The New AB1 GG 4-1000A Linear Amp - Regulated Screen and Grid

[K1JJ]

Well, after some work, here's the all-new  4-1000A grounded grid linear amp.   Just finishing up the wiring, though it needs cable dressing, cleaning up and initial testing.

Featured is an electronically regulated screen supply using a pair of 11N90 MOSFETS. Also using zener diodes in the cathode for grid bias.  The 500v regulated screen voltage will cause the screen to draw a lot more current than when at DC ground and provide better IMD linearity. I figure 13dB of gain or so, so my 200w  pep  FT-1000D should make a loafing driver.  I'm hoping the grid will draw zero grid current at about 1.5 KW pep output and be at its cleanest. I will be using about 4KV.   We will see.

It will be able to switch to class C for better AM efficiency using the zener string.  (See older thread about this)   It will run AB1 too, of course.

I put in a pi-network for the INPUT tuning using a roller inductor. Look for it mounted underneath the RF tank. No homo fixed coils and caps for Barry. This variable input will give 1:1 on all bands.

In the plate tank, vacuum variables are used and I'm going to give that copper strap roller inductor a try.  It should cover 160 to 10M, though I plan only on 160- 40M for operation.

This was designed to be a silent amp - #1 priority.  I'm using a pair of super-quiet muffin fans mounted directly below the tubes sockets. They are mounted on foam and put a reasonably good air flow thru - and are quiet. I'm hoping between this breeze and radiation cooling to have a silent amp for AM. (or ssb)   I figure an easy 500 watt carrier with little stress on the tubes when running near silent.

#2 priority - easy to work on.  Broadcash style layout where most everything is accessible.

The filament choke consists of two #6 wires thru a string of ferrite cores. I measured the inductance to be suitable down to 160M. I use this technique in all my linear amps.

I'm thinking of naming it 'The White Barry White'   in honor of Barry.      This thing certainly has a lot of soul .....

T

[K1JJ]

Input tuning

Regulated screen supply

Back view - notice the 50 watt zener string on the heat sink at the back.
The filament Variac is a one-set at 7.5V and leave.

[K1JJ]

4" viewing window

Look closely to the right-background and you will see the 4X1 final in Fabio, the plate modulated AM rig.  

There are now six 4X1's in service around here. Just love these tubes.


T

[w8khk]

Beautiful construction!  The homebrew plate choke looks strapping.  A pair of 4-1000As should loaf when you turn up the wick!

[K1JJ]

Beautiful construction!  The homebrew plate choke looks strapping.  A pair of 4-1000As should loaf when you turn up the wick!

Thanks Rick.

You need to post some of your stuff you sent me. Talk about FB work!

BTW, I want to thank Rick publicly for his generosity.  I was looking for a roller inductor for my input tuning and was more than willing to pay for it.  He would not accept any $ and just sent it out - including shipping.  What a great example of hams helping hams.

The roller has worked out FB so far, Rick.

T

[w8khk]

Beautiful construction!  The homebrew plate choke looks strapping.  A pair of 4-1000As should loaf when you turn up the wick!

Thanks Rick.

You need to post some of your stuff you sent me. Talk about FB work!

BTW, I want to thank Rick publicly for his generosity.  I was looking for a roller inductor for my input tuning and was more than willing to pay for it.  He would not accept any $ and just sent it out - including shipping.  What a great example of hams helping hams.

The roller has worked out FB so far, Rick.

T

Pursuant to Tom's encouragement, I am posting some of my recent homebrew activities here:

http://amfone.net/Amforum/index.php?topic=34544.0

[KA2DZT]

Tom,

When did you build this rig??  I thought we were still working on Fabio II.

I think you've been pulling our legs.  How many workers do you have there getting all this stuff built in so little time??

I don't get it.  I fall asleep for an hour and you got another rig built when I wake up.

I've got to step up my game.

Fred

[KL7OF]

Nice Tom!  .....I like the plate choke..The braid on the plate caps speaks hi voltage....I will be interested to see how the muffin fans play....The tuned input is great.....Regulated screen voltage is cool as well.....Maybe this winter I'll get to hear some of your rigs....

