The AM Forum

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 ..... ;D

T

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.

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

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

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

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

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,
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   

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....  ;D

T

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

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

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

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

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

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

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

Thanks Tom,

I'll be doing some tinkering...Jeff

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

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.       

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.

Those are  the 'finals' supposed to be used in the Philips 2KVA audio amp here. It will have to be 4-1000's or something here.

The QB5/1750 should be interesting to hams because it uses a large grid bias unlike the 3-500 or 3-1000 and it also requires only simple fan cooling, no special arrangements. The plate connection is a 9.5mm solid rod for better heat removal. It's a big tube but has a 500W dissipation, built like a tank.

Thanks for these details!

I am not thinking of building one of these big 2-tube amps, but a 1-tube is possible and should be OK for AM. I don't have a steady supply of good 4-1000's.

I found one of those B&W tanks, It ought to be enough. Made for KW DC, right? There is also a 2KVA plate transformer here.

I would like to make something physically like the old Elenco Commander, if I could find the schematic, but improve it with the screen and bias, like in your amp and run AB1. No grid current plus 'low'gain to avoid working VHF.

I would like to see the  11N90 MOSFET screen regulator schematic, I see something's been posted, I need to look at it. I'm interested in the input tuning too, to keep things cleaner.

The goal is to replace the NCL-2000. It is a fine amp but not best for AM.



It's as well to repeat the specs on those 5 blade fans from ebay since you say they are less noisy. Ball bearings means long life too.

Dimensions (As Pictured)   Electrical
Height   4.75"   Voltage   115 VAC
Width   4.75"   Motor RPM   2800
Depth   1.5"   Motor Watts   18
Operating Life   10,000 Hrs.   Motor Amps   .2

Thank you,
Patrick

T

Yes ... good to cross paths again OM.

In the end I was IMD limited by the transceiver (K3) driving the amplifier ... couldn't look much below -38 dB IM3. Started at Q = 1.0 (knowingly too low) and was working my way up to Q levels >3 ... lack of time and the need to get it on the air (for Es season) cut further testing short.  It was rather inconvenient testing as the components in the input pi net were fixed and had to be changed out for subsequent tests. Your mention of the roller inductor in your amplifier peaked my interest ... as I think you have a 'real time' IMD test setup.

As far as the 'chain' effect ... if there's some clean drive to spare one could insert a power attenuator between the xcvr and the amplifier to 'dampen' the effects of the amplifier input match tuning.

Anyway, figured this testing might dovetail with your eternal quest for lower and lower IMD levels. Wish others were as interested in this aspect of transceiver/power amplifiers ... especially on 6 meters! There are some really crappy signals on that band.       

Hi Thomas,
Excellent on the White Barry White, It is now time to build a BIG SBE rig,
Tim WA1HnyLR

Yo Tron!


I always wanted to build an 833A SBE, but yours provides plenty of on-air puking for both of us....  ;D

Can you come up with a good name?   Maybe something that refers to a drunk as he SBE/weaves down the street.


I'm making a cabinet for the White Barry White now. Then the panel lettering and testing. Shud have it on the air soon.  Seez ya then.


T

Use a lab sig gen to provide carrier. A little PM can be added for SBE.

how about a free for all to suggest names..

"Harvey Sidebanger"

"Ethanol Viper" got to keep an alcohol theme??


welcome to the drunken style kung fu of radio.

so how much 'deviation' has to be added for the desired noise reduction while remaining plausibly deniable in the culpability department?  deviants.

Here's some pics of the completed unit. Got the panel lettered, cables tie wrapped and cabinet built and painted.

Total self-contained except for the HV power supply.

Still need to test everything and put the cabinet on.  

I decided to call it "Rico Suave II."      The original Rico was the class E rig. That is torn down, though may be back on the air soon with another ham.

Rico Suave:
http://www.youtube.com/watch?v=jKT-o7fg2Rg


Pics:  I added a C2 loading 160M .002 uf padding cap and relay.  Coverage down to 1800 Khz.

Rico is quite a handsome amp in person - the pics don't do justice. Using the caster wheels, Rico can dance well too.

T

Here's the aluminium cabinet - just painted today.   I'll take a shot with it all together soon.

That's the front and rear view.  I plan to add handles. The cabinet slides off in seconds for easy servicing.  The HV is completely blocked off for safety when the cabinet is in place.


T

Hey Tom,

    Do you do anything to try to cure the paint or just enjoy the scent? I'm asking because I have a pair of racks which will soon be spray painted to match the rest of the stuff in the new shack.

DC

I think that a classie chassie needs a demure, but at the same time, stylish accessory.


klc

Hi Dave,

I just spray it outside on the stone driveway and let the sun bake it for a day or two.  It dries to the touch within 2 minutes, so no dust sticks.   There appears to be little to no outgassing after 2 days in the sun. 

I use ACE Hardware sprays, probably a lacquer based paint.  My cabinet choice is wildflower blue, satin gloss.


What color are you matching up?

T

Thanks Tom. That's roughly what I will be doing. Matching some semi-gloss black.

Today I slid the cabinet on and tuned up Rico Suave II on 160, 75 and 40M to run some initial tests.  Dummy load of course.

I found it ran best at about 400V of regulated screen voltage .  Also, 50 watts of carrier drive gives about 20ma of screen current, 10ma of grid current and about 2000 watts of output.  That's about 16dB of gain, not bad. This is FB, since the grid is drawing little current, even in GG, thus better IMD figures.  At 500 watts carrier, there is no screen or grid current at all.

I have the amp biased to slightly below cut-off to take advantage of the class C efficiency effect.  At 500 watts of carrier there is barely a glow on the plates.

With 4KV, the amp will do an easy 2500 watts pep. So, running it at 400 watts of carrier will give plenty of headroom for X6 audio overhead reserve.

To my dismay, on 40M I found at certain input tuning and output tuning settings it took off - was unstable.  I realized I forgot to add the old GFZ trick and put in a 100pf in series with a 100 ohm 2 watt resistor across one filament pin to ground for each tube. This loads the filament choke for VHF parasitics. That did the job and now the amp is completely stable, at least from 160 - 40M.

I still need to run some 2-tone IMD tests, but because the screen voltage keeps the grid current down, I expect the amp to be at least -35 dB 3rd or better running at this conservative level. After all, ONE tube will do 500 watts carrier at 2500 w pep - with a higher voltage.... ;D

Both the input and tank roller inductors are working well. I can easily match 1:1 on the input. I find that the tank inductor gets luke warm after an old buzzard, but not excessively.  I can set tank C1-L1-C2 to the desired band rather quickly using the turns counters. The presets are written down and efficient.


I need to monitor tube exit temperature to see how the quiet muffin fans make out, but so far it seems to stay quiet and cool at 400 watts carrier out. Driven by the FT-1000D, this is my "fast get on the air- switch bands" rig.

It uses the common 6-step sequencer system as Fabio II. However, I'm using just the HV, ant relay and audio steps for it.  It was easily integrated into the station and fast to select for use.  I really like this amp and hope to use it often.

Look at the pic below and you will see the op position is getting crowded. I am surrounded by white Fabios, Rico Suaves and Dr. Loves.  The making of a good B horror flick..  

T

***   BTW,  Here's the 4-1000A plate modulated by a pair of 4-1000As project:

http://amfone.net/Amforum/index.php?topic=33816.0


And the view of two 4X1's in cascade: