Amps
[Top] [All Lists]

Re: [Amps] Switching Supplies for High Current Filaments

To: amps@contesting.com
Subject: Re: [Amps] Switching Supplies for High Current Filaments
From: "Roger (K8RI)" <sub1@rogerhalstead.com>
Date: Wed, 06 Oct 2010 23:31:47 -0400
List-post: <amps@contesting.com">mailto:amps@contesting.com>

On 10/6/2010 10:40 PM, TexasRF@aol.com wrote:
> The problem with switchers or using any regulated supply is the cold
> resistance of the heater/filament is typically 10% of the hot resistance.
Isn't it worst than that for the tubes like the 4CX3000A7 and 3CX3000A7 
or the imported GS35B
>   That
> issue is automatically handled by the current limitations of a typical
> filament  transformer. The transformer becomes pretty much a constant current
> source until  the heater temperature rises enough to get the transformer out 
> of
> saturation.
>
> On big tubes requiring many amps of filament current this can be a real
> problem. If the transformer has to supply say 30 amps, then it will supply
> many  times more current than that for short periods of time. A cold tube can
> be  subjected to very high stress under these circumstances and can even have
> the  filament burned open, just like a 100w light bulb blows out. No
> problem with the  $1 light bulb but loosing a big tube that way is heart 
> breaking.
> (Been there,  done that, have multiple tee shirts).
>
> The solution to the over current problem is reducing the startup voltage
> enough that the 10% filament resistance still draws rated current. That takes
>   10% voltage according to ohm's law.
Hence the step start or using a triac to ramp up the voltage. I really 
like the idea of ramping up the voltage, but with the big tubes I'd use 
a separate ramp for the tube and high voltage.
>
> So, for any regulated and current limited power supply, there needs to be a
>   method of starting at low voltage and ramp it up as the tube warms up.
>
> Being the concept guy, I will bow out at this point and leave the answer to
>   others. I have the answer for transformer filament power sources but no
> clue how  to do it with a switcher or any other regulated/current limited
> supply.
I really don't like the idea of using a switching PS for the filaments 
on "big tubes".  I don't think DC on the filament is a good idea and 
putting the square wave through leads and chokes to the tube(s) opens up 
a whole new can of worms with noise that can be very difficult to 
eliminate. Depending on the circuit you can end up with the old cathode 
modulation.

Although not trivial, I doubt creating a "dual voltage" switching supply 
would be much of a problem. Computer power supplies now run well over a 
KW capability (those aren't the $15 dollar ones) and have multiple 30A 
"rails" at 12V and other voltages as well.  Those typically use the 
"push, pull", high capacity, low RPM, 140mm fans.  Since those came out 
I can actually carry on a conversation here in the den between the two 
full size towers that have 5 of those fans...each! That and I only have 
to clean them about once a season instead of monthly or even weekly.
> If it  was actually a current limited supply, not a current shutdown
> design, all would  be well and the tube could warm up with a constant rated
> current source. A  regulated current design instead of voltage would be
> wonderful.
Back in the "old days" of computers I had one system so heavily loaded 
that just plugging in the small fan in the side would cause the PS to 
shut down.

Then there was the "big stuff" (not computers) We used regulated current 
AC supplies with a pair of LARGE (water cooled) SCRs.  The controlling 
sample was across a length of about # 12 nichrome wire about an inch 
long.  This received its current of 5A from a series of very large 
current transformers on the secondaries of the main transformer. The 
SCRs were running something like 1200A(or more) @ 480 VAC single phase.  
That's substantially over half a megawatt. I watched an attempt to lift 
one of the main transformers put a fork life right up on its nose and we 
had rows of those things. Of course that was over 23 years ago so I can 
only imagine what is there now.

But getting back to powering up one of those "big tubes" you are looking 
at as much iron in the filament transformer as there is in the plate 
transformer for some of the "economy amps". The tube requires CCS 
ratings while the amp can get away with IAS with about the same KVA ratings.

Kind of a sobering thought as I have several of those so called "instant 
on" tubes.

73

Roger (K8RI)
>
> By the way, that 10% resistance figure applies to light bulbs as well. The
> hot resistance can be estimated by r  equals voltage squared divided by
> wattage rating. That is the well known power = e squared divided by
> resistance formula turned around backwards. The cold resistance can be 
> measured  with
> a VOM or DVM.
>
> 73,
> Gerald K5GW
>
>
>
> In a message dated 10/5/2010 11:01:14 P.M. Central Daylight Time,
> jim.thom@telus.net writes:
>
> Date:  Tue, 5 Oct 2010 07:34:41 -0700
> From: Patrick Barthelow<apolloeme@live.com>
> Subject: [Amps] Switching Supplies for High  Current Filaments
>
>
>
> Are there any sensibly priced switching  supplies (even surplus)  that are
> small and light, that could power some  of the QRO tubes
> with high current filaments?   Would, say,  a  5v, 30A switcher which could
> power a pair of 3-500Zs, be economically
> competitive with a 5V 30A transformer, and even if it is not directly, it
> might still be worthy of consideration
> if it was small enough, and light  enough to significantly lower size and
> weight in an RF deck?
> Would cold  start surge current be a problem for switcher implementation
> for  filaments?
> Best Regards,
> 73, de Pat Barthelow AA6EG   apolloeme@live.com
>
> ## This was 1st tried on a  GS35B  6m  amp.  The inrush current was so
> great that the
> switcher thought  it was being cro-barred... and went into shut down mode !
>    A  bigger
> switcher was used..and that worked,  but the 2nd one was almost  double the
> capacity
> of the 1st one.  Also, I believe my buddy had to  use a Dc-dc  isolation
> converter in their some where, to
> solve  yet another problem.
>
> ##  In the end, with numerous other  problems, like heating of the air
> variable load cap, the RF deck was  scrapped,
> and  a new one designed from scratch.  2nd version used  2 x GS35B's... and
> this time, 2 x separate conventional
> fil xfmr's used,  one per tube.  The problem with  a pair of 3-500Z's
> is.... the B-  return is via the CT  of the conventional
> fil xfmr.   We  gave up on the  notion of using switcher's  for fil
> supplies  on oxide tubes.  Some have had better success,
> so it can be done   with oxide tubes.  Then you have an adjustable,
> regulated  supply.   They will not work on a directly
> heated thoriated  tungsten tube.   On some of the switcher's... hash and
> noise   was across  some of the HF bands...and
> a bitch to filter out  completely.
>
> ##  hammond makes a 5V @ 30A  [with CT]   conventional fil xfmr...for 2 x
> 3-500Z's.  Runs barely luke warm.   It's   priced
> way below what a dahl is.... and readily available  in the usa, through the
> usual distributor's.   Comes in 2 x  version's ,
> with or without bell end housings, IE: bell housings or open  frame.   They
> measure 5.5 V  no load.. and 5.0 V    with a 30A load.
> Last I checked  a few years ago, it was aprx $45.00  .  Their is only one
> primary, 115 vac.   You can also get it in  5v  @  15A .
> They also make a 7.5 vac  @ 21 A.    A friend  in W6 land got his  YC-156
> fil xfmr  from  hammond.    The YC-156 fil xfmr was a
> torroidal type.
>
> later... Jim    VE7RF.
> _______________________________________________
> Amps mailing  list
> Amps@contesting.com
> http://lists.contesting.com/mailman/listinfo/amps
>
> _______________________________________________
> Amps mailing list
> Amps@contesting.com
> http://lists.contesting.com/mailman/listinfo/amps
>
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps

<Prev in Thread] Current Thread [Next in Thread>