The AM Forum

Yo!

I've been dinkin' with various HV power supply layouts for the latest incarnation of "George" and "George Jr" which I'm calling "Georgette" based on the final 4-65A PA on page 186 of the 1957 ARRL Handbook.  I gave up on the single 4-125A 350 watt George and now am finally building something in the neighborhood of 200.  I posted awhile back wondering about bleeder values and finally came up with this.

All the iron is matched Thordarson.  We discussed the plate transformer pros and cons in another post.  With 115 volts on the high side of the primary, it is rated at 4360 VCT.  All iron is matched.  Swinging choke is 4-20 Hy at 300 ma followed by a 12 Hy at 200 ma.  Since in an AM mode the 4-65A operates at 120 ma, I'm good to go with the lower current draw smoothing choke.

I'm using a pair of 4 MFD @ 3 KV matched oil caps with a 250 watt, 60K Ohm bleeder. I'm measuring 2050 VDC output. Now, am I right in computing about 34 ma bleeder current draw (34.1666 actual)?  The bleeder wattage is higher than needed but it is new and ready to serve.  The plate iron is rated at 115 volts in but my measurements were taken with 120 on the primary.  I think I can live with that.  Will keep you posted!

Once the PA load is added I should have close to my goal of 1800 VDC.  ;D

Happy New Year from frigid Wisconsin!  And to think I had an offer to manage a greenhouse in Arizona back in the day and turned it down!   :'(

Mike
W9ZSL

The 2nd choke is a waste.
You only need one choke.

I've tried every combination of filters and caps including a single choke both swinging and fixed with choke input filters and capacitive inputs using every value and combination of caps from 4 to 8 to 10 to 12.  This gives me the exact unloaded voltage I need.  If it ain't broke, why fix it?

   You might try some LT P-Spice simulations of that filter. Here is a similar two stage filter designed for a much higher ripple frequency:

http://www.eetimes.com/author.asp?section_id=183&doc_id=1280238

   My experience with those old circuits from the day is that they were designed when chokes were cheap, capacitors were expensive, and rectifier tubes were very limited in peak current ratings, like the 816. Everything stated is different today, chokes are expensive, caps are cheap, and rectifier diodes have very high peak current ratings.

   All this hits home with me when I had an RCA AR88 receiver. This receiver has multiple stages of c-l-c-l-c filtering. I beefed up the audio output stage in an attempt to get audio down to 80 hz or so. Things were not working, and then I saw what the Power supply was doing. The B+ on the scope with a slow sweep looked like the bouncing ball over the captions of an old cartoon. The scope at 50V / division had several divisions +/- as I boosted the audio gain when tuned to an AM BC station with a disc jockey that had a deep voice..

   Your two stage filter will have multiple resonant frequencies depending on load current. If CW keying were done at a rate to excite that resonance, the power supply output could soar to 3 kv or more after a key up. As the power supply for a plate modulator, if the modulator passes a lot of lows, the filter resonance will be excited, and the modulator will modulate the power supply about as much as the RF PA tube. With filters such as this, they are OK if heavily bled, and for plate modulator usage, limit the lowest audio frequencies at a high level to something way above the filter self resonance.

   Following the EE Times article presented, the resonance issue can be tamed if the output choke is shunted with a resistor (about 9K) to minimize second stage filter resonant peaking, and to increase the output capacitor to something higher, such as >= 25 ufd.

Jim
Wd5JKO

well put, Jim ..... all boat anchor operators/restorers should remember this

Why the goal of 1800 volts?

Some extra voltage never hurt anything, and most things run better with more voltage.
My 3x4D32 rig (plate modulated) runs 1250 volts on the plates, with the 4D32 rated at 600 volts in plate modulated class C service.
Been running (with the same tubes) for many years with no issues.
The tubes work much better at that voltage.

For an RF deck, you load it up a little lighter, for a mod deck, you bias it up a little.

When I was young and stupid, I built power supplies with two big chokes, those UTC commercial series chokes rated at 7000 volts and 500 ma (CCS).
So now I would have to remake the things because I would have a big hole in the chassis where the nice 2nd choke was.
More capacitace is a lot better then another choke, and on the later stuff I use one choke and about 70UF of oil filled caps for energy storage.

The 2nd chokes are doing nothing but cause the regulation to get worse from the DC drop in that 2nd choke.
The first choke is needed to reduce the voltage and improve regulation, and make inrush current light, the 2nd choke is just making things worse.

I did bypass the 2nd choke, but there they sit, doing nothing but making sure the power supply is stable in a tornado, and covering some holes.

If you only have 2 or 4 uf to work with, then you need the 2nd choke to kill ripple, but that is no good for AM.

It was a different world in the old days, and fidelity and peak power was not on the radar, and oil filled caps must have been rare and expensive, or else why would so many rigs use small electrolytic caps in series at close to their rated voltages as filter caps?
The later 32V3 series rigs had an 8 uf oil filled cap (I put 10 in), but that is the only rig I know of that used them.
Maybe that was why it cost 4X the other rigs....



