The AM Forum

Hi All,

I've never seen 4-1000A GG linear specs showing IMD figures. Eimac doesn't publish them nor have I seen a ham post them.  

Today I ran some tests using my 4X1 in GG, 3500 volts, idling at 150 ma,  and driving with a clean FT-1000D.  At lower powers, the FT-1000D  shows near -40db 3rd IMD, so it is a reasonable driver for this test.  (The FT-1000D is about -31db at full 200w power)  I was using a 2.8 kc ssb transmit signal and measuring 3kc or more away.

I used my HPSDR SDR receiver with a 3" wire antenna as a spectrum analyzer to sample RF and kept the receiver level as low as possible to eliminate overload, etc.

I ran three types of audio programming to see the worst results - a two tone, a pecker pulse and my own voice thru the audio to get an average of various IMD conditions.

* These results apply to either ssb or AM linear operation.

The results were surprising but expected in some areas too.  On average, putting out 2KW pep, the 3rd order IMD was about -31db.  At 1200 watts out it improved to -33db.  At 750 watts out, it was very close to -35db!  At 750 watts it is barely out of class A and certainly shows it in cleanliness.

Bottom line is the 4X1 in GG  is acceptable at 2KW pep out, being near -31db 3rd.  At 1200W it is a very clean amplifer compared to most modern SS transceivers on the market. At 500-750 watts out, it is a super star.

It was interesting to see how the general noise floor out 5khz or so would come up as the drive was increased - and how the main signal started to compress and become less linear.  

By increasing the voltage above 3500 volts may or may not have much effect on IMD, but it might be worth a try.  Maybe more plate voltage will produce less grid current and allow the amplifer to run at a higher power level for a given IMD.

I was also surprised that by loading the amplifer VERY heavily (less C2 loading) that this had little to no effect on IMD.  It just reduced the power and was the same IMD when brought back to the original power level.  I found that it was best to peak the power out using both C1 and C2 and then slightly decrease C2 (heavier loading) so the power dropped maybe 50 watts to be sure we are on the correct side of the loading and transfer curve.  Again, adding further loading past this point had little effect on IMD.  It is still a valid practice in most cases, however.

I want to try a 3-1000Z (when I find one) and see if a tube designed for linear service is better (IMD) than the 4X1 in GG.  But bottom line is the 4X1 in GG, run at a reasonable level, is an economical contender.

This again proves the old philosophy that to run a clean signal in linear operation, use a tube that is at LEAST twice as big as you intend to run it for power.  For example, for 1500-2KW  out pep, a 3CX-3000A7 loafing along would be a great choice.   Sure, we can run a 4X1 in GG at 2KW out, but we pay the price with more side splatter and this can be a problem when the band is quiet and crowded.


More tests later when I get a 3-1000Z to try.  Anyone have one to sell?  :-)

Tom, K1JJ

Cool stuff, Tom - thanks for posting it!  :D

Thanks Todd.

I think over time, general ham web searches may find this thread of use.


BTW, after running more tests on my other amplifiers, I am finding that the limitation in my setup is the FT-1000D driver!   Whenever I pull more than about 10 watts of drive from it, the IMD drops quickly from -37 down to -31db.   This probably has a lot to do with why the 4X1 IMD is dropping off so fast at higher power.   The 4X1 amplifier may be cleaner than I first thought at higher powers.  I will add a clean IPA and see what happens.

So, at least the figures I posted are minimums - and they show that the 4X1 is a pretty decent tube in GG.

T

Despite the issue with the exciter's distortion, the 4-1000's distortion may also have to do with the plate voltage or indirectly the percentage of the amp's rated power that is actually used. The Elenco Commander had a GG 4-1000, 6 KV on the plate, and was rated 4KW PEP input (1500 AM), with some of their advertising literature saying the higher plate voltage in combination with running the amp at normal 'ham' levels, helped to keep distortion lower. It was made from 1958-1959, perhaps a magazine ad or article could be found to support this.

I don't recall where I saw the ad. Unfortunately some amateur mags got a little skittish when gear was able to blow past the limit and there are fewer ads for ham-purpose things like this. A few more though for commercial gear like TMC and other big stuff.

This aligns with your measurements at 3500V anyway, lower distortion when demanding a lesser percentage of available power, even if it was the exciter's amp.

I'll bet you could get the 4-1000 distortion down to the level of a 3-1000 by applying screed and control grid potentials.

I heard some W6 dude going on and on explaining to a new ham that he should always run his transceiver at 100 watts and not less in order to avoid distortion............................

