Filament Startup

[captainbeefheart]

Then every single person I sell an amp to has to send me their tubes in order for me to set the filament voltage to not deviate substantially, then I need to write a manual on how to open the amp up, hook up a multimeter, and make adjustments if they want to roll tubes or if their tubes wear out. This seems like a huge mistake for an amplifier designer, but I haven't been doing this all that long.

Not necessarily. The simplest of Current sources yes but with a little more work it can be done automatically like with Tent labs filament supply.

Also not true. For example, I used an L200C regulator in a past product and set the maximum current output to 2x quiescent current, but the chip itself is setup as a voltage regulator.

Yes it's a voltage regulator and you still had 2x quiescent operating current.


A voltage source will short the cathode (audio signals) since it needs to be decoupled through a large electrolytic capacitor for quality sound. A current source will not short the cathode since it intrinsically has a high output impedance and will not affect the AC signal.

 

[captainbeefheart]

Won't that overpower the filament? is it safe to ignore the actual wattage being consumed?

Good question. I have read arguments for both voltage compliance and current compliance and it boils down to what the tube manufacturer states, which is usually voltage compliance. So in those cases I adjust the current source to achieve the desired voltage compliance designated in the datasheet.

 

[paul_b]

Won't that overpower the filament? is it safe to ignore the actual wattage being consumed?

Based on what I've seen from old texts and Jac, that is correct.

 

[captainbeefheart]

Taking that last excellent question further;

Yes I would think that wattage of the filament would be much more appropriate since tolerance of filament resistance will not be consistent. So if you take the datasheet values of say 5v at 1.2a that's 6 watts. Now if you have a voltage source set to exactly 5v but measure 1.3A that's giving us 6.5 watts of filament power, I would think it would be better to adjust the voltage lower until you reach the desired 6 watts of filament power and possibly run some plate sweeps on a curve tracer and compare them.

 

[TriodeLuvr]

Taking that last excellent question further;

Yes I would think that wattage of the filament would be much more appropriate since tolerance of filament resistance will not be consistent. So if you take the datasheet values of say 5v at 1.2a that's 6 watts. Now if you have a voltage source set to exactly 5v but measure 1.3A that's giving us 6.5 watts of filament power, I would think it would be better to adjust the voltage lower until you reach the desired 6 watts of filament power and possibly run some plate sweeps on a curve tracer and compare them.

I have always assumed that tubes are intended by the manufacturer to be run at their rated voltage, without undue regard for the exact current that creates. I say this because transmitting tubes produced in the '40s and '50s were intended for use in equipment with the specified voltage provided by a voltage source, either a battery or a line-powered transformer. I suppose it's more of an issue now due to the cost, and we all want to get as much life from the tubes as possible. Still, even reissue tubes need to be compatible with older gear, and I would expect manufacturers to design them with that in mind. Not taking sides here with what's best, but I would only use a constant-current supply if the manufacturer of the tubes I buy said that current - not voltage - is their standard for filament power.

Regarding wattage, it might be interesting to design a PIC that would monitor voltage and current. It could perform the math and adjust current or voltage (or both) automatically to achieve the rated consumption.

Jack

 

[captainbeefheart]

I have always assumed that tubes are intended by the manufacturer to be run at their rated voltage, without undue regard for the exact current that creates. I say this because transmitting tubes produced in the '40s and '50s were intended for use in equipment with the specified voltage provided by a voltage source, either a battery or a line-powered transformer. I suppose it's more of an issue now due to the cost, and we all want to get as much life from the tubes as possible. Still, even reissue tubes need to be compatible with older gear, and I would expect manufacturers to design them with that in mind. Not taking sides here with what's best, but I would only use a constant-current supply if the manufacturer of the tubes I buy said that current - not voltage - is their standard for filament power.

Yes, as I said when using a current source I also follow the datasheets voltage not current, so I adjust the current until voltage compliance is met. We are all saying basically the same thing in that keep the voltage 5v across the 300b filament, as long as you do that your tubes will last a very long time.

The conversation has morphed from talking cold start surge current to operational current/voltage and it's relation to service life. As I said, we all agree that the it's best to stick with datasheet voltage compliance for the filament normal operating point for optimal long service life of the tube. My initial concerns with using a voltage source is although yes you will get some current limiting due to source resistance you still get more than 2x the operational current through the filament when it's cold and the weak spots in the filament can eventually get stressed to the point that they burn open during start up. That's why for my amps I prefer a current source since current is always limited to the operational current when filament is warm to mitigate any stress in the filament from surge current. EML labs even states AC filament resistance will limit it enough but that's not my experience. Typical 5v winding rated for a 300b will have a resistance of .5 ohms, 4 amp load will droop only 2v, and 10 amps will droop the entire 5v. You still get large current surge with AC filament power. With a voltage source you need to AC short the cathode/filament with a large capacitor, the high impedance of a current source means you don't need a capacitor.

 

[M17XR2B]

RCA published this for tungsten filaments, just as an example.

Apparently running tungsten filament lower than rated can drastically improve lifetime. This is with the caveat that full power isn't required.
Someone did a trace of 01A normal vs starved and the curves look normal at the typical operating point for input duty:

And lower voltage can even lower distortion tho I'll keep my non tungsten tubes at the rated voltage

DHT with starved filaments.

 

[targeteye]

From the 1936 WE data sheet for the 300A, The 1950 datasheet for the 300B doesn't state but its likely they didn't think they needed to. I have never seen a tube ever rated for a given current only for a given voltage with a resultant current. I found also a combined WE spec sheet for the 300A/B and it also state the below identically.




So if you want to use sand regulators and your concerned about startup current then a 5 Volt regulator that is current limited is in order. which i think most already have that, just pick one with the specs you need. Then you just need to obsess about what to do for regulator failure conditions.

Otherwise if your anal about the statement in the spec sheet and you don't want silicon in your amp you need something like a rheostat coupled with a functional voltmeter on the face of the amp in order to accommodate variance in line voltage. Perhaps a thermister on the a/c line input for inrush help.

 

[M17XR2B]

I have never seen a tube ever rated for a given current

Browsing enough datasheets you will find a handful, 101F being one of them:



Another option I've used with great success for really slow startup on my 46 amp, LCL. Not power efficient, used 12V 4A transformers -> 200mH 1.7Ω -> 10mF -> 400mH 3.4Ω. The last choke needs to have a reactance ideally 10 times lower than the filament resistance at the target frequency i.e. 30Hz.