[K1JJ]

Nice Tom!  .....I like the plate choke..The braid on the plate caps speaks hi voltage....I will be interested to see how the muffin fans play....The tuned input is great.....Regulated screen voltage is cool as well.....Maybe this winter I'll get to hear some of your rigs....

Thanks for the comments, guys.

Steve: Yeah, the only reason for the braid is for mechanical stress release on the plate caps.  As you know, a solid mechanical connection can actually break the seal or glass when expansion occurs.  I set it up so the cap slides on naturally aligned with no side force.

Fred: It's ben so darn hot that I suspended testing on Fabio a month ago. He's working pretty well. I just need to ramp up to the next level and break some parts.  So in the meantime I wanted to build a long-time goal project ... a silent linear capable of a clean 1.5KW out. I have several linears that will do 1.5KW now, but they are all noisy with squirrel cage fans. I want to run audio processing on AM, which requires a silent room.    Fabio is reasonably quiet on AM, but not dead silent.  The White Barry White should be a silent toilet clogger.

I'm finished with new projects for now, so don't expect any surprises.. No new plans on the board except to perfect what I have and get on the air when it cools down.

T

[ka1bwo]

Fred,
Don't let Tom Vu fool you! This stuff is being built at Smokey's CB shop in Danielson, CT they are on Tom's payroll.

Tom,

When did you build this rig??  I thought we were still working on Fabio II.

I think you've been pulling our legs.  How many workers do you have there getting all this stuff built in so little time??

I don't get it.  I fall asleep for an hour and you got another rig built when I wake up.

I've got to step up my game.

Fred

[ka1bwo]

Tom,
You really do excellent work and a lot of it. The best part is you can actually hear you on the air with your home brew equipment and the on going process of tweeking things in. I remember the night of the maiden voyage of the four tube linear amp using 3-1000s or were they 3-500s.
Joe   

[K1JJ]

Problem corrected:

I used Chinese sockets that lowered the tubes by about 1/2".   Not good.   By the time they were mounted for air clearance, the tube sockets were 1" below the chassis.  This caused the air gap between the chimney and tube to be too big at the top. This created a poor air flow (low pressure) at the top seal - running too hot.

(Look at old picture #3 back in my original post to see the huge gap at the top between the tube and chimney.)

I remounted the tubes ABOVE the chassis and now there is a uniform chimney air gap all the way to the top of the tubes.  This lets the exit air blow on the top plate cap and seal.  The chimney needs to HUG the contour of the tube, especially at the top outlet area.

Now when the filaments are on, I can hold my fingers on the plate caps. Before, it got too hot.

The Chinese socket has a design defect in mounting the tube too low. However, by mounting it 1/2" above the chassis, more air flows by the socket, thru the lower pin holes and past the top seal.

(This tube sub-chassis has been used for many 4X1 projects in the past.)

Pic below is the new mounted version and air flow is much better.

Joe: Thanks for the comments.  Yep, ya can't go wrong at Smokey's CB shop....  

T

[Jeff W9GY]

Can you post a schematic of the GG linear arrangement ... I'm particularly interested in the screen hook up (bypassing, etc) , and if you remove the screen voltage in standby , or just maybe apply a negative voltage to the grid and leave the screen supply up? Thanks, Jeff

[K1JJ]

Can you post a schematic of the GG linear arrangement ... I'm particularly interested in the screen hook up (bypassing, etc) , and if you remove the screen voltage in standby , or just maybe apply a negative voltage to the grid and leave the screen supply up? Thanks, Jeff

Hi Jeff,

I haven't drawn up a schematic yet, but it should be easy to describe.

The screens and grids are both bypassed to chassis at the socket pins using door-knob style  5KV at .001 caps. I use two at each screen and two at each grid. 

I don't have any DC bias on the grid, just a connection to the meter which then goes to ground. This monitors DC grid current.  I obtain grid bias by using three 50 watt zeners in the cathode circuit. They are 24V, 33V and 90V.  All three in series gives class C and  just the 24 and 33 V in series gives class AB1.