I just saw something somewhere yesterday that claimed you shouldn't run a modern 100 watt rig at 50 watts or less because it could destroy the finals.  ::)

' I heard some W6 dude going on and on explaining to a new ham that he should always run his transceiver at 100 watts and not less in order to avoid distortion............................ '

ya shuda broke in and said that " It's O.K. if you use Bird Watts. "

KLC

thats the watts I use

It's not all necessarily BS. I've heard some fairly reputable people claim the IMD on certain rigs was get worse as the power was lowered. This had something do to with how it was done in those rigs.

Interesting.

It's amazing to me that in this day and age, most modern solid state transceivers (riceboxes) still exhibit -31db 3rd order.  Most except for the rare exceptions that permit class A use at reduced powers.  The bottom line is if a transceiver is only -31db, then no matter how clean the next linear amplifier is, the cumulative IMD will always be LESS than -31 db. For example, if the linear amp is -40db and the driver is -31db, then the overall result may be -29.5db or whatever.

Yes, some rigs do degrade as we reduce power. I understand that bi-polar SS finals exhibit better cleanliness when reduced in power. I know my FT-1000D does. Whereas MOSFET finals appear to degrade or stay the same as we reduce power.  The 1000D uses bipolars. The Ft-1000D, running alone, is clean as a whistle, like almost -37db 3rd at times until we increase to that 10 watt level and it slowly degrades to -31 at 200w.   The whole noise floor starts coming up out +-5kc using a pecker pulse source.  This is normal.  So no matter how clean our 4X1 or 3X1 or 8877 amp may be, we are fooling ourselves with marginal drivers.

With MOSFETs and possibly some configurations, I think as we reduce power to very low levels  the cross over distortion become more prominent.

I am thinking of building up a class A linear to take the clean 10W 1000D level and go to 100 watts. Maybe a single 3-500Z in class A would do it. This wud give a clean 100W signal (-40db) to drive ANY amp in the shack - and let it shine.  A 3-500Z at 1500V at reduced power is said by Eimac to approach -45db+ third.

The 4X1 with higher voltage will exhibit more gain, thus will require less drive, which is a good thang. Though, I have been warned that anything over 5500V produces indirect electron bounce?  problems which can cause distortion.  The 4X1 sure is a robust tube and I have a good supply. Wish I cud find a 3-1000Z to try in its place.

I'm still toying with the idea of adding MORE  negative feedback to the GG stage. I understand that adding a small resistance and parallel cap to ground from the grid is the way to do it. I have the grids strapped to ground right now and prefer this for stability..  

But why won't a 100 ohm resistor in the cathode work as degenerative NFB, just like a single ended audio stage? It must have something to do with the GG configuration, no?  I once tried it and thought it didn't do much, but cud be wrong.

T

Tom,

A cathode resistor in series does work.  Denny Had used it on the Dentrons, and I've heard of it on a few others.

10 to 25 ohms is the value that comes to mind.




On a different note, what type of mod xformer are you using on the 2X1?  I'm contemplating a pair of 4-400s modding my 4-1000....

--Shane
KD6VXI

Thanks for the info, Shane.

OK on Denny using it in his amps. Now that I have a spec analyzer set up, think I will try the cathode resistor again and see what happens.  Now, does it get bypassed with a capacitor of some value or does the resistor sit in series alone?  I also have a string of diodes in there as cathode bias.

On the 4X1 plate modulated by a pair of tetrode-connected 4X1's, I am using an RCA 1KW broadcash mod transformer. I think it originally used 4-400As.  It weighs about 140 pounds.

T

Here's a link to Measures and a few others smarter than I disguisting it...

http://groups.yahoo.com/group/ham_amplifiers/message/6143?var=1



I KNOW I did a virgin MLA2500 here that had both R in series and shunt across the input. Don't recall, and don't have the old work orders to find out if it was an A or B. The shunt was the gain killer (to keep within rules and regs of the time I'm supposing) and the series served two purposes:  It increased input R while dropping distortion.....  Sounds like feedback to me.. :)


--Shane
KD6VXI

I've seen and heard the same thing. It could easily be true for certain circumstances. Problem was the "elmer" making it into a blanket statement.

Barring gain control loops and quirky bias schemes more headroom should always be cleaner.

Mr T,

With the amount of 4-1000's, iron, aluminum, and plexi for wide open viewing youz got ..

I say a PAIR of 4-1000's in GG.. Loaf with low 3rd order..

Yours in amplitude,
Jeff

That's actually a great idea, Jeff.  Running a pair at 1500w out wud be as good as it gets. The challenge wud be to find a driver as clean. A 32S3 uses negative feedback with its 6146's IIRC, so that wud work out to be an easy -40db 3rd driver..   