 

Why 1800 volts?  That's pretty much a "bench mark".  I can't get the voltage much higher than 2300 volts no load because of the plate transformer's ratings.  I also have to keep the input to around 200 watts because my UTC mod iron (VM-3) is rated at 125 watts.  I could try a different PS configuration on the bench again, this time with just one choke (swinging) and cap input using either an 8 and a 10 or a 10 and a 12 mfd all of which are rated at 3KV.

While I'm at it, I'll stick a meter in the bleeder circuit to double-check the current draw which I understand should be at leat 30 ma. This rig, actually an amp, is going to be used primarily for AM.  I have other rigs for CW...namely a DX-20, DX-60 and a Kenwood 440.

For AM, use choke input, and as much capacitance as you can get.

And mod iron tends to work better with less current through it.

I was never very fussy about the bleed current, the RF deck will bleed voltage, the bleeders help prevent the voltage soaring under no load, and removes the charge for safety.
A stiff bleed might help a bit for a modulator, but if run off the same supply as the rf deck, its not as important.

The only good supply for an AM rig is a choke input with plenty of capacitance at the output.
Any cap input supply has very poor dynamic regulation.

I would use whatever a choke input supply gives, if it is far off, I would look for different parts or change a design to use the voltage and current it gives.

There are designs that use full wave bridge for higher voltages, and the center tap for lower voltages, and with a variac, that would give a very wide range of dual voltages.

A swinging choke is best used for modulator service, the RF deck does not present a varying load.

Well, this is a KISS supply I just bench tested.  I got 2150 VDC and 32 ma bleeder current with 75K Ohms.  I can live with that.  I'm going to try putting the 8 and 12 in parallel that would give me 20 mfd.  Any problems with that idea?

As for the modulator supply, using the 4-20 Hy swinging choke, does a choke input supply fill the bill or should I go with capacitive with a 4 mfd - choke - 8 mfd?  I could also use a pair of matched 4 mfd.  I'm only going to have about 1KV and don't especially want to use the 8 because it is huge.

I would go with more then 12UF, and are you going to key the supply, or leave it run all the time.
The smart way (not the way I did it) is to leave the supply on and key the cathode of the tubes.

I did step start on all my stuff as I do not do break in much....

AM only.  I put the 8 & 12 in parallel for 20 mfd.  With the 60K bleeder I get 2050 volts with bleeder current 38 ma.  Using a 75K it's 2150 and 32 ma. 

Couple of new questions.  I've spotted a 75K @ 100 watt adjustable Ohmite resistor on eBay.  My test resistor is only 50 watts.  My calculations tell me I can use the tap for the screen voltage on the 4-65A and at 400 volts / 50 ma we are talking about 20 watts max.  I won't be running the screen that high...probably more like 250 volts.  Eimac shows screen volts being taken from the plate supply so I surmise taking it off the bleeder tap would work?

Second question.  You mentioned a "soft start" system.  I've looked through all my handbooks and can't find a circuit.  What say??

For the rf deck, you can get away with less, I think I have 10 uf.
For the modulators, I have 55 uf I think.

I have 60 in a rig that runs both off one supply, never got around to building a supply for the modulator.

I picked up some nice 40UF at 3000 volt oil filled caps a while ago, like when they were $10.00 each...

Choke input filters on both supplies?  I'd like to work with the capacitors on hand 'cause I've spent too many $$$ already.  I have 3 x 4mfd @ 3KV, 1 x 8 mfd @ 3KV, a 10 mfd @ 1.4 KV and a 12 @ 2 KV that I can parallel for the modulator supply for 22 mfd and finally a 12 @ 4 KV that can be paralled with the 8 for 20 mfd.

So...I can build the RF supply using the 20 mfd in a 12 Hy choke input with a 75K bleeder.  As for the modulator, swinging choke input with 22 mfd output and bleeder to be determined.  OR what for the modulator?  I'd have the 3 x 4 mfd caps that could be used anywhere.  The modulator will use a pair of 807s or 5933's.

Here is what I use on my Henry 2K-4:

This one samples the 240VAC going to the transformer primary.

If you are using 120V primary transformers, just select a relay with a coil for 120VAC and change the resistor to 10 ohms.

OK so far!  I just finished a bench test PS for the modulator using a 4-20 swinging choke input filter with 14 mfd cap (4 + 10) and a 60K bleeder.  I suspect I need more mfd.  Came up with 1025 volts and need 750 for the 5933's. So, like the PA supply using a choke input filter, how about the modulator?  I notice that if I test a cap input filter, naturally I get a higher output voltage.  I'm using a variac for the tests and really want to run the system without it.  The Choke input filters get me where I want to go with 120 on both primaries...and yes, I have several 120 volt heavy-duty relays for switching.

For the rf deck, use the choke and 10 or 20 uf.
For a modulator deck, use the swinging choke and as much capacitance as you can fit.

