Filament voltage and affect on Rp

[Jeb-D.]

The filament voltage meters 5.7V, which is the minimum voltage for the tube being used. It's not a directly heated triode. Does any one know if this will raise the Rp of the tube? Opposed to running it at 6.3V. It seems to bias at a significantly more positive grid voltage; for a given amount of current, when compared to the curves. I've tried with 5 tubes and all act the same way. Anyone else noticed this, or am I just trippin'?

[Brian Beck]

Generally, yes. Starving a tube's heater will lower cathode emission, which will lower cathode current, which will reduce transconductance, gm. Since mu is more dependent on mechanical design and relative electrode dimensions, and since rp = mu/gm, then rp will increase as gm drops. There are complicating factors, which I'm sure others will bring up, but that's the gist of it.

[cerrem]

The behavior of a valve with variation in the heater is a case by case situation...
6.3V is a mean value for heater voltage...
It's hard to imagine this nowadays, but many tubes were original designed for use with Battery supplies and portable applications..mostly portable radio equipment as in ARMY communications..ect..
Heaters are to be in normal operating mode with +/- 10% variation... So in many tubes...the space charge is still adequete at 5.7V .... ie, the space charge density acts as a cloud of reservoir of electrons...so you still have plenty in reserve..The valve is not in starvation mode...

Chris

[rdf]

Originally posted by cerrem
It's hard to imagine this nowadays, but many tubes were original designed for use with Battery supplies and portable applications..mostly portable radio equipment as in ARMY communications..ect..

Easy to believe! I spent a couple of evenings last week investigating the distortion spectrum of a WW2 military DHT acorn triode. The filaments were powered the whole time by a single AA battery (1.25 volts, 100 ma.)

[EC8010]

As Brian Beck says, ra can be expected to rise if heater voltage is reduced - it's not a question of whether the valve is near saturation.

But I measured an ECC808 on my AVO with Va = 250V and Vgk = -1.9V and, assuming mu = 100, calculated ra...

5.7V 0.65mA 1.40mA/V 71.4k
6.0V 0.68mA 1.45mA/V 70k
6.3V 0.82mA 1.70mA/V 58.8k
6.6V 0.91mA 1.80mA/V 55.6k
6.9V 1.00mA 1.85mA/V 54.1k

[sorenj07]

Are there any negative effects to having a filament voltage that's a bit too high, like around 6.9-7VAC? I'd think that it'd protect the tube extra against cathode stripping, but can it also burn it out faster? I remember reading about this guy who had an amp that kept blowing output tubes until he lowered the filament voltage to normal.

[Tweeker]

DH tube life drops exponentially with higher filament voltage. Not sure how badly oxide cathode tubes react, but they dont like it either.

[Brian Beck]

In “Vacuum Tube Amplifiers” by Valley and Wallman, on page 421, reference is made to the effects of heater voltage variation on tube parameters when plate current is small compared to cathode emission (well below saturation, which is our usual condition in audio).

An interesting quote: “Variation of the cathode temperature of a vacuum tube in which the current is limited by electrode potentials and space charge rather than by cathode emission has the effect of varying the average initial electron velocity and therefore also the electrode voltages required to obtain a given electron flow. … It [the effect] may be expressed in terms of the amount by which the cathode voltage must be changed relative to the other electrode voltages in order to hold the current constant as the cathode temperature is varied. For oxide-coated unipotential cathodes, this amount is approximately 0.2 volt for a 20 per cent change of heater voltage about the normal value, whether the tube is a diode, a triode, or a multi-grid tube.”

I visualize the space charge cloud as having a smoothly varying distribution of electron velocities and electron densities, with the outer reaches having the least electron density, populated by those electrons which had the most initial velocity. Other electrons have less initial velocity and stay closer to the cathode. The outer electrons are the ones that are most easily lured to the plate, creating a plate current, simply because they are closer to the positive plate, and they “feel” the repelling push of other electrons boiling beneath them. If the cathode temperature is lowered, the AVERAGE initial electron velocity is reduced. The space cloud will shrink a bit, and the outermost electrons are reduced in number and lower in average “altitude” above the cathode. This has the effect of a change in cathode voltage, as discussed above. There are still plenty of electrons in the space charge cloud to supply very large plate current values, even enough for values well in excess of those normally being considered. But it will take more pull from the plate (by increasing the plate voltage), or a more positive grid voltage to restore the same plate current as before the cathode temperature reduction.

With a 10% drop in cathode voltage (about the drop from 6.3 volts to 5.7 volts that Jeb-D asked about) we can estimate (by interpolating the figures above) that a decrease in cathode voltage of about 0.1 volt would be required to restore the plate current to the 6.3-heater-volt level. Since a 0.1 volt change in plate-to-cathode voltage (Vpk), would have only a slight effect on plate current by itself, we can concern ourselves primarily with the 0.1 volt effect on cathode-to-grid voltage, Vgk. We can use the gm parameter to estimate the change to plate current due to a heater voltage reduction of 10%. For a high-gm tube, using as an example a 6DJ8 with gm = 10mS, a 0.1 volt change in Vgk amounts to a 1mA change in plate current, if plate voltage change is not a big factor.

EC8010 tested an ECC808 (thank you) across a 19% change of heater voltage. Over that range, we can estimate a 0.19 volt equivalent change in cathode voltage, which, when multiplied by the midrange value of gm of 1.7mS, yields a plate current change of 0.32 mA. EC8010 measured 0.35mA, which agrees well with our prediction. Then, as he goes on to show, gm varies with those changes in plate current as expected. Therefore plate resistance values must also vary - across a range of 54K to 71K ohms in this example.

For other tubes, the guideline of a 0.2 equivalent cathode voltage change per 20% heater voltage change will still apply; however the effect of the 0.2 volt cathode voltage change on plate current, gm and rp will be dependent upon the specific values of mu and nominal gm, and to a degree upon plate loading.

[EC8010]

Thanks for the Valley & Wallman snippet - it clearly justifies rereading.

[Brian Beck]

Further note on plate loading: When the plate load is very high (such as with a CCS), the change in effective cathode voltage due to heater voltage changes cannot affect plate current (since it’s constant after all). Instead the plate voltage will change (increasing when heater voltage is reduced) by an amount necessary to account for the change in equivalent cathode voltage while holding plate current constant. Gm is no longer the relevant parameter; instead mu is. The 0.2 volt cathode change would result in a plate voltage change of about mu*0.2. In the ECC808 case that would result in a change of about 20 volts on the plate. In the CCS plate load case, the plate resistance will still increase with reduced heater voltage, but to a lesser degree than in the case when plate current is allowed to vary in response to heater voltage variation. The change of plate resistance as a function of plate voltage at constant plate current is variable among tube types, but it always increases at higher plate voltages (where the plate characteristic curves lean or slope farther to the right). In real world cases where the plate load is not infinite, both gm and mu will be factors in determining a tube’s response to heater voltage changes.