The problem with two tubes for me is that the amp is already built and can barely fit one tube. The fil xfmr is also just for one, etc.

The good news:  I just bought two used 3-1000Z pulls today with a money back guarantee if they are not good. I pick them up tmw locally.   Can't wait to see how they do.   Now I can take my time looking for a NOS 3-1000Z.


T

T,

NOS
http://www.ebay.com/itm/NEW-3-1000Z-in-EIMAC-factory-box-8164-NOS-boxed-HF-linear-transmitting-tube-/170954958669?pt=US_Radio_Comm_Tubes&hash=item27cdb5a74d

you could build a few amps up for that price!

Have fun, and great info BTW!

Jeff

Yeah, I've been looking at that one for a few days now, but it's in Greece- plus he wants a left BA for shipping.

These kind of things work the same way... suddenly 3 or 4 new ones will pop out of the woodwork for a good price all at once and I will regret buying the two pulls... :-)

T

I looked for an inexpensive 3-1000 for a couple years for my Harris RF-103.

It's currently using a 4-1000.  The 3- tube is about worth it's weight in gold.


--Shane
KD6VXI

I guess they will eventually find their way to market as time goes by.

Well, I just ran some tests using a 50 ohm glowbar in series with the input to the cathode of the GG 4X1 linear. (after the tuned circuit) It should give about 3db of improvement in IMD.  I had a hard time seeing any difference.
I was told that it just adds attenuation to the input signal, no NFB.  My conclusion is that it's probably not worth it considering it's only 3db and we would need to double the drive power to make up for it. <aybe I'll try the resistor series cathode method later.

I expect the 3-1000Z to have another 3db of gain over the GG 4X1, so it might make sense to add the 50 ohm 3db IMD improvement later.

Using a big tube and reducing the power output to 1/2 it's capability is probably the easiest way to grab a quick  IMD improvement.  I will revisit this later once the levels are figgered out.

T

I'm trying some different audio inputs for the IMD tests.  I had been using the pecker pulse and voice for general measurements.  I find that these audio sources, especially the pulse pecker, tend to produce more generous IMD numbers.

I just ran some serious two tone, three tone and four tone tests from a computer generator and find the results are much harsher.  The two tone shows the 4X1 in GG is at least -31db 3rd, but rarely gets above -34db 3rd for most tone combinations.  I made some changes in my numbers for past posts to reflect this so the archives will be more accurate.

Still, it seems the limitation is still in the 1000D driver as usual.   So far, the 4X1 does not seem to be adding any meaninful IMD degradation to the overall signal, which is a good thang.  The 3-1000Z tests will be interesting to see tomorrow.

T

The highest B+ Ive run 4X1 IMD tests on is 4500V and yes, it does improve IMD when tuned up for 2300W output and the Load cap set for a bit higher loading until the output power is reduced 50W or so.

The IMD 3rd runs in the -33 range at full bore and gets better with less drive until the TS-940 output is below its 75-80W out sweet spot where it is in the low -40's and drops to -36 at 120W. The 4X1 is then around -36 at 1500W. This was with an almost new tube that came from some scrapped industrial gear; the usual tired pull likely would not be as good. I do have a NIB 4PR1000A and a 7KV PS with a Variac for whenever I get another urge. 8)

The TS-950SD is a bit better for IMD plus it has 150W specs but I didnt have it back when I tested that amp back in the late 80's and I dont like pushing the 940 past its usual 100W for more than some quick tests.

I believe I also used the generic ARRL 2 tones back then which is really cherry picking for best IMD.

My 4-1000 GG amp ran 6900 volts from a 2.5 amp PD transformer.  I would like to see more voltage and the results.

I see now why the top line rigs have the Class A option.

Tom, I wonder what kind of input the amp has?  Tuned input?  Fixed 50 ohm?

Cant wait to see the 3-1000 numbers.

Good to see you back Shane.

C

OK on the comments, Carl and Clark...

On a friend's suggestion, I did the glowbar in the input lead test again with the more stable two-tones, looking to add NFB - and am hard pressed to see any difference except for input attenuation. Just wanted to set the record straight. I will try the series resistor in the cathode and see what that does for NFB.  

Clark, I am using a standard pi-network input to the cathode. Variable C1 to ground, L1 roller in series and a second variable C2 to ground. Works on all bands for a 1:1 match and supposedly a better IMD via flywheel effect, if that really exists. Your 4X1's must really sing with that voltage.  A good layout needed, thats for sure.

