Building a new amplifier

[ka1tdq]

Now you're just being a silly-willy 

On to the RF deck!

Jon

[ka1tdq]

I just ordered the last parts for the RF deck (minus the toobs... $600 for three 3-500s, ouch!). 

Actually, I just wanted to post a picture of this thing with the lights on.  I like it.

Jon

[ka1tdq]

I'm trial fitting everything.  The ceramic insulator on the right will be a homebrew plate choke.  I didn't have enough room in the chassis for a massive loading capacitor, so I'm using a smaller value one in combination with a doorknob capacitor.  The Ruskie vacuum capacitor was new in the original sealed plastic bag with the manufacturing stats. 

I'm using a Home Depot PVC box to pressurize beneath the tube sockets. The whole thing is on a 4-foot high welded steel frame.  The 12" x 12" PVC box will hold (3) 3-500's nicely in a triangle pattern.

Jon

[ka1tdq]

I got the holes drilled and I'm just waiting on my sockets/chimneys to arrive from China.  I'm hoping that they didn't get lost in the mail... it's been like over 3 weeks now!

I did order three high performance plate caps from www.73cnc.com.  He has a YouTube video comparing his custom plate cap to an ordinary plate cap, and the difference is remarkable.  He takes temperature crayon dust that melts at 225 degrees F and places it on top of both plate caps.  Then he cranks up the power.  The dust on top of the ordinary plate cap melts very quickly and the other doesn't melt until about 1.25KW. 

Three 3-500's draw 45 amps of filament current, so I'm going to use two MFJ 30 amp filament chokes in parallel (Also waiting on those for delivery).

Jon

[WA9NQW]

Hi Jon,

Shipments from China have been disrupted for a few weeks. It is Chinese New Year time again.

73,

Jack WA9NQW

[w1vtp]

<snip>

...Three 3-500's draw 45 amps of filament current, so I'm going to use two MFJ 30 amp filament chokes in parallel (Also waiting on those for delivery).

Jon

Will that "half" the inductance of the chokes since they are in parallel?

[ka1tdq]

That's true, I didn't think about that.  MFJ lists the frequency for a single filament down to 160 meters.  My guess is that two in parallel would still be fine for just 40 meters. 

I've attached a picture of the homebrew filament choke I used with my last linear (using just a pair of 3-500's).  The ferrite rod is 6-3/4" long and 5/8" in diameter.  The wire is tightly wound solid #12.  I think it'll handle the 45 amps fine enough, I'm just worried about the voltage drop.

Jon

[ka1tdq]

I've said it before, and I'll say it again... God bless the Chinese.  My tube sockets and chimneys just arrived.  They are absolute masters of packaging efficiently.  I had to destroy the packing boxes because they shipped everything in tight boxes with expandable foam.  But, it made it from China a-okay!

Jon

[PA0NVD]

I didn't have enough room in the chassis for a massive loading capacitor, so I'm using a smaller value one in combination with a doorknob capacitor.

Hi Jon
Take care with this set-up, The caps have relatively high capacitances so the interconnection wires should VERY short and better be copper tape to reduce the inductance. I had several problems with parallel resonance at higher frequencies this way that gave a high load impedance, high harmonics or incorrect loading.

So I had to remove the fixed caps and went back to a high value loading cap.

[ka1tdq]

I'm definitely going to use copper strapping on the output network.  And the way things worked out on the layout, the lengths aren't going to be too long.  The loading capacitor lead lengths in particular are going to be very short.  Also, I have a little bit of fluff-forgiveness factor built in since this will only be used on 40 meters and not the higher bands.

I've got most of the major components mounted now.  I'm using two Hammond transformers in phase for the filament power.  Heck, if I used them out-of-phase it'd be a quick show.  And the bias will be around 9 volts with 12 1kv/3amp diodes in series.

I'm waiting on one more part to arrive and then I'll whip out the soldering iron.

Jon 

[ka1tdq]

I'm about 90% done except for buying the tubes.  I need those to measure the placement for the parasitic chokes.  I have another day of wiring underneath the chassis and that'll be all done. 

I would've made the plate choke longer, but I ran out of #20 wire.  I ended up with 6" long and 1" diameter.  At 31 tpi, that works out to be about 135uH, or about 6200 ohms on 40 meters.  It's good enough, so I'll just go with it.

Jon

[K1JJ]

Very pretty, Jon!  See how good your building skills are becoming?   A few rigs and you're an expert.


Questions:  

1)  Did you leave room for a turns counter for the vac variable?  A turns counter is fast - counting revolutions is slow.  


2)  Is that air variable the loading cap in the back? About 600 pF?    You have 3 tubes which makes the impedance even lower - if so, you may need up to 2000 pF for 160M. 1200 pf for 75M at times.  Maybe you plan to pad it.

( I see a couple of fixed caps back there)


Use a heavy duty rotary switch to add in incremental stages of padding, if that's the way you go.  Don't use TV door knob caps. Try to get as much of the air variable meshed as possible when in use.  (compared to the % of padding)   The air variable is much more efficient for circulating current and heat than most fixed caps.  That is a resonant circulating circuit where Total RF current  = Q * I.

