The AM Forum

"Hollywood" is finally ready for his first grand appearance.  After the photographers were finished we put him back up on the bench.  So far all his AC, PTT, RF input and outputs look good.


What is missing in the pictures is the twin muffin fans that mount above the four 813s to supply a quiet breeze. You are looking at both the modulators and RF finals - with the mod iron mounted in a different cabinet.  The idea was to make everything viewable, easy to work on and easy to carry.  The RF input is broadband toroidal without tuning with a shield blocking it from the output.

I am building a new HV supply (old BC transformer) that will furnish 2100 and 2800VDC at 625 mA from taps and also be controlled by a big Variac.  The mod transformer weighs 59 pounds and may be a 500 watter from RCA BC vintage. I have about 30H of mod reactor inductance and a 30H screen choke.

I am building a second GFZ MOSFET audio driver specifically for this project. Heavy audio negative feedback is planned. The only audio transformer will be the mod transformer. That is real glass on the viewing window. Notice the silver plating on the RF tank areas?  Hint: purely cosmetics.

The missing screws on the bottom right panel support the internal aluminum shield to the RF input broadband network.

Will the layout make it to 10M?  We will have to see... :-)  I usually use my linear amps up there anyway.

Hollywood's basic metalwork was recycled from an old 8877 linear I built in 2010, so the panel layout and symmetry is not exactly what I would have done using a blank chassis and panel.  But it worked out in the end, just like a Frankenstein movie.  98% of the parts came from the cellar dungeon.

Hollywood has a blue metal cabinet, though I am considering making a viewable Plexiglas enclosure. It will ride on a wheeled platform to save my back. Pretty compact as are all my rigs. I simply run out of space is the real truth...


T

Pics to follow:

I still need five working 813s to start real dummy load testing... anyone have some good used 813s?

Yaz IV is now 3 years old and has been very patient watching Hollywood progress.

Not even finished and I'm already thinking about a 4D32 rig...

In the third pic below, the two white insulators at both ends of the panel support the two missing muffin fans above the tubes on a horizontal rail.  The brown cord is the 120 VAC feed.

That's two 450K power resistors for the audio NFB -  and an extra one to bleed the HV PS in case the HV bleeder fails in the supply.  WTF.


Easy to get at any part in case of failure. I've always liked broadcash building style, but still use a sub-chassis whenever I need a blower for a pressurized tube in my other linear amps.


I have a choice of running these 813s with no air or turning on the switch. I direct the air down across the four tubes and it bounces off the front panel and chassis floor, creating a 360 degree cooling pattern. With a Variac on the muffins, it is very quiet.

I'm in the presence of greatness at its best.  Nice job, Tom!  Love the compactness and layout and how fast you put this together.

Question:  In the image below, why is the 3rd 813 from the left positioned different from the rest? Odd ball tube?

http://amfone.net/Amforum/index.php?action=dlattach;topic=45867.0;attach=64988;image

Tnx, Bob -


Yep, I wondered the same thing. In addition, the filament glows at about 1/4 brilliance and generates no heat. It's like a 25 volt filament or something. I have a lead on a few 813s so that one will be replaced for sure.

As far as layout, in hindsight, I should have mounted the tank coil vertically and the lead length to C1 would have been shorter.  But, I only use these rigs up to 40M and it shud be OK there.  I was kinda stuck mounting C1 up high due to the pre-existing holes.

But, looking it over, I could probably remount the tank coil vertically in a couple hours work and be done with it... hmmm...  I can do much better and make all RF leads much shorter... they are ridiculous now.  Later tonight.   Funny how a picture puts it in perspective.


T

Nice work Sir Thomas! Gonna look it over for a while here.

But when you get the tank coil vertical it needs some orange wire!

Yep, some Dentron orange wire is just what it needs... :-)

Well, I couldn't take looking at these long RF tank leads and completely redid the tank circuit.  I mounted the coil vertical, thus, all leads are a few inches long at most. The tube plates are closer to the network too. The vacuum variables almost touch the coil ends now like it shud be.  I also elevated and remounted the bandswitch so that it is next to the coil.  I'm thinking the rig might make it up to at least 15M now.  You won't believe how different it all looks.  I had to remount the muffin fans on the bottom since the coil took the top area and left a big space on the bottom.


Shuda done it right in the first place.  I'll post some pics when it's finished tmw.

T 

Looks very fine Tom....  But it seems you did not look carefully at that third photo!


Yaz sez "Is the snow melted yet?  Let's go sailing, Tommy!  You can mount that inductor on Hollywood when we get home."

