3CX800A7 Performance in an Alpha 89

[W3RSW]

Attached are Excel spreadsheets showing performance of a new pair of 2013 date code 3cx800a7's in W3RSW's Alpha 89 linear amplifier.

One spreadsheet shows outputs vs. inputs if the Pi/L net is tuned and loaded first at 1500 watts output, kept constant there, and inputs reduced.

Next spread sheet shows outputs vs. inputs if the Pi/L net is tuned and loaded for best output at each input lesser than 1500, say you only wanted to optimize output for 1000 watt max.

Most readings of RSW's data taken with two tone input generated by a Elecraft K3.  Two tone peaks were observed for maximum peak sharpness at maximum vertical deflection on a scope.
These data are not super accurate test apparatus originated but the inputs and outputs are
taken with an LP-100A Digital Vector RF Wattmeter.  The K3 can be depended to limit output without ALC, hence no ALC line is connected at any time.

Also attached for fun is a spreadsheet of what an Anan 10 can do when driving the Alpha 89 with the new 3cx800a7's.  

[Pete, WA2CWA]

Also attached for fun is a spreadsheet of what an Anan 10 can do when driving the Alpha 89 with the new 3cx800a7's.  

In this spreadsheet, something is not not clear as to what's driving what. In the upper left corner, it  says Anan 10, but all the rest of the headers and charts say K3.

[W3RSW]

Tnx fer catch Pete, old description headers on same used again Excel formula cells but new Anan driving data. Regardless, the input wattage is the constant.  I just wanted to see if the Anan connections worked so I ran it up and down through the values anyway for future pure signal base line data without feedback.

Same K3 mislabeling in the Anan Excel charts also. Actually only two are relevent, other two left over and should have been removed, but it was a quick "Let's see" job. Btw the cells should be active so you can input stuff too.

I should have put thread in Technical Topics so should copy it there. But left it for now since I'd promised in earlier thread.

I think the interesting thing is the high dB gain at lower input levels. I should add IM3 and IM5 values.  That'll be a fun project for both the K3 and Anan.

[Pete, WA2CWA]

I'm still confused. OK on the two K3 charts in the Anan spreadsheet.
But in the Anan spreadsheet:
http://amfone.net/Amforum/index.php?action=dlattach;topic=42047.0;attach=53207
1st column below; the Anan output is Mfr spec'd from 1 to 10 watts out. Your column shows 11, 13, and 17 watts. Is "K3 set watts" really suppose to be "Anan 10 set watts"?


ANAN 10

K3 set
 watts

3.1
3.9
5.1
5.9
6.9
7.7
8.9
11
13
17


Now leaving tune and load constant


17
14
12
10.1
8.6
6.9
5.1
4.1

[W3RSW]

Revised nomenclature spread sheet
Pete, hopefully I got all matching including the Pset on HPSDR vs. Pactual as shown by the LP-100a for the Anan 10.

I really didn't need to graph input power set vs. LP100a's idea of actual, just playing with charts to show a linear deviation converging at higher inputs, probably just error of small values through real machines.

I'll go back to original post and insert the 'corrected' spread sheet

[W3RSW]

Oh and meant to mention Pete, those are really Anan10 watts.  The original HPSDR Apollo amp was spec'd at up to 20 watts, literally copied for production. I gather Anan reduced to 10 because low output on 6 meters. Many fellows are using 15 watts as set point for SSB , CW service on 160 through 10 meters but no more than 4 or 5 watts carrier for AM.

So especially on AM, further amplification is desirable.  As you know iM products are quite a bit less at 10 watts PEP than 20 too, so that's the next experiment to run if someone hasn't already done it.

[n2bc]

I'm not sure what you are trying to show....  Why adjust the PA tank for 1500W then lower the drive without retuning?
The plate load at 1500W is not the same as the plate load at reduced power - the PA tank needs to be retuned to the reduced plate load... unless you reduce both the PA plate current and the PA plate voltage.    No?
Just curious,  73,  Bill  N2BC

[W3RSW]

Because that's exactly what's going on when you tune and load a Linear Amplifier for 1500 watts then "speak into the tube" with all sorts of lesser input values.  
That's why they call it a linear amplifier.  You set tune and load once at max. desired, then use it. The  reduced input values just simulate in discreet steps what a real varying signal is providing.  Hope that is clear.  

The other spreadsheet shows exactly what you posit,  tuning and loading for maximum at each peak lesser than what the amp is capable of, say tuning and loading for 1000 w pep, or 500 w pep.  You'll note that if you do, the gain for tubes increase.

Say you want to run only 1kw pep output (and the Alpha is quite capable of much more), then you tune and load either two tone or constant carrier to achieve the 1kw pep.  Now each syllable or whatever will have values all over the place on input but the max. output is
still tuned and loaded for a 1 kw peak.  CW of course will have a constant input, no variation except in on/off states.

There may be a difference because of HV power supply drawdown between two tone and
constant, key down carrier in where tune and load are set, but that begins to disappear at outputs lower than what the amp is rated for.  Now we're not talking efficiency here.  That's another story, but gain is interesting at levels less than an amp's capable of.

One of the reasons I included the Anan 10 data is to show what happens if you don't have a driver radio capable of running the Alpha at 1.5 kw output.  So you can see in a real world case what a smaller radio will do.  Another reason for the tables is what, say a 100 watt driving radio, operated at only 20 or 30 watts with much reduced IM can do amplified by a high gain amplifier, such as a loafing Alpha, hence adding little to the IM produced in combo.

Efficiency is of course is not optimum at these lower levels.  Also the pi/L net's not optimum, but look at the data to see that the Alpha has no trouble tuning to these lower levels.  Optimum P/L and input networks or not, doesn't take much input to produce pretty remarkable outputs.

The data is real; some of the points aren't exactly in line a/c errors of some sort, picking an exact nice peak, not flat-topped in two tone testing, etc.
 

[K1JJ]

Hi Rick -

A good way to protect those expensive linear tubes from grid current overdrive is to use a hall effect device in series with the grid current to open a fast LATCHING relay to the RF input.  A "fast" mechanical relay is not fast enuff to protect solid state stuff, but appears plenty fast for even fragile grids like my 4CX-350J.

I did this for my 8877 linear, 4CX-350J linear and 4-1000A AM rig.  Also protected for excessive plate and screen current too.

There are tiny Hall effect devices available from the robotic venders on the web. To sample grid current with a directly grounded grid circuit, just float the -HV supply and use a sampling resistor for grid current. This circuit is available on most linear schematics.

Bottom line is when I get foggy and stupid and tune up my rig like a drunk, a bunch of relays go off and protect me from myself... This is needed as time rolls on - believe me. It's YUGE!  

T

[W3RSW]

Hey thanks Tom. Just got back to reading thread. Neat circuit idea. I'd used GM3SEK's triode control board when I retrofitted my HB 8875's amp.
His circuit uses a Schrack 6 milSec relay tripped by a voltage comparator / flip flop across the grid current measuring meter.  I imagine his tetrode control board uses the same idea for screen protection.