The AM Forum

Holiday Greetings.

I intend to build a linear over the next year and need some advice on tubes.

I have  10watts of drive from a HX-50 and am  interested in 160m-80m-40m only.  I realize HX-50 is not a very good  performer, but I will dial it in ok so it sounds good.

I am looking at 2 ways of skinning this cat.  Mybe a 6146 driving into a pair of tubes, or one maybe 4CX800A tetrode tube.

A pair of 572B tubes are about 100 bucks and the 4CX800A tube is about $140.  These prices are from Tom's tubes and I'm fairly certain they are from Russia.

Are these reasonable ways to go?  One better than the other?  What should I expect for output power.  Are 811 or 813 tubes another possibility?

best regards,
Bob
N9NEO

Hi Bob;

How about a single 3-500Z grounded grid?.. Inexpensive tube, readily available, easy to build, don't need a leaf blower to cool it, good IMD, stable...

I've been using a single 3-500Z for several years now for amplifying my AM exciter. 10 watts of drive gets me about 125 watts out. I would have gone with a pair of 3-500Z's but I couldn't get 220v in the shack in my last house. Not the case with the new QTH  :D
 

Another great tube/option I forgot to mention in my last post is the Russian GS-35B triode. You can get them new for $79 plus shipping from the "QRO-Stuff Store" You'll get 2kw+ in a grounded grid configuration with around 3kv on the anode. Don't need a socket, they can be mounted directly on your chassis. Best dollars-per-watt choice.

Here's a link to the E-bay store that sells them;

http://cgi.ebay.com/Tube-GS-35B-new-in-box-2500-W_W0QQitemZ5841508470QQcategoryZ48710QQtcZphotoQQcmdZViewItem

And lots of info on building an amplifier here = http://amp.w4zt.com/

...and here = http://home.comcast.net/%7Ewb8wju/

Hola Bob,

Lots of ways to go.

You being a very technical guy, this may interest you. I've been getting a good edumacation from the Amps Reflector about using tetrodes with negative feedback. Normally a GG amplifier is good for -32 to -36db 3rd order IMD according to the amp manufacturers in the real world.

But by using drivers that are running in class A, low level and driving a tetrode with voltage only, you can achieve upwards of 50db+ 3rd order with the drivers.  Then put some RF negative feedback around the tetrode power final to bring it up with the rest of the chain.

I like this design and plan to try it out. Read my post about IMD tests and you will see the difference it makes when tuning across a signal going from a -32 signal to a  to -55db 3rd order signal. [Estimates]
 
This applies to both AM and ssb, of course. With input baseband frequencies out to 6kc on AM, a little IMD can really raise havoc compared to the narrow <2.8 ssb audio.

Starting with only 5 mW into a OPA695 op amp at 500mw, I plan to use a 6550 running class A putting out LESS than a watt, driving a 4CX-350 running class A also putting out a watt or so [mostly for voltage] driving a 4CX-3000 with about 20db of negative feedback from the 4X3 back to the 4CX-350A.

We'll see.

A 6550 driving your  grid driven 4CX-800 with neg feedback could be a nice, clean way to go. No grid current in the 4CX-800 - Running 350V on plate, 200V on screen and -37V on grid of the 6550 gives deep super clean class A operation with enough voltage swing to drive the 4CX-800 easily.


BTW, check out your 10W IMD output from the HX-50 first. You want that to be at least 10db cleaner than the amplifier to make sense here. You may be able to do this with some serious work.

Here's a schematic using a 4X10 and a similar scheme. Notice the ease of the neutralization and the negative feedback. Simple.  You can scale this for many types of tetrodes.  ALL of my amps here are triodes, and I feel they could be cleaner, thus the effort.

Copy and paste this into your browser and go - it does not work by clicking on it as a direct link for some reason:

http://wc6w.50webs.com/wc6wamps/popp.html?h10a1.jpg


Just another approach to consider, OM.

73,

T

The answer depends on what you have available for a power supply. No matter what:
572b's are overpriced for what you get.
For AM the name of the game is plate dissipation.
6146's should be declared unconstitutional.

Bob,
There are many possible ways to skin the cat. The all time cheapest way to go is use 813s. This may be a safe way to learn. Being a smart guy I'm sure you could easily build something bigger. Just start collecting parts. and see where it takes you.
The Russian tubes sure have become popular.

