Looking for ideas for a ~25 watt plate modulated homebrew Pissweaker rig

[K1JJ]

More:

There is one more post of this series back on page 5.

You will see Fabio II  the 4X1 X 4X1s plate modulated,  Hollywood the pair of 813s X 813s plate modulated, Yaz the 813 X 813s series modulated, Summer Breeze the 4D32 plate modulated by a Hammond xfmr and SS amp, Baby Blue a single 3-500Z linear, Rico Suave II the pair of 4X1s in linear,    Fabio I the single 8877 linear -  and Dr. Love.


First pic:  4X1 X 4X1s plate modulated rig. Sits on a rotating pedestal for easy access serving. The top cabinet is the RF deck and the bottom cabinet is the modulator.
Last pic: The MOSFET audio drivers and power supplies.

[K1JJ]

More:

Pic 1 -    a cool lineup view.

[K1JJ]

First pic - Pair of 4X1s linear:

[K1JJ]

More:

First pic: 813s X 813s plate modulated.


Second:  8877 Linear

[K1JJ]

More:


First pic: The awaiting harem.


The audio drivers and Yaz all need Plexiglass cabinets.  Why the Plexiglas cabinets?  I enjoy viewing the insides to refresh my memory over and over.

I'm running out of room and workspace. It's like a submarine in here.... help!

T

[w8khk]

Tom, thanks for the shack tour!  Those blue panels really look sweet!  My original dual 4-400 linear had a darker blue panel, almost the same color as my Air Force Blues, but I like your choice of blue better!

A couple questions -  Is the safety choke at the output loading capacitor heavy enough to survive a shorted plate blocking capacitor?  What fusing or safety glitch resistor did you use on the plate supply to the final?  And were the parasitic suppressors necessary on the modulator plates, or did you just use them for consistency?

I guess that is more than a couple questions, but they are all related....

[K1JJ]

Tom, thanks for the shack tour!  Those blue panels really look sweet!  My original dual 4-400 linear had a darker blue panel, almost the same color as my Air Force Blues, but I like your choice of blue better!

A couple questions -  Is the safety choke at the output loading capacitor heavy enough to survive a shorted plate blocking capacitor?  What fusing or safety glitch resistor did you use on the plate supply to the final?  And were the parasitic suppressors necessary on the modulator plates, or did you just use them for consistency?

I guess that is more than a couple questions, but they are all related....

Hi Rick,

Yes, the safety choke is too small. I usually wind my own using #22 wire so it will withstand full current from the HV supply. But got lazy. I will add one shortly.

The glitch resistor is 10 ohms and wirewound as shown in the picture.  Up to 25 ohms would be OK for this rig.

I've seen both plate bare resistors and parasitic chokes used for commercial audio use.  I guess if there are problems with audio instability I can experiment, but it's pretty stable for now.

A flat medium gray might be my next panel color.  

I've been thinking, dreaming and looking over the Dr. Love linear amp (the 6' rack)  and thinking of converting it over to a big AM PDM rig.   A 3CX-2500F3  PDM'd by a 3CX-2500F3 would be quite the machine. I already have the big PDM coils, one wound on a 5 gallon paint can.   But just yearning for a new big rig after building all these pissweakers.  I'm just a building fool lately.  But Dr. Love is such a nice linear it wud be a pity to tear it down.  I don't have the room in the shack for anything else without replacing something.  

T

[w8khk]

I've seen both plate bare resistors and parasitic chokes used for commercial audio use.  I guess if there are problems with audio instability I can experiment, but it's pretty stable for now.

I've been thinking, dreaming and looking over the Dr. Love linear amp (the 6' rack)  and thinking of converting it over to a big AM PDM rig.   A 3CX-2500F3  PDM'd by a 3CX-2500F3 would be quite the machine. I already have the big PDM coils, one wound on a 5 gallon paint can.   But just yearning for a new big rig after building all these pissweakers.  I'm just a building fool lately.  But Dr. Love is such a nice linear it wud be a pity to tear it down.  I don't have the room in the shack for anything else without replacing something.  