Carl, looks like we are seeing similar IMD numbers - at least when I am using the two-tone test. I was getting fooled by the pecker pulse numbers which always looked better. Why is this so?  I would think a pulse is rich in harmonics and crud.

If a pecker pulse is hitting the same peak power at a quick 10% average power pulse, why would a tone-tone at a more constant 50% power average create worse IMD numbers?   I would think the tube would see the same internal dynamics of non-linearity whether the audio is a low average or high average. But that's what I am trying to learn...

Today I decided to pull out the 4-1000A GG and replace it with an 8877.  (external anode 3CPX1500A7)  The 8877 will be able to loaf along even better than a 3-1000Z with 1/2 the drive. It's a higher gain tube.  It probably has better IMD specs too. There are now Chinese 8877's called 3CPX1500A7's on ebay for <$250 these days. The word is they work as well as the Eimacs for both power  performance and IMD.  I already have an Eimac 8877, so will try that for now.

So, I will not be buying the two 3-1000Z pulls and instead going with the easily obtainable 8877.  I'm not gonna  play the unobtainium tube game with 3-1000Z's anymore.


More later -

T

Limit the frequency response especially the low end to cheese out some lower IMD.

Same issue I found.  I'd love to have the glow, and the  gain of the 3-1000, but at the same time, it's hard to spend more on a tube thats already absolete when for a lower cost you can run a ceramic.  I've got a 3cx1000 here that's going to go into the harris when the 4-1000 is a dud or gets converted to the plate mod'ed rig.  Not to mention, it takes me 120 watts to get the same Pout that 60-65 gave before.

If it's the Greenstone tubes, I've had good reports on them vs the Eimacs.

Clark, good to be back.  A full strapping 10 watts of carrier!  My antenna is good for DX, locally it sucks..  But, TimTron comes in during the daylight hours full quieting.  Listening to you on 40 right now, matter of fact :)  Just made a contact with you, a full 10 watts carrier!

--Shane
KD6VXI

S9 into PHX Shane on 10 watts.. I was on 18 watts from the FT101EE.

C

Below are a few comments from Frank/ WA1GFZ. He's been busy lately but was nice enough to send me an email with the best explanation I've heard of why two-tone tests are used to measure IMD vs: other methods like a pulse pecker, etc...

T
----------------------


Frank:  BTW the ARRL two  tone test gives you an automatic 6dB better number



Tom:  
OK on the glowbar .  I did the test again and see no difference this time. It's tough to get readings that close (3db) I guess.


Frank:
There is no @ucking  feedback loop with a resistor on the input

 
Tom:
Why doesn’t a 10% avg pecker pulse, with wide band crud, work as well as a 50% two-tone?   Does the avg power have an effect on IMD?

Frank:
The whole point of the two tone test is to mix the signals which generate mixing products when the stage goes nonlinear. Remember you can’t plate modulate a class A final because you need the non linear final to generate the upper and lower sidebands.

When testing a linear amp you don’t want it to go nonlinear so this is a method of measuring how nonlinear the stage is.  As you increase the power input of the two tone test at some point you hit zero which generates IMD and then you flat top again generates IMD.

When you have no IMD the two signals go through the amp clean with no mixing products.

A pulse is just gating the amp on and off, it will generate distortion when saturated but also generate imd when pulse turns off. A non linear input so it is useless for IMD. A good test if you are building a radar system though. There is a test method for that in the spec I sent you.

-----------

Makes a lot of sense to use "two tone" vs square pulse for a distortion test.

Gating is good to quickly tune an amp for peak power into a dummy load.

A nicer test keyer or pulser for tuning a linear amp would shape the drive pulse somewhat Gaussian and avoid the sharp amplitude edges that can generate a plate current mess well outside the most efficient pi network bandwidth. - It's still only good for tuning, not distortion measurements.

Yes, you have to have TWO tones to intermodulate.

Just remember that the two-tone test is a static test. It shows the IMD at a given power, so it is FB for characterizing tubes. But it says nothing about the dynamic regulation of the power supply which can have some impact on the real world IMD of an amp (the input/output doesn't remain constant). Dynamic or transient intermodulation distortion measurements are more common in the audio amp world than the RF amp world.

Throw in the ALC effects in some transceivers and the actual on the air performance is probably far worse in many cases than the lab IMD tests.

Good point. Yep, I agree about the dynamic test as well as testing the WHOLE station infrastructure.  That's why in the past I like to play through audio voice tapes and tune the band up and down looking for crud levels 3kc, 5kc and 8kc away.  It's amazing how much trash can be generated by a raspy "Yallo."   It's all relative, but this is a good way to check a transceiver and then add on the linear and see how much it changes the crud level. (Assuming the signal is normalized and no overload)    If there is a sagging power supply, ALC, audio parasitics, or other dynamic problems, it will show up with a voice program acid test.  

