[Amps] What tube?

[Manfred Mornhinweg]

> you can be sure that tube life is adversely affected by frequent  
> on/off cycles of the filament.

Well, then it seems to be a bad idea to do so. Which puts tubes some more points 
behind solid state devices for me, because in my particular case the power 
consumption during RX is an important matter, as I typically listen a lot more 
time than I transmit, but want to have the amp ready to use at any time, and in 
summer I'm very short on energy, given that my electricity comes from a small 
turbine in creek that every summer seems to have less water than the summer 
before. At this time (southern summer, and it hasn't rained even one drop for a 
full month), I'm barely generating 450W, and have to run the whole home on that. 
I use a battery buffering system, that allows me to draw 3kW for peaks during 
SSB transmission, but spending 200W all the time during RX for keeping filaments 
hot is a problem.

In winter instead power is no problem here, and it's free.

So I will have to keep looking into solid state amps, or just operate with 100W 
only during summers.

> One of the current tube manufacturers (Burle 
> I  think) has a paper on his website that suggests that one on/off cycle of 
> the  filament power is roughly equivalent to reducing tube life by 60 hours. 
> If you  expect say 18000 hours, the 300 on/off cycles pretty well uses that 
> up.

That seems overly pessimistic to me. I must have switched my NCL-2000 on and off 
many more than 300 times, and the tubes are still good. But it's very reasonable 
that filaments should suffer some wear from power cycling.

> This was for a larger transmitting tube so perhaps the 60 hour figure is  
> not accurate for a 3-500Z or similar but the principle applies.

I think so. Also a lot must depend on the amount of inrush current.

My idea about an amp that powers down the filaments during RX was based on using 
a switching power supply. This has at least 3 advantages: Lighter, cheaper, and 
provides a regulated voltage. But in the case of filaments, there is an 
additional advantage: Current limiting. If I build such an amplifier, I can 
precisely set the filament current limit. The filament would come up at that 
current, which could be as low as 120% of the nominal current, until getting 
into voltage regulation when it's hot.

But that will obviously slow down the heating! The amp's power supply would need 
to power up the high voltage section once the filament current has become the 
nominal one, meaning that the filament is up at operating temperature. Two 
question arise here: How long would it take to bring the filament up to full 
emission, at a current limit of 120, 150 or 200%? And would such a controlled, 
slow heating avoid all or most of the power cycling damage? I fear it wouldn't, 
because the heating and cooling filament will still expand and contract, causing 
some friction in its supports, which might abrade it over time. But maybe the 
life span would be long enough to be acceptable, and maybe the startup process 
would even be fast enough to be acceptable. But I tend to doubt it.

Lacking data on it, probably only an experiment would show.

Gerald,

> he had the right idea; except they last longer if you never turn them on!

Well, that's pretty much what I do to make mine last!  ;-)

Carl,

> If you use care in selecting the filament transformer it can be self current 
> limiting. 

Yes, but in 2015 it's simpler and much cheaper to use a switching power supply. 
You don't even need to build it! You can buy ready made switching supplies in 
many voltage and current ratings. Many of them have adjustable voltage, and 
adjustable current limiting. Power supplies to feed 5V 15A filaments, or 12.6V 
10A filaments, cost like 20-30 dollars, new. Much cheaper than a decent filament 
transformer. Much lighter too. And you get a precisely regulated voltage, with 
wide tolerance to line voltage variations, plus adjustable current limiting! 
It's really the way to go. I wouldn't put a big chunk of iron and expensive 
copper in any new amplifier.

If you are willing to make some simple modifications, you can use old PC power 
supplies to power either type of filament. Such power supplies are typically 
available for free. I have a bunch of them here, obtained from old junked PCs.

The high voltage supply instead is not as easily or cheaply available in 
switching technology, so that's a thing one should build oneself. It's not hard, 
though. The usual approach is to rectify and filter the line voltage into 
roughly 320VDC, apply that to an IGBT bridge, that is controlled by a PWM chip 
and powers several compact ferrite transformers in parallel. The secondary of 
each transformer is full-wave rectified with ultrafast diodes, filtered by a 
simple LC circuit, and the DC outputs from each section are all placed in 
series. Add some RFI suppression, and that's it. It's more complex than a 
conventional supply, but _much_ less expensive, hugely lighter, 
short-circuit-proof, and delivers a regulated HV, tolerating a wide range of 
line voltages. And if desired, it's rather easy to add power factor correction. 
That's useful if you want to run a big amp from any normal power outlet, or from 
a tightly sized generator.

> The 4-400 and GU-81M can both be run in GG. In a SB-220 with100W of drive 
> you can expect 1000W due to a bit lower gain than the 3-500Z.

Good to know. But somehow I don't feel comfortable doing that! I mean, a tetrode 
or pentode was designed as such, and not intended for GG with all grids tied 
together! Given that a 3-500Z is slightly less expensive than a 4-400 of the 
same quality, I would certainly use the 3-500Z if I built a GG amplifier and had 
to choose between those two only.

And the GU-81M seems to work at very high grid and screen voltages, compared to 
other tubes. I wonder how it behaves in GG. I can imagine it being very hard to 
drive in that configuration.

> Crank up the voltage at the same 100W and 1500W is likely attainable even on 
> 10M with full emission tubes, 

That also forces an increase in the Q of the tank circuit, requiring very good 
components - typically vacuum caps, big silver plated coils and the like. 
Perfectly doable, but not really optimal.

> those are often $20-30 NIB military surplus 
> into the 70's. Sockets are the same as the 3-500 and the flat glazed ceramic 
> Johnson style are $15-30 at US hamfests and Fleabay. The Eimac SK-410 runs 
> about twice that from the same sources.

OK.

> Im using GU-81M's as AM modulators and the sockets are often more expensive 
> than the tubes from most European sources. 

That's what I saw too. And a good socket for those tubes isn't very easy to home 
brew. Anyway, we are talking far less money for a GU-81M plus socket, than for a 
3-500Z alone!

> A tube that Im surprised you and others overlooked is the QB5/1750, popular 
> in BCB and SW broadcasting.

Probably because it doesn't show up in ham literature! I had never heard of it.

Well, interesting musings. Now if there only was enough interesting activity on 
the bands, to justify building a new amp! It's ages, really, since I last found 
another ham on any band with whom it was possible to discuss technical things. 
Most seem to know at most something about the current weather in their area, and 
can rattle down the model numbers of the radios and antennas they use. And 
that's for local contacts. If it's DX, it's almost invariably "hello - five nine 
- thanks for the contact - next station please!"

Oh well...

Manfred


========================
Visit my hobby homepage!
http://ludens.cl
========================