#39382

OK, the 2? (brand new)? x Ameritron ALS-606's only put out barely 300w on 6m (and only 400w on 75m). Ready to tear our hair out. Plan C was the new ALS-1306...which actually does work.

Cut to the chase. 10 kw CXR out with aprx 600 w of drive. (brand new Coaxial dynamics 25-60 mhz, 10 kw calibrated slug). Line section with the 10 kw slug / meter has the pep kit installed. Line section with 1 kw slug, between ALS-1306 and input to the tube amp....(to measure drive into tube amp). Line section with 25 watt reflected slug between ALS-1306 and input to tube amp....(used for tweaking the PI tuned input, with it's 2 x air variable caps).

Jennings ceramic vac tune cap....and a Swiss made (Comet) ceramic vac load cap. All testing done on 50.125 mhz.

The big dummy load consist of 32 x 1600 ohm globars (18" long) in parallel to make the 50 ohm commercial DL. 7-16 Din on output of tube amp and also downstream.

B+ sagged quite a bit with a CXR. But with every 1 volt AC going in, he has aprx 25 vdc coming out. The line sagged 6 vac from the street, so we lost 150 vdc. Then another 125 vdc through the 40 ohm glitch R. No load B+ was 5900 vdc. Loaded to 3.125 amps of DC plate current, it drops to just 4300 vdc. (It would have been 4575 vdc without the above pair of V drops). Then you have the typ 5-10% drop in the B+ supply anyway. Then the small V drop across any 2 of the 4 x strings of diodes in the FWB.

On SSB / CW, it drops very little. This is the difference between a static load....and a dynamic load. Total of 30 uf of filter caps (2 x 60 uf @ 4 kv oil caps in series), so we don't have a huge amount of energy stored in the caps. (522 joules).

Plate current = 3.125 amps
Grid current = 500 ma
Loaded B+ = 4300 vdc

DC input = 13,437 watts.
Power output = 10 kilowatts.
Anode Dissipation = 3437 watts.
Eff = 74.4 %

Bias can be toggled between 50 ma idle and 200 ma...(Class AB). (toggle switch changes the tap on the series diode string, used for bias).

Special thanks to John Lyles, K5PRO, for his assistance on Saturday, conducting some SRF (self resonance freq) tests on some doorknob caps, on his Vector impedance meter. (Old HP 500 khz to 108 mhz unit). We changed out the bypass caps at the cold end of the fil choke.

Scott designed and built the HV supply.? I designed the unique RF deck. (Scott? also built the RF deck).?
The 80-15m 3CX-6000A7 amp will be the next project....already in the works. I have most of the RF deck designed.

https://www.youtube.com/watch?v=W87ySwzL_-I

Later.... Jim VE7RF

#39383

开云体育

Good job, congrats.

But efficiency is less then 74.4%, you forgot to deduct the drive power which is fed through.

?

73

Peter, DJ7WW

?

-----Original-Nachricht-----

Betreff: [ham-amplifiers] 10 KW CCS ON 6M...USING THE 3CX-6000A7... PART 30. Success at Last !!

Datum: 2023-08-01T09:01:31+0200

Von: "Jim VE7RF" <jim.thom@...>

?

DC input = 13,437 watts.
Power output = 10 kilowatts.
Anode Dissipation = 3437 watts.
Eff = 74.4 %

?

#39384

With the aprx? 500 watts of feedthrough power deducted, eff drops to 70%.? I still find that hard to believe.? ( 625 watts drive applied).

Perhaps the HV meter is reading too low.
Wattmeter is reading too high.
Plate current meter is reading too low.
Combo of above.?

I was expecting IF we got 60% eff, we would be ecstatic, and overjoyed.? ?I never subtract drive power.... I call that ..'free eff'.?

?

Typ the secret is in the tuned input.? Get that wrong, and it all goes to hell.

?

The 10 kw? L-PI output is a unique design I cooked up.? ?1.125" OD? Tubing was out, as was a? strap coil, using 1.75"? wide strap.?

I calculated out the required uh for the main coil, using strap in a straight line....aprx 8.3".? Then bent the straight line into a 1/2 circle.? ( same 8.3", but 1/2 the circumference of a circle).? ?Knowing 1/2 the circumference, the rest was easy, just double the 8.3"? to find the total circumference.? ?Divide circumference by PI to get the diam (4.99" inch spacing between vac tune and load caps).? ?The amount it bulges out is just the radius, or? 2.5".?

?

The idea was, the strap loop conducts? Current on BOTH sides.? ?Strap coils only conduct on the outside.? ? Tubing coils only conduct on the outer periphery.? ?Kintronics? models a tube as a length of strap...that has been rolled into a tube.

?

Short of it is... we needed something that would handle a stupid amount of RF current at 50.125 mhz.?

?

