Has anyone ever hooked up a transformer this way?

[KD6VXI]

In my experience, if you are using a pass regulator (like a typical Astron, etc) then you'll need to dump have the exoect d peak screen current in a resistor, depending on the tube.  Some tubes go negative on screen current actually dumping electrons and causing screen voltage to increase under some tuning conditions.

A shunt regulator eliminates that problem.

Keep this in mind.

I used a bipolar device for my shunt regulator out of a stereo.  Iirc they where good to 900 or 1200 volts.  I also had some mosfets I played with but wasn't happy with them and went back to bipolar.  For the life of me, I can't remember why though.


If your tube(s) won't go negative on the screen, a series regulator with a small amount of bleed R will be fine.  250B style tubes will need more bleed current.


--Shane
WP2ASS / ex KD6VXI

[K9MB]

In my experience, if you are using a pass regulator (like a typical Astron, etc) then you'll need to dump have the exoect d peak screen current in a resistor, depending on the tube.  Some tubes go negative on screen current actually dumping electrons and causing screen voltage to increase under some tuning conditions.

A shunt regulator eliminates that problem.

Keep this in mind.

I used a bipolar device for my shunt regulator out of a stereo.  Iirc they where good to 900 or 1200 volts.  I also had some mosfets I played with but wasn't happy with them and went back to bipolar.  For the life of me, I can't remember why though.


If your tube(s) won't go negative on the screen, a series regulator with a small amount of bleed R will be fine.  250B style tubes will need more bleed current.


--Shane
WP2ASS / ex KD6VXI

Thanks for the advice, Shane.
I actually have some high voltage NPN power transistors coming in.
They are ON FJP5027 high voltage devices in SO220 packs.
They are good for 1100v C-B and 800v C-E
Data sheet below:
The SOA graph is also there. I will be using about 300-350volts on my 4-1000A tubes. The DC graph shows about 40-50mA at 300 volts and maybe 30-40mA for 350 volts.
I figure that I can put 3-4 in parallel for my screen supply and ballast the emitters with (10ohm?) resistors to balance things and stabilize them. Three of them would give at least 100mA at close to 400 volts, I think, which is more than the 80-90mA I will see in Grounded Grid operation at about 4500 volts and about 350 screen volts.
Comments on this device?

Yes, I am very familiar with the 4CX250B, 4CX300A and 4CX350A tubes.
Screen currents would almost always go several mA negative in AB1 for SSB. I used 0xx cold cathode tubes as shunt regulators and they would brighten with modulation on a pair in AB1. I still have a box full of 0B3, OA2, etc…
Nice blue glow like 866 rectifiers. I do not use them- opting for the souless but compact silicon stacks….😉
Have you parallel bipolar pass transistors- like this with emitter ballast for higher currents?
A pair of 4-1000A tubes will not go negative, I think, but if I decided to push them, the screens could pull 0ver 200mA, though that is not my intent, beimg focused exclusively on low IMD, not high output.
73, Mike

[K9MB]

Hi Mike,

Good decision on using a GG 3-500Z as the driver for the pair of 4X1s in GG linear.  You should not have any issues with getting good IMD results from the 3-500Z itself.

A few suggestions:

Be sure to have more than 1650V available for the 3-500Z in case you need it.  You do not want the tube drawing much grid current for best cleanliness. The grid current demand rises as you try to pull more power out of the tube.  More HV will reduce the grid current requirement and make it cleaner.  I am aware of the -45 dB 3rd order datasheet claim at 1500V, but not sure I believe it. I did not not see a noticable change in IMD when I ran my single one as a driver.  I ended up using 2200V for less grid current as discussed above and it worked out better.  IE, at 2200V, I just had to tickle the 3-500Z to get good power out as a driver for my pair of 4X1s in linear service.

For the 4X1s, use regulated screen voltage with minimum grid current in GG.  (I think that is already your plan)  The 4X1s will be the bottleneck for clean IMD so do everything you can to make them clean.  Expect to have low efficiency and more cooling because of heavily loading the plates and higher idle, thus two tubes for more dissipation to hit your final output requirements.

