First try @ a homebrew actually amplified for the first time!!

[KJ4OLL]

Hi,
It works!!!

It's been a few years to get to this point. Learned lots of good stuff.
However, as the XYL frequently complements me "He has a mind like a steel sieve"

Had to do over the power supply to address "Critical Value of inductance"
-Doubled filter choke to 20H, increased bleeder current:


 
Had to ground the grid (first design had it floated)



Redesigned the cathode drive (replaced Micas w/ doorknobs), replaced the filament choke w/ a donor from some commercial rig:



Added a "glitch fuse":



Discovered the 3CX3000 needed an input tuner:




Took a while to find the sweet spot @3885:



www.youtube.com/watch?v=jXi3WUr5UWg

Now for the hard part, building Don's T/R Relay sequencer (ER #198)!

73
Frank
KJ4OLL

[WB4AIO]

Nice work! Can't believe it's a first try...

73,

Kevin, WB4AIO.

[flintstone mop]

The video was too quick. I thought there would be some views of "Bird Watts" going into the dummy

[KJ4OLL]

Hi,
First time the automatic fans have ever powered up on the Palstar DL5K!!

www.youtube.com/watch?v=XA-D0rD1sPE

73
Frank
KJ4OLL

[Opcom]

2KV is kinda low for that tube but its nice to know it can work there.

[KL7OF]

Congratulations.....The filament choke looks like it came from an FRT-39  (GPT-10K)...Are those ferrite rods in the center of the choke?...Steve

[W3RSW]

Agree that 2kv is way underutilizing a 3cx3000. If the meters have no other scale factors and I read them right in the Utube, then approx. 600 ma at 2kv is 1200 watts input or about 600 watts output.  Somewhat more than the filament power. 

Might try a voltage doubler with that transformer. See if it can still deliver sufficient current as a doubler to make it worthwhile.

But very nice work there. Maybe a little sprucing up of the front panel, maybe some enamel paint, some transfer lettering and Voila!, a very kuhl and professional rig on the outside to match a lot of good work on the inside.  Good Job. Bet you had a lot of fun getting it built, debugged and working.

[K4RT]

Frank - Thanks for posting the photos & video links.  How about posting your schematics when you get a chance. I think the front panel is nice looking the way it is.

[W3GMS]

Frank,

Great job and you don't throw an amplifier like that together in and evening. 

Very nice construction work and enjoy the amplifier.

Joe-W3GMS 

[w1vtp]

Agree on the plate voltage.  But then was the tank designed for that power level?

Nice work.  Make longer videos so we can drool and have time to wipe off our chins   

Nice picture quality.  Never did like those postage sized pictures.  Keep up the good work

Al

[K1JJ]

Hi Frank -

Neat amplifier and easy to work on with a lot of room.

Pick your favorite  color,  spray the front panel and add some transparent lettering to make it really shine.

Looking at the "glitch fuse" picture:  Even at 2KV, I'll bet if the glitch wire opens, you will have a sustained arc walking all around the area. It will probably blacken the insulators and might even touch the round gray object sitting next to it.  A short circuit will create a huge spark as I have found here. You need wider spacing so that the circuit gets broken quickly and does not linger.  I use glitch fuses that are spaced about 2.5" apart - and even they can get sooted up sometimes.  ( I use 4KV on my amplifier with a single thin outer shield strand from RG-213 coax)

The correct way, but expensive method is to use a real high voltage fuse rated for 10KV. The fuse is in a vacuum and will give instantaneous cut off. But I've seen them offered for $50 each, so I opt for the wire as you have.

That's a very elaborate RF input tuning circuit and it should give you a nice 1:1 match throughout the bands. I use a similar input system.

I agree to raise the plate voltage. It will make the tube cleaner and make better efficiency. That tube will do at least -40db 3rd order IMD. You should have at least 4KV to achieve this.  Otherwise it requires more grid current that always reduces cleanliness.

Of course, if you run the amplifier at 1500 watts pep out with little to no grid current, the cleanliness will be very good.

Oh, BTW, in case you forgot - that gray wire connection to the right is not soldered as of July 2 ...  


Tom, K1JJ

[WBear2GCR]

Nice job on the "desktop KW"!!  

Cool work... how about a wide shot of the guts, I guess from the rear??

One method WRT the chokes is "resonate" the chokes with a small cap across the choke.
In reality this turns the choke + the following (or preceeding cap, depending on if it is a choke input
filter or cap input) into a "eliptical" filter. In particular this makes the choke + (small) cap into a bandstop element, which is a notch type filter. This can put a pretty deep notch at a frequency of interest (120Hz, for example) and improve the performance of the filter where the notch is set substantially over the 2nd order or 3rd order
filter that one gets with a choke + cap, or cap + choke + cap configuration.

Looking at it, you can see that to get substantial rolloff of a filter at 60Hz (or 120Hz) one needs to start the rolloff (the -3dB point) a number of octaves below. The rolloff of a 2nd order is 12dB/octave, so working backwards, if you want 24dB of ripple reduction at 60 Hz, and 36dB at 120Hz, the -3dB point needs to be 15Hz. More rolloff, move that point lower in frequency... which means "make the choke bigger and add more capacity".