Actually the pecker pulse test is not a bad test to check relative side channel crud levels cuz it is a constant level and makes measurements easy. Also, we can run the amp full peak power at only 10% duty cycle - less heat.  I can easily see the effects of loading, overdriving and other parameters using the pecker side trash.  I'll bet a white noise generator would work well too.

But to get a standardized measurement that can be compared to the real whirl, we need to run the two-tone to compare apples to apples.


* Pulled out the 4X1  and got the new 8877 mounted in the chassis and wiring it up now...

T

Tom, et al,

***I'm probably in over my head on this one but I'll give it a try.***

I believe there is a relationship between IMD performance and the 1dB compression point of amplifiers. Researching this is giving me a headache as all the papers that I can find about it are full of difficult (for me) mathmatical equations so I can't put a number on it here.

Regardless, I was wondering if it may be useful to do 1dB compression point tests in evaluating an amplifier as the test method is easy and not subject to the IMD performance of the driving source. 1dBCP measurements are simple, require less test equipment, and reliable. This may be an aid during amplifier tuning/adjusting with a two-tone test done afterward to confirm the results.

For those without the equipment to do two-tone testing it may be the only way to guestimate an amplifiers IMD or, at the least, to know how hard they should (or shouldn't) drive it.

Just throwing it out there.