This series of pics is focused a lot on the process of relocating the tank components to achieve shorter RF tank leads and better stability on the higher bands. Compare to the earlier pictures and you will see a much better layout.  I should know better by now and shuda done it correctly the first time.


Notice that C1 and C2 have short straps to L1 as well as the bandswitch being relocated right next to the tank coil.


I added the muffin fans and fired the fils up.  813s do not require forced air cooling, but a good convection and breeze helps. Without the fans on I could easily burn my skin by touching the 813 tube glass. With just a faint, slight breeze, the glass stayed luke warm. I experimented to find the best fan positioning for both the modulators and finals so that air bounced off the panel and chassis to create a better cooling around the back of the tubes too.

Notice the front panel band switch location is changed and rather crowded, but anything for shorter leads!

You will see pictures of the HV supply being started as well as the GFZ audio board parts.  The 813 rig itself has been tested and all circuits appear to be working.

A friend sent me an email with a contact -  I have ten tested 813s on the way!

T

** If the pic is sideways, just click on the blue link:

More:

More:

More:

More:


100 Pound plate transformer and mod xfmr all needing cleaning and painting.



GFZ audio board and mod reactors.


Yaz wondering when his next run in the woods is coming up....

What are you using for a sequencer?

Hi John,

I haven't decided yet. But I have a old SS  4-step sequencer that I may use.  

However, the new technique I use now is to use a vacuum relay to put a 5K 50 watt power resistor ACROSS the mod reactor during unkey to quench the spark. It worked so well in the 4X1 rig that I've often wondered if I really needed such an elaborate sequencer that I now use.  Maybe that simple technique will make it as easy as a couple of relay steps.

BTW, I'm still not satisfied with the latest second tank build on Hollywood.  I'm thinking of simplifying it with no bandswitch and a larger coil to cover 160M without padding. It's too busy and crowded now.  A trap door to manually change the coil tap may be in its future with even shorter RF connections.  I hardly change bands from 75M anyway.  It's gotten too hard to work on the tubes and tank area since the recent changes.  The first way was easy, but the leads too long.. oh my, what to do!

T

OK, this is the third and FINAL rebuild on Hollywood. On rebuild #2 the rig was starting to get crowded and ugly so I just HAD to do it over again. Disregard build #1 and #2.  Number 3 is the goods!  The RF layout "flow" is cleaner and easier on the eye.

I switched C1/C2 positions for shorter lead length. I found a beautiful silver plated copper tank coil in the cellar that I didn't know I had. Now I can run a Q=12 on 160M without a padder. The last coil was too small.  I mounted the bandswitch on its own fiberglass stock bracket. The fans are repositioned.  The leads from the bandswitch to the tank coil are as short as I could make them, considering the parts size and heck, it's a big rig.  I can also get at all the parts easily, including to change the tubes in seconds.  I added a strapping HV safety choke - I believe in it being robust enuff to pop the breaker without smoking like a little 2.5 mH does.  

The tank coil connections still need to be soldered once I find the optimum taps using tone peak power tests. They should be close to Q=12 the way they are positioned now.

This is my best effort for now. I'm really happy with the front panel look and the overall potential to be RF stable.. though we will see in testing.

I get emails every so often asking about building up an 813 rig, likely from the 813 schematic  posted on this site.  These pictures will help others to see what they are in for if they want to build up their own. It's a perfect size to put out a big signal but not quite the heat generator of a big linear or broadcast transmitter.

Tom, K1JJ

BTW, here's the tank C1, C2 and L1 calculations for 160-20M.  A 4" diameter coil, 9" long and 22 turns  works out perfectly. 160M is almost the whole coil (about 26 uH) whereas 75M is about 1/2 the coil and a  1:1 dia/length ratio.  I used 2800V, 400 mA, and 3500 Ohms plate impedance allowing for a K=2, the class C constant.   Q=12.

pi-net calculations:

            C1         C2          L1
160M   289 pF   1750 pF   26.5  uH

75M     140       848     12.8 uH

40M   75       450   6.8 uH

20M    38    230   3.5  uH


I'll bet this 813 tank, the way it's built now, could handle a class C  4-1000A.  


** I've got eleven tested 813s coming in soon!    Thinking about a little rig to replace my 3-500Z linear.  A single 813 cathode or screen modulated with a MOSFET circuit.  

More:

More:

More:

*** The RF deck and modulator deck is done for now.  No fourth rebuild.   ;D    Now onto the HV supply, mod iron and GFZ MOSFET audio driver.  ****


T

I will be interested to see if you have any audio distortion with this transmitter.  The reason being: I have built the 813x813 transmitter and cannot for the life of me
get rid of the audio distortion.  I have tried 3 different modulation transformers but all give the same results, audio so distorted as to be unreadable.  Maybe it's like
Einstein said, "making the same mistake each time but expecting different results."  Anyway, good luck with yours and please keep us posted.

