The AM Forum

Thought some of you would find this interesting....

I did a test tonight and measured the RELATIVE IMD levels of the FT-1000D
10mW output, the barefoot FT-1000D 200W power amp, and a pair of 3-500Z's.
I used an FT-102, well isolated, as a receiver with a pad attenuator so that I could set each signal to a normalized S9+40 over. Each transmit signal was dumped into a Bird dummy
load. The FT-1000D uses an aftermarket 2.8kc ssb TX filter.

I have detailed readings, but here's a rough summary:

When I ran the 3-500Z's at ~1200W out, with S9+40 on freq, the ssb trash was
S9+5 at 3kc up.  At 3.5kc up =S9.    4kc up=S7.   4.5kc up = S3      I could
still hear the trash up 5kc, but not moving the meter.

Now, the FT-1000D barefoot is advertised at -36db IMD for 150W, I believe. This
is running it at 100W out:

On freq: S9+40.
3kc up= S8.    3.5kc up=S7    4kc up=S2     5kc up= barely heard anything.
Better.


NOW... the FT-1000D with the low level 10mW output tap:

At 3kc up I could barely hear any trash! At 3.5kc it was totally gone.  I
would estimate it was upwards of -55db++  3rd order assuming the FT-1000D at
100W was -36db.

I did notice that if I ran the drive levels and audio up so that I was
getting the max 10mW out, it did deteriorate the IMD.  Running everything
about 1/2 full level was VERY clean. [~5mW out]  Tuning across the recorded
voice was so sharp it was like hearing a CW note. In contrast, tuning across
the 3-500Z's showed a definate side channel trash that would show up on a
quiet band.

The bottom line is after seeing how clean an ssb OR AM  linear signal can be, there is no
doubt that I want to pursue building a -55++ db 3rd order IMD amplifier with negative feedback using the FT-1000D 10mW output. It appears WELL worth the effort!

73,
Tom, K1JJ

Tom,
You need to send a two tone signal through the TX to measure the IP3 products.
The synthesizer crud is a clue of how good it could be though. I don't know the FT1000 synthesizer but I would think you need to use a different radio to monitor the crud. I don't trust S meters so a step attenuator would provide more accurate information. The FT102 may have lower phase noise since I don't think it has a synthesizer. A step attenuator ahead of it would be a more accurate way to measure noise coming out of the FT1000. Be careful the FT102 has a lot lower dynamic range
I think I read 84 dB.
It is very hard to measure phase noise and dynamic range. I never trust my numbers and have made many mistakes catching synthesizer spurs and the usual brain fats.
I usually have to get the same number three times on three different nights.
Also there are a few different methods. A cool one is in the HP8640B manual using a double balanced mixer. You used the ARRL method which works. OH to own real high performance test equipment......or at least take my projects to work.
BTW you also have to correct for bandwidth of the monitor radio subtract CF=10 Log BW in hz.  300 Hz filter you subtract about 25 dB from the phase noise measured.
Danger you could get receiver disease playing with this stuff!

OK, Franz,

Notice that this was a RELATIVE test checking the readings between the 10mw output, 200w module and 3-500Z's.  I used recorded voice programming which may be better than tones according to some for getting a rough idea of overall perfomance. [not for raw IMD db numbers].

All I can say is the difference between the 10mw normalized trash and the 3-500Z trash was unbelievable. I'm almost embarrassed to get on a quiet band with that amp, though everyone is using the same thing more or less... :-)

There's no reason now not to build a super clean amplifier that will complement the clean 10mw signal.

BTW, the FT-1000D phase noise is way down... it disappears below the FT-102's own generated noise level. No problem.   It's all a matter of IMD amplifier generated trash after the 10mw stage.


T

yes it should sound better since you are comparing a 10 dBM signal to a 63 dBM signal.
The leakage through the coax shields is a big deal. I've been burned by this many times.  You might sonsider building a directional coupler to put in line with the coax to the dummy load. I actually use a high power directional coupler when I connect a scope to look at the signal. This way you have a 50 ohm source and can send it through 50 ohm cable to a scope loaded to 50 ohms or a step attenuator and RX.
You could be fooling yourself monitoring leakage.

