Date: Mon, 11 Jul 2011 00:10:05 -0400
From: "David H Craig" <davidhcraig@verizon.net>
Subject: [Amps] GS35b FWB issue
I lost a diode string on an 8x4 string 1N5408 FWB on my single GS35b 6m HB
amp. To whit, 8 x 1N5408s per leg of a FWB for a total of 32 diodes.
### with a 3 kv sec + 245 vac fed into the 240 primary, ur no load B+ =
aprx 3062 x 1.414 = 4330 vdc. If you want to do this right, you want
12-13 x diodes PER each of the 4 x legs..and use 6A10's. If you look at any
of these commercially made FM broadcast single phase B+ FWB assy's... they
all use triple the no load B+
Presumably a lightening strike / blown lightening arrestor, but now I wonder.
## Install a heavy duty lightning arrestor right at the 240 vac service
panel....like the Joslyn series..which is an industry standard. Consist of 2 x
big mother MOV's. [ 130-140 vac rated..EACH]. Each MOV is wired from hot
to NEUTRAL. The neutral is bonded to the grnd connection inside the main 200A
panel. A fuse/breaker is used between each MOV and neutral. That will solve
abt 99% of any glitch's /transients/spikes/surges. Then you are protected
from tranisents/spikes coming down EITHER hot leg..or across both hot
legs..which is why they use 2 x 130/150v rated move's in series.... with the
CT going to neutral.
## use KM1H's suggestion , and put a .0047uf @ 10/15 kv disc cap across each
leg [ 4 required]. They are made by Ceramite. Both henry radio + Mouser carry
them. Put 1-2 more in parallel, between B+ on output of lytics and chassis.
Put a couple more at base of plate choke. [ these are 4700 pf each]
## use step start, with a 10 sec timer. If commercial AC power goes off for
any reason.. then comes back on say 5-30 mins later, the B+ supply will
simply go through it's normal step start sequence again.
## Wire a single HV fuse... between SEC of plate xfmr ..and input to FWB
assy. This HV fuse can go in EITHER leg of the plate xfmr sec.
## Ur 12 x 450 vdc @ 800 uf caps are not overkill. 12 x 450 vdc = 5400
vdc. Ur no load B+ [ with 245 vac fed into pri] is 4330 vdc. 4330/5400
= .8018 or 80% of their max ratings. This is hardly overkill. I never run
em more than 70-75% of their max rating. 800 uf /12 = 67 uf. 67 uf
isn't over kill either. [ I have tried 450 vdc caps in 2300/2500/10,000
uf...and discovered the dynamic regulation is beyond superb, provide you have a
decent size plate xfmr....and minimal V drops to main panel]
## It appears you have used 6 x 27 K resistor's in series [ 162k] across
each HV cap. That's just asking for trbl! You have 6 x 12 = 72 resistor's
to mess with, and if any of the 72 ever opened up..then kaboom.
Use a SINGLE 100 K @ 3 watt mof 1% Tol resistor across each cap. [
VISHAY brand. Mouser part # 71-CPF3-F-100K-T2
Mouser cat # CPF-31003FB14 VISHAY PN CPF3100K00FHB14 . Use stakon
full circle type crimp connectors, and crimp a lug onto each end of each 100k
resistor..then solder em as well. These 100 K 3 watt mof resistors are
superb, and no need to ever play matchup ever again. They are well under 1%
tol. Mine are all just .2% tol. You will end up with a precise V
across each HV cap. The combo stakon crimp tool/wire stripper is standard
hardware store stuff. [ home depot/electrical supply, etc].
## OK, now you require a 6A10 across EACH of the 12 x hv caps. Each
6A10 is RVS connected across each cap. Each diode requires another pair of
stakon crimped and soldered connectors. [ don't try and stuff the 100 k
resistor + diode into just one stakon]. Use an internal tooth
lock-washer..and I assume you are using 10-32 machine screws for each hv cap.
Then crank em down good and tight. The RVS connected 6A10 diode across
each hv cap is a safety diode. If any of the 4 x legs in ur main FWB assy
ever shorted out, you don't want to be pumping raw AC into the string of hv
caps on 1/2 cycles, and destroying the caps.
## Wire some bolt down /block type 250/277 vac MOVS across the 240 primary
of the plate xfmr itself. Wire a bolt down block type 130 vac rated MOV
between each 120 vac hot leg..and neutral. [ 2 x required]. Mouser has loads
of these bolt down type movs. Typ 2" wide at the base x 2.5" tall x 1/2"
thick....with 2 x machine screws on top for the electrical connections. The
base is bolted to the chassis/metal sidewall... with 2 x machine screws.
## get rid of that 250 ohm 50 watt B- float resistor. Wire 6A10's RVS
connected, between B- and chassis....again using crimped sta-kon connectors,
and solder em. I use 2-4 diodes in parallel, to make one big diode assy.
2 x such assy's are used... so 8 x 6A10 diodes used in total. OK, now ur
B- can never float more than +/- .7 vdc from chassis potential.