Don

~~~
Reference sources for those with a willingness to subject themselves to unpleasant experiences.
(Take 2 aspirin and get out your slide rule.)

"A Simple Technique for IIP3 prediction from the Gain Compression Curve"
http://wami.eng.usf.edu/Conferences/WAMICON/2005/Electronic-Materials/posters/tp-1.pdf

"Gain Compression or Expansion, Distortion and Other Large Signal Power Amplifier Related Phenomena"
http://www.av.it.pt/nbcarvalho/docs/cn4.pdf

Which is another reason to lease or borrow some lab equipment.

All this work could go on for a while. Later on Mistla Vu find out his existing test method not so fb.

Hehehheh.....   Well, at least there's progress of some sort.  Before I used to call the Huzman and axe him to listen and see if he hears any splatter.  At least now I have an idea what is going on in a relative way.


Interesting on the compression tests, Don. Yes, I can see how that could work because if both a driver and amplifier are perfectly linear, there would be no compression throughout the full range - until it ran out of headroom - and so forth.

As I sit here typing, the new 8877 is idling for a few hours burn in before I throw the HV switch and test it out. I have high hopes for better IMD with this tube.   But I think going from a glowing 4X1/ 3X1  to a "no soul" external anus 8877 is a lean to the dark side on this BB....  ;D  (pun intended)

I did like the 4X1 cuz it's partly radiation cooling and let me run the blower way down. The 8877 needs all air and is louder.

T

Well, I fired up the 8877 (3CX-1500A7) and ran some tests. This is a 1500 watt dissipation tube.  It is in the same sub-chassis as the previous 4X1, but I redesigned the tank values, etc. plate  Q=12 with input cathode tuning.  

This tube is definately a cut above the 4X1 in GG for linear service. At 3500V it put out a robust 2500w pep.  It is very stable with not a hint of instability. At about 16db gain, it takes just about 35 watts to make 1500w out.  This means the driver can run at a lower level to improve cleanliness.

I ran it at around 1500w output for these IMD tests.   I took notes on the various IMD levels compared to  the 4X1 GG  and also the transceiver barefoot, so I have relative numbers.

Using voice, I see at least 5-8db better/lower  crud levels above 3kc than the 4X1.  The average noise floor is lower when speaking.  The pecker pulse side crud measures out to be about 10db better at times. I was seeing the side crud above 3.1kc down about 45db in many cases.

The two tone test showed about 38db 3rd. My 1000D at 200w and the 4X1 GG at 1500w were much worse than that with the two-tone at about -31  to -33db.  

So, bottom line is I will probably never change back to a glass tube again for linear use. A similar tube, the 3CPX-1500A7, is on ebay for under $250 these days and these inexpensive tubes are supposedly good according to web posts. No excuse now. The 8877 I am using is an old MRI pull I got for $300 ten years ago. It puts out 2500w in my 6M amp, so it's got good emission.


These are preliminary tests, but I am a happy camper so far.



More later.

T

Pro Tek ?? Built in a hot rod shop?

I wouldnt be too concerned about any IMD that is at least in the mid 30's as its usually buried in the band noise at least for 20M and below.
On 6M it can make a big difference.

Yes, 6M is a bear. Even "clean" signals can be heard far up the band when the noise floor is S1.


Frank brought up the point that it may not be a fair test between the 8877 and 4X1 cuz the exciter needs more drive for the 4X1 -  so the 1000D is not as clean.  This is true.  However, I did run a number of tests with the 1000D at 10 watts drive for both amplifiers. Even though the resultant output power was higher with the 8877, it was cleaner than the 4X1 GG amp.

Heck, today I thought that using a Collins 32S3 as an IPA - pair of 6146 finals with heavy NFB could be an interesting way to get a good 100Mw to 100 watt  -45db 3rd driver.  I could use the low level output of the 1000D and use the 32S3 as a linear.   Increase the NFB a little and we're there.


T

Ive never heard of Pro Tek and Id be suspicious of ones labled Made in USA since Eimac is the only source here.
I just noticed they have 3-500Z's marked PRC ;D

http://www.dandcelectronics.com/protek.html

Pro-Tek®  is a wholly owned and operated subsidiary of D & C Electronics Co. in Brooksville, FL. Our products are designed, engineered, and produced in the United States and other locations around the world.  Final testing and quality control is done at our facility in Brooksville, FL.  Our products are being used in almost every industry including but not limited to broadcast, military, industrial heating, medical, and scientific research. Many OEM's and end users around the world have come to rely on our products outstanding performance as well as our unparalleled customer service.  We build a full line of high performance triodes, tetrodes, and pentodes along with the smaller audio and ham radio tubes. We also build sockets and socket repair kits, and can custom built to suite your application.   

Jim
WD5JKO     

Take a look at the IMD transmitted signals from a 32S3, regular ricebox and a class A FT-1000MP Mk IV driving an 8877.

Looks like the 32S3 with NFB adds about 6db over an FT-1000D.  The regular ricebox looks nasty.   But that class A and 8877 combination is quite outstanding.   The MkV, class A, barefoot, is a dream. (5th order down -70db!)
 Why don't any of MY solid state amps work that well in class A?

T

Go down about 1/3 of the document: (starting at page 17)

http://www.sherweng.com/documents/NC0B-Contest-U-2008-9.pdf

WOW..  Nice Data Tom.  That thing sounds clean with the Class A radio!   

C

Translation: A rebrander using NOS or low use Eimac tubes that have been washed and relabled. Otherwise they are from the PRC with our own brand painted on.

Wasnt there something in the news recently about a US outfit selling Chinese junk to the US military rebranded as Eimac, etc ?

