Homebrew 3CX3000 on the 4kv plate tap!

[KJ4OLL]

Hi,
More mods to seek the smoke...

Moved the plate transformer tap off of the 2.6kv, up to the 4kv tap:


Changed the bleeders from 100K ohms to 40K ohms.
Bleeders now operate at 220 degrees, vs 120 before.
B+ rock steady no matter the load!

Spaced out the glitch fuse insulators.
AWG #30 teflon coated wire.


Power went up a little bit:
www.youtube.com/watch?v=sYZS2Dm0iXE

(only have 70 watts drive available from the 32v-2)

Now on to the PTT and other operating circuits!

73
Frank
KJ4OLL

[WD5JKO]

(only have 70 watts drive available from the 32v-2)

     I'd say reduce the drive from the 32V2 to bring the grid current down on the amplifier, and move the loading of the amplifier (less capacitance) till the plate efficiency drops to about 35%. That might give you about 750-1KW RF output. Then switch the 32V to AM, and modulate. Optimize the plate loading of the amplifier for best positive peaks. Better yet, move the operating bias closer to cutoff. This will reduce the gain a DB or two, will raise the efficiency some, and will enhance the modulation. Something like 80% modulation from the driver will result in 100% at the linear amplifier output.

Jim
Wd5JKO

[K1JJ]

Now you're cooking Frank!

4 KV is a comfortable voltage for a 3CX-3000A7.  It could even take more if you had it.

70 watts of carrier drive is more than enough. With ~13dB of amplifier gain, that is about 1400 w carrier and 5600 w pep output.

Being a 3KW dissipation tube in linear operation,  normally we expect a maximum of  1/2 the tube dissipation for the output carrier.  What you are seeing now at 1400 w output carrier makes sense.   Just be sure that you have enough headroom for your audio peaks. Most male voices have more positive peaks than 100%, so if the amplifier shows some flat topping on the scope, just back down the carrier drive until the peaks are clean. You can probably do this by simply loading the 32V3 heavier. (Less output loading capacitance)

Next, run an audio tone through and see how the scope looks - and if you have a spectrum analyzer or SDR receiver, look at the IMD figures.

Last, get on the air with it and axe your buddies to tune around and check the bandwidth.  You will find that loading the amplifier heavily (again, less C2 loading capacitance) until the output power drops about 7%, will give you the cleanest roundest peaks and best spectrum readings.

I hope the amp is located in another room or far away from the mike.  Mucho blower noise is hard to tame, especially on AM if you have any processing / compression.

Very cool project, OM! Congrats!

[Whoops, looks like Jim and I posted a minute apart and said basically the same thing...   ]  BTW, Jim, good idea about increasing the amplifier bias to improve a couple of things. He can even run it close to cutoff or slightly below, when using an AM carrier.  Of course, on ssb it needs to run with some idling current.

T

[Opcom]

Very nice, looks like 65% efficiency, nice indeed. So there is an amp that will never burn out on AM in amateur service.

[KJ4OLL]

Thanks very much for the tuning and loading tips!
It's one thing to build the critter, quite another to learn how to best operate.
That is what I really enjoy about this hobby, always learning more.

There is a 4.5kv tap available on the plate transformer, probably not going there at this time.

I have a 1kw sensor from http://radioassociates.com/, hoping to be able to use it to better visualize the modulation.

There is also this Bird meter and probe to provide input to the Spectrum Analyzer.
But other than spurs and birdies, not sure what I am looking for.



The operating plan is to roll the amp out of the shack when in use.
There is a concrete pad behind.
240vac is already there for the radial arm saw, just need to extend the LMR-600 and RG-58.

Must remember "Don't use the radial arm saw while PTT"

73
Frank
KJ4OLL

[ka1tdq]

That's funny... I can't use my amp while the kitchen oven is on.  House manufacturer used every slot in the breaker box.

Great amp by the way!  That's a real coax melter.

Jon

[K1JJ]

"There is also this Bird meter and probe to provide input to the Spectrum Analyzer.
But other than spurs and birdies, not sure what I am looking for."