T

OK on the modulator parasitic suppressors.  I have seen only resistors used on larger audio amplifiers, but I suspect you would not need any with your parallel 813s in the series mod configuration.  With the low impedance drive from the GFZ board, I would think it would be extremely stable with just straight copper wires to the plates.

Regarding Dr. Love, I understand the space situation in the shack.  I have a smaller shack upstairs, which also has my library, including QST from the 1920's to current, taking up so much space that I have no room for even one 6' rack, so my 8877 amplifier, in a rack 64" high, 22" wide, and 33" deep weighing over 700 pounds is relegated to the finished basement shack, along with all the other rack rigs on casters.  Yes, it is big and heavy, but it will not suffer melt-down if left on at full power for days.  I posted some photos of it here:
http://amfone.net/Amforum/index.php?topic=34544.0

It would be a shame to decommission such a nice rig as the amorous doctor, but sometimes building is more fun than operating.   And if it is truly redundant to your needs for legal limit amplifiers, why not reorganize?  I still have not finished my 4-1000 rig, as other, older rigs were more pleasurable to restore and redesign.  I remember you asked about that one quite a while ago, it may no longer be on the bucket list, even though the final deck is thoroughly completed.

My original plans for the BIG series modulated rig were identical to what you are contemplating:  3CX-2500F3 modded by another one.  I determined that would probably not work out quite as well as I had hoped.  Thus I changed my design to use a 3CX2500F3 or 3CX3000F3 modding a pair of push-pull Eimac 304-TLs.  Looking at the curves, I think a 3CX tube of that stature might be anemic as a modulated RF final amplifier with only around 2KV at carrier.  But I could be wrong and it might just play fine.  I come up with that number because I am not comfortable putting more than 5KV on the modulator, and this is the minimum voltage necessary to achieve 2KV on the final, and still have some positive peaks greater than 100%.  That final is not designed to have the filaments elevated to thousands of volts, so the modulator will need to be on top of the final, working as a cathode follower, complicating the audio driver circuitry.  If there is no challenge, there is less builder satisfaction.

Over 5000 volts can get pretty hairy to manage as far as insulation is concerned.  Pretty much my comfort limit for HV when it appears throughout the entire enclosure.  That is why I picked up the all the parts necessary to assemble a 5KV 2 amp CCS power supply from a Collins 5KW FM broadcast transmitter - just bolt the parts down, wire it up, done! 

I chose the 304-TLs due to their enormous emission, with the filaments running at 5 volts, 26 amps each, they will play well at QRO power levels with less than 2KV on the plates, and I have already restored the push-pull RF deck that I watched dad build in 1949, when I was only two.  My only regret is that he did not put any peek-a-boo windows in the front panel, and I do not have the heart to modify it to that degree!  It may not be that well suited to the higher bands, but my main interest for this rig is just the 75 and 40 meter wavelengths.

You may come up with different results when you model the big 3CX-2500 bottle at only 2KV, and if you do, it will also make a sweet rig, legal limit, no mod iron, DC to daylight response! 

Whether we use one or two of these tubes, the primary goal will be to use the voluminous six-foot cavern to create some plenum ductwork that will muffle all but the slightest breath of the cooling system, making it a pleasure to operate right near the microphone.

[KD6VXI]

I completely understand about the voltage limit after having taken 6.4kv to the hand.  However, on a technical level I can say the 3cx3000 will take a lot more than 5kv.  I've seen guys running them at 8.5kv.

So if you wanted to crank the thing up later, it's capable of it.

--Shane
KD6VXI

[K1JJ]

Yo Rick,

Yes, a 3CX2500F3 ( a 4KW diss tube) would need a lot of voltage to make it worthwhile.   380 watts just to light one filament. It likes 5KV, so figure a 12 KV main HV  power supply for PDM to make 5 KV across the RF tube and 7KV  across the modulator available for modulation..

