I was thinking about a RVS connected diode across the choke..... but Hammond/dahl read me the riot act, said...'don't even think about it'.? ?They are correct of course, since the normal Voltage across the choke will swing up/down a bit like +70 to +100 vdc...then swing negative the same amount.? The diode would conduct on 1/2 cycles....and short out the choke.?
A MOV may well work, BUT if it fails, it will fail shorted every time.? ?MOV's really should be fused...but that would require a proper sized HV fuse.? Modeling in software, the peak V can easily be through the stratosphere, when excess amounts of fault current are drawn through a choke, like almost triple the B+ in some cases.
The spark gap is dead simple.? And yes, probably? ? .005" is ample.? ?Actually, .032" is good for 1100 V..so that amounts to 1100/32 =? 34 volts per .001".? ?.005" = 172 volts.? ?I believe mine is aprx .005"?
The thickness of a 8.5" x 11" sheet of printer paper is typ .005" thick. (.13mm)
?
On PSUD-2,? I don't use the resistive load, I use the constant current load.? ? And? then also have the option to toggle between any 2 x distinct current loads, like? say 200 ma, then 800 ma... OR the other way around, like 800 ma, then down to 200 ma.? ?It won't simulate? ssb,? ( but sorta).? ?When u toggle between 2 x different loads, it's like? right now, asap, instantaneous.? ?On CW, you have a 5 msec rise and decay time.? On PSUD, it's 0 msecs.?
?
PSUD-2 will not handle a resonant choke.? ?The problem with the resonant choke, like used in Henry amps is.... the B+ is only .9 X sec AC voltage.? ?With a C input filter, it's 1.414? x sec AC voltage.? ?IE: 6000 vac sec on a 8K ultra = 5400 vdc no load.? ?You coulda got the same 5400 vdc, by using? a FWB + just a (5400/1.414)? =3819 vac sec AC winding.? ?If a FWD was used, the sec is then just 1909 vdc.? ?The resonant choke setup also requires a huge amount of bleeder power at all times.? ?The 8k ultra uses? 5 x 20K @ 100 watt wire wounds in series.? ?That mess sucks 300 watts? CCS, or 60 watts per resistor...and all 5 of em run stupid hot.? If just any one of em opens up, the B+ soars to? 6000 x 1,414 = 8484 vdc.?
With the bigger choke values, like your 10H, they require a? smaller resonating cap...which is good,? BUT the 10H choke has a lot more dc resistance vs a 1-2-4 H choke.? ?All that wire has to fit the core, so they have to use smaller ga wire = less current handling capability.? ? The issue with the resonating cap, like when a 4 H choke is used is.... the choke + cap form a 120 hz notch filter.... fine, BUT all the harmonics of 120 hz will? sail right through the resonating cap.??
On a 10H choke, it requires a .1759 uf resonating cap.? ?A 4H choke requires just .4398 uf.? ?(20 hz will sail right through a 2 uf cap).? 240 hz will sail right through a .1666 uf cap.? It just gets worse with the higher harmonics of 120 hz...like 240-360-480 hz etc.? ?Then u have the issue with now 2 x dc currents flowing through the choke, the main current, and also the circulating current in the choke.? The smaller the choke value, the higher the circulating current.? It's bad enough with a 8 H choke, but with a 4H choke, the circulating current is as high, or higher than the main current.?
To kill the harmonics, ideally the C1 cap needs to be a decent value.
John Lyles, K5PRO, when he worked for Broadcast engineering, designing single phase 5 kw? FM broadcast PA's, told us that the FM pa's had to meet this spec, called an ...."AM noise spec" .? The supply used a resonant choke.? (resonating caps were a series pair of .9 uf oil caps, like used in microwave ovens).? Caps were sent to the choke manufacturer, and the choke was hand wound, so it resonated around the supplied caps, .45 uf for the series pair).? In some cases the choke is made to a larger value, so it can be used on 50 hz.... then tapped down to a lower value for 60 hz..so resonating cap pair stayed constant.?
?
Problem was, the resonant choke would not meet the....."AM noise spec'.? (I forget how big the C1 oil cap was, think it was like 16uf to 20 uf).? ?His fix was to wire a LC filter onto the end of the existing resonant choke + C1 cap.? This time the added LC filter does? NOT have it's choke resonated.? It's there to kill the harmonics of 100/120 hz.? IE: it ends up being a (resonant) L - C1? - L2 - C2? setup.?
?
A simpler? ?C-L-C woulda done the same thing, and also coulda used a much lower secondary Vac.??
I also wanted to try yet another config... using the same? C-L-C? setup, BUT resonating the choke.? Got a buddy in W5 land to simulate it on LT SPICE for me.? It results in a lousy 4 db improvement at 120 hz.? The main drawback was, the existing, superb harmonic attenuation? was ruined by the resonating cap.? Same deal, harmonics sail through the resonating cap....and attenuated only by whatever value my C2 cap is.?
The consensus was to not mess with a resonating cap on a C-L-C config.?
As far as testing your choke for heat, like with your proposed? 2.0 to 2.5 amp? CCS load,? that's? easily tested by using a small lab supply, with current limiting, like a 0-160 vdc @ 5 amp variety.?
With 850 ma into 28 ohms =? 20.23? watts.? ?2.0 amps into 28 ohms =? 112 watts.? ?2.5 amps into 28 ohms =? 175 watts.? When more than rated current is drawn the dc resistance increases, then it slowly goes into thermal runaway.?
?
But for FT-8 use, like (I think)? 15 secs on, and 15 secs off, duty cycle is 50%, so may well get away with it.?
?
Another option is to use a lot more C....since ripple is inversely proportional to uf.? Doubling the uf results in a 6db? reduction in ripple.? ?But you would need one helluva lot of uf.?
Another option is a smaller choke, like 1-2-4 H, but with a lot higher current rating.?
Mine is rated for 3 amps? CCS, and is 4 H....and good for ( I think) 15 kv..... but it's big, 127 lbs.? Stone cold with 3 amps? CCS.? With 6 amps CCS, it runs pretty hot, (195 F)? right where the gapped cores reside.?
