Text 267, 120 rader
Skriven 2005-02-26 04:06:01 av Roy J. Tellason (1:270/615.0)
    Kommentar till text 265 av MIKE ROSS (1:167/133.0)
Ärende: Fan Of True Multitasking
================================
MIKE ROSS wrote in a message to Roy J. Tellason:

 MR> "Roy J. Tellason" bravely wrote to "MIKE ROSS" (24 Feb 05 
 MR> 20:06:46)  --- on the heady topic of "Fan Of True Multitasking"

 RJT> MIKE ROSS wrote in a message to Roy J. Tellason:

 MR> One would suppose the collector to base junction would have to be
 MR> a lot thicker, perhaps this is what makes it less sensitive to
 MR> influence from the base?

 RJT> The junction being a boundary layer I don't see how it could be
 RJT> "thicker".

 MR> Even a solid capacitor has a dielectric breakdown limit. Similarly,
 MR> to achieve a 1500 volt Vcbo the semiconductor material must be
 MR> thicker, if just for the parasitic capacitance alone. The junction
 MR> isn't just at one specific location but is diffused throughout the
 MR> bulk. IOW it is a gradual voltage gradient not like the sharp
 MR> boundary you are describing. Point-contact transistors are usually
 MR> associated with sharp boundaries. Read up on "diffused"
 MR> junction(s).

Yeah,  I can see your point...

I downloaded a book a while back that I had originally thought to be about
"designing analog circuits".  Well,  it was,  but they were talking about
designing analog _chips_!  It was still full of lots of useful info,  a bunch
of interesting history,  and was worthwhile overall.  It explained a lot.  I
probably oughta get back into it again,  but there's only so many hours in a
day.

 MR> Horizontal Output transistors are also relatively slow and the
 MR> best rated are good at switching at perhaps a maximum just shy of
 MR> 100KHz.

 RJT> I've seen f(sub)T ratings going as high as maybe 15 MHz,  but a lot of
 RJT> small-signal stuff goes way higher.

 MR> Ft is not a very useful spec for power devices because what matters
 MR> is the power bandwidth. Though one might measure a very high ft in
 MR> a device, in reality it will not output useful power at that
 MR> frequency limit. The reason is that the current gain of power
 MR> transistors falls off with increasing collector current in addition
 MR> to frequency. Thus the useful bandwidth of power transistors is
 MR> considerably short of ft. You will note that a HOT doesn't have an
 MR> ft spec but rather Horiz Out. 

Yeah,  those usually seem to be spec'd in terms of some time constant rather
than ft.

 MR> They tend to be able to carry quite a bit of current as some have
 MR> ratings of 15 amperes or more.

 RJT> Yep.

 MR> Most these days have a built-in low value resistor (~10 ohms)
 MR> across the base to emitter terminals.

 RJT> I'm thinking 47 ohms is more typical.  Makes it a bit tough to do
 RJT> resistance checking to see if it's okay...  :-)

 MR> This is for damping the input waveform which is typically injected
 MR> using a small pulse transformer. The low resistance also helps
 MR> swamp the Icbo (collector to base leakage) to ensure cutoff in
 MR> case there is a runaway condition.

 RJT> I didn't know about that part.

 MR> There may be 47 ohms internal but the loaded circuit will be ~10
 MR> ohms. The 10 ohm resistor will be in the circuit across the B-E
 MR> junction. 

I'm just going by what I see in the data sheets.

 RJT> Some of them also have built-in damper diodes.

 MR> Many have a damper but the odd one doesn't because the damper
 MR> current is useful for generating a boost voltage. Often secondary
 MR> supplies are taken off the flyback but if this power is taken ahead
 MR> of it then the flyback can be smaller. An external damper diode
 MR> allows doing that.

I never did get it clear why some had it and some didn't.  I thought pretty
much took power off the flyback these days,  it seems to be a pretty common
approach to stuff any more.  Though it's been quite some time now since I
really sat down and studied schematics for that stuff -- I don't work on tv and
haven't for years,  and I have no schematics for monitors that are all that
recent,  though I may be getting a hold of some.

 MR> Maybe I can help fix that dead monitor? If the HOT is shorted this
 MR> is often due to a bad drive waveform. Either the frequency was
 MR> wrong or the peak control network off the collector is open.

I suspect that I'll have to pull that transistor out of there in order to test
it properly.  And even then if it's one of those with the integral parts it's
not going to be easy.  Too bad I don't have enough space in here,  I could use
my curve trace and scope,  but at present there's absolutely no place to set
them up.

 MR> Sometimes a powerline spike can fry the HOT or short out the
 MR> regulator. You can probe the circuit without the HOT by tack
 MR> soldering a rectifier and parallel resistance to simulate the B-E
 MR> junction. The drive should have clean edges, going positive to 0.7
 MR> volts and negative ~2 volts. 

There's a thought.  I just hope it isn't one of those setups that kicks it once
to get it started and then relies on the HOT to keep things going after that. 
I guess it depends on where the H oscillator gets its power from.  Though there
_is_ a pretty hefty power FET in there for the LV supply...

Right now my available work space is taken up by a MB and monitor and a couple
of drives,  I was testing ram sticks (and did find one bad one),  and just last
night tested out one 10G HD that I'm probably going to add to my setup,  I just
need to decide what else I'm going to test and go from there,  maybe get that
monitor back there shortly...


--- 
 * Origin: TANSTAAFL BBS 717-838-8539 (1:270/615)