So probably a better choice would be  multiple  lower voltage tubes in parallel, triodes preferred, to eliminate the floating screen supply.   There's a lot of good choices that like 3-4 KV at dead carrier.

Even a bunch of parallel 813s would put out some soup at lower voltage.  I'd have to think more about it.  My plate modulated 4X1 and 813s rigs are good for now.  It's like having a garage full of old cars... they all get you there, but each in a different way.

I've always liked running a tube PDM rig. It has a certain mystic about it and runs really cool with the RF final at 80% and the modulator approaching 90%.  

It wud be EZ for you to convert your new big rig over to PDM rather than series modulated. That 3X3 could be reduced to something 1/2 the size.  Use Steve's PDM driver board with a driver chip.   The crossover modulation point gets squirrelly so cover it up with the built in neg peak limiter.

I'm in no rush - just thinking about future winter stuff for now.

T

[w8khk]

Yes, a 3CX2500F3 ( a 4KW diss tube) would need a lot of voltage to make it worthwhile.   380 watts just to light one filament. It likes 5KV, so figure a 12 KV main HV  power supply for PDM to make 5 KV across the RF tube and 7KV  across the modulator available for modulation......snip

Yes, a great deal of voltage is needed to make the 3CX play QRO RF.  As Shane mentioned in the previous post, it is rated around 5KV, and he has used it up to 8.5 KV.  But since I do not have an abundant supply, I would rather run it conservatively, and that is of course why I am using a 3KW dissipation tube for a ham legal limit modulator.  

That said, I would like to comment on the 12 KV numbers you mentioned.  If we have a 12 KV power supply, we do not simply divide it such that 5 KV is provided to the RF stage and 7 KV across the final.  Unfortunately, the series modulator MUST be able to withstand the ENTIRE supply voltage, which is why my design is limited to 5 KV, and of course because a 5 KV supply almost fell into my lap, thanks to Dennis, W7TFO. (But my lap had to be in a MINNESOTA SNOWSTORM in March for that to work out, hi hi.)  Remember the series modulator is like a series resistor, and when almost saturated, it drops very little voltage, applying the majority of the supply voltage to the final amplifier, making nice positive peaks, more than double the steady state, unmodulated RF amplifier plate voltage.  We must stay in the most linear portion of the curve, so we do not push it to saturation, keeping that sweet spot to minimize the IMD.  But when we create our negative peaks, the series modulator approaches cutoff, thereby dropping ALL the power supply voltage, while none of the voltage is applied to the final.  So in your case, the 3CX3000F3 would probably not be very happy with 12 KV between the plate and the cathode (filament).  

So, looking at the numbers realistically, with a power supply of 5 KV, and a dead carrier input of 2 KV at 500 mA on the final (1 KW DC Input) we see the series modulator dropping 3 KV at 500 mA, thus dissipating 1500 watts (1.5 KW), which is HALF of its rated dissipation when there is no modulation.  This dissipation is reduced when it is modulated, and instead of wasting power in the series modulator, the modulation ADDS to the steady state DC power applied to the final, actually providing useful output power in the form of modulation. Unfortunately, it appears to me we will not be able to come close to 4 KV or even 3KV plate voltage on the final amplifier, unmodulated.  2 KV seems to be the practical limit if we want any positive peaks above 100 percent, using the 3CX3000F3 series modulator.

But the bottom line is the series modulator tube maximum voltage specifications must be considered because the entire power supply voltage will appear across the series modulator tube during negative modulation peaks.

When comparing PDM and Series Modulation, and traditional Class-B Plate Modulation, each one provides the same modulated plate voltage and plate current to the final RF amplifier.  So it really should not matter which RF stage configuration is used.  For a given resting plate voltage and power level, the final amplifier sees no efficiency difference with the various modulation schemes.  It is all about the modulator - efficiency or lack thereof, simplicity or complexity (series v PDM) reliability (fragile iron v none) drive requirements, etc. etc. etc.