The screen is fed with the regulated 500V DC thru a meter.

I key the screen AC primary supply as well as the HV pri.  There is a provision to key the  cathode fil CT if I choose, but may not need to.

The cathode RF is fed thru a 0.01 cap and a pi-network with roller inductor.


Hope this helps.

T

[Opcom]

Could you show more of the cooling arrangement underneath? The idea of a quiet 4-1000 while not violating Eimac's cooling suggestions is really cool.

Can you reveal the fans'make and model?

Can you say the inductance of the filament choke and type of ferrites?

Is it right to say that GG here means cathode drive, grounded control grid, and a RF-grounded screen with its supply floating between cathode and screen, or ground and screen?

[K1JJ]

Could you show more of the cooling arrangement underneath? The idea of a quiet 4-1000 while not violating Eimac's cooling suggestions is really cool.

Can you reveal the fans'make and model?

Can you say the inductance of the filament choke and type of ferrites?

Is it right to say that GG here means cathode drive, grounded control grid, and a RF-grounded screen with its supply floating between cathode and screen, or ground and screen?

Hi Op -

Here's the eBay link for the muffin fans:

http://www.ebay.com/itm/New-120mm-38mm-Case-Fan-110V-115V-120V-AC-110CFM-2-Pin-Ball-Bearing-Muffin-PC-4-/271226288637?ssPageName=ADME:L:OC:US:3160

At full speed they put out the required cooling with a slight noise. However, I plan to run the tubes at about 1/2 power so I can lower the fan speed.  This is with no cabinet, out in the open.   Once moved away and with a cabinet, shud be pretty quiet.

The two fans sit on the main aluminium plate over a pair of 4.5" holes to suck air in. The are mounted on foam weather stripping so there is no mechanical vibration.  Each sits directly under its respective tube socket. They run off a 120V AC Variac.  Pretty simple, really.


For the fil choke I am using type 43 ferrite cores, the same ones used with the class E rigs. I've never measured the choke inductance in mH, but using the MFJ-259B I find it is >650 ohms on 75M and about 400 ohms on 160M.  Being a 50 ohm circuit (two 4X1's in parallel) this shud be FB.    I'm using 7 cores on the outside of the chassis plus one big one - and two underneath the chassis at the tube pins.   I wanted the majority of cores outside to keep the sub chassis clear for good air flow.  I may end up shielding the outside cores with braid since it can produce 0 degrees feedback if it sees the plate.  I doubt there will be any instability problems, however.

Yes, both the screen and grid are at zero RF potential, grounded.    The screen supply is fed as normal - the plus to the tube pin and the negative to chassis ground.  The cathode DC goes thru the choke, thru the fil xfmr CT, thru the meter and finally to the negative of the HV supply.

From my experience using 4X1's as tetrode modulators with screen voltage, I've found they run very cleanly until we start to haul down grid current.  In GG, the 500V of screen voltage (90ma screen current on peaks) shud keep the grid current near zero under moderate power conditions.  Actually, up to 1000V screen voltage is permitted, but I think 500V is optimum from tests I've run.


Are you thinking of building one?   I can post the simple 11N90 MOSFET screen regulator schematic if you are interested.

T

[Jeff W9GY]

Yes, Tom, I'm interested in seeing the screen reg. schematic.  thanks Jeff

[Detroit47]

Tom I like your construction with one exception. In my experience with hi power dielectric presses and welders. We got away from using braided strap for any hi power RF. The reason being as the braid acts like a hundred little instructors in parallel. I have seen the stuff get red hot when 5 kw is put across it. We went over to using flat copper strap. The strap is still flexible without the inductance. Just my two cents. It probably wont matter at the power levels you are running. But I just wanted to share my experience. 