Hi Frank,

Here is a helpful thread on this board about testing a rig for IMD and THD. Both plate modulated or linear amplifiers can use the  same procedures, basically:

http://amfone.net/Amforum/index.php?topic=33555.0


My own linear amp testing using the class A driver stages:

http://amfone.net/Amforum/index.php?topic=33505.0


Look at the spectrums generated by a single audio tone (for  AM tests) or  two tone (for ssb tests) and you will get the idea. Basically, you want as little crud as possible extending outwards after the main tones.  We can tell how linear the amplifier is by how far down in dB the unwanted mixing signals are.

Building an amplifier is quite a challenge, but testing it, troubleshooting it and making it so clean that you blend into the background with smaller amplifiers on the air is what counts.  You have a great start using a tube that is as clean as they get. Now it's up to you to do the rest.

While looking at your spectrum shots, play with the various input and output levels, biasing on both the linear amplifier and driver or modulators, loading on the driver and amplifier, etc. Tank Q that is higher than the common 12 figure usually shows better IMD figures at the expense of efficiency.  You will find the OVERALL sweet spot combinations to making the amp cleaner, sometimes by as much as 5 to 8 dB 3rd order IMD better, depending on where you started...  

I personally struggled for years trying to make my various homebrew amplifiers cleaner, but usually failed to get them operating pristine. Some were downright nasty when driven hard.  Not until I started using tubes designed for linear service, paid attention to the driver lineup I was using and got a spectrum analyzer to run REAL tests, did I finally build amplifers I could be proud of using. You have a great head start there with a FB amplifier, OM..  Your limitation is now the 32V2 driver. It is probably 10 dB less clean than the 3CX-3000A7, but that is another subject. That is your present bottleneck and will probably limit your 3rd IMD numbers to -30dB 3rd or so. But that is acceptable. A signal is only as clean as the highest IMD level in the overall chain. (actually slightly worse)  They all add to the degradation.

But remember the rule: If you are 6 dB louder than the average QRO amplifier on the air, then you better be 6dB cleaner than the average amplifier on the air - or expect to be noticed - and not in a positive light.

Here is another thread of my trials and tribulations finally settling on a 3CX-350J driver for my main linear. The "J" version tube is the cleanest on the market and expensive. It is made to precise parameters. It will do -55dB 3rd IMD at 50 watts output in class A.  Fair Radio has a bunch.  Before this I use a class A one watt lab amp pre-driver rated at -75 dB 3rd IMD.

http://amfone.net/Amforum/index.php?topic=33505.0



In the end, you want to have the confidence that you can plop down anywhere on the band, with reasonable spacing for the mode used, and not be noticed by your band neighbors.  Once you develop an ear for amplifier cleanliness, you could tune the band at random and easily tell whose amp is generating crud and who is clean. It's very easy to see.  It funny how often the dirty signals on the band have operators who are defensive and usually deny that they have a problem. While the guys with clean signals will immediately look into a problem if you suggest it.  Be in the latter camp and life on the air becomes easy when QRO...

T

[KA2DZT]

Tom,

You state that the rig will only be as clean as the lowest IMD in the chain.  Should it be the highest IMD or am I not understanding something.

Fred

[K1JJ]

Tom,

You state that the rig will only be as clean as the lowest IMD in the chain.  Should it be the highest IMD or am I not understanding something.

Fred

Hi Fred,

Yep, you are correct -  I  stated it backwards because the numbers are negative. It can be cornfusing. ...but to be clear, the highest numerical number in dB is the poorest performer in a chain. Low IMD (low distortion) means a low negative number.  I corrected it above, thanks.

For example, if the pre-driver IMD is -40db 3rd, (good)  the driver is -20db 3rd (very poor) and final amplifier is -45dB third, (very good) then the overall chain will be about -19dB 3rd (very poor).  It can never be better than -20dB because the poorest stage distortion will always limit it. Because the other two stages are not perfect, the overall results will be somewhat worse than -20 dB 3rd.   IE, the chain can never be better than its poorest performing stage.

That's why using a poor driver and clean final in a linear amplifier chain is a waste.  Tubes like a 3CX-3000A7 and YC-156 will do -40 to -45 dB 3rd if set up correctly, but most of the riceboxes today are lucky to be -30dB 3rd unless they are among the few that have a class A feature.

