On Nov 5, 2006, at 2:06 AM, pentalab wrote:
--- In ham_amplifiers@..., R L Measures <r@...> wrote:
On Nov 3, 2006, at 5:10 PM, pentalab wrote:
Gent's
Some poor fellow on the 'other' amp reflector smoked the 25
ohm 25
watt wire wound glitch R on his Henry 3K .
What was not mentioned was that Henry installs the glitch R in
the B-..... NOT the B+.... no big deal... cept the glitch R +
all
wiring in/out of the glitch R has to use 10 kv wire.. and
glitch R well insulated from chassis... on stand offs.
RICH SEZ.... Not only that, but the filament transformer must to
be able to withstand the full anode V. This is why Henry Radio
amplifiers need circuit improvements before they are fired up --
sometimes quite literally.
### Point well taken !! I drew it out on paper... ur right...
the fil xfmr will spike to full B+.... with the glitch R.. in
the B- lead. What were these guys thinking ?
They probably weren't.
Easy to fix.
indeed
Another problem is that this particular resistor usually has an
adjustable tap / slider -- and that makes it not suitable for
glitch service.
### Why install an adjustable wirewound ?? Plane nuts.
It was part of the over-current relay ckt. -- something that
thoriated-tungsten filament tubes do not need because their emissive
layer (di-tungsten carbide / W^2C) can not be damaged by too much
cathode current - as is the case with oxide cathode tubes such as the
8877.
Try
this for an experiment... if u are brave enough. Install a 25 ohm-
25 watt wire wound wound... in a metal bomb proof box... fed
with 10 kv Red /Black Belden HV wire.... Hook the other ends of
the HV wire DIRECTLY to the B+ and B- of ur favourite HV
supply..... like a 3-K, etc....[NO HV fuse installed] Install a
heavy duty vac relay or better yet.. a vac contactor in the B+
lead. Turn on the HV supply.... stand back... and activate the
vac relay..... applying 3-8 kv across the glitch R...... then
lemme know IF the glitch R remains intact.
A 25W ww resistor is c. 2" long. 3000v is on the ragged edge of too
much V-gradient for a 2" resistor. Since 10w ww resistors are also
c. 2" long, it would be preferable to series two of them for glitch
service at 3000v or greater. The ultimate solution is to use a surge-
rated - in Joules - type resistor. The glitch resistors we current
supply with our low-Q suppressor retrofit kits are rated at 120J,
which is slightly more than what a SB-220 HV-PS is capable of during
a serious glitch.
He also smoked the safety diode between chassis and one end of
the glitch R. He shoulda
also had safety diodes across both the grid + plate meter. On a
8877... it really needs electronic fast grid overcurrent
protection.....
RICH SEZ.... The problem with 8877s is not grid damage from too
much HF grid-I, it's gold-evaporation from the grid caused by
occasional UHF parasitic oscillations that causes large bursts of
UHF grid-I that evaporates off thin layers of gold, which in turn
forms gold melt balls that create leakage paths in the vacuum. It
is my opinion that the way to stop gold sputtering is to limit peak
discharge current from the HV filter C with a sturdy glitch R in the
HV+ lead. As I see it, 8877s have so much UHF gain that they can be
a pain.
which I doubt would even work with the safety
diode shorted between chassis and his glitch R/ B-
agreed
### Here's the deal..Now this IS important. . A buddy of mine
brought over a QST article a few yrs back... in the stray's
dept... it depicted ur typ "electronic grid overcurrent"
protection scheme.... seen in many handbook projects... and also
used in many commercially produced linears. In many handbook amp
designs.... you will see a pair of diodes REVERSE connected...
between the chassis and B- Only one diode is really needed...
the worse thing that can happen usually.. is a B+ to chassis
short.