Tom,

    I have been playing with the 8877 tube for a number of years with a Dentron MLA-2500 conversion from two 8875's to a single 8877. In the process I've learned a few things on what not to do, and as a consequence I have had two Eimac 8877's fail on me.

    Mistake #1: I used a filament dropping resistor to drop the 6.3v (for 8875's) down to 5v, but it was a bit high, at 5.3 volts (Standby). It would be better to be closer to 5V standby, and then maybe 4.8v at full strap. Lots of debate on proper filament voltage, but for sure too high is as bad as too low.

    Mistake #2: Dentron in the MLA-2500 only drove the 8875 cathodes. The filament had one side grounded to chassis ground. It is better to use a trifiliar filament choke and drive the cathode and filament pins common mode. Put a .01 uf cap between the three pins at the tube socket. Don't ground the filament supply, let it float at cathode potential instead.

Why did my 8877's fail? Each time the amp was in standby, and I heard a flurry of sparks inside the amplifier. 

My theory: Dentron uses a 47K cathode bias resistor in standby to provide cutoff bias to the tube. So, in standby mode, that might be 200v bias on the 8877 cathode. With the filament grounded, and with the filament running HOT (5.3v), a heater to cathode short will result in zero bias, turning the AMP ON when the PTT relay is not energized. The AMP takes off in a parasitic. This even bypasses the zener diode for setting 8v bias.

So what happened to my tubes? In both cases the filament opened. On the 2nd tube, it took out the 0-1 amp cathode current meter. I found that I can still use the tubes by first making a fixture using a 12V charged 100,000 mfd capacitor across the filament pins, and then tapping the tube with a ball peen hammer until the filament connection at the base makes contact. At that point it spot welds itself, and the tube works fine so long as I bring up the AMP with a variac. Once the tube is cold, if I turn on the amp suddenly, the filament opens again, requiring me to pull the tube and repeat the spot weld process. So far I've done this many times, and both tubes can be restored to 100% functionality.

   Also you might look at the Rich Measures low Q parasitic suppressors:

   http://www.somis.org/Price-Info.html

Jim
WD5JKO

Jim,

Great info on the care and feeding of this tube series.  The trifilar choke is something I would not have initially considered.  Most of my experience has been on grid driven class C amps.

Joe, GMS

Jim,

Thanks for your experiences using the 8877.

I just went thru this whole thing about cathode and filament feeding with Chuck, K1KW.  He recommended the same thing - instead of a DC connection between one side of the fil and cathode at the tube pins, to capacitively couple both fils to the cathode with .01's.  And that's what I did.  

BUT, I grounded the center tap of the 5V fil xfmr.  I was worried about the DC difference between the two as a result of the 47K keying resistor and bias diodes in the cathode. Without a DC connection at the pins, how do they equalize for DC?  I understand that there is a 150 V rating from cathode to fils and the "dreaded" fil to cathode short is a real thing.

If we do not grouund the fil CT,  the cathode will equalize to the fils when driven with RF, but how will it do so for DC?  Thinking more, maybe the cathode shud float for DC. I'm gonna take the fil CT off ground.

But, I also wonder why connecting the cathode to the fil pin and running it thru a bifilar choke is not a good idea? I also wonder if tying the one fil directly to the cathode could induce a small amount of hum/buzzies to the signal. I plan to test this today. Check the sample schematics below.

There are at least five schematics out there for 8877's showing five different fil/cathode configs. It can be confusing.

BTW, I did some more IMD measurements on the 8877 and it is by far my cleanest amplifier. Much better than even the pair of Henry 3-500Z's too. If I can just get the fil/cathode issue in a safe mode, it shud be FB. The 8877 can be a fragile tube if mistreated, but run a long time if run correctly.


Here's some 8877 schematic samples - everyone has their own way of doing it... :-)

http://wc6w.50webs.com/wc6wamps/index.html?fr40.html
 
http://www.w8zr.net/images/W8ZR%208877%20Amp%20Schematics.pdf
 
http://www.bing.com/images/search?q=8877+scehmatic&view=detail&id=4C0A4E3A6F498096371F696A39630E829A5A3653&first=1
 
 
This guys ground the fil xfmr CT.
 
http://www.qro.it/amp/schemi/pdf/90hb3046.pdf
 
 
This guys connects the cathode to the fil:
http://www.wv7u.com/yc156amp/skems/RF_DECK.pdf

I was curious about Pro-Tec tubes, and whether comments within this thread were true or not, so I contacted D & C Electronics, the folks who market Pro-Tek tubes ( a subsidiary of D &C). I got this response via email today (re-posted with permission):

"From: Dick Gross <sales AT dandcelectronics.com>
Sent: Tuesday, December 18, 2012 8:25 AM
Subject: Re: Pro-Tec 3CPX1500A7

Hello Jim,

    We realize that there are a lot of things floating around the message
boards online, and you can't believe eveything you read. We have been in
business and a US Government contractor for over 30 years in the same
place and do not and have never rebranded or otherwise manipulated an
Eimac tube for something it's not. We are one of the worlds largest
stocking distributors of new Eimac product, and we have no need to damage
our relationship with the factory with any of these shenanigans. Aside
from the fact that it is illegal, it's just plain bad business. If it's a
new tube, we sell it as new or NOS. If it's used, we test, clean, and sell
it as such. Your thread makes reference to a company that has been in
trouble for rebranding Eimac tubes. This is not us. The California based
company referenced has been completely debarred from any government
contracts and the owner of the company spent a year in jail. All of this
