Why some people's amplifiers don't work and get hacked up.

[Opcom]

The topic is a little misleading. It may have worked, but not very well.
Pictures of the specimen in its 4 FT rack are here:
http://amfone.net/Amforum/index.php?topic=36305.msg278979#msg278979

So, why is this, or was this, a bad amp?

Here are its main parts, all pretty decent:
B&W FC-30 30-amp filament choke
Tank coil B&W 850A (disintegrated plastic forms)
Loading capacitor: 5-section "BC RX spacing", 1400-1825pF max.
Tuning Capacitor: 320pF EFJ 154-38-2 320E20 (same as Johnson Thunderbolt, a 2200V amp)
Plate transformer 120V 0.2 Ohms DCR primary, 6400VCT 372 Ohms DCR CT secondary, Z=317 Ohms
Filter Capacitors: 2x) 3.5uF oil
Bleeder Resistor: 40K 200W
Filter Choke 7.5H 33 Ohms DCR

The transformer there, PSUD says that with 6400V can only make 3KV at a light load. Trying to push 600-700mA pretty much ruins that, and you get 2600V and probably way more grid current than is healthy. PSUD seems to be pretty accurate. The only thing it doesn't tell is where the transformer is going to saturate.

At 2600V with the given plate current, the control grid current on the 4-1000 would be higher than the 170mA in the Eimac Engineering Newsletter FCJ-59D2: "4-1000A Typical grounded-grid triode", which probably should be taken as a max. This is in the 3KV column with a 1475W output and 130W drive. The value of 160-170mA is also given by Orr in his Eimac Amateur newsletter AS-10: "The 4-1000A in Grounded grid", describing a 900W DC input input amp running at 3KV. That is a lot of current on a 25W grid. The 4000V data in the newsletter gives a 105W drive for 1870W output and 150mA control grid current, and the 5000V column is even better needing just 70W, 115mA, and and no worry about losses in the input tuned circuits.

For AM, the whole setup might have been better with a 3-1000Z or a pair of 3-500Z's, and a B&W 852 tank. Those tubes perform better at lower voltages. It might be OK on SSB, but contrary to the space taken up, it's no monster-amp, and can't really make good use of the 4-1000A. The sad part is that even with a pair of 3-500Zs, it is 5 times the size of an SB-220 for about the same oomph.

The transformer looks like it is good for about 1KW DC output at 10% off the no-load voltage, and 2KW DC at 20% off the no-load voltage. At 2600V the 4-1000 has a hard time making a lot of output.

So the place where it all goes wrong is the match-up of the power supply with the RF deck.

Maybe it's the biggest iron the original builder could muster, but a lot of the disheveled or badly made home-built stuff I end up with, is stuff that has been sitting in garages, not working for 20 years, with what seems like the proverbial 20 lb whistle on a 10 lb boiler. This is the most common mess-up I ever see.

If the transformer could be used with a bridge, then the amp might behave like the legendary Elenco Commander, a GG 4-1000A SSB amp with 4KW PEP input, until something on the RF deck shorted, So that's out.

The center tap terminal on the transformer has the same huge insulator as the start and finish terminals.
http://amfone.net/Amforum/index.php?action=dlattach;topic=36305.0;attach=42626;image
Maybe it is one of those better ones. Just thinking about things.  No loss, parts is parts.

[KD6VXI]

In GG,  you have the screen and grid taking up grid current.

Eimac and Orr recommend using something to evenly split the grid current between screen and control grid.   I've never seen it utilized in that regard.

A triode probably would have worked better.   But,  the terraces where and still are available as pulls for a hundred bucks or less.   The triode is a lot more.

--Shane
KD6VXI

[Opcom]

Good point on tube cost! And the division of drive between control and screen grids.

Schematics have floated around using NI resistors, a tap on a toroidal transformer or tank, connecting the grid to the cathode and driving those as one against the screen, and the 'correct' method according to Eimac which retains a normal tetrode setup complete with grid bias and screen voltage, except that the drive is applied to the cathode.
I think one example is the Special Design Products SDP-1000L, a KW input amp designed for MARS. It is very similar to the Eldico SSB-1000 except the Eldico has the cathode grounded and drive is to the grid. So, both are very close schematically having a bias supply and a screen supply, with the difference being what is grounded and where the RF is put in.

Here is one set of ideas applied on 4CX250's. Just tie the control grid to the cathode..
http://home.comcast.net/~k7rld/4cx250bamplifier.htm
http://www.hellocq.net/forum/read.php?tid=256399&page=1
The above are 4x250 type amps, the point is to reduce the control grid dissipation in GG service.

I don't know what really works on a 4-1000 to divide up the currents properly. The amp of this post had no input network at all. In AB1 or AB2, the 4-1000 will work in a conventional circuit at low voltages like 2500-2000 but is not so efficient due to the G2 needing to be 500V or more to get the current up, and when driven hard the G2 current is high also.