Now the Flex 5000...   Did you run a two-tone IMD test with it yet to prove out the favorable IMD expection?   If you expect -35 dB 3rd out of the 4X1s, you will probably need at least -38 to 39 dB 3rd IMD out of the Flex 5000 at its required driving power.  Being a solid state final, I don't know if this can be done, considering most solid state ham RF power amplifiers are around -30 db 3rd or so.  Is the Flex 5000 an exception?

As you know, the final result cannot be better than the dirtiest link in the chain.  Now is the time for design mistakes to be made, not later...

And of course, your overall safety net is still using adaptive pre-distortion later on if this does not work out.  

T

Hi Tom,
Since you mentioned Adaptive Predistortion in this discussion, I have casually tried to find a way to add adaptive Predistortion to my Flex 5000 transceiver.
I found some comments in forums that there were operators who had adapted it to the Flex5000a with excellent results.
I can find no real application information or interfaces that can sample the amplifier output and a board that feeds the signal back into the transceiver and software to make instantaneous changes to correct non-linearities to improve 3rd order IMD of the amplifier output.

I am proceeding with the construction of my single hole 3-500z and my 2x4K or 2x3K amplifier.
The single hole design is complete and I am ready to mount components.
I will be doing everything possible to improve the inherent IMD of the single hole amp, but the ceiling is in the high -30s for this amp and that is better than my transceiver.
I could lay out $4500+ for an ANAN with pure audio built in to push it up to -60 or so, but the question is- can even these pricey radios accept sampler input from a linear amplifier and where can one buy or build that interface?
I wish that there was more information available in a how to guide, but I cannot find it.
Can anyone give me some clues on this?
Thanks in advance for any advice?

[K1JJ]

Hi Mike,

I recently did some reading trying to figger out if the older Flex radios will do predistortion.  (You mentioned that some clever guys have figgered it out) Otherwise, the Anan line (and associated kits)  are the only ones with the software, and hardware available for hams.  The Anan 200 seems to be a great used buy thes days, if you can find one.

The HPSDR "kit" radios will do it too, (HPSDR, SmartSDR) being closely associated with Anan's technology.

However, I would contact Rob/ W1AEX to get his latest opinion.  Probably the cheapest method would be to get a Hermes, (or latest equivalent) wind your own toroidal RF sampler for your 4X1 amp and feed it into the Hermes predistortion set up. Any linear amp will work.  There are likely pre-made samplers available.

I may be doing something similar in the future too. Predistortion is certainly the path to take for the cleanest signal.  

BTW, the dual GG 3-1000Z amp driven by the single GG 3-500Z sounds like your best bet for old school technology.



Flex's positions on predistortion:

Old:
https://community.flexradio.com/discussion/6512026/official-position-on-adaptive-predistortion

Newer:
https://www.systems4silicon.com/ip-cores/digital-predistortion


Does it mean only the latest Flex rigs apply?


Since you're doing the legwork, let me know if you find a complete plan for using a Flex 5000 to do the job...  


** And here is an excellent post about setting up a complete station using a Flex 6700:
http://www.nn4zz.com/FLEX6700.htm


T

[K9MB]

Hi Mike,

I recently did some reading trying to figger out if the older Flex radios will do predistortion.  (You mentioned that some clever guys hae figgered it out) Otherwise, the Anan line (and associated kits)  are the only ones with the software, and hardware available for hams.

The HPSDR "kit" radios will do it too, being closely associated with Anan's technology.

However, I would contact Rob/ W1AEX to get his latest opinion.  Probably the cheapest method would be to get a Hermes, (or latest equivalent) wind your own toroidal RF sampler for your 4X1 amp and feed it into the Hermes predistortion set up. Any linear amp will work.  There are likely pre-made samplers available.

I may be doing something similar in the future too. Predistortion is certainly the path to take for the cleanest signal.  

BTW, the dual GG 3-1000Z amp driven by the single GG 3-500Z sounds like your best bet for old school technology.