No way dude!.. Isn't Irb using them in his dixie 100??  ;D ;D

Exactly. YOU are suppposed to do as Irwin says not what Irwin does.

Ok, then it's settled ;D

73
NEO

Bob,--N9NEO,
    I have some "info" and skizmatics on grounded grid 813 amplifiers, which I
     have collected as I am in the process of just starting to build an 813 amp
     for 160, 75 and 40.
     Let me know if you would like/ or need info, and will get try es get some
     of it together and I can snail mail it to you.

                                    73, Rene, K1MVP
 
 P.S.,With a pair of 813`s, u should be able to get at least 150 watts out es
       while its not 300, you will still be only 3 db down from running 300 watts,
       and the 813`s are "dirt cheap" and thus you will get a lot of "bang for
       the buck".

 P.P.S, I think a Valiant with three 6146`s will give you 150 watts, so you would
          be on a par with a Junkston Valiant, --not too shabby, and you could
          run an exciter with "good quality" audio into it .     

Tetrodes are a major pain in the backside. You have to work around screen bypass capacitors which must be low inductance and usually expensive. A 4CX800 either needs to run in grounded screen or have a relatively expensive screen bypass capacitor built into the socket. Then you have to have a solid screen supply that can sink negative screen current. Then you should have glitch protection like gas discharge devices so your complicated screen supply doesn't blow up WHEN the tube arcs internally. THEN the distortion will probably not be too good unless the tube is cathode driven or grid driven with degenerative feedback using some resistance introduced into the cathode.
Any tube(s) cornsidered for a new linear amp design should be priced including the proper socket chimney and blowers if used.
Then take the whole thing as dollars verses plate dissipation. If you really like the tetrodes then try a GU-84 or a GU-78B.
http://www.nd2x.net/tube-1.html
The trick is to find the sockets with a built-in screen bypass. Screen bypass rings can be fabricated at home using Beryllium Copper finger stock purchased from Surplus Snails and mulitple "chip" capacitors.
The olde standby best kept secret would be 813's. Triode connect 4 of them and you will have a great low band only amp. Parts count will be up with this and you might want extra toobs on hand in case some aren't up to the standards of the others in parallel.
A single 4-1000 is nice too. Cathode drive it and if your plate supply isn't enough to get 1500 Watts P.E.P. out then put some negative bias on the control grid and a little
positive on the screen. You get the glass with the Orange glow and all the lead inductance problems that are married to glass toobs at the same time.
If none of the above posts including this one help then you could roll dice. Maybe a fortune teller......
A single GS-35B is nice but don't forget a socket fixture and chimney for the $ figure.
10 Watts will not drive one of these to it's potential. You will also want a B+ around 3500 Volts under load.

813s make nice zero-bias triodes when the screen and grid are grounded.

for more power, its easy to go to 3 or 4 or 5 or more... 813's ...........I always wanted to build one that has a row of 813's so I could just plug in more tubes for more strap..... 

Yeah, that would be fun.  It's called a "barefoot linear", cuz you don't need a blower and it runs silent. You can tell 'em you're barefoot... ;D

I know a ham who collected twelve 813's with sockets to do just that. Intended it for 160 and 75M. He never did and sold them off.

Then there's my buddy who put four in a linear with 4KV - got out 2KW pep.

So, twelve would be good for 6 kW pep, multiplied out.

Great tube, those 813's.

T

output C is a bit high for 20 and above but GG with a pair would do good power 4 should do 2 kw in. Lots of emission. i have a box full that I want to use some day. fc

I have an old (4) 813 deck that belonged to a friend. The thing made every bit as much power as a pair of 3-500Z's for years. Once in a while He would replace a toob.
Don't let the 125 Watt plate dissipation rating fool ya. It's got the same sized plate as higher rated toobs.

4 with air on them is good for legal limit. Make it 5 to be different.

Can you still buy a new 813??..... If you use 4, don't you have to match them do a certain degree??
push-pull, parallel?
If not, one's gonna be huffin and puffin!!!

You don't need a socket.....and nothing wrong with a cheap PVC chimney..

That's right, 10 watts won't drive it to 2.5kw... but you can't run 2.5kw on AM, right??

Matching is not really needed, unless one is really dead.

You can run 2.5 kW on AM. You can run as much input as you want. It's the output the FCC measures. 8)

1500 Watt carrier, 100,000 Watt upper sideband, 100,000 Watt lower sideband exactly 180 degrees out of polarity with the other sideband. Conforms to current part 97 regulations.