## we are not done yet. Between B+ and ur output HV connector.... u need
the 2nd HV fuse + glitch R . From B+ on output of string of caps, wire to
HV fuse 1st. Then wire from output of HV fuse to input of a 50 ohm /
50-225watt wire wound glitch R. You always want this 2nd HV fuse to precede
the 50 ohm glitch R. 50 ohm @ 225 W CCS wire wounds are dirt cheap @
Mouser. I have also used 2 x 100 ohm resistor's in parallel....and also 4 x
200 ohms in parallel. [ all in 225 W CCS variety] I have also used 4 x 200
ohm globars in parallel. [ 1" diam x 12" long] . If ur sucking 2-3 A of
plate current, key down...even for just 10 x secs to either tune an amp, or
just to get some steady state plate / grid current readings.. u need the big
stuff. I also fan cool some of the glitch R assy's.
## I use a magnetic-hydraulic breaker in the 240 vac line.... where the 240
vac 1st enter's the HV supply box. Use a super fast one. The fast ones
have NO oil in the internal resovoir and are ... 'instant trip'.
## Wire a B+ meter directly between B+ and B- of your string of
lytics...[ inside HV supply]. Wire a 2nd B+ meter inside the RF deck,
between B+ and B- NEVER wire any HV meter between B+ and chassis. You
also have to bond the RF deck chassis and HV supply chassis directly
together.... and not via a star grnd either. Use 6-10 ga wire with big lugs
on each end....and terminate each end with 1/4-20 SS bolts on both
chassis's. use SS flat washer's / split ring lock washers etc.
## OK, now you are done. Everything is now FULLY protected.
## IF you blow the HV AC fuse between plate xfmr sec + FWB... BOTH HV
meters will drop to zero. IF you blow just the HV DC fuse [ just b4 the
50 ohm glitch]... the HV meter in the RF deck will read zero.... but the HV
meter inside the HV supply will STILL read 4330 vdc. A simple 0-500 ua
dc meter + 10 x series 1 meg @ 2-3 watt mof's in series will suffice for a
0-5 kvdc meter. A 0-1 ma with the same 10 x resistor's will read 0-10 kvdc.
## IF any individual HV cap shorted, nothing will happen..except ur max cap
rating went down by 1 x cap. IF you have a short across the sec of the plate
xfmr, the 240 vac breaker on input to plate xfmr will trip open. IF you have
a short, or grnded B+ in the string of caps, the AC HV fuse [ between sec of
plate xfmr + FWB diode board] will blow open.
## IF u get a short from B+ inside RF deck to chassis/metal... the 50 ohm
glitch R will LIMIT the fault current to just 4330/50 = 87 A. The DC
HV fuse just prior to the glitch R will blow open < 2 msecs. You have now
INTERRUPTED the fault current completely. In that case, the fault current
path is from B+ to chassis..then from chassis up through the 6A10 safety
diodes, and back to B- If 2-4 x 6A10 diodes are used in parallel, their
combined surge rating is an easy 800-1600A. They won't open up..since the
glitch R limited the fault current to just 87 Amps.
## I'd use more than 2 seconds for the step start delay. I use 10-15
secs..so as to get the b+ up to it's highest level, to minimize any sec
surge. Now this all depends on what size step start resistor u used, and also
what size hv caps. [ total uf] etc.
## Now u are protected from anything and everything. If you or anybody else
requires HV lytics, I have caseloads of em. All 450 vdc @ 2300uf [2.5" OD
x 4"]....... and also 2500 uf [2" OD x 4"] CD and also united chemi con
brands, all new, complete with 10-32 machine screws,etc. I can also toss in
the stakons and 6A10 diodes and also the 100 k @ 3 W @ 1% tol resistor's if u
require them.
## The above scheme is 100% bomb proof. It's what's I used on 5-15 kw
out HF RF decks. It's used on a 2 x GS35B 6m amp as well. [ 5.2 kw out]
later......... Jim VE7RF
I replaced the diodes on the blown string & tested again- worked fine with
just HV trans running to the bridge. Then, put back together w/ cap bank-
blown fuse again & no solid HV. Checked again, another leg of the FWB blown.
Replaced those 8 5408s & tried again with the cap bank inline. Ergo, another
string was blown. Replaced them, and then tested with just HV & the FWB. No
problem. Then connected the cap bacn (overkill, 12 x 450V in series, nominal
800 MFD Mallories w/ 6 x 27k ohm Rs across each cap) WITHOUT the RF deck
connected, and another audible glitch. Tested all the 5408s in the bank, and 3
of the (just) replaced diodes show x Meg ohm in the WRONG direction (yes, I
tested all before & after installation before poweup). All the rest were fine.
Then, tested each cap in the 12 x 800 MFD bank, and discovered nothing (none
shorted), but on a hunch, tried to tighten each screw, and low and behold,
after 2+ years in service, EVERY screw needed 1/8 to 1/4 turn
to get tight. Question: Could the combined losening of the screws create
enough resistance to draw enough current to blow a diode string in the FWB? I
have checked everything else in the PS, and the common denominator seems to be
the cap bank. Am I missing something?
Without load at least, the high current relay works 100% & the HV
soft-start relay is working 100% as is the 24 volt ps that works the relay.
The 2 second TDR that controls the soft-start is similarly fine.
The 250ohm 50w B- float resistor is also fine (tested).
Having isolated the problem to the HV supply, I am baffled. I plan to
replace the now-bad diodes, again, but am not sanguine it will work, and I am
running out of 5408s... I don't want to waste bigger diode blocks on this if
they'll short too.
Any advice or suggestions would be appreciated.
73 Dave N3DB
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|