... which would be the case if the glitch-R was not capable of
limiting I-pk to the diode's peak-current ability. (typ. 200A-pk in
3A diodes or 400A-pk in 6A diodes)
### The tech note went on to show what really happens... IF
reverse connected diodes are installed from chassis to B- AND
electronic grid protection is used. They re-drew the entire
schematic.... and what is now depicted is the 2nd diode [not
needed].... ends up DIRECTLY in parallel... with.. u guessed it....
the small value resistor installed between grid shunt and
chassis. This small value resistor, [typ 5-15 ohms] has DC grid
current flowing through it all the time. It also has a V drop
across it.. typ 1-3 volts... which is fed to the wiper of a
pot... used as a V divider... then applied to a transistor...
activates a relay.. etc.. kicks amp offline, lights LED...
latches... etc. Trbl is... If the un needed 2nd safety diode
is used.... the V drop across the resistor is directly in parallel
with it... and turns on the diode !!!! DC Grid current is then
partially or fully diverted through the un-needed safety diode....
and u guessed it... the dc grid meter is now reading too low !
Use a 0.5¦¸ grid-current meter shunt-R and one can read up to 1A of
grid-I without exceeding a Si diode's 0.5v threshold point.
### The fix is... to remove the un-needed safety diode.
### On my 3000/6000A7 designs... NO electronic grid overcurrent
protection is used.
Does the 3cx6000A7 have a gold-plated grid?
A simple fast 3agc fuse is installed
between neg of grid meter and chassis. [with nothing across the
fuse either].
I would put a transient suppressor diode across the fuse. They are <
a dollar.
RVS connected 6 A diodes ARE used [I use 2-3
6A diodes in EACH direction... 4-6 in total].. wired between
chassis and B- I don't install any 10-1000 ohm resistor
between B- and chassis in either RF deck.... OR HV supply.
good
The
RVS connected diodes will never allow the B- to float more than
+/- .7v from chassis potential[grnd]...(grid fuse open). With
the grid fuse intact... the B- ends up at chassis potential
anyway... since the neg of the grid meter is fed via grid fuse...
to chassis.
### Originally, we installed a 100 K 3 watt resistor across the
grid fuse holder... thinking IF the grid fuse opened up... the V
drop across the 100 k resistor would bias the amp off.... it
doesn't ! With 800 w of drive applied on a 3x6... and the grid
fuse OPEN... all of a sudden you get 60 w reflected power on the
input of the amp [normally zero]... and abt 20% of normal power
output... [2500w instead of normal 12 kw]. We removed the 100 K
resistor... leaving a wide open (if fuse opens up).... and now with
drive applied.. and an open grid fuse.... u get ZERO watts
output.... and of course... ZERO DC grid current. [desired effect]
### The BIG question is.... WHERE the hell does this 800 watts of
drive end up ??? It's still being fed to cathode. Does it
just sit there and cook the cathode ??
How can you measure P if the input Z goes bananas?
## In any event the grid fuse concept works superbly..
everytime. Some reason's for the grid fuse blowing are.. excess
drive with a dead cxr... HV fuse blown... underloading, etc. I'm
now thinking that since all bugs are out... and all steady state
grid/plate current/ fil V/ plate V measurements have been done....
.. and now we just pulse tune it... then talk.... OR just dial up
the vac tune load caps... "by the pre-set numbers".. we can
probably REDUCE the size of the fast grid fuse from the current
750 Ma... down to something like 400-500 ma.
If it works, why fix it?
### Also.. my scheme of paralleling a PAIR of 50A magnetic-
hydraulic breaker to make one big 100A breaker's works quite
well.... one such assy is installed in EACH leg of the 240V line.
The kicker here is... we removed the tie bar on each assy. With
steady state dead cxr stuff... all 4 x poles are turned on.... with
ssb.... u can safely shut one pole off per assy. [now it's a 50 A
breaker] THEN.. if something goes amuck... with a lower value
grid fuse.. and pri breaker [ HV fuse remained same value.. 3
A] ... while on ssb.... both the pri breaker and lower valued grid
fuse will blow even faster.
Is faster better?
Later... Jim VE7RF
R L Measures, AG6K, 805.386.3734
r@..., , rlm@..., www.somis.org
Yahoo! Groups Links
R L Measures, AG6K, 805.386.3734
r@..., , rlm@..., www.somis.org