Flex's positions on predistortion:

Old:
https://community.flexradio.com/discussion/6512026/official-position-on-adaptive-predistortion

Newer:
https://www.systems4silicon.com/ip-cores/digital-predistortion


Does it mean only the latest Flex rigs apply?


Since you're doing the legwork, let me know if you find a complete plan for using a Flex 5000 to do the job...  


** And here is an excellent post about setting up a complete station using a Flex 6700:
http://www.nn4zz.com/FLEX6700.htm


T

Many thanks, Tom.
I found a number of articles for build your own couplers all based in an article by W5QN.
They look simple to build.
That is the easy part.
I have also looked ar the Hermes 14bit board and it is attractive and cheap. It has inherently low 3rd order products at -53dB.
The only rub is the 500mW output and it should be run at half that.
This might be a problem without the Adaptive Predistortion system in place, but given that- an amp that is not too expemsive could be used to bring it up to a constant 25-30 watts carrier (100 watt amp) and then that could be used to drive my single hole 3-500z to 150watts carrier and then the big jugs.
I agree that the 3-1000z X 2 amp is the attractive option and I now have 4 “good” ones that cosmetically look new.  I wanted to get a set and some spares before completing the build.
I will-however- make plenty of headroom and elbow room and air flow for changing to the 2x 4K setup if needed. I have 6 good 4xKs to work with and if Adaptibe Predistortion works, those tetrodes run in AB1 with 350 screen voltage and appropriate cathode bias to clean them up.
I will likely run the 3-1000z x2 at about 4kv but my variac can go up to 4500 if I run the 4XKs.

I will certainly share anything I discover regarding the adaption of my FLEX radios.

By the way, I noticed thatyou were back on the audio improvement track.
I will send you some info I got from working with Jim Tonne in 2021 that you may find useful. He focused on antialiasing in these sharp multistage filters and has some nice analog answers using discreet op amps that are configured for easy cutoff changing based on changing only the resistor values. Will semd that along you to look at.
Thanks again for your counsel.
73, Mike



https://www.collinsradio.org/wp-content/uploads/2015/05/Build-a-Quality-RF-Power-Sampler-Jackson.pdf

http://www.n4ga.com/50db-hf-rf-sampler/

https://www.qsl.net/on7eq/projects/rf_sampler.htm

http://www.lyonscomputer.com.au/Test-Equipment/50-Ohm-HF-RF-Sampler/50-Ohm-HF-RF-Sampler.html

[K1JJ]

Very good on all, Mike. Sounds like  a good plan.  Starting with a clean amplifier and then adding predistortion will only improve things more, vs: starting with a rat to clean up... .  

As far as audio improvements, Rick/ W8KHK rebuilt my old Switched-Capacitor Filter (SCF) based around two Maxim MAX295 8th-order Butterworth filter ICs in cascade.  I was using the homebrew L/C analog filter but didn't feel it was sharp enuff for the big 4X1 plate modulated rig. (Fabio 2) Fabio is pretty clean for IMD as is.   But the audio skirts rolled off at 5-6 KHz, but still had some slight fuzz out to 7-8 KHz.   The Maxin 295 chips rolled the audio off much sharper. It is also variable, so has an advantage to adjust on the fly.  

Rick plans to make these filters available in the Max audio Processor or as a stand-alone filter wired / or as a kit.  I think the filter alone could be a very valuable asset in the shack to tame big rigs that are inherently clean, but need some additional skirt management.  For example, using the MAXIM filter, the spectrum analyzer pics I did of the 4X1 rig look almost like predistortion skirts.

I think Rick's processor and filter will become popular once guys see what it can do.

T

Pictured below are three waterfall bandwidth snapshots of about +- 4.5 KHz,  +- 6.5KHz  and  +- 7.5KHz  audio using two Maxim 295 chips in cascade.   The 4X1 modulated by 4X1s rig is flat out to about +- 12 KHz, but is reigned in by the filter.  The filter is placed just before the limiter near the end of the audio chain.  The top line-waveform shows the +- 7.5KHz response using a real voice into the mic.