John N8QPC

[pa0ast]

Just another lin amp for fun. 2 x QB 5/1750 ,  200 watts  ( for fillament and drive power. )  Va = 5kV  Ia = 1A peak

[K1JJ]

Tom I like your construction with one exception. In my experience with hi power dielectric presses and welders. We got away from using braided strap for any hi power RF. The reason being as the braid acts like a hundred little instructors in parallel. I have seen the stuff get red hot when 5 kw is put across it. We went over to using flat copper strap. The strap is still flexible without the inductance. Just my two cents. It probably wont matter at the power levels you are running. But I just wanted to share my experience.  

John N8QPC

Interesting, John.

I never liked the look of braided strap anyway. I'll replace it with some copper strap.  As long as the strap has a gradual bend to it and is custom pre-formed, it will not put a meaningful stress on the plate cap.

In defense of the braid ...  there is not a lot of current at the plate cap, just the 500ma pulses that feed the tank circuit. The huge circulating current (20A or whatever) is in the tank which uses copper strap.  In fact, if you look at it, in series with the braid is the parasitic suppressor which uses thin wire, even nichrome, to add losses on purpose.  But again, I hate the looks of braid on the plate caps and will dump it...  Just too JS looking.

PA0AST:  Very nice looking amp!  Neat layout and construction.  Thanks for posting it.

Jeff: I think I posted that regulator schematic somewhere, maybe in the Fabio thread. Let me look for it...

T

[K1JJ]

Yes, Tom, I'm interested in seeing the screen reg. schematic.  thanks Jeff

http://www.tubecad.com/2007/06/blog0109.htm

Jeff,

Here's the link to the shunt regulator. It's about 1/2 way down the page and uses TWO FETS. (Just above circuit 2)   I used the 11N90s rated at 900V each for about $3 each.   I ramped it up to 500VDC with no problems.  I have two units in service.  Both worked right away.

T

[Jeff W9GY]

Thanks Tom,

I'll be doing some tinkering...Jeff

[K1JJ]

I couldn't stand seeing those braids anymore, so replaced them with the copper plate cap straps.  I feel better...


I must admit they look much better.


Also checked out the tank circuit on all bands.  I injected the MFJ-259B into the 50 ohm output and put an 8K across the tube plate caps to ground to simulate a plate load.  Tuned FB.

Also checked out the pi-net input circuit and was able to get a 1:1 on all bands using a 100 ohm resistor load from cathode to ground.   Now it's just a matter of interfacing it to the station, adding HV and giving it a go into the dummy load.

I ran out of HV wire, so got some on order.

T

[W1VD]

T

If you're up for some experimentation ... it might be interesting to check IMD vs. input network Q. The 'rule of thumb' is Q=1.5 is sufficient to provide 'acceptable' IMD levels. In a recent project here (SB-220 converted for 6 meters) higher input Q values seem to offer some improvement. With your pi net roller inductor and variable Cs you can adjust for a wide range of Q. Seems like you have the perfect test bed and measurement setup to have a look.       

[K1JJ]

Hi Jay!

Nice to hear from you.

Cool on the 6M SB-220 experiments.   How much improvement did you see and what do you estimate the Q you could obtain?



Good idea on the Hi-Q for the input.  I usually try to set the coil for as little inductance as possible and still achieve a decent match. This will give max Q, of course. But sometimes there is a limit when a 1:1 match needs a certain min amount of L or there is a poor match.  It's a matter of degree to how far off the match can be and still get the highest Q using minimum L.   I'm sure you know what I mean.


Back when I was optimizing my multi-chain class A linear system for best IMD, I found by playing around with the input networks I could get the IMD peaks to reduce. It sometimes didn't correlate to a voltage tuning peak in the network - it just seemed that the previous stage liked a slightly reactive load to work into for best IMD.

This was the difference between getting -45db 3rd or -55db 3rd IMD when everything was tweaked.  It did require a custom tune-up for each freq, but the results were outstanding for a chain not using NFB of any kind.

So what you are saying is well worth while pursuing.

T

Link to Rico Suave, a pair of 4X1s in GG linear for 160-20M using regulated screen and grid voltages.

http://amfone.net/Amforum/index.php?topic=34543.0

Below:  A pic of the fil choke. To cover 160M I used 10 cores outside and  10 underneath the chassis.