I think it is just a matter of choosing a final RF amplifier configuration that can produce the output power desired with the available voltage and current provided by the modulator at hand.  With the series modulator using a 5 KV power supply, it is clear that we must limit our RF final resting plate voltage to 2 KV or less if we wish to achieve 100 percent positive modulation and some nice peaks beyond 100 percent.  So choosing a tube or tubes that give us a kilowatt input power at around 2 KV and 500 mA meets that requirement, and the RF portion of the system may be designed, tested, and implemented separate and apart from the modulator design and implementation.  

By the way, my load test for the series modulator is a string of between 16 and 20 each 60 watt incandescent light bulbs, also strung together in series on a pine board.  A 60 watt bulb requires 120 volts at 500 mA.  At 600 watts input, that is 10 bulbs, 1200 volts, still 500 mA.  Rise to one kilowatt, 16.666 bulbs.  (16 bulbs is 1920 volts at 500 mA.  They are a perfect fit for testing the 3CX3000F3 modulator, to determine the maximum undistorted modulated DC available within our defined IMD limits.  The entire modulation system may be tested, sans any RF components.  Using 20 bulbs provides some headroom with modulation.  But being this is DECEMBER, I am reminded the fun I had as a kid trying to fix a series string of bubble lights.  Now that I am alone and retired, no tree, no series string lights, no problem, Yuengling Time!

Get that part out of the way, then take a tried and true final amplifier, meld them together, blast some atomic yaaaaaloooos, and look at the spectrum and envelope with no fear of arcing any enamel or fish paper!  It just doesn't get any better!  Don't need no stinkin' iron in this rig!

If you can get legal limit out of a pair of 813s at 2000 volts and 500 mA (And I am SURE YOU CAN) they should work just as well as my 304-TL final.  So Tom, my question for you:  Now that you have the PW series mod rig singing, are we BOTH up to building QRO series mod rigs using the 3CX3000 bottle as the modulator?  Let's put the RF amplifier below the modulator, where it belongs for safety.  And we will design the modulator and driver together.  As Yaz says, "Let's start building, Tommy!"

[K1JJ]

HO-K on all, Rick.

I was talking about the 3CX-2500F3 which is really more a class C AM broadcash service tube. It will take 5-6KV fully  plate modulated or equivalent to maybe 12KV peak under modulation.  Not to be confused with the 3CX-3000A7 or linear amp tubes.    I have some 3CX-2500F3s and chimneys.  I also have a 3CX-3000A7 and YC-156, so many possibilities for class A series service.

http://www.tubecollectors.org/eimac/archives/3cx2500f3(67).pdf

I would consider a series modulated rig using one modulating one though it would mean really big HV, like 12KV to put 5KV across the RF final and 7KV across the modulator.  I have limited my HV supply to 2- 4KV these days.   4KV on a 3CX-2500F3 PDM or series rig is a waste of big tubes.

But your idea about using the modulator "on top" still interests me.   I've seen your descriptions but I really need a schematic to understand how the modulator floats above ground. I can see using a fiber optic rated at the HV, but otherwise I don't get it.    This could also open the door for a big PDM rig without the need to run the RF deck at floating HV.  (like my present 813 series modulated rig is now)    Can you post a hand-drawn version that is specific to what we are doing?  I've seen power supplies and other related stuff but no actual on top ham rigs.

I'm also thinking about how I could use my spare 3-500Zs in the class C final and PDM modulator. That wud be roughly the class of a pair of 4-400As plate modulated.  It would work well at 6KV.  

A 3-500Z will run class C plate modulated with 3KV, so maybe 7KV peak may work.  How about two or three  3-500Zs in the final with a single 3CX-2500F3 series modulator with 7KV across the whole thing?  3KV RF   and 4 KV modulator divided.

http://www.umich.edu/~umarc/station/docs/3-500z.pdf

I should check with Frank about the MOSFET driver. The latest one for the 813 rig is single ended and good for only about  160 V peak-peak audio drive output.  That is way too low for what we are proposing.  Maybe a linear type high-mu tube is a better choice as the modulator.