The ace in the hole is the new SDR rigs that use pre-distortion. We can then take a dirty linear amp and use this software to clean it up using a sample of the linear output. I wish this was available before I spent a year of my life building up the latest conventional class A/ AB linear system I now use... [sigh]

BTW, all of this IMD stuff also applies to audio and modulators, since they are linear amplifiers too. RF is nothing special, except perhaps that audio negative feedback may be easier to implement around a series of audio stages - easier compared to when RF amplifiers are physically spread out. It is best to start with any series of stages as clean as possible before adding negative feedback.  The cool thing about grounded grid RF linears is the built-in negative feedback, though the driver usually is outboard and almost impossible to make a loop of NFB stable unless on board.  (enter pre-distortion)

T

[KJ4OLL]

Hi,
Tom, thanks very much for the links to the IMD testing threads!
Particularly intrigued with the design changes you made, and subsequent effects on IMD.

Have been trying to find a way to test the linearity of the 3CX3000A7 homebrew, as a first step.
Had some test equipment, tried some different configurations, but no luck.

Imperative to avoid this situation:



Thus, obtained some additional test fixtures from http://www.cleanrf.com/products.html, specifically "Splatter View" (Perfect product name)

Set it up with the 32V-2/Alpha to gain experience, now on to connect the homebrew in place of the Alpha and see how it looks.

https://www.youtube.com/watch?v=wpCmnjTogz0

Finally, I hope to be able to figure out a way to connect a sample output to the Spectrum Analyzer, w/o blowing the front end out of it, so as to see the IMD products.

I'm guessing the SA configuration for the amp output should be something similar to the way I test Collins R-390A filters?




Frank

[Opcom]

To revisit this, how is this amplifier performing today? Has the 4500V tap been tried?

[KJ4OLL]

Hi,
I am using the 4kv tap, because it provides adequate energy to produce a 1500 watt carrier.
Also, the "Critical Value of Inductance" numbers work for the 4kv tap, would not @ 4.5kv.

Currently working on PTT and cutoff circuitry.
73
Frank
KJ4OLL

[w1vtp]

Thanks very much for the tuning and loading tips!
It's one thing to build the critter, quite another to learn how to best operate.
That is what I really enjoy about this hobby, always learning more.

There is a 4.5kv tap available on the plate transformer, probably not going there at this time.

I have a 1kw sensor from http://radioassociates.com/, hoping to be able to use it to better visualize the modulation.

There is also this Bird meter and probe to provide input to the Spectrum Analyzer.
But other than spurs and birdies, not sure what I am looking for.
<snip
73
Frank
KJ4OLL

I have that same sampler - it's a very important part of my station monitoring equipment

Al

[wb3eii]

Please keep posting updates and pictures, this is the crux of the hobby.
 (Personal disclaimer, I have a stalled 3cx3000a7 project, need some ambition and encouragement.)
 73, Ed

[KJ4OLL]

Hi Ed,
Will do, I want to help in any way possible!
Fully aware of the motivation issue.

When N2ZAB was kind enough to critique the design, (After I thought I was almost done..)
I did not feel like working on it for months after, there was so much to un-do and start again.

But eventually the desire returned to finish it enough to put a signal through and see what happened.
Seeing the power meter move into the WOW range for the first time was worth all the re-working.

In my case, all the metalworking was the easy part, have been mangling metal for all kinds of hobbies for
many decades.
Amateur Radio is a new hobby for me, and so the electronics is my weak point.
The challenge of building an amp from scratch was balanced by me wanting to learn the technology.

(Even though I forget it all immediately, but I make lots of notes, which I know are around here somewhere,
if I could only find my glasses...)

Also, being 65 means my ancient, rusty brain is barely able to comprehend anything complex.

But it is all so much fun, I will just keep going as long as I can!
73
Frank
KJ4OLL

[wb3eii]

 Frank, Thanks for the reply.
 I am especially interested in the ptt and idle bias circuits, I have nothing in this area.
 I am converting a Henry 2000D rf generator, it was originally used at 27.125 mhz in a plasma torch for a spectrum analyzer.
 Power supply is huge, cabinet has plenty of room for tank coils and such. I just need to get it back to work bench and re-work some of the previous mods I had made.
 I had the grand idea that I was making an all band linear, 10 through 160.
 I found huge rotary switches for the Pi-L net, equally large vac caps, and some roller inductors that would not look out of place in a battle ship.
 All good components for a world beater amplifier.
 I find my operating style suits 80 meter fone, so the amp will probably get rebuilt to a single bander, maybe 40 will be added later.
 Hope to finish this some day. 73 Ed.