OK about the possible coax leakage.  I will set up some kind of 50 ohm tap off the dummy load and feed it directly into the atten pad and RX.  Do I really need a directional coupler?  What is an easy way to make one for the job?  The only type I have is a common Bird directional coupler for swr readings, etc...

But explain to me this: If the cable is leaking, it's leaking both the signal and the IMD trash at the same time. Since I normalized the pad for S9+40 over on the RX for each test, why would I hear more or less IMD compared to the signal?  ie, why sould the IMD leak more than the signal?... ;D

T

There is more RF floating around at high power. This means there are possible leakage paths say between the attenuator and the 102. Here is a thing to think about. Say you need 10 dB on the attenuator at 10 mw out. How many db do you add to the attenuator when the power is increased to 1 KW. That is a power increase of 50 dB. I would trust your number if the attenuator setting increase in db matches the actual power increase in dB. My point is to be careful in how you measure this stuff it is easy to introduce error.
Also cosider this the FT102 dynamic range is about 84 dB I read. Say S9 is 54 dB above the noise floor and you are at 40 dB over meaning 94 dB above the noise floor. 10 dB above the dynamic range of the radio. You are pumping enough power into the 102 to have third order spurs appear 30 dB above the noise floor. A third order spur changes 3 db for every 1 db increase in input power.
The most simple directional coupler is the mono match. Slip a pick up wire in a coax
and terminate the end toward the load with 50 ohms to the shield. The other end goes to the 50 ohm attenuator through 50 ohm cable. The length of the pick up wire determines how much power is coupled for a given frequency. Couplers can also be built using ferrite cores if you want something more broad banded. You now can use the FT102 as a power monitor ranter than a diode detector. There is a Bird 43 slug that has a bnc sampler output. That is a part I'm looking for.
You could aslo ask Big Al to monitor your signal. He is close enough to hear 10 mw
well. Nothing beats far field as a final test.
I ran into this trying to measure 100 dB of dynamic range. My 2 generators were not isolated enough and blew the number until I added extra pads between them.

OK Franz -

I'll make up a directional coupler with a terminated probe as you suggested.

Before the test, I made sure that when I unplugged the coax from the receiver when running 1200W, there was NO signal. ie, the FT-102 is well shielded with no external paths in except for the SO-239.

BTW, the results I've seen are very much in line with what we would expect. The tansceiver at 150W is a little cleaner [-36db as avertised] than the linear at 1200W, and the 10mW is much cleaner, like approaching an estimated -55db as you suggested it might.  So I will try the test again, though I thnk the relative results will be similar...

The Big Al test 10 Miles away wud be difficult since his noise floor is S5 and a 10mW signal at his place would be down around S8 based upon tests we have already run with Mr Ugly the other day.  So the test bench is the only way.

BTW, can you see any reason why a two tone test would not work as well for an AM transmitter [double sideband with carrier] as a ssb rig?  What two tone frequecies wud you suggest for a hi fi AM rig?

T

Tom,
Your numbers make a lot of sense. My point is you can't trust what you read having burned myself so many times. The trick is to know and control your possible error.
-55 dB of crud off the ssb generator makes sense. Unwanted side band should be down at least that far also. You want to put your attenuator ahead of the 102 and check the s meter.  The combination of attenuator setting and receiver input will extend the dynamic range of the 102. BTW RG58 cable has very poor shield. RG223 is much better. Strong signals on 75 have trashed my numbers due to cable leakage.
I can hear signals with a 4 foot cable through the shield between a test RX and generator.

OK, I found the Collins slopbucket manual.  Here's a few examples how Collins did it.  And no smartass comments or the COllins Collectors Association will send you a lump-o-coal for xmas.

edit - these images didn't turn out too well on the forum.  You can get the original by right clicking on the image, select "view image"

A 3-stage feedbaclk amp ending in a 4X5:
(http://i19.photobucket.com/albums/b153/throbbo/rf-feedback.jpg)

A 2- stage feedback amp, medium power

(http://i19.photobucket.com/albums/b153/throbbo/feedback2.jpg)

And finally a typical layout for a NFB amp

(http://i19.photobucket.com/albums/b153/throbbo/feedback3.jpg)

Tnx, John!

