Junker Amp Plate Voltage

[W7SOE]

The Junker amp is moving along again...

The power supply uses a transformer producing 1150 VAC, unloaded.  The doubler boosts this to 3200 VAC, unloaded.  I am using ten caps instead of the original six, and I have a couple more diodes in the diode chains.

I am running a pair of 813's.

3200 is too high for an 813.  I assume the voltage will drop under load but it would have to get down to 2800.  I am not sure what do do atthis point other than continue and see what it actually is under load.

Rich

[K1JJ]

3200 is too high for an 813.  
Rich

Hi Rich -


The 813s are in class C, plate modulated, right? If the surrounding parts are designed to handle it, why not?

In the past I have run 813s X 813s  at 3KV, class C.   A pair will load up to a 800w - 1 KW carrier out. (with 140% modulation capability) At 3 KV both the finals and modulators do not show color, but are approaching full dissipation. I needed some bias on the modulators in the form of diodes in the cathode lead. I use chimneys and forced air with carbon 813 plates. Doing so probably  raises the plate dissipation to about 175 watts rather than 125W.

I realize the data sheets show lower voltages, and some guys think otherwise,  but this is a real whirl example. Never had a flashover or lost a tube at that level. They are very, very rugged tubes.

So, when you key up your power supply, if it drops down to 2800V under dead carrier load, you should be FB.

To do this, the rig must be running near perfect - efficient  without instabilities or arcing. You will need 1 KW mod iron or it will be at risk. It is best to start at 1500V and slowly work your way up using a Variac.

T

[W7SOE]

Thanks Tom.  I am not going to use the amp on AM, at least initially, only SSB.  I don't think the PS has enough strap for that.  It is a a grounded grid amp. I do not think there is any bias, either on the grids or cathode, the CT for the filament transformer is grounded though I could add diodes.  Schematic here:

http://www.w7ekb.com/glowbugs/projects/JunkerAmp.pdf

Not sure what class it will be operating in...

I am using two 813s instead of the one specified, perhaps that will help drop that plate voltage.

Rich

[w1vtp]

Is a 12.5% drop in HV PS OK for linear service?  That seems to be a lot

Al

[w4bfs]

Is a 12.5% drop in HV PS OK for linear service?  That seems to be a lot

Al

good question ....depends how much crud you (and others) are willing to put up with

[W1ITT]

And speaking of crud, your neighbors on the band might be happier if you included a tuned circuit input.  I don't know what the cathode input impedance of a pair of 813s might be, but it would only be dumb luck if it turned out to be 50 ohms non-reactive.  WD7S lists it on his site as 270 ohms per tube, so cut that in half for two tubes as a starting guess.  Back in the days when that article appeared in QST, all of the usual exciters used tubes and pi network output stages and could be tuned to operate into a wide range of impedances.  The typical modern rice box only wants to operate into a nice 50 ohm load.
A tuned input circuit will reduce IMD, as well as making it easier to drive the amplifier. WD7S lists his TU-6B tuned input circuit kit for about $60.  I have one for an amplifier and it's a well made kit.     http://home.earthlink.net/~wd7s/TU-6B.htm   
I know that buying things like this takes you out of the true Junker category, but there's no reason you couldn't cook up your own on a strip of Bakelite for authenticity.  The use of Fahnestock clips for input and output connections would only add to the aura.

[W7SOE]

And speaking of crud, your neighbors on the band might be happier if you included a tuned circuit input.  I don't know what the cathode input impedance of a pair of 813s might be, but it would only be dumb luck if it turned out to be 50 ohms non-reactive.  WD7S lists it on his site as 270 ohms per tube, so cut that in half for two tubes as a starting guess.  Back in the days when that article appeared in QST, all of the usual exciters used tubes and pi network output stages and could be tuned to operate into a wide range of impedances.  The typical modern rice box only wants to operate into a nice 50 ohm load.
A tuned input circuit will reduce IMD, as well as making it easier to drive the amplifier. WD7S lists his TU-6B tuned input circuit kit for about $60.  I have one for an amplifier and it's a well made kit.     http://home.earthlink.net/~wd7s/TU-6B.htm   
I know that buying things like this takes you out of the true Junker category, but there's no reason you couldn't cook up your own on a strip of Bakelite for authenticity.  The use of Fahnestock clips for input and output connections would only add to the aura.

I plan on using the amp with a tube transceiver (Drake).  That said am going to (eventually ) build in my B&W turret coil assembly with variable cap to serve as a tunable input.

Rich

[K1JJ]

Hi Rich,

A pair of 813s in linear service?  Now that's a whole different thang.

I had a friend who had four in parallel in GG linear service with 4KV on them. It did 2KW pep out.  As long as you can idle them at about 1/3 dissipation, they will be fine.

When ya think about it, running them in class C plate modulated service with even 2500 volts means about 5-6KV peak under full modulation. So 3 -4 KV is nothing when it comes to flashovers in an 813 linear.