Hi Roy,

Interesting.  Don't worry - we will help you find the problem. If the audio is so distorted it's unreadable, it should be a problem that can be identified by using a scope to trace thru the stages and look for the point where it gets bad. There are other tests we can run to separate and isolate the culprit.

I know of about ten homebrew 813s X 813s plate modulated rigs on the air. Some sound average, some sound pretty good and a few sound absolutely top shelf pristine, like a class E rig. The last one I built ten years ago, the TimTron said it was the best AM rig he ever heard me on.  I used triode-connected 813s and heavy audio negative feedback using Frank/GFZ's MOSFET driver. It was a big effort to optimize everything.  As I have often said, it can take as long to optimize a plate modulated AM rig as it takes to build sometimes - and it will greatly depend on a builder's experience.

That said, tell me more about your efforts to find the problem. I'd say it's not the mod xfmr for now, since you replaced it a few times. How do the early audio driver stages look on the scope (with tones, swept sine and 50 Hz triangle) and what are you using for driver components?  How about the voltage regulation and modulator linearity?  Give some more rig fine details.

Again, since there is bad distortion, we shud be able to identify the area and go from there. That rig should sweep well and look good when we figure out the main cause. Then fine-tuning efforts will bring it into the hi-hi FB class.    I am going to be at that point soon with my new "Hollywood" 813 rig and will post some on-air recordings when I do. It will be a slow process, but I guarantee it will sound great when done.  The overall design is fine once we find the problem.

* Please post some detailed pictures...***

T

Those tank values are a bit off!  This site uses the Eimac values for calculations.  These values match what I've measured.

http://people.physics.anu.edu.au/~dxt103/calculators/pi_tank.php

BTW, the 28 pF combined output capacity plus the minimum C1 and strays will drive you to a Q of 40 or more on 10M.  BUT, no worries, better harmonic reduction AND the efficiency isn't reduced by much.  I'm down about 90 watts on 10 M compared to 160 or 75, getting about 770 watt carrier with 1 KW input on 75M.

More to follow...

http://people.physics.anu.edu.au/~dxt103/calculators/pi_tank.php

Chuck,

77% efficiency on 75M is hi-hi FB on your new 813 rig!


http://people.physics.anu.edu.au/~dxt103/calculators/pi_tank.php

Yep, that's the calculator I used. I just ran 75M to be sure - and it shows the same 140 pF, 848 pF and 12.8 uH as I posted above.

Probably the difference you see is my assumption of a K constant = 2 for a class C stage. This was from another website.   I came up with 3500 ohms for the plate impedance using K=2, 2800V and 400mA IIRC.  I've heard to use K=1.5 too, so maybe you can advise.



The site says: "3. The Plate Current (Ip), is calculated by the formula: a. Ip = Pin ÷ Ep 1. Plate Voltage (Ep) 'maximum', for the Tube's class of operation, is obtained from the Tube Manufactures Data Sheet.
 4. The Class of Operation uses a 'K' factor for a specific Amplifier design which is: a. For Class A Operation it is (1.3 ~ 1.4)  b. For Class AB Operation it is (1.5 ~ 1.7)  1. For Class AB1 Operation it is (1.5) 2. For Class AB2 Operation it is (1.6 ~ 1.7) c. For Class B Operation it is (1.8 ~ 1.9) d. For Class C Operation it is (2.0)
5. The Plate Load Resistance (RL), is calculated by the formula: a. RL = Ep ÷ K x Ip 1. Plate Voltage (Ep) 'maximum' is obtained from the Tube Manufactures Data Sheet."


http://rfcec.com/RFCEC/Section-3%20-%20Fundamentals%20of%20RF%20Communication-Electronics/04%20-%20AMPLIFIERS%20-%20RF%20POWER%20AMPLIFIER%20BASICS/RFPA%20-%20Design%20Formulas%20(By%20Larry%20E.%20Gugle%20K4RFE).pdf

I didn't account for stray from the circuit and tubes which will change things too.

But the bottom line is I planned on getting the taps in the general ballpark and then run tone peak and power tests to see where the fine tuning lies.  Notice I didn't solder the tank straps on and will use a strong clip to hold them tight during initial testing.   I'm finishing up the new 3KV supply now and then the mod iron and GFZ MOSFET audio driver next.