Still thinking.  That's half the fun.

T

[w8khk]

Tom, I am reviewing the curves and verifying my numbers, and will have a response within a few days.  The diagram will be a preliminary version covering the basic concepts, but of course I cannot give a detailed design before some of the prototyping is completed.  I think it will be wise to provide a negative peak limiter at the input to the driver stage, and a carrier level control, which allows adjustment of the plate voltage to the final.  Current will be determined by loading, of course.

Looking at the specs on the 3CX2500F3, it looks like 5,000 volts for a plate modulated class C stage, which would imply 10,000 volt peaks, so running above 6,000 volts should not be an issue.  For a class C oscillator or AF amplifier, 6,000 volts is the max in the spec sheet, but we know we can go well over that voltage.  I still wish to keep it down to 5,000 volts, and run the final amplifier at around 2,000 volts as a target, and up to 500 mA.  The supply and the series modulator can of course do a lot more than this voltage, and gobs greater current if we wish.

I have two of the 3CX2500F3 bottles, untested.  Will design the plenum and cooling system before deciding on the chimney.  No socket makes it very easy.

We can do lots of brainstorming on paper, but I would rather nail down the prototype with 810 modulators and the 8000 final before delving into the super QRO version.   Even though you have a plate modulated 4-1000, I believe a series modulated rig of similar power, using lower voltage tube(s) at higher current would be somewhat superior.  Perhaps one or two, or maybe even three 3-500Z tubes in the final, with the single series modulator tube.  Of course, I am not really focusing on an all-band rig, I primarily am interested in 75 and 40, and possibly 160, but that could be a separate final in the same rack, sharing the power supply and modulator.

Perhaps three separate single-band finals with tank circuits specifically designed for each band, all in a single six-foot rack would be the ultimate......  no-tune band changes!

[KD6VXI]

Years ago I got a 3cx3000A1 from Bill in Colorado.  I always wondered what to do with it, but for the price he wanted, NIB, I couldn't pass it up.

Maybe a PDM modulator?  Have it mod a 3000 or 4cx5K?

--Shane
KD6VXI

[K1JJ]

Yo Rick,

FB on the info.

From what I see, I think putting the modulator above the RF adds some more complexity and I think I will stay with the original method using the modulator on the bottom.  There is really no performance to gain.   I am not sure yet about what specific tubes I'm gonna use and whether to entertain a tube PDM or series modulated rig yet.  It’s all about major parts and infrastructure right now.  A major obstacle is my 4 KV max supply.  I could always build it and use it at 4KV and maybe later build a bigger supply if it works out I suppose.  Or maybe it will be a lower voltage, high current rig.

So, I'm getting ready to build a new big rig.  I cleaned out the 8877 linear position and will use it for parts. That amp is about 10 years old, original tube,  and I just don't use it much compared to the 4X1s linear. I ran some IMD and power output tests and the 4X1s linear is actually better than the 8877 for my use. A pair, GG 4X1s with regulated screen and grid, grid driven, is a nice, clean amp when run conservatively at 1500w pep legal limit with 4KV. Somewhere between -38 to -40 DB 3rd IMD.  Nice. I left the 8877 HV, control PTT, coaxes and modified a few things so that I can run the new rig where the 8877 was before.  

So, looks like I’m going to throw in and join you for a new big rig, series modulated or PDM, OM… 😊


Shane:  FB on the 3CX3000A1 and 4CX-5000 PDM combo.  It sure can become a Frankenstein project real fast...

Later

T

[K1JJ]

Rick and the gang,

After tearing down the 8877 resulting a nice supply of parts;  vac variables for C1 and C2,  a big copper roller inductor for L1, turns counters, relays, etc.,  I am leaning towards this tube line up:

A single 4-1000A class C in the final, PDM (class D) modulated by a single 4-1000A modulator.   The initial problem was the HV power supply being only 4KV.  I really need 7-10KV to allow 4 KV on the final and 6 KV on the modulator at dead carrier. This will be like a plate modulated 4X1 with 4KV.    At 10 KV peak I should be able to see 150% positive, or higher with a lower carrier.