That third diagram is the first one I've seen using all GG triode tubes.

Interesting that they use a neutralizing cap on the GG final and a loop all the way back to the first tube.  And all the power adds to the output... except for the grid current.

I'm trying to think of what wud be a logical choice of hi mu triodes to use. They all require power to drive the next stage.  Nothing cheap. A 3-500Z driving a 3-500Z driving a 3CX3000.

I'm thinking that maybe the tetrodes chain I've described before may be the cheapest and cleanest system to use.  The extra regulated screen and grid supplies are the drawback.


TNX again OM.

T

I worry that with tetrodes you will lose so much linearity that you will lose the gain advantage because of the feedback you will need.  But some tetrodes have pretty good IM performance.

Tetrodes have much lower distortion in ultra-linear mode.  I've never seen this applied to RF linears, though.  Maybe some negative rectified output signal added to the screen voltage (giving somewhat less screen voltage at higher output levels). 

I like the idea of a small impedance in series with the screen to improve linearity, but for this application the top end response requirement is severe, and I'm not sure about screen impedance behavior in a tube with high secondary emission.  With the secondary emission of a big tube, you would want a screen choke with a medium value resistor across it (I'm guessing around 1K), and the iron and stray capacitance of the choke would limit the top end response of the feedback.  This may be a job for a small choke with 100 KHz response, in series with a bigger one good down to 300 Hz or lower.  Low frequency rolloff won't matter much, because it will only affect the distortion reduction at low frequencies, not adding significant bandwidth to the output signal.  But high end rolloff will kill the distortion reduction.  The RF bypassing of the screen would need to be minimal as well.

Because of the above issues, it might make sense to use an active screen driver for this.  Possibly a level detector looking at drive level, possibly some diode-break linearity adjustments in the baseband feed forward signal path, feeding a screen modulator for the final, producing a small amount of screen modulation to tweak linearity.  This would be baseband signal and would not require large or expensive power stages.  Then with the large gain of tetrodes, you can put some RF feedback around the whole thing.

The Collins book has some tricks to greatly decrease IMD on tetrodes.  One is to feed both the grid and the cathode with a capacitive divider, and ground the screen.  Monday I'll try and copy some of the book and send it to you, Tom.

Great, John -

Each new batch of info seems to help.

One of the gurus has mentioned that I'd be better off finding a replacement for the 6146. He said some of the audiophile guys might know of a cleaner 50W tube to use.  Anyone here have a suggestion?

He also thought that the 4-400A, even running in class A was not good enuff. Said to stay with the 4CX-350.  I don't have any, so will have to look around.

Sounds like lots of ways to play with the screen, etc as Bacon has said. I think I'd rather stay with conventional neg feedback and super clear drivers to get there. Unless I had a real whirl circuit that worked. Otherwise I'd be experimenting and blowing parts up like my last attempt to get an 813 driving Mr Ugly to work with feedback. That was a bust.

I may start on the low level 10mW > 1 Watt module first and get that clean and move up from there.  He said I wud need to get that at -75db IM3 to make it worthwhile.  IMD products do not ad linearly, so need to be 10db cleaner than the next stage.  This is going to be harder than I thought.... ;D


T

Tom,
I don't know the specs right off hand but have you checked out the 4CX350Y's? I seem to remember these being around all the fest tables a while ago. Most were new. The only thing I didn't see was the socket they were to be used with. I guess they are the commodity. But if you could get the socket, I don't think you would have a problem with finding a tube.

A 4CX1500B into a pair of 4CX1500B's into 8 4CX1500B's.
Hey, the CB amp builders have it right after all!

Dave Boot had tubes and sockets at Hostraders.

In receivers hard class A gets things linear and feed back sets the operating Z and makes the stage stable. The Norton stuff I sent you from Dallas is by far the best performance I have measured in a RX amplifier. I have a push pull stage with a pair of 2N5109s at 60 ma each that would need amplifiers on my generators to determine the IP3. Your 10 mw could become 1/10 watt. A second stage with 2n3375 with even more bias could get you to a watt. Holding -75 dbc third order takes a lot of bias. at 10 mw out you are talking 50 dBM third order intercept point. at 1 wATT out
and -75 dBC crud you are talking IP3 of about plus 80 DBM. Class A running at 100 watts to make a watt might do it. Our 1 kW amp at work is running 1000V at 1.25 a
resting and about 7.5 amps to make 1 KW out. so figure you will need to run at least 10 times the output power as class a bias.   and maybe more in the lower level stages. Tom Vu will have a warm shack this winter.

