Those tubes are expensive. It might be worthwhile to also use a ferroresonant transformer for the filaments. The rheostat or pot would then be pretty much set-and-forget.
I had to use one for the 3CX3000 filament transformer the summer mains voltage intermittently drops and rises, seen it as low as 109 when both HVAC are on. Hooked up the rectifiers fil xfmr to it too.
Hi Patrick,
Thanks for the heads up. I had not thought of line variations being an issue though our local power grid voltage varies from 240 to 255 at times and it os mafe worse by my 20K Grid Tied Solar Array in the middle of the day in Summer when loads are up and down.
I have a few questions on the effects this will have on my tube filament voltages.
1- Is the concern about transient voltage spikes or just a gradual variable rise and fall in voltage?
2- The specs for 4-1000a and 3-1000z tubes is 7.5 volts +- 0.37 volts
In my setup, that corresponds to a variation of 120-131vac input to the filament transformer, give. The 16:1 stepdown ratio and the variac I will use to set voltage to 120vac at 0.96 of the 125vac average grid voltage.
My protocol will be to constantly monitor filament voltage on the front panel and begin at zero voltage on startup to prevent a surge to s cold filament. The variac will be adjustable on the same front panel.
If a surge would not cause immediate failure, I could just watch it and keep,the voltage close to 7.5vac.
2- A constant voltage (ferroresonant) transformer is heavy and expensive, so how about just hooking a cheaper 1kva UPS to a cheap surplus isolation transformer and get a regulated 120vac that will be immune to variations? I have some 1kva UPS and a couple of 2kva isolation transformers in stock. Will assume a 0.7 power factor but only need 300vA for filaments anyway.
Comments please. Thanks fir your help on this.
73, Mike
-----------------
Reply from Patrick:
TL;DR
My situation with line cariations from 109 to 125VAC, slow transients from large loads including motor-starting, occasional high frequency spikes (when the power fails and the backup diesel kicks on or off), and not wanting to ride a vvariac moment by moment, makes a constant voltage / ferroresonant transformer a good choice for me. A UPS is not a good choice in my opinion, unless it has an always-on low-distortion 'true sine wave' inverter that supplies 100% of the load at all times.
The case for the ferroresonan transformer:
My concern is keeping the filament voltage within the specified range according to the tube datasheet and the Eimac 'care and feeding' tube manual. Knowing that my line voltage varies widely due to others' loads because of the cruddy undersized pole transformer and being at the far end of the HT line, plus my shop voltage jumps up or down a few volts when the condenser turns on and then back off, with an extra dip of up to 10 volts when it starts, makes me want to give the tube filament a constant and soft ride.
The transformer blocks spikes well, as a byproduct of how it regulates. If the current goes much above the transformer's rating, the transformer will limit the current (dropping the voltage).
A key to using one with a resistor-type, relay-timed soft-start is to pick the resistor value to prevent more than 130-150% inrush current to the tube, and oversize the transformer's capacity to accommodate that.
-OR-
Choose transformer size based on known minimum line voltage and desired inrush overcurrent to match up a tube and transformer, and forget the soft start (I don't like that way but if the gear is always on like a radio broadcaster or semiconductor plasma generator then it is cheaper without a timed soft-start). So there is some little science or art to it.
An operating characteristics document is attached that should explain these transformers' reactions to disturbances better.
The constant voltage transformers are a little heavy but used ones are cheap. Mine is loaded to 65-70% capacity and is a 1KVA unit, but it is inside a double rack amplifier, mounted to the backside of a steel rack panel along with a large 'set' rheostat.
The alternative to regulating the filament voltage would be to consume front panel space with a 10A variac and then ride it, but I'd always miss the moment when the a/c kicks on and off - subjecting the tube to unwanted over-and-under voltage conditions on the filament. That is the kind of spike or surge the ferroresonant transformer eliminates for me. There is a very slight disturbance on the filament voltage but does not exceed the tube spec for filament volts. The waveform is a sine with maybe 3% distortion at most, and harmonics arwe attenuated (unless you get a very, very old ferroresonant transformer). I can trust my filament voltage meter with that as I would trust it with a variac.
UPS's:
An AVR-type UPS -might- do better, depending on how it is designed. The cheap computer store ones do not seem to be smooth in their regulation, but have a 'step' or coarse steps in their boost or buck. They turn the inverter on only when adjustment or full substitution is needed. How rude!
Case in point an 800VA one here is like that and it was not very cheap, but rather cheaply made and a somewhat inconsiderate design. The variac would still be needed is nearly all cases like that. Beware units that make a square wave or pseudo sine wave (just more square steps) when tripped on. Because your filament current is going to look like that also and your filament voltage meter may not indicate true RMS power to the filament.
A commerical, precision regulating sine wave UPS would be better but are $$$. By that I mean, a UPS with a synthesized (high frequency switching power supply) inverter that is always on and always supplies 100% of the power to the load. Such units have typically a bridge rectifier and capacitor input filter for making DC, and this DC powers a so-called 'pure sin wave' inverter that is very well regulated. They have all kinds of hash filters of course but do need proper attention to grounding and cable dress. Tripp Lite among others makes them.
UPS also use transient suppressors to try and eat 'spikes', whereas a ferroresonant transformer should block spikes. When the transient suppressors or MOVs have had enough, they can fail open, and there's no indication the protection has been lost. Or they can short out resulting in smoke and an entertaining electric-fire display until the fuse/breaker opens.
No doubt I have some errors or misconceptions in the above, but I have really done my best to study the situation. Others please correct any errors.
