If any of you intend to series parallel Oil caps for greater
V ratings... BEWARE. I did just that with 4 x 50 uf @ 4500 V
units. These were each 12" tall x 8' wide x 4" thick. I
got em years ago from W6RR, Roger Mace... now sk.
I had one of em flash over [6900 vdc no load] . The arc
occured between the steel case and the threaded rod that
comes up outa the oil !! Even though these units had large cone
HV insulator's out the top... the weak spot is BELOW the
insulator /rubber O ring.
The fix was to FLOAT all 4 of em... both from the chassis...
and from EACH other ! I used a huge sheet of 1/4" thick HV
Red Micarta under them... and 1/8" thick sheets of micarta
BETWEEN em. End of problem. Even though mine.. and most of
these oil caps have a nice grnd lug welded to the steel case...
DON'T BOND it to the chassis. By floating all the cases... you
will RELIEVE the stress on em... they will never arc. Treat em
like lytics.
There is now an abundance of huge size 350 V / 450 V / 500V
lytics on the surplus market. 1600uf to 5200 uf. Lotsa
joules ... which is what we want.
The 1600 uf @ 450 V lytics I obtained from Fair radio yrs ago
[$5.00 each] were all measured between 1900 uf and 2000
uf. They were all 3" diam x 6" long... and used # 10 machine
screw terminals.
I mounted 20 of em horizontally [4 rows of 5].... between two
huge sheets of 3/8" thick UHMW [from the local plastic
shop] We used a 3" hole saw in a modified drill press... and
drilled out both sheets at same time.
It all fits in a nice 19" wide short rack... with front and
rear panels. The HV meter is a 0-1 ma with 10 x 2 watt, 1
meg, 2% tol series resistor's.
This is my "external cap box". I used Millen Red/ Black HV
connector's on it. IF using the Millen's.... either use
nylon mounting screws or mount em on Micarta. Both Kings and
ROWE make better HV connector's imo.
Measured C is 100 uf. I built it years ago and used 50 k,
5% tol, 25 watt wire wound's across each cap. If you build
one of these huge lytic stacks... use Rich Measure's 100 K @ 3
watt units across each lytic. For faster bleed down time...
use a PAIR of his 100K 3 watt units across EACH lytic.
With the wirewounds, I used 'L' shaped copper straps on each
screw terminal... to dissipate heat. One end of the 'L' goes
to the screw... the wire wound is screwed to the other end of
the 'L'.
With just a single 50 K resistor across each one.... HV bleeds
down slowly.. a few mins. With 100 K units.... it takes
longer.
I recently obtained cases of 2500 uf @ 450 V lytics from that
fellow in Florida. Now these newer design caps are slick...
only 2" diam x 4" long.... and a whopping 9.8 A CCS ripple
current rating. ESR is aprx 50 MILLI ohms each.
I have 72 of em.... so I'm building 3 x banks .. with 24
caps per bank. I can use 1 or 2 or all 3 banks in parallel.
To protect all this mess.... we used a BUSS HVU-3 HV sand
filled fuse [ SSON has ton's of em... so does Buss... Buss
has em in a huge variety of ratings]. These units are 13/16"
diam on the ends... and fit a standard 3/4" fuse clip snugly.
I used the 3/4" from SSON that have end retainers on em... so
once inserted, there is NO way they can migrate out.. esp when
mounted vertically.
The surplus HV fuses from SSON are rated at 3 A CCS.. and
are sandfilled. $3.50 each... in bulk. The SSON units are
glass bodied.... the newer ones from Buss are Ceramic bodied...
other wise, they are identical. The 5" long HV fuses are
rated at 5 kv AC RMS.... and work at 8 kv DC no sweat.
They also make 10" long fuses... rated at 10 kv AC RMS / 16 kv
DC.