A similar calc was done to design the 'L' coil? ( that goes between the block caps and C1 tune cap).?

?

Next up is IMD and harmonic suppression tests.? ?Then tube removed, and amp + HV supply crated up.?

#39385

Typ the secret is in the tuned input.? Get that wrong, and it all goes to hell.

I found this out about 45 years ago. ?If the input is broad band and the swr shows 1 to 1 it has to be investigated if further tweaking can add another 10% or so to final output. ?Got to remember that if you are running at 12db gain every additional?watt of drive power is multiplied by 12db. ?The input is extremely important and is often overlooked. ?Here in could lie additional efficiency you may be losing.

Show quoted text

On Tuesday, August 1, 2023, 4:14 AM, Jim VE7RF <jim.thom@...> wrote:

With the aprx? 500 watts of feedthrough power deducted, eff drops to 70%.? I still find that hard to believe.? ( 625 watts drive applied).

Perhaps the HV meter is reading too low.
Wattmeter is reading too high.
Plate current meter is reading too low.
Combo of above.?

I was expecting IF we got 60% eff, we would be ecstatic, and overjoyed.? ?I never subtract drive power.... I call that ..'free eff'.?

?

Typ the secret is in the tuned input.? Get that wrong, and it all goes to hell.

?

The 10 kw? L-PI output is a unique design I cooked up.? ?1.125" OD? Tubing was out, as was a? strap coil, using 1.75"? wide strap.?

I calculated out the required uh for the main coil, using strap in a straight line....aprx 8.3".? Then bent the straight line into a 1/2 circle.? ( same 8.3", but 1/2 the circumference of a circle).? ?Knowing 1/2 the circumference, the rest was easy, just double the 8.3"? to find the total circumference.? ?Divide circumference by PI to get the diam (4.99" inch spacing between vac tune and load caps).? ?The amount it bulges out is just the radius, or? 2.5".?

?

The idea was, the strap loop conducts? Current on BOTH sides.? ?Strap coils only conduct on the outside.? ? Tubing coils only conduct on the outer periphery.? ?Kintronics? models a tube as a length of strap...that has been rolled into a tube.

?

Short of it is... we needed something that would handle a stupid amount of RF current at 50.125 mhz.?

?

A similar calc was done to design the 'L' coil? ( that goes between the block caps and C1 tune cap).?

?

Next up is IMD and harmonic suppression tests.? ?Then tube removed, and amp + HV supply crated up.?

#39386

Lou, why would a 1:1? input swr raise alarms ?? ?I did notice on the heath SB-220? tuned inputs, the C1 + C2 cap values are very low? vs the L4B.?
IMO, the Q is way too low on every band.?

I gave up on tuned slugs years ago, like the infamous Millen? 69046 slugs.? You have to sit with a box of silver mica caps to get em right on each band..pita.? ?The ferrite slugs will only take such much power.? AC0C burned up the slugs on his highly modified? SB-200 amp...running RTTY.? ?The wire on em is puny as is....ditto if T-50's used...unless bigger roids used.?

If? ?T-50's etc are used, and mating ARCO compression trimmers ( and padded on 160..and maybe 80m).... per KM1H, they are good for 200 watts ssb / cw.?

On these metal tube amps, the pair of air variables and either tapped coil or roller ( 8 ga wire)? u can use any Q u want... always flat swr... and covers the entire 1.8-30 mhz.? Length of coax to ipa / xcvr ..and also? bias / b+ level / drive level, etc, is easily compensated... just tweak both caps for a match.? With the q on the high side, and on 80m, it won't be dead flat across the band.? ? No big deal,? ?I have one setting for the lower 80m portion..and another setting for the upper 80m portion.?

The concept of trying to design and build nine (9)? high powered, and bandswitched PI? ?tuned inputs was the final straw.? Now that's the ultimate pita.?

#39387

Here's the issue with the V regulation on the 10 kw 6m amp. Total length (1 way) from plate xfmr primary, to the new 50 kva pole pig xfmr across the street is a whopping 267'. And of course, double that or 534' for the complete loop. Then add in any non perfect splices or crimps. The power company uses aluminum wire for the drop line into the home...and 1 gauge smaller as well. If aluminum is used, it needs to be 2 x gauges bigger, not 1 x gauge smaller. Everything between poles is aluminum, and that also includes the 14.4 kv line as well as the 240/120 stuff.

Scotts driveway is real long....and that part of his drop wire is all underground, in conduit. It comes out, out of the grnd at the road, up a 20' pole, then across the street..at an angle, and terminates on the 40' pole with the pole pig xfmr.

Ideally, new cu wire, and bigger gauge needs to be pulled through the existing conduit, replacing the old, undersized junk. If existing conduit is too small, bigger diam conduit needs to be installed. Ideally, a new pole needs to be installed on his side of the street, and the 14.4 kv line wired across to his side of the street...with a 75 kva xfmr on the new pole.