is public record from the US District court.

  With respect to the used tubes showing up on Ebay and elsewhere online,
we assume no responsibility for other peoples representations of them
(new, used, etc.). We own the Pro-Tek trademark, and every item that bears
the name was a factory new tube when it left here. We deal with several
different factories around the world, and although a lot of tubes may be
assembled overseas (China, Russia, France, etc.), all final processing,
testing, and QA are done here in the US.

    I give you full permission to repost this, and we may institute legal
actions upon the person(s) who are making these false statements.

Regards,

Rich Gross

Government Contracts (1R4E7)

D & C Electronics"

That still doesnt explain the Made In USA on their 3-500G's.

Nor other Chinese tubes labeled and sold as a US brand.

Not much available on Google but this and government actions may not show up.

http://www.bbb.org/west-florida/business-reviews/electronic-equipment-dealers/d-and-c-electronics-company-in-weeki-wachee-fl-90118156/complaints#breakdown

Do you believe this?

http://www.dandcelectronics.com/protek.html

I might have to put my waders on but I'll maintain an open mind for awhile.

At least we have experts in consumer complaints available here for comment.

Apparently clairvoyant too.   ::)

 ;D   Meanwhile....

Well, after a lot of mods, changes and tests, I ran a full system IMD test on the whole amplifier chain.  Chuck suggested I increase the 8877 Q to over 20. Using about 490pf for C1 and 3,000pf for C2 and about 5uh, I was able to get about Q=25. The power output was about the same as when Q=12.  Being a single ended stage, Q is important for best IMD, just as it is in the input tuning. I'm setting up my other amps for higher Q too.

I used recorded voice audio to best simulate real whirl.  I was amazed.  The spectrum analyzer tests showed upwards of 40-45 db 3rd.  I then listened to a remote receiver located 100 yards away from the shack and found that at S9+30, the side crud fell into the noise when up 3.5 to 4.0khz. This verifies the IMD test in the shack. It is definately cleaner than other amps I've used in the past.

I do not see much difference in IMD when idling the amp at 50 ma or 200 ma or 300 ma.  Some orders of IMD get better while other shift the other way. So I will run it lower for less heat.  I know that W8JI has written this to be true too, which kind of flies in the face of traditional practice.

The next project is building a tube amp to go from 10mW to 100 W at -45db 3rd or better, to drive the 8877.   The bottleneck now is definately the FT-1000D. It should be 10db cleaner than the 8877 to do the system justice.  I'm not sure what the tube lineup will be yet.

In addition, the blower noise MUST be reduced. On AM this is mandatory and on ssb it is important. I am putting mucho air thru since I learned my lesson about failing tube seals. I was thinking of some kind of sound proof barrier in from of the amplifier rack, like soundproof curtains or  panels, or even some styrofoam sheets built into paneling. If I can drop the noise another -20db, a new RF speech clipper can be used on ssb. The RF clipper, common in the 60's 70's,  is a board from Germany that converts audio to 455khz, clips it at the desired level, and then converts back into audio for the rig. On SSB, it produces the classic 6db of average increase in voice density, but all the audio harmonic trash is filtered by the 455 filters, etc.  Right now my average density is low, but it needs to be until I get a quiet shack.

So, there's lots of upgrades planned here.

T

 

Tom,

If you do an audio chain, you can notch filter the blower out.  That's what I did, upon the suggestion of VE7RF.  A simple notch filter placed BEFORE the noise gate did wonders. 

Mine is in software, so I can set it for a VERY steep curve, JUST taking the blower noise out.  It's almost undetectable when it comes to my voice, since the voice is several dB higher than the blower noise regardless.  Just enough notch to kill it, and you're FB again.

Or, get a BIGGER wheel.  Turn it mo slow.


--Shane
KD6VXI

4CX350FJ?  Kinda scarce

I dont know what you can drag out of a 8072/8121/8122 at a real low B+. The old Signal One ran a 8072 at around 700V if I remember.

Shane:  I have tried the notch idea, but the blower noise is too wideband white noise and it takes too much audio with it.   Today I picked up the lumber and insulation needed to make three 4'X8' moveable walls that will seal the noise out, I hope.   Nothing like a QRO, quiet shack.

Carl:  I wouldn't mind trying a lineup similar to that. I need 30db of gain, 10mW > 100w.


I'm taking a different approach down the hill:

Tom how about one of the preselectors that follow the radio around via the tuner cable?  That would clean up the tranciever.   They give you 70 DB!..

http://www.arraysolutions.com/Hamation/bandpassfilters.htm


C

Hi Clark,

Thanks for the suggestion.  That looks like a nice bandpass filter system - something I'd like for my HPSDR radio.

It would be nice if it were that easy, but in this case we are talking about two different things.  That filter would help the harmonic suppression far removed from the operating freq.  (X2, X3, etc)  Notice the filter is still flat at 15 mhz on 20M.

We need better IMD numbers - from distortion that is generated CLOSE to the operating frequency by a non-linear amplifier. (splatter)    No amount of 2nd harmonic suppression will help that, I'm afraid.  The FT-1000D is what it is and that's about -31db third order IMD at 200w output.  We need something that will do 15db better for close-in cleanliness to do justice to the nice 8877 amplifier IMD numbers.


T

T,

The old ways are the best.

klc

I use a Class A 2N5109 to boost the 10mw from a TS-830 transverter port to 100 mw or so to drive a HA-6 transverter to the 15W needed to drive the 1200W amp. Needed a slip on heat sink due to the bias needed for that mode at 12V.