If 4-1000's were cheap, OK that's fine. Just wondering what to do with the amp now. My few 3-1000Z's are not going to be used in that. No 3-500's lying around, and who wants just another 1200-1500W amp, if it is huge and 4 FT tall?
Fix the tank, for sure. Maybe put the power supply back together and add a 1000V 500-700mA rectifier power supply in series wit the original CT. Get the volts up to 3.6-4kV there and make it sing.

[KA0HCP]

In GG,  you have the screen and grid taking up grid current.

Eimac and Orr recommend using something to evenly split the grid current between screen and control grid.   I've never seen it utilized in that regard.

A triode probably would have worked better.   But,  the terraces where and still are available as pulls for a hundred bucks or less.   The triode is a lot more.

--Shane
KD6VXI

If you are referring to the 'Super Cathode' method of floating the control grid with a capacitor divider see what Tom W8JI says about his history with Orr on this.  Pages 23-27.
http://rfcec.com/RFCEC/Chapter-3%20-%20Fundamentals%20of%20RF%20Communication-Electronics/04%20-%20AMPLIFIERS%20-%20RF%20POWER%20AMPLIFIER%20BASICS/RFPA%20-%20High%20Frequency%20RF%20Power%20Amplifier%20Stability%20at%20VHF%20%28By%20Charles%20T.%20Rauch,%20Jr.%20W8JI%29.pdf

[W3RSW]

Very informative article particularly the difference between triodes and tetrodes when it comes to proper methods of grid grounding in a cathode driven amplifier.

With respect to Orr:

There comes a time
 When we discover
Our parents' feet of clay,
 As will ours become,

       Someday.

[KD6VXI]

The Orr handbook I have shows it with a tapped coil.

Interesting,  Tom is using Charles now.   Too much bullshit between him and measures brought them both down.

As such,  I read what both of them say,  but don't take either as gospel.

I would like to see WHAT Eimac says is the preferred method...

--Shane
KD6VXI

[KA0HCP]

That power point lecture is ages old.  I don't think Tom has changed his name because of Rich M. 

[Opcom]

I won't want to be the judge of topologies with those people. Attached is Eimac's word on it. That's what I would go on first.

There is an Eimac article about using Tetrodes in GG service. Rather than comment on it, there it is.

The 4-1000 in GG was also discussed here:
http://amfone.net/Amforum/index.php?topic=18634.0
and built.. does not run on 2600V.

[WD5JKO]

From that PDF linked written by Charlie:

"The notion
a healthy system can go along
for hours, weeks, or years and su
ddenly break into an
uncontrolled oscillation that dam
ages components is highly unli
kely unless a major component
significantly change
s characteristics"

Some amplifiers using heater cathode tubes in G-G like the 8877 can have this problem. I converted a MLA-2500 to use a single 8877 instead of a pair of 8875's. Dentron used a 50K resistor in the cathode return to bias the tube off during standby operation. Then the fatal combination of that 50K, and the tube filament being grounded on one side was put in place. Alpha also grounded the 8877 filament on the 76..

So I lost a pair of 8877's the same way. In both cases there was a spark festival in the amplifier while the amp was in standby. In the 2nd instance I lost the cathode current meter. In both instances the fuse blew before I could react and turn off the amplifier. Both tubes had an open filament afterward. These are all facts.

Here is where FACTS become speculation. I think what happened was that there was a heater / cathode short while the AMP was in standby. Since the filament was grounded (one side), the cutoff bias was lost, and the tube had cathode current. Since the T/R relay was in the receive position, the 8877 input and output was not terminated with proper impedances. A parasitic followed leading to the carnage.

Both tubes could be brought back to life, and I used them for several years before making amplifier changes, and getting a 3rd tube. I found that filament continuity could be momentarily brought back with a mild hammer tap at the right spot.   This lead me to a special fixture which consisted of a socket, the dead tube, and a 100,000 MFD capacitor across the filament pins. The capacitor was charged to 12 volts. Then the appropriate hammer blow onto the tube at the right angle would spotweld the filament wire inside the tube to the pin. Afterward, bringing the amplifier up on a variac, the tube worked normally. I might get 5-10 uses of that tube before having to repeat the sequence.

Imagine the Ebay seller, "this 8877 put out full power the last time I used it". The tube would be DOA and the seller told the truth.

Jim
WD5JKO

[KA0HCP]

...I converted a MLA-2500....

Operative words...

Tom's discussions are about well known factory designs that operate for years/decades without problem. 

Rich's failing is his magical thinking, unsubstantiated by hard, cold facts and methodical engineering measurements and calculations.

[KA0HCP]

...I converted a MLA-2500....

Operative words...

Tom's discussions are about well known factory designs, produced in the thousands, that operate for years/decades without problem.  

Rich's failing is his magical thinking, unsubstantiated by hard, cold facts and methodical engineering measurements and calculations.  He refuses to acknowledge that component aging, improper operation, ceramic switches contaminated with sloppy application of solvents, contacts burned by hot switching are possible explanations for failures or signs found 'after the fact'.

Rich has never presented an engineering analysis of a single amplifier design showing that the design is fatally flawed.  Nor offered measurements of an amplifier in stock condition with components in 'like new' condition showing that fatal oscillations can be predicted and demonstrated at will under normal operating conditions.