The 6550 is apparently a much more linear toob than a 6146.  Central used 'em in the 100V and 200V.  You can get Russkie 6550s pretty damn cheaply,too.  I'd be willing to donate a Russian 300B triode to the cause, T - prolly good for 10-15 watts in Class A service.  I built a P-P audio amp outta those bad boyz, <.1% THD and IMD which is what, about -60 dB from a pure tone.

John,
was that .1% voltage or power distortion? Power would only be 30 dB.

All those special amps aren't going to make you one dB louder or allow you to work one more station.

total harmonic distortion (THD): Of a signal, the ratio of (a) the sum of the powers of all harmonic frequencies above the fundamental frequency to (b) the power of the fundamental frequency

6146 is the LAST toob you would ever want to spec for a linear amplifier. The only reason these were used in slimeband rigs was marketing. Hammy Hambone went around saying that sweepie toobs are no good and the Japanese rigs were junk.
6146 was designed for class C use.
6146 works well in class C.
6146 works fine in linear service with R.F. feedback.
Nobody designs homebrew amplifiers using R.F. feedback (except for what you get with cathode driven designs) anymore except Tom Vu.
If you absolutely have to have a 6146 for some reason then consider lower B+ loke 450 Volts.
I dislike 6146's as linear amplifiers.
Repetition can be used to make a point stick.
I'd rather eat worms than drive a 6146.

Dave,

So, 6146's are your favorite tube, huh?  I've already decided to go with a 6550 instead of the 6146. It must have been a brain seizure, solly. The 6550 curves are quite good and they are cheap on the web. I'm surprised the audiophools haven't run them up too badly yet. I still see some cheap ones.

I measured about 5mW coming out of the FT-1000D low level.. very clean. I wud estimate at -75db 3rd order after hearing it again today. Then driving an OPA695 chip for 20db giving 500Mw. That chip will do -75db 3rd order according to specs.

Then into the 6550 class A, driving the 4CX-350 class A [gotta find one] and then into the 4CX-XXXXX  or whatever, AB1. Heavy Neg feedback. That's the plans as of now. Gonna start on the low level chip stuff soon.

Steve - you axe, why bother for a cleaner amp?  This is mainly for the 75M DX window. At grayline and late afternoon, the signals from DX stations are like S3. Low noise floor. The locals are S9+50 over. Tune 4 kc away and the local is still S9, covering up the DX station EVEN if he is clean. That's with these -32 to -40db 3rd order amps everyone uses. Most are far worse. I've been tuning across signals lately and can pretty well estimate IMD from experience playing with my own signals in my RX. So, I'm gonna TRY to raise the bar a little and show em how it's done, caw mawn.

The last time I tried with Mr. Ugly and an 813 - when I added the neg feedback I blew up half the amplifier. Tried again and kept blowing parts. But I'm gonna try again, this time with a little more Vasoline.  If you like, you can be on the telephone when I key it up.

T

I know you are aware of this Tawm. Still the unwashed masses want to hack sweepies out and replace with those damnable 6146's. Did I mention that I dislike them for linear service???

Da mud ducks ain't never gonna respect yo wid a clean amp like dat!

You tell those Mud Ducks there's a new Cash Heat Sheriff in town. He's pissed and looking for Skull Krushas, Muddy Wata Slappers and Big Boomers to body slam, caw mawn.

 

This great thread you start Mastah Vu.

When I built my 300B audio amp I used a Heapshit IMD analyzer and a British (can't remember the brand) THD analyzer.  That's a good question, voltage or power.  These instruments measure voltage across a load, which implies that they measure power.  But I guess it all depends on whether they use a 10log scale or a 20log scale.  At any rate, .1% is considered Fine Business for an audio amp and most agree that anything under .5% is undetectable by the human ear.  However, as we know, most receivers (except 51S-1s and SX-117s) have much more dynamic range than the human ear.