T

[W1ITT]

Jon
From where did you get that mount for the vacuum variable capacitor ?  I have a few nice ones without mounts and, as simple as they look, it's not always easy to fabricate something that pretty.
Norm
W1ITT

[ka1tdq]

Thanks Tom! 

To answer your questions, this will be used on one frequency on 40 meters so I don't need to count turns on the vacuum variable.  Just tune it and park it. 

That is a 150 pf loading capacitor in the back that I have paralleled with a 680pf (I think) Ruskie doorknob cap.  Again, once I find the right value, it'll be parked. I went this way because I ran out of room in the chassis for a larger sized air variable. 

Norm:  I was on eBay at the right time.  I found those vacuum variable mounts from a Ukrainian seller.  He got them as a pull from old military equipment over there.  I overpaid for them (I think $65 shipped), but they're perfect for the job. 

Jon

[Opcom]

In case it has not been mentioned and the cathode circuit not completed, three 3-500Z in grounded grid have an impedance of 38.33 Ohms according to the attachment.

[ka1tdq]

That's true.  A pair of 3-500s has a nice 50 ohm impedance match, but three don't.  I'm going to use an LDG auto tuner in line before the amp. 

Jon

[ka1tdq]

All done except for buying the tubes.  Here's how it'll look in the closet.  The chassis does have a top that I'll put on once I insert the tubes.

Jon

[KG0MN]

I am in awe of this build, it truly is a thing of beauty and not many things impress me except when something is designed very and look nice when it is finished.  Do you have an estimate what the amp would cost for someone that has an empty junk box. For now, I will continue to use my  Drake L4B and dream of a home built amp at night in my dreams.

73 and Good Luck with the amp,
Kevin - KG0MN

[ka1tdq]

Thanks Kevin!

Well, it still wasn't cheap.  I bought everything for the project except the parts for the HV power supply (I already had the plate choke, filter and caps).  After I buy the tubes, I'll be in it for over 2 grand I'm sure.  I'm too scared to add it up. 

It's still somewhat of a bargain though.  The Ameritron dual 3-500 amp goes for over $3000.  And you get all band, compact size and a factory warranty.  I only operate 40 meters though and I wanted the extra tube.  Plus, it's a little more rugged than Ameritron, which is good for AM service.

OR...

You could go a lot cheaper for the same power with class E.  My building ability just isn't there yet.  I'm still stuck in the building furnaces stage.   

Jon

[w1vtp]

Thanks Kevin!

Well, it still wasn't cheap.  I bought everything for the project except the parts for the HV power supply (I already had the plate choke, filter and caps).  After I buy the tubes, I'll be in it for over 2 grand I'm sure.  I'm too scared to add it up. 
<snip>

Let's see:

 $2,000 ... Parts
-$1,000 ... Fun Factor discount
--------------
 $1,000 ... Total cost

Sounds like a bargain to me!  Enjoy Jon  Maybe we'll hook up on 40 some time

Al

[ka1tdq]

Ha! True. 

I talked with Steve one time using my old 2-holer, so I'm sure I'll be able to make the trip with 3.

Jon

[K1JJ]

Ha! True.  

I talked with Steve one time using my old 2-holer, so I'm sure I'll be able to make the trip with 3.

Jon

The difference between 2 tubes and 3 tubes is about 1.5 dB.... :-)

You'd have to go to 8 tubes to see a meaningful 6dB increase. (1 S-unit)

You'll be fine with 3 tubes.  Once you get above  1 KW, it gets harder and harder to get meaningful increases.  Just like adding horsepower to a drag car. Diminishing returns.

T

[WD5JKO]

   Question: Doesn't the gain stay the same whether it is 1, 2, or 3 tubes? I've never used a 3-500Z, but I'd presume something around 10-13 db gain depending on circuit conditions. If the driver can only drive one tube to full output, then adding 1 or 2 more won't do much. A tuned input will help increase the gain. Perhaps the LDG tuner if close enough to the tube cathodes will act as a tuned input.

Jim
Wd5JKO

[K1JJ]

Yes  that's right - if the high voltage, idle current per tube, Q and general circuit parameters are the same, then the amplifier stage gain remains the same when the number of tubes is increased.

I wasn't referring to stage gain. I meant that by increasing the number of tubes the amplifier would have more power output capability before flat topping -  AND of course, it would take more proportional drive to do it.  IE, twice the tubes would require twice the drive to double the output. (3dB more power out, with the same stage gain of 13 dB or whatever -   and 1/2 the plate load impedance)


BTW, Jon, I have run an outboard tuner into GG amps like you plan and it has worked FB. Be sure to use as short a coax cable as possible, like less than 2'.  Even mount the tuner inside the amp near the sockets if possible.  A 'T' or 'Pi'  tuner circuit works well for 50 ohms into  30-100 ohm type loads.   A  Q=3 or so is all that's needed.

T

[KD6VXI]

In regards to an external tuna. 

Your efficiency on the upper bands will be helped even more by putting 100 or so pf at the tube socket (or,  appropriately sized caps divided up to equal about 100 pf).

This moves the return pulse path to the cathode pin  rather than down your tuned input coax.

This being a 40 meter mono bander it won't help much,  but maybe some 2 to 350 puffers  (or,  75 pf per fil pin on this 3 holer)  may help.

--Shane
KD6VXI