:

I thought we were talking about outpoot...???. 8) 8)

A few months back I checked a large quantity of 813's. All were at least 40 years old and some over 60. They were HiMu triode connected as modulators during the test. I checked idling current and ability to make positive peaks. Some of the toobs had been pretty beaten. I found that pairing them is not really too difficult. Even different brands work together.

I suppose if we were really worried about wasting electricity, we wouldn't be on the air at all, or even have our receivers powered. ;D

In the words of the late W2WME:

All non-essential communication is just that, non-essential.

I'm not worried about wasting electricity, but 20% effeciency is a bit silly!  :o :o

You lost me. What is 20% efficient??

Who cares about efficiency when the basement shack is under 65 and heat rises.

Excuse me, not 20%, but  17%  ...2500 watts input 375 watts output.

Steve's comment

"You can run 2.5 kW on AM. You can run as much input as you want. It's the output the FCC measures."

How you say.... tounge in cheek?  :D :D

I think you are planning to run a linear there Glenn. Guess what, 20% efficient will be about it on AM. You can go for 25 or 30 but this leave little headroom for the audio peaks. Of course, this is 20% efficient only when there is no modulation. The efficiency goes up with modulation.

You wanna hear even more ridiculous efficiency?

My new linear amp is gonna have three driver stages, all running pure class A, grid driven.  The efficiency is down around 10% for them.  Usually, take the power dissipated in the plate and multiply it by 10% to come up wid the power out in class A.

A 1W op amp at 500mV out, driving a  6550 tube  running 20W resting idle, putting out 10 volts swing, driving a  4CX-250 running 125W resting idle, putting out 300V monkey swing - all sweating it out to cleanly drive a 4CX-XXXX with no grid current.  The final runs AB1, thank God -  with neg feedback.  Sick, huh?

Caw mawn.

Exactly what I was thinking Steve when I hit the enter key... PEP vs CW carrier...

1500w pep output w/2500w input... dats looks better... so it's 17% efficient with no modulation and 40% efficient on modulation peaks??

Whats the anode voltage and current on that 4CX-xxxx  T?  also, what's the max carrier output?

Glenn,

Well, the 6550 driver tube will have  250V and the 4CX-250 will have 1500V on them. The 1500V is the same regulated supply that is also used for the 4CX-XXXX screen... kinda slick.

Notice how conservative the demand is on the drivers... the 6550 is a 50W? tube with 250V - and only 10V swing needed, no power -  the 4CX-250 is a 250W tube with 1500V and only 300V swing is needed out, no power.  We can get clean IMD figures in the -60db++ range like this, so I'm told.

The 4CX-XXXX will run xxxx volts at x amps = x * xxxx  watts in AB1 at ~60% eff..   ;D  I'm hoping it will be up around -50 to -55 db 3rd IMD once 20db of feedback is looped back to the drivers. We'll see. 

An interesting point:

IMD does not add linearly between stages. The driver chain needs to be at least 6db cleaner than the final so that the final IMD degrades only about 1 db. ie, if the final is -50 db  and the driver is -56db, then the overall IMD is -49db. 

If the final is the same IMD as the driver, then you will see a whopping -6db degradation overall. [-44]  So, you can see why I'm putting so much effort into a clean driver.

T

Vely interesting!... That's a lot of effort... I have to wonder if there will be a noticable/audible difference compared to the standard pair of 3-500Z's... the artifacts were gone after 5kc from what I remember reading in your post..

 Real world amplifier numbers. 1000 Watts DC input gives you 300 Watts carrier out and enough headroom for 100% positive. If not then something isn't right.
2500 Watts P.E.P. input will give you 1500 Watts P.E.P. output AM or ssb.

Check your calculator batteries.

Something like this. At 100% modulation or so, you should achieve the maximum efficiency capability of the amp (most GG AB2 linears are in the 50-60 % range). Theory states a Class B amp can have, at best, 78% efficiency. Real world loss keep that down in the 60% range (with AB1 and AB2 lower due to higher idling currents).