Tom Vu the audiophools definitely know about 6550s, but mostly for replacement stock.  Ever price a Tung-Sol 6550?  I scored a quad of them at a hamfest for a quarter a piece, I'm counting on them to fund my kid's college.

Like everything associated with Art Collins that SSB book is very pricey.  I saw one on ebay a few days ago for $100.  I should have invested in real estate in Cedar Rapids; I'm sure the Collins Collectors will discover that next  >:(

73 John

Yup John, Tom Vu thread ties right in with my receiver disease.
We all need a 1/2 a mil of test equipment in the shack to really know what is going on.

Yes, this can be an interesting thread.

Once all of us get on the air and happy with the basic rigs we have, then what's left but to talk -    or later go back to improve upon the rig's performance?

It will be fun putting this new amp on the air for AM too - driven by the clean 5mW from the FT-1000D on AM.  The 1000D's output uses a direct connection to the balanced modulator, with no audio restrictions. If it's anything like the ssb test I made, after the audio frequencies, the "crackle" side trash should drop off into the abyss within a kc or so..  A TX amp is easy to demonstrate whereas a receiver needs a personal visit..

Marv/WC6W has been suggesting using a "feed forward" system for the FT-1000D 200W power amp. He seems to feel it would be all outboard.  ie, sample the 5mw, and feed the 200W module of -32db 3rd in and get 50W of clean -65db+ 3rd  out. I gotta check more into that if it's just a mater of building a "goesinta, goesouta" box.  All I see on the web are complex block diagrams from the ivory castles with no practical circuits or values.  So, it might just come down to starting at 5Mw and going up from there with conventional tubes, etc..

Frank has offered to design up a a pair of 2N3375's pair running a Norton config to get the 5mW to whatever is needed for the 6550 tube.  He thinks that may be the cleanest way.

The thing with this project is that IMD doesn't add linearly. ie, a -55 stage driving a -50 stage is not -52.5, but rather something less than -50.  Each stage must be at least -10db cleaner than the next one.  The real IMD problem maker will be the final tube that needs to generate power. I hope to add a cathode NFB resistor there AND the highest NFB loop around it to the driver I can get away with. If the final tube will do -35db spec alone, then adding -20db will bring the final up to the desired -55db IMD goal.

It will be fun starting from low level and proofing out each stage. I need to set up a 2 tone generator and directional coupler for numbers tests.  But first more planning and thinking to do.

T

Tom look at the norton schematic. All you have to do is change the emitter resistors to get the resting current you want. 3375 want to run about 100 to 120 ma. each.
TO5s start getting pretty warm at 60 ma. The 3375 is a stud mounter device 10-32 thread. 3632 is another good one that does more power.

Frank,
Is there an advantage to using a 3.8 mhz resonant tank(s) with this Norton circuit or do you use resistors or ferrite transformers and make it broadbanded?  Maybe an output tank wud help cuz the Norton may be 20' or more away from the big linear input, I dunno. I see your info used broadband transformers in and out.

BTW, what schematic would you suggest I build? Maybe you sent me one, but all I have is the multi page article on Norton amps...

I'm not sure yet if this linear amp will be a mono band 75M unit or will go form 160-20M.  It all depends upon the feedback used and other factors.

T

I would stay with the transformers for broad band feedback and impedance control. You could hang a low pass after it. Harmonic energy should be low but a low pass wouldn hurt. Check the aritcle again there are schematics. I can resend if you need me to. I did the 10 dB gain push pull with 2N5109s. That is the only one I tried so far.
The next logical power level 1 watt or so would be an MRF type driver transistor like a MRF466 or MRF 433. BTW Pete SOV uses a CTI driver 2 stages class A in his Rig.
The bias had to be cut back because it over heated but worked fine. Check out the CTI thing. TO220 transistors. MRF 475 and 472 if I remember. They also have a couple others worth checking that were clean. The 2N3375 is older but should work.
The trick is to use a device with high Ft. The 2N5109 is 1800 MHz Ft if I remember.
The gain of the stage is set by the transformer ratios. there should be a chart.