The whole key is now cleanliness, IMD. As already mentioned, an input tuned circuit is important for lower IMD. Running grounded grid is good too, as you plan, because of the inherent negative feedback.  Try to drive the amp with as little grid current as possible. The least input drive power you can get away with is best. Grid current ALWAYS degrades an amplifier's IMD, so if you put a nice high voltage on the plate, you will see more power output BEFORE you start drawing grid current.  AB1 is always better for cleanliness vs: AB2.

The power supply could cause IMD problems if it fluctuates too much from no signal to full signal. At 2800V, it probably should not drop more than 150 volts or so, or your band neighbors will know about it.  A voltage doubler multiplies the regulation drop by X2, so be ready to go either conventional supply design or add some big mo-fo filter caps and a choke.

When set up right, 813s have proven to be good linear tubes when run in GG and driven lightly with heavy plate loading. (This means less meshed loading cap so power drops about 7% from peak tuning) 

Be absolutely sure the driver is clean, since this could easily be the IMD bottleneck, no matter how clean the 813 linear is.

T

[W7SOE]

Tom,
Thank you for the help, you are super informative as usual.  I will keep an eye on the voltage sag, now that I understand that it will introduce distortion I hope it is not too much.  I assume that keeping the grid current low means just driving the input with less power...

Rich

[K1JJ]

Tom,
Thank you for the help, you are super informative as usual.  I will keep an eye on the voltage sag, now that I understand that it will introduce distortion I hope it is not too much.  I assume that keeping the grid current low means just driving the input with less power...

Rich

Good luck on the project Rich. Hope you post some pics and specs of the results when finished up.

Yes, less power drive to the cathode input would result in less grid current.  You will find that maximum plate voltage, with heavy plate loading and minimum input drive for a given power output will give  the cleanest signal out.  But there is a price to pay since heavier loading decreases the amplifier efficiency somewhat and less grid current means less output power.

In contrast, if you were to peak the plate tuning and loading for maximum power output like many hams do, and run the drive up for balls-to-the-walls grid current, you may see another 25-40% output in power, but the IMD may climb into the low -20's dB area, which becomes close to problematic.  An IMD of at least -30dB or better is desirable and is achievable running 813s conservatively as discussed above.

It is a science and an art (and a compromise) to design and adjust a linear amplifier for efficient power output and be acceptably clean at the same time.  It is somewhat easier with tubes designed for linear service like the 3-500Z, 8877 or 3CX-3000A7 since they give us a head start of at least -5 to -10 dB better IMD than an 813 - an older design not as adaptable to GG linear service like the newer tubes.



T

[W7SOE]

Thanks again Tom, I am a ways off from testing yet, but will report.

Rich

[K1JJ]

Nice job, Rich!

That is no junker... :-)

T

[Opcom]

Thanks again Tom, I am a ways off from testing yet, but will report.

Rich

I like those Gothic arches! Cane metal windows for them perhaps?

[W2PFY]

Nice amp Rich, it would be great if you can find another good RCA with the so called meatball logo. I have read somewhere that tubes with that logo are pre WW2 or before 1940?

I've only seen that logo on 813's and 833A tubes. The 833A's with the logo are very rare. Let us know how it works. I think B&W made amplifiers with two 813's and there was also another brand that is lost in my brain matter.   

[W7SOE]

Nice amp Rich, it would be great if you can find another good RCA with the so called meatball logo. I have read somewhere that tubes with that logo are pre WW2 or before 1940?

I've only seen that logo on 813's and 833A tubes. The 833A's with the logo are very rare. Let us know how it works. I think B&W made amplifiers with two 813's and there was also another brand that is lost in my brain matter.   

It would be most aesthetically please to have another one like that, I agree!  Not sure I could ever find one though.

I have a question about the window.  I am planning on using a sheet of polycarbonate with a stainless steel mesh behaind it.  Is the mesh neccesary, as an RF shield?

Rich

[W2PFY]

Hi Rick, The RF shielding may keep out RF into your audio chain? Maybe once you have it up & running you could run it without the screen in there and see what the effects are?

I would be very careful about keeping smartphones and other such devices directly in front of the windows. In the past I have destroyed at least two VCR's due to strong RF fields. I may have an RCA meatball tube around but I am not sure if it's a 810 or an 813.

It may be a couple months before I can locate it as the crap has hit the fan here and have been busy with stuff that is not fun. OTOH I may stumble across it very soon?Maybe someone else has one out there? 

So how soon will you be ready to test it?

[W7SOE]

Hi Rick, The RF shielding may keep out RF into your audio chain? Maybe once you have it up & running you could run it without the screen in there and see what the effects are?

I would be very careful about keeping smartphones and other such devices directly in front of the windows. In the past I have destroyed at least two VCR's due to strong RF fields. I may have an RCA meatball tube around but I am not sure if it's a 810 or an 813.

It may be a couple months before I can locate it as the crap has hit the fan here and have been busy with stuff that is not fun. OTOH I may stumble across it very soon?Maybe someone else has one out there? 

So how soon will you be ready to test it?

If you could take a look for such a tube I would appreciate it.  Of course I'll pay the going rate.  I have another spare 813 to use in the mean time.  The Penta Labs 813 I have is suspect.  It was from eBay and I think there was someone selling bad Penta 813s...

Rich