Apparently, the preferred method of installing 400 amp service to the home is to install a new 400 amp meter base. Then TWO, 200 amp cable assy's are wired to the new 400 amp meter base. One pair of cables goes to the existing 200 amp panel....and a 2nd pair of cables going off to a NEW 200 amp panel. Then the new install is easy peasy.

At that point, the V drops have been minimized. Only other option after that is done, is to add more uf to the filter cap assy.

This is the problem when using high powered PA's...in residential area's. The B+ supplies are typ just a FWB + a C filter. They are UN regulated to begin with.

On a side note, PSUD-2 will factor in the DC resistance of that 534' total loop length. That just gets ADDED to the plate xfmr's primary DC resistance. DC resistance of the plate xfmr primary on the bigger xfmr's, is typ just a fraction of 1 ohm.

You start increasing the resistance on the pri side, you can see the effects asap in PSUD2 / spice / LT-spice. It screws everything up.

The peak current, every 8.3 msecs will make u gag. It's 60 amps every 8.3 msecs on an Ameritron AL-1200 amp, running at 1.5 kw cxr...on a 240 vac / 60 hz circuit. On a 10 kw out amp, it's typ 400++ amps every 8.3 msecs. Do the maths. The peak V drop, every 8.3 msecs on that 534' long loop is just wicked.

And all that peak current also has to flow through any mains fuses, breakers, relay contacts, contactors, fuse holders, etc. This is why all that stuff has to be oversized.

Then when the HV filter uf is increased, peak currents also increase...or try to. I have found that greater uf used in the HV filter is always a benefit...and results in better dynamic regulation on ssb / am. And also slightly better static regulation. It also results in less ripple, as ripple is in inverse proportion to filter C. (double the uf, and 1/2 the ripple etc).

Later Jim VE7RF

#39388

Here's the issue with the V regulation on the 10 kw 6m amp. Total length (1 way) from plate xfmr primary, to the new 50 kva pole pig xfmr across the street is a whopping 267'. And of course, double that or 534' for the complete loop. Then add in any non perfect splices or crimps. The power company uses aluminum wire for the drop line into the home...and 1 gauge smaller as well. If aluminum is used, it needs to be 2 x gauges bigger, not 1 x gauge smaller. Everything between poles is aluminum, and that also includes the 14.4 kv line as well as the 240/120 stuff.

Scotts driveway is real long....and that part of his drop wire is all underground, in conduit. It comes out, out of the grnd at the road, up a 20' pole, then across the street..at an angle, and terminates on the 40' pole with the pole pig xfmr.

Ideally, new cu wire, and bigger gauge needs to be pulled through the existing conduit, replacing the old, undersized junk. If existing conduit is too small, bigger diam conduit needs to be installed. Ideally, a new pole needs to be installed on his side of the street, and the 14.4 kv line wired across to his side of the street...with a 75 kva xfmr on the new pole.
Apparently, the preferred method of installing 400 amp service to the home is to install a new 400 amp meter base. Then TWO, 200 amp cable assy's are wired to the new 400 amp meter base. One pair of cables goes to the existing 200 amp panel....and a 2nd pair of cables going off to a NEW 200 amp panel. Then the new install is easy peasy.

At that point, the V drops have been minimized. Only other option after that is done, is to add more uf to the filter cap assy.
This is the problem when using high powered PA's...in residential area's. The B+ supplies are typ just a FWB + a C filter. They are UN regulated to begin with.

On a side note, PSUD-2 will factor in the DC resistance of that 534' total loop length. That just gets ADDED to the plate xfmr's primary DC resistance. DC resistance of the plate xfmr primary on the bigger xfmr's, is typ just a fraction of 1 ohm.
You start increasing the resistance on the pri side, you can see the effects asap in PSUD2 / spice / LT-spice. It screws everything up.
The peak current, every 8.3 msecs will make u gag. It's 60 amps every 8.3 msecs on an Ameritron AL-1200 amp, running at 1.5 kw cxr...on a 240 vac / 60 hz circuit. On a 10 kw out amp, it's typ 400++ amps every 8.3 msecs. Do the maths. The peak V drop, every 8.3 msecs on that 534' long loop is just wicked.

And all that peak current also has to flow through any mains fuses, breakers, relay contacts, contactors, fuse holders, etc. This is why all that stuff has to be oversized.
Then when the HV filter uf is increased, peak currents also increase...or try to. I have found that greater uf used in the HV filter is always a benefit...and results in better dynamic regulation on ssb / am. And also slightly better static regulation. It also results in less ripple, as ripple is in inverse proportion to filter C. (double the uf, and 1/2 the ripple etc).

Later Jim VE7RF