I found TWO  2KVA  HV power transformers in the cellar. They are 3500V at 600 mA each. (no center tap) I can put the 115VAC primaries (with 240V) in series and the 3500V secondaries in series to have 7 KV DC under load using a filter choke or 10KV DC with the choke switched out.  I already have a 60 pound 70 uF 10KV filter cap -  and a bridge diode assembly.  HV problem solved, caw mawn.

I have my old PDM coils wound on 5 gallon plastic pails ready to go.  The HV power supply will be built outboard, but the 4X1s and big PDM coils will fit in the same 3' high rack cabinet.

Want to take my time and order some fresh aluminum for the chassis and front panel just like the last Yaz rig.   The cabinet is a heavy duty 19" rack, IBM blue, very nice looking and strong.

Looking for another used 4-1000A and chimney.    Gotta look around but might need two sockets, but will look on e-Bay.   I already have six 4X1s in service in various rigs, so getting low.

This is not yet cast in concrete, but leaning this way.


T

[w8khk]

Sounds like a good plan, Tom.....

I wonder if the insulation in those two transformers can withstand operation with the secondaries in series?  Perhaps a safer approach might be to put the secondaries in parallel, and use a voltage doubler.  This could be done in a manner that would not cause excessive voltage between windings or to the core.  If the secondary voltages are not exactly equal, the two could be isolated by separate diode stacks.  

With the efficiency of PDM, one transformer might be sufficient, but might as well double up and build conservatively.

I had thought about building a tube PDM rig, but after discussion with Steve, the requirement for analog compensation caused me to think otherwise.

I am still proceeding with the series mod approach, and I decided to start with the modulator driver circuitry, and finish the GFZ board and regulator before having all the HV stuff spread all over the lab while I do the small low voltage stuff.  That single ended GFZ board seems to be first on the to-do list right now.  John has mailed a couple regulator boards for me to populate.  Thanks, John!

[K1JJ]

Good points on the HV xfmr ratings in series. I did think of that though I have run it like this before using other transformers with no problems. I don't like voltage doublers.   I'll have think about this.  It will really depend on the individual xfmr insulation, of course.

Or, as you suggested, using a doubler.  I have only one 10KV @70 uF  and one 5KV at 53 uF cap for use so one cap will sag down a little more than the other.  (and the 5KV at 53 uF will have more voltage across it than the 70 uF.


I mailed out your GFZ board a few days ago.  Shud be there soon.

Below, a doubler.  The problem is the caps are different. I don't have identical filter caps.    Using two identical xfmrs,  I have the 115VAC primaries in series with 240V   and the secondaries in parallel for 3500V AC out.  It appears the disadvantage is that each cap is being charged like a 1/2 wave rectifier, thus the poorer regulation. A class A series rig would be OK, but he dynamic plate modulated PDM rig suffers.

T

[w8khk]

I mailed out your GFZ board a few days ago.  Shud be there soon.

Thanks so much Tom!  I am sure it will be here soon.  I know the mail is very slow right now, USPS claims that they are being overloaded with holiday mail.  The last priority 2-3 day delivery package I sent on Nov 30 was just delivered Friday afternoon, twelve days in transit up the east coast!

[w8khk]

Good points on the HV xfmr ratings in series. I did think of that though I have run it like this before using other transformers with no problems. I don't like voltage doublers.   I'll have think about this.  It will really depend on the individual xfmr insulation, of course.

Or, as you suggested, using a doubler.  I have only one 10KV @70 uF  and one 5KV at 53 uF cap for use so one cap will sag down a little more than the other.  (and the 5KV at 53 uF will have more voltage across it than the 70 uF.

