 
  • Like
Reactions: 4 users
Cathode is prematurely worn (stripped) if current rating is exceeded. If a tube is operated at, or close to, maximum steady current rating (about 70 mA for Type 45), then the positive side of filament will be in excess of current rating, and will wear faster than negative side. In a typical Class A1 SET, the typical idle current is about 40 mA, well below the limit, so uneven filament wear shouldn't be a problem.
 
Do you have some tube literature to demonstrate this positive side of the filament story ?
If the filament voltage is adding and substracting from bias voltage like a voltage source in series with the cathode , how the tube could work using AC filament without massive hum ?
For a 12V filament the bias would be so skewed on one side that the effect would be very big and obvious . Then only a small portion of G1 would be used and plate too ... the tube must have different loadlines
 
Last edited:
Depanatoru,

1. How can a type 45 DHT work with an AC powered filament and have very little hum?
Easy . . .
Let's connect one end of the 2.5V filament to a 25 Ohm resistor, and from there to the self bias resistor.
Let's connect the other end of the 2.5V filament to another 25 Ohm resistor, and from there to the self bias resistor.
(virtual center of the filament effectively connects to the self bias resistor). The other end of the self bias resistor goes to ground.
Now, connect 2.5VAC across the ends of the filament.

2.5VAC = 3.536V peak
3.536/2 = 1.768V peak
1.768/2 = 0.884V

At 0 degrees, 180 degrees, and 360 degrees, both ends of the filament are at zero volts, and the filament center is also at zero volts.
The effective bias voltage from the filament to the grid is steady and even all along the filament.

At 90 degrees and 270 degrees, one end of the filament will be at + 1,768V; the center of the filament will be at zero volts; and the other end of the filament will be at - 1.768V.

At 30 degrees, 150 degrees, 210 degrees, and 330 degrees, one end of the filament will be at + 0.884V; the center of the filament will be at zero volts; and the other end of the filament will be at - 0.884V.

Do you see how this is reasonably well balanced from one end to the other end, along the length of the filament wire?

That is a well working practical circuit.

2. At a finer level, there is another effect, the Intermodulation sudebabds of 2 X the power mains frequency on each musical note, and musical harmonic.
That is because the Transconductance is not constant at all bias voltages. So, the transconductance is not the same at one end of the filament that is at the positive peak, versus the other end of the filament that is at the negative peak, of the AC filament power.
And, the transconductance along the filament is higher at the more negative filament end, and lower at the more positive filament end (but constant at the center of the filament wire, because the center is Always at zero volts).

We have to look further down to see this secondary effect, sometimes 60dBc, 80dBc, or 100dBc, but the effect IS there.
Welcome to one of the inner layers of the onion.
Now you know why DC DHT filaments are used, not just to reduce hum, but to eliminate the 2X power mains frequency upper and lower sidebands on each and every musical note and musical harmonic.
 
  • Like
Reactions: 3 users