I install ONE of these HV fuses... in ONE leg.... between the
SECONDARY of the Dahl plate xfmr and the FWB diodes. The
2nd HV fuse goes BETWEEN the B+ output of the lytics and the
HV "Glitch Resistor's"
If something ever shorts out in the lytic banks.. the 1st
HV fuse between sec of plate xfmr and HV diodes will save
the diodes every time. With fault current... they blow
REALLY fast.
The 2nd HV fuse, located in the B+ line is always located
BEFORE the glitch Resistor... never after. IF u flipped em
around... and something happened to the glitch R [arc, dust,
etc] the fault current from the lytics could easily damage /
finish off the glitch R assy.
The "Glitch Resistor" for 2.5 to 3.0 A of dc plate
current can be built a couple of ways. My buddy used 2 x
paralleled 100 ohm @ 225 watt wire wound resistor's. A fan is
used on em to keep things cool at all times. With 3 A
keydown... diss is a whopping 450 watts ! On ssb/ cw... it's
only 1/4 of that... or 112 watts. [66 watts per resistor]
The 2nd method employed uses 4 x 200 ohm 275 watt globars...
each 1" diam x 12" long. I had loads of these older, 60's
vintage globars.... TYPE CX. BEWARE, the older CX type
globars are NOT glass bodied.. and are NOT suitable for oil
immersion. Since I couldn't use em for a QRO dummy load... they
make perfect... "glitch resistor's ".
On my buddy's latest project [I designed it, he built it],
while ironing out some bugs... One or both HV fuses opened up a
few times. The paralled 100 ohm-225 watt wirewound's didn't
flinch. The surge current is limited to aprx 158 A..... which
will blow a 3 A rated fuse in <2 msecs. The glitch R and
the C filter will also form a 'RC' time constant. Without a
glitch R... the current from the bank(s) of caps is HUGE...
only limited by the ESR of the 20-30 lytics... and the ESR of
the plate choke(s) 1.25 ohms + 1.2 ohms.
I always use 6A1O diodes... rated at 6 A CCS @ 1 kv PIV...
400 A surge. These babies are exactly double the diam of a
1N5408. They will drop right in where a 1N5408 once stood.
Dahl figured this one out too... and makes all his diode
assy's from 6A1O's. I looked at HV diodes at 8-10-12 A
ratings.... and all of em had LESS surge capability than a
6A10... and all of em required a heatsink.
The 6A1O's are used every where.... from meter protection to
series strings of em... used for BIAS. [tapped with a 20 x
position rotary switch....slickest thing yet]
Don't mess with .01uf disc caps and resistor's across each
diode... waste of time/ money /effort/SPACE. The "IN" way of
building strings of HV diodes is to use ... no load plate
voltage x 3. IOW... for a 7900 V supply... use 24 diodes
per EACH of the 4 legs in a FWB.
Now both Dahl and other's use the "ultimate" in diode
protection. They parallel a 510 v MOV across each diode.
[510LA40A.. available from Mouser, Digi-Key] I also use a 275
V bolt down MOV across the plate xfmr PRIMARY [Movistar,
available from Dahl]... and 130 V bolt down MOV's ... between
each hot leg and neutral/grnd. [available from Mouser/Digi-key]
Some will even install 6 x 1000 v [in series] MOV's across the
plate xfmr sec.
IMO... the huge banks of lytics will "swallow" any
transients, spikes,surges, that manage to get through all of
that. The lytics will look like a low Z short to any higher
V spikes, etc.
Of course... this mess of lytics has to be step started...
properly. We use a 0-30 delay timer.... the type that comes
in a 8 pin octal base... and use the mating flange mount
socket [no holes required for the mating socket... you can
flush mount em to anything] These delay units all use a DPDT
mech relay with 10 A CCS contacts. I get mine with a 120
vac coil. One 1/2 of the DPDT contacts activate the step
start contactor. You can use the other 1/2 of the DPDT
contacts to turn on a NEON.... or close the key line to the
linear..[via a pair of RCA connector's.] This way... the big
amp can't be keyed UNTIL the step start cycle has been
completed. On our step start contactor's that use 220/240v
windings.... both halves of the DPDT relay inside the delay unit
must be used. Fortunately.... those contactor's had AUX
contacts... and the AUX contacts can easily be used to close
the key line to the big amp + turn ON a neon.