Below, a doubler.  The problem is the caps are different. I don't have identical filter caps.    Using two identical xfmrs,  I have the 115VAC primaries in series with 240V   and the secondaries in parallel for 3500V AC out.  It appears the disadvantage is that each cap is being charged like a 1/2 wave rectifier, thus the poorer regulation. A class A series rig would be OK, but he dynamic plate modulated PDM rig suffers.

T

There are so many ways to design a power supply, and many, many ways to design a voltage doubler.  I did not know until a year ago that it is possible to make a full-wave voltage doubler with just a couple extra diodes and capacitors!  But that would be overkill for this application, and very expensive for stacks of oil-filled capacitors.

That said, one often overlooked fact is that, in some cases, a half-wave voltage doubler provides BETTER regulation and BETTER filtering than a conventional center-tapped full wave supply, or even a full-wave bridge supply.  The issue addressed by the voltage doubler is the drop in voltage due to the resistance in the HV transformer secondary.  If we use half the turns, for half the voltage, but at the same time double the cross sectional area of the wire for double the output current, then we have one quarter of the secondary winding resistance, thus much less voltage drop, and better regulation with dynamic loads.  This is why many of the amplifiers today use doublers, with two capacitors.  This results in a transformer either the same physical size, or possibly a bit smaller because there may be less insulation and space between turns.

Large value capacitors, and diode stacks with high PIV and high surge-current ratings are common-place today, where back in the old days a 4 uF cap was expensive and considered large, in combination with a swinging and smoothing choke.

So let's look at what we can do with the parts you have, and "run what you brung".  I will attach two circuit snips, and being lazy, I just plagiarized a piece of and old Heath HP-23 power supply circuit.  

Doubler1 is essentially what you have posted, albeit it is drawn a bit differently, the circuit is identical.  In this case, the problem of exceeding the insulation resistance is not solved, because the bottom end of the transformer secondary floats at half the output DC voltage, and the top of the secondary (of the second transformer in series) has peaks close to the output voltage.  So this voltage doubler is no better for the transformer than the bridge, and the components you have do not match the requrements (capacitors).  Putting the transformers in series increases the voltage drop on the secondary, reducing dynamic regulation, and also increasing transformer heating due to IR losses.

Look at Doubler2, which is the low voltage (250 or 300 volt) section of the power supply.  That is the top portion of the circuit. Ignore the bias supply circuit at the bottom.  In this case, the bottom end of the secondary is grounded, so the top end will not see any more voltage than the rated secondary voltage.  I believe this is safer than floating both ends of the secondary, as in the full-wave bridge circuit.  If you use this version of the doubler, both secondaries could be tied in parallel, reducing IR drop in the winding.  

The first capacitor, between the transformer and the junction of the two diodes, charges to peak, or 1.4 times the transformer output, under no load.  It AC-couples the transformer output to the second diode, adding the peak DC voltage to the input of this diode, charging the second capacitor to 2 x peak.  Of course it will drop down under load, and, even with PWM, the carrier load is always there, while the modulation load is dynamic.

You could easily use your 55 uF 5 KV capacitor as the first capacitor, and your 70 uF 10 KV as the output capacitor.  If your bridge rectifier does not have separate diodes that can be used individually, both halves of the bridge stack may be used in parallel without alteration.  Simply connect the negative output to ground, the positive output to the 70 uF 10 KV capacitor (output) and tie the 55 uF 5 KV capacitor between the transformer output and BOTH of the AC input terminals of the bridge.  Now you have both halves of the bridge sharing the load.  )To truly achieve balancing, it might be necessary to add a small resistance in series with each AC input port of the bridge.)

The down side of this voltage doubler is that the transformer current is passed through the first oil cap.  If it has a high ESR, this could cause heating and capacitor failure.  But considering the values in uF you have, and the low current load at 10 KV, I do not believe this is a problem.  This circuit is much more demanding of the electrolytic capacitors used at lower voltage in the old Heath (and other) supplies.