So, if you can't or don't want to add more Henries or microFarads, by "resonating" the choke, one can get effectively the -dB point of a much larger filter combination - but only for the resonant frequency. As much as double or more attenuation is possible.

In practice if one was using two chokes, one could put a cap between them and "resonate" each one for a different frequency... two deep notches.

Also, there are some nice computer based freeware design programs for power supplies - or you can build it in
LTSPICE - that show exactly what the results of different combinations of L & C will result in. The filter shape
does vary with the L & C combination and with respect to the load and source impedances. In practical applications, the filter shape for a typical PS filter doesn't have much and effect and probably does not need to
be taken into account.

The whole sheebang revolves around "Q".

</braindump>

                      _-_-

[WBear2GCR]

Tom,

How about putting the HV fuse wire inside a length of teflon tubing??

                      _-_-

[K1JJ]

Tom,

How about putting the HV fuse wire inside a length of teflon tubing??

                      _-_-

Bear,

I'll bet that would help to contain the arc  away from the surrounding area, but because it is not in a vacuum, the arc may sustain itself even more in that cavity environment.  We'd have to test it.   A real vacuum HV fuse is the correct method. Maybe there are some cheap Chinese HV fuses available.  

T

[KD6VXI]

I've encased a piece of #30 (iirc on the gauge)  in a bunch of sand before.   Use heat shrink,  soldering iron tip to one side.   Then a dab of hot glue to completely seal.

Now,  fill with sand.   Heat gun the heat shrink,  and a dab of glue on the opposite end. 

Poor man's sand filled hv fuse.

--Shane
KD6VXI

[KJ4OLL]

Hi,
Thanks very much for all the kind words and comments!
It is very exciting to have this critter actually work, w/o drama involving a fire extinguisher.

Lots of the parts came from the local surplus place http://www.skycraftsurplus.com/
Highly recommend visiting if you are in town for the Orlando Hamfest or Dizzy world, etc.

Most of the ideas were from http://w2dtc.com/3cx3000f7.htm
Subsequently modified with input from my Elmer, Joe Lang, and N2ZAB.

I'm just the Chimp with the soldering iron, milling machine and TIG welder, following instructions.

In the 1950's I was in the Orlando Electronics Club, and in later decades, returned to RF electronics (industrial).
But time and brain cells fade away, and so I had to pick up and start over with I=E/R again.



Blower is a three phase Rotron Centrimax.
Used a VFD (Variable Frequency Drive) to operate it on single phase power.


Here is the current state of the interior of the cabinet:



The filament choke was about 4uH as delivered.



I stuffed it full of type 43 rods, hoping to use this amp on 160M someday. Close to 30uH now.



The plate has three taps on the secondary, we started with the "2600".
Unfortunately, the other two taps are are "no way" and "ludicrous".

This transformer was all I could obtain at the time, would have been better if there was something in the 3k range.



I will try the next tap later (Hey, 1KW carrier is not enough for you guys???).

It will be necessary to do some minor mods to maintain the critical value of inductance w/o changing the choke any more.
(Can't fit more than 20H in there)

Resonating the chokes might be in the plan if I can't get the numbers right, trying the simple way first.

So the only other easy variable to modify is the bleeder current.
The bleeders operate at about 120 degrees now, at a small amount of bleeder of current.
For the next tap 2600-->4000 (I have not done the calculations), might need over 500 mills, so fans will be needed for the bleeders.
They won't melt until about 700 degrees;)
(It's hot enough in the shack with this thing now....)

The tank is some Collins commercial thing from the 1940's.
(Even older than me)
VERY well made, all on ceramic standoffs, not worried about breakdown.



These are the windings it really likes for 75M


The weak link is the oil caps, rated @ 5kv.
But if I can keep the critical value of inductance in line, there won't be any weird voltage spikes, so that might work
OK @ the next tap on the plate transformer.
 
Thanks for the advice on the glitch fuse, I'll fix it. The gray shape is the fan housing (plastic).

Also need to add the PTT & bias parts of this circuit from w2dtc, so as to cutoff the PA.
http://w2dtc.com/w2dtc-3cx3000f7/2011-0811-schematic-cathode.jpg

I'm looking forward to retiring next year so I can concentrate on even more fun projects!
73
Frank
KJ4OLL

[Opcom]

I have the same filament choke and a 3CX3000 project waiting to be started. I'm glad to learn that adding rods for inductance works so well, and 30uH is good.

It is a GPT-10K choke for a 4CX5000A.
The manual calls it, L915, a 5uH choke.
The operating range is given as 2-28MHz.

In the original equipment, the reactance might have been tuned out by the driver's pi network output capacitor. Here are tidbits learned from the GPT-10K manual.

The 4CX5000A is a normal setup with grid bias and screen voltage and is cathode driven by a PL-172 at 1000W.

The PL-172 stage is designed for a 50 Ohm load and can be used in an emergency (final blown up) at 1KW power.

The driver has a conventional pi network output circuit that goes through a cable of just a few feet (and connectors) right to the 4CX5000A filament.

Reactance: At 2 MHz it is 62.8 ohms. At 1.7Mhz it is 53.4 Ohms -that is very significant when adding it to something that is supposed to have a 50 Ohm input.

Thanks for this proven experiment that the ferrite rods work well on that!