10M?  I'd probably be at the end of that coil - 20M is probably it ... it wud need a smaller 10-15M stout coil switched in and C1 at min to pull it off on 10.

Glad you took the time to look over the new rig, Chuck!

T

Just an update...


Roy tells me via email he has solved the distortion problem and all is well with his 813 rig.  Turns out it was a wiring error on his schematic involving the modulation reactor coupling cap. He is now using the correct wiring from the AMFone schematic and the audio is now FB.


Roy, I just finished my HV supply and all that is left is to wire up the WA1GFZ MOSFET audio driver.  Then it will be testing time and real on-air trials.  You might consider Frank's audio driver with audio negative feedback. (NFB).  I now use one on my 4X1 rig and can't wait to see how the second one performs with this new 813 rig. I'll let ya know when I am ready and we can have a 813 to 813 QSO on 75M at night.  I am making it a goal to get this 813 rig sounding absolutely perfect to set a high bar for the guys who've already built them - but still need some more audio work to get that "sound."

T

My mistake...I didn't see the Q=12.  I run a Q of 20 to 25 on all rigs and have these values almost memorized.

The reason I do this is that at a Q of 12, a mismatch could lower your Q quite a bit!  A 2:1 SWR with a load impedance high and reactive, it could go as low as 6 with values selected for a Q of 12 into 50 ohms.  Ever run into a situation where the rig "just won't load up"?  The resulting low Q at an attempt to match something well off 50 ohms load probably did it.  Raising the Q by lowering L will fix it.  Changing the feedline length is another hack to fix it.

So....I use a Q of at least 20 to avoid this and still get decent second harmonic reduction.

The reduction in efficiency at Q=20 vs Q=12 is not measureable on a Bird watt meter.  Perhaps it's a needle width but likely less.  If the Q of the inductor is above 200 or so, the loss should be minimal.  A decent inductor will often hit 300 but when you start shorting turns on a tank inductor to change bands, it will drop quite a bit.  That's where the loss comes from.  Dedicated single coils for each band is the way to go.

Chuck

Very interesting.   I still have the chance to move my tank switch inductor taps to give a Q=20... nothing is soldered on the coil yet.

I can see that if an antenna matcher is used, then there could be a 1:1 all the time. But in my case (and yours and many others) we are using coax or hardline so that the swr will vary.  I will try the higher Q taps on this rig and see how it works.  I have always liked using higher Q tanks better anyway. The sharp tuning gives me a charge....  8)

T

Instead of shorting the coil, what if we simply tapped the coil and left the extra turns open?  I realize we would have Tesla coil action, especially on the higher bands and possibly arc over the band switch or coil itself.  But what if we had a 160M coil that was tapped 1/2 way for 75M with no arcing? Would this help preserve the Q? Or would the auto-transformer circulating  current effect still take its toll on Q even when not arcing?   Tapping a coil rather than shorting it always looked cleaner to me, assuming we didn't do it on a big coil above 40M or so.

T

As my final post on this thread:

Finally - Success!     At 12 Midnight, Don, K4KYV was calling CQ on 3875.  I fired up Hollywood for the first time on the air and gave him a call.

I've spent a few days running tone tests and making some mods to get Hollywood sounding the best I can. I believe a rig can sound excellent before ever seeing real time QSOs.

Don was QSO #1.  He said it sounds exactly like the 4-1000A plate modulated rig. I didn't touch the audio chain, EQ or anything related to audio except the level.  It was a summer night and the shack remained cool. Great tubes, those 813s.

I'm real pleased with the sharp +- 6 KHz  bandwidth audio skirts I see on the receive SDR spectrum, the REA modulation peaks as well as general sound in the mod monitor.  It sweeps cleanly from 19 Hz to about 10KHz. I haven't added in NFB yet due to some circuit mods that need to be done, so it shud get even better.

I put the cover on and consider it a done project for now.  As I was working Don, I was laying out the chassis and front panel for my new 125 watt summertime rig; a 4D32 plate modulated by a pair of 6LF6 big sweep tubes - or maybe plate modulated by a 75 watt SS 8 ohm amplifier using a 1642SE  Hammond  53 Henry 28 pound super hi-fi output transformer. The little rig for summertime.

But for now, Hollywood ran a couple of hours last night without a single problem at 1/2 gallon out. (500 watts)   I have a HV PS that will do 1500-3KV.     2200 V is a great voltage for the lashup.

So there's now two plate modulated rigs in the harem with a third to be added soon.  I shud be on the air with it quite often over the next few weeks, so give me a call.


T

This thread continues:

http://amfone.net/Amforum/index.php?topic=45941.0