Glenn,

Well, as you know, there's no way to hear the difference on frequency between a -30db or -55 db 3rd order signal.   Someone here once mentioned that -30db 3rd equates to 0.1% distortion level.  It's all in the sideband trash.  I was listening last night to my 5mW ssb signal and it went from S9+40 over to below S1 only 3kc away. That is probably -65db++ 3rd.  But when I turned on the FT-1000D 200w amp and ran it up to only 10W, the trash [normalized] was S9+5  3kc away. Huge difference.  I could still hear it up 4.5 kc.  The 3-500Z amp is even worse than the stock FT-1000D, by about 5db or so.

Just to prove it was not receiver overload, I was able to turn the 5mW level up to 15mW and found the IMD degraded to the same as the 10W amplifier. Running class A and then demanding only 1/10 the power capacity is the ticket to cleanliness. But, the problem is the final. To get 1500W out of class A final requires a 4CX-15,000. Big waste. So a smaller final has to be run class AB1 for a tetrode or AB2 for a hi mu triode. Neg feedback seems to be the only way to bring that tetrode final up to the clean driver IMD level...   At least the GG amp has some inherent NFB.

Anyway...  I think one thing not many guys think about is the varying input impedance of a linear. Just like driving a triode modulator tube, you need a driver [or exciter] that can handle the load variation when going into a cathode driven GG amp. That cathode may vary from 50 to 30 ohms or whatever. I saw it on my own 3-500Z amp when I stuck a Bird in the input.  This will degrade IMD.

So, I'm thinking that an AB1 tetrode has this advantage... no grid current, thus a stable input grid load for the driver, unlike a GG cathode driven amp.

T

"1500w pep output w/2500w input... "  67% efficient...

yup yup yup yup!!

I guess I was thinking that you are hearing those sidebands depending on the bandwidth of the receiver ??

Try another set of new batteries.

I don't know Dave, maybe your calc is in reverse polish notation mode,  ::) 2500w input 1500w output = 67%

2500w input 1250w out = 50%... maybe I'm not wearing my tinfoil hat today??

Glenn,

Well, once you do a particular RX bandwidth setting, you keep it there and do relative comparisons against the standard.  You can use whatever bandwidth you want as long as the good and bad products sound totally different.

In my case, I used 2.8kc bandwidth on the receiver. The transmitter had a recorded voice as a program. When clean, I could hear the voice as clean tones that disappeared into the noise after it's ~3kc audio normal bandwidth.  However the trash has a distinctly different sound. When it is there, it is a raspy, buzzy sound that extends at and past the normal 3kc channel.   As you turn up the drive and the signal goes from clean to poor, the raspy trash products begin to get louder and creep higher up the band.

There are some that say a good reciever with a good ear is a BETTER instrument than a spec analyzer for identifying  an IMD problem. I happen to agree with that. The front end and other parameters can be better in a good RX than the spec anal  too.  Heck, your receiver can be used just like a spec analyzer-  using a two tone in the TX and identifying the 3rd and 5th tone mix where it should be on the dial. Use the atten pad to find it's value from one of the peak two tones to the 3rd or 5th you identify. You can actually come up with real IMD numbers. There is an actual  procedure written if you're interested.

Try listening on your own signal and see.  For a simple setup, put a 6" piece of antenna wire on the back of your receiver and dump the ssb TX into a dummy load.   Use a step pad and terminated 50 ohm pickup for the RX  if you have one for better control.  Set the RX at S9+40 over and see how quickly the trash disappears as you tune on your own signal.

Then listen to signals on the air that you normalize to S9+40. After awhile, you can tell who is clean and who is not... in a RELATIVE way. I would guess signals on the air probably average around -31db 3rd IMD. Some much better, some far worse.  You can hear it when someone is running a -45+ db transmitter. Like tuning a CW note... :-)

T 

you need to use the cw filter when using a rx as a spectrum analyzer this is because the bandwidth is too wide. then correct for the bandwidth. dB=10 Log BW in hz.
subtract correction factor  from reading. a wider filter has a higher correction.

Yes, that makes good sense.  Taking actual measurments on weak tones requires sharp selectivity.

T

A Step attenuator and RMS volt meter on the audio will provide much more accurate numbers.  You know you have a bad method when nimbers change from day to day.

I understand... If you use a wide bandwidth filter, You are hearing (seeing) everything (the trash + voice) at once, with a narrow CW filter you can tune across the signal to listen specifically to the IMD products.

So IMD is more of a co-channel interference issue, not so much distortion to the actual tuned-in signal/voice

I know with audio, it's all within the "passband" of what your listening to...