Perhaps you might want to do a load test of this circuit, using first one, then two transformers in parallel.  Putting the primaries in series for 240 volt feed, and the secondaries in parallel for lower resistance and better regulation should work very well.  Another advantage to this configuration is the fact that, if both transformers are not absolutely identical in turns ratio, having the primaries in SERIES eliminates any potential failure caused by different voltages in the secondary, which could be a serious problem if the primaries are in parallel.  The series primary arrangement is self-equalizing.  The outputs in parallel MUST be equal in voltage, but the primaries in series might each take slightly different input voltage, at equal current, from the source.

I hope this gives you some food for thought.  Swapping and bartering parts for a better fit is cool, but shipping is expensive for this stuff, so fitting to the task what you already have on hand is a bonus!

[w8khk]

Tom, for comparison's sake, you might want to look at the Rockwell (Collins) power pebble 828C-1, the tiny 1KW AM sibling of the 5KW Power Rock.  You are probably already familiar with this PDM broadcast transmitter.  It is discussed in this thread:

http://amfone.net/Amforum/index.php?topic=42096.0

It has a bridge and single oil cap filtered power supply, providing negative 8.5 KV to the modulator, and about - 4 KV to the final.  The plate circuit is the positive return directly to ground.  Of course both 3-500Z filament transformers have HV isloation.

It has only 20 uF filter cap at 10 KV.  Also, there are two each 150 ohm series resistors, totalling 300 ohms, between the bridge rectifier and the filter capacitor.  There is additional series resistance between the positive side of the rectifier and ground, for overload sensing, current meter shunt, etc.

Your 70 uF at 10 KV filter cap should have no trouble providing clean, stable DC!

This rig runs a single 3-500Z as the modulator, and a pair of 3-500Zs in the 1 KW output final, 24/7 CCS.

Your 4-1000A modulating a 4-1000A with PDM will have almost unlimited headroom with a 10 KV supply, conservatively making legal limit.  Note also the fact that the total plate voltage requirements for a tube PDM rig are substantially less than for a similarly-sized series modulated rig, as the PDM tube switches between cutoff and saturation. 

Perhaps this project no longer fits in the 25 watt pissweaker thread category.  At least consider moving it up to 50-watt Fred status.

[W3SLK]

This may be a stupid question but have you considered a HV switching supply? We use several of them for our ozone generator which are good for a couple of amps(8~10KW). I know RFI hash may be an issue but it would certainly reduce the size need and you could probably have more control over your HVPS. The only issue I do see is that our supply iron goes from 220VAC to something like 385VAC. But I was really surprised how small it actually was.

[w8khk]

This may be a stupid question but have you considered a HV switching supply? We use several of them for our ozone generator which are good for a couple of amps(8~10KW). I know RFI hash may be an issue but it would certainly reduce the size need and you could probably have more control over your HVPS. The only issue I do see is that our supply iron goes from 220VAC to something like 385VAC. But I was really surprised how small it actually was.

Mike, that is an interesting alternative that I had not considered.  Although I will not be building anything like this, I am curious as to the cost and availability of the power supply you suggested.  Do you have links or further information?

I would think some switching hash during transmit periods would not be an issue, as the power supply will be idle when receiving.

[KD6VXI]

1.  There was a gentleman many years ago who marketed a 3.5kv switching supply for a short time.  He is now SK so getting info from him would be an issue.  He designed them to be used by dxpeditions, etc.  And because iron was starting to get on the expensive side.

2.  If you're going to build a switch mode supply, might as well just modulate it.

3.  What about a pair of supplies.  Each xformer feeding its own bridge, then having the bridge outputs in series.  I had a 2 tube amp that had 4kv on it.  Needed some more voltage as the drive wasn't enough to reach peak output.  I put a 240 to 600v 6kva control xformer mounted on glastic to the original 3.8kv Dahl.  The control xformer fed a doubler that sat on the output of the FWB of the Dahl supply.  Instant 6.something kv! This amp has been running trouble free for 3 years.

--Shane
KD6VXI

[w9jsw]

You guys should start a new thread. You are light years away from pissweaker with this discussion. Suspect more may be interested if the subject better reflected the topic.