The Neon can be used as a visual to let one know that the step
start conatctor /TD relay has operated.
We use 2 x paralled 50 ohm 100 W resistors for the step
start resistor assy. They can be either metal finned bolt
down units... OR wire wounds. If one ever opened up.... the 2nd
one is the back up. If a single resistor was used... and opened
up.... you could shunt a wide open !
The step start assy [25 ohms] is inserted in ONE LEG only of
the primary of the plate xfmr. 50 ohms is too much
resistance... and will produce a bigger current surge when
shunted. Install a 0-50 A AC ammeter or a 0-100A in one leg of
the 240 V primary.... you will get an eye opener ! The 50 A
and bigger AC ammeter's use current xfmr's.... which simply go
around one of the big 120 v hot conductor's. Small 18 ga
wire is then used between the actual meter movement...and the
current xfmr.
Now here's the kicker. Joules [watt seconds] goes up to the
SQUARE of the voltage! When a 8 kv pwr supply is charged up to
4 kv... the caps are only ONE QUARTER charged.. and have 1/4 the
energy of a fully charged cap. The same caps charged up to
5656 volts is only ONE HALF charged up !!
The trick here is to get the HV up as HIGH as you can... BEFORE
shunting out the step start resistor. This implies NOT
sucking huge amounts of bleeder current. Also BEWARE... all
these Dahl / Pole pig plate xfmr's typ suck 1.4 to 1.8
Amps of ac magnetizing current on their 240 v primaries.....
even with NOTHING connected to the HV secondary !! [The
magnetizing current will always be flowing through the step
start resistor... until it's shunted] .
We typ set the TD delay for 14-20 seconds... with 'typ' 100-135
uf C input HV supplies. And longer with 200-300 uf supplies.
34 uf will typ give you aprx 3% ripple. [6900 V @ 3 A]
100 uf will give you 1 % ripple. Ripple is one thing....
dynamic regulation is another [as opposed to static
regulation] . To get good dynamic regulation on a FWB C input
HV supply.... you want to shoot for 1% or less ripple.
These Dahl 253 lb xfmr's I use have a .08 dc ohm primary...
and a 8.2 dc ohm secondary ! My 10 kva Hypersil pole pigs typ
have 16-18 ohms of sec dc resistance.
Dahl tells me there is NO problem whatsoever using 150-300 uf
C input filters with his xfmr's.
If modifying a 3 pole contactor as one big SPST-DM contactor...
after paralleling the 3 contacts on each side.... you go in on one
corner... and exit diagonally. [enter top left... exit bottom
right] This will distribute the current evenly. This trick
is also used on paralled doorknobs, diodes, bandswitch wafers,
breakers, broadcast caps, etc.
On these latest bunch of 135 A rated 3 pole contactors... we
paralled all 3 poles with just aluminium 1" x 1" x 1/8"
thick 6061-T6 angle stock. This also makes it EASY to
terminate the 2 ga wire used [and also 3 x ought] Also makes
it easy to simply bolt two angles together... when the output
of one contactor is feeding the input of the step start
contactor. One modified contactor is installed in each hot
leg. The 135 A rating looked a little flaky to me... and we had
plenty of contactor's... so instead of 2... we used 3 x modified
ones. I also would suggest silver plating the contacts
with 'cool amp'... available from the cool-amp co... in Ore. I
use that stuff to silver plate everything from contacts to
tank coils to loads of copper strap. Use latex/vinal gloves..
and use ur fingers to rub it in. Contact resistance will drop
to zero every time. It works EXTREMELY well... so does 'conducto
lube'... used on sliding contacts.... pure silver in grease.
Sri for the diatribe. Sorry for boring anybody.
Later... Jim VE7RF