What voltage on the modulators?
Tubes like 807's like 750 volts, so the voltage is lower, and the current is higher, calling for more capacitance, but at a lower voltage.
1000 volt caps will be fine for 807's that run 700 to 750 volts on the plates, 1500 volts will be great.
807's will make 120 watts of audio in class B, needing driving power, not so easy to make it clean using tubes and iron drivers.

A single 4-65 makes 140 watts out at 1500 volts, and 230 watts at 2500 volts.
A pair of 807's is a poor choice of modulators.
They will modulate up to about 100 watts of carrier output, so you are short of audio power.
A good rule of thumb is as much audio power as rf output power.
But you can run the 4-65 at 100 watts out...say 1800 volts at 100 ma.

I can use a pair of 811A's into the UTC CVM-3 if necessary.  I'm working from  1949 diagrams out of a West Coast Handbook.  One shows a modulator with 807's and the second with 811A's  The PA shown in the second should be ignored because I'm going with a different circuit.  The main feature of both modulators is they work with the UTC CVM-3 which I have.  The speech amp / driver will be on a separate chassis from the modulator and it will be able to drive either of the two.

If you can, use the 811's, no screens to worry about, plenty of power output, and they work with about 1000 to 1700 volts. Above 1300, you need to put some bias on them, easy to do using diodes in the cathode.

An easy way to drive the modulator grids is a solid state audio amp into an 8 ohm to 5000 ohm transformer.
Quite hi fi, but maybe not as fun as a bunch of tubes and a D104.

Actually 811A's will work with voltages as low as 750 with 145 audio watts out.  I have a recording stuio with a HB limiter amp that has more than enough drive to do the job into a Kenyon 500 ohm balanced line to grids transformer.  Lots of possibilities!

Am xmtrs should never be using cap input filters.  Use choke input filters for both of the supplies.  Don't go crazy over the bleeder current, it is not that critical.  The bleeder resistors need to be high wattage.  A rule to follow is that the wattage should be no less than twice the wattage being dissipated.  Example, 1800 volts with say 60K bleeder.  The bleeder current is 30ma.  The power dissipated is IxIxR which is 54 watts.  In this case you need a minimum of a 100 watt resistor.  You can use two 50 watt resistors in series, anything in the 60K-100K range total.  In my HB xmtr I use two 50K 100watt resistors in series, the supply is 1500 volts.

Bleeder current can be a little more for the modulator supply and not as much for the PA supply.  The PA presents a constant load on the supply so no need for a lot of bleeder current.  The importance of the bleeders is to keep the supply voltage from rising above the voltage ratings of the caps if there was no other load on the supply.  If you key the supplies on and off (transmit/receive) the PA load will keep the voltages in check.  If you leave the PS on all the time and key the cathodes (something I don't do, better to key the supplies) then the supplies will have little load on them in stand-by.  This is where the bleeder current is more important because you can't exceed the voltage ratings on the caps.  If you key the supplies you'll need a step-start (also called a soft-start) relay.  This circuit is simple, it uses a relay and a resistor (usually about 10 ohms).

For the modulator, forget the 807s, use the 811s as you need a lot of audio power.

Caps can be 10uf-20uf for the modulator supply and 20uf-30uf for the PA supply Or whatever you can muster together.

I use two chokes in my single supply.  I run the modulators off the first choke/cap and the PA off the second stage.  But most only use a single choke and I may be changing my supply to only use a single choke.

Fred.

For running both modulator and rf deck off one supply, you can have one output section (swinging choke and caps) go to the modulator, and the regular choke and cap/caps for the rf deck.
No point in having all the current flow through the first choke.

Seperate supplies are much better, that way you can have the modulator voltage a bit higher which makes for good audio peaks and a better match with most iron.

I am not sure how clean 811's are in class B at lower voltages.
With some tubes, you can adjust the bias for lower plate voltage to avoid crossover distortion, with zero bias tubes, you would need to add positive voltage at some point to avoid that...

I think the tube manuals show the 811 starting at 1000 volts, up to 1200 volts no bias, minor amounts of bias above that, and you can run them at 1750 with enough bias.

I never found any distortion ratings for the 811 at various voltages....


If you need any parts, let me know, I am always interested in helping anyone who is actually BUILDING something...

If bleeder current isnt that critical then I have it covered except maybe for needing more caps to get the values high enough for the modulator supply.  If a 12 Hy choke followed by a 12 mfd cap will work for the PA, then I'm home free.  I take it that even though the modulator transformer is only rated at 125 watts, having the extra headroom in reserve isn't critical unless the driver level is so high that the iron is saturated.  I can easily get 1KV or better using my second Thordarson plate transformer. A single supply wouldn't be practical in my case because each plate iron is only rated at 300 ma.  A pair of 811A's at a KV (pardon the pun) sounds good to me!  N2DTS, emal me at W9ZSL@yahoo.com please re: a few parts I may need...like another 811A.  I have one.

I just finished with a bench test of a PS for the modulator using my second Thordarson plate iron with the swinging choke followed by 20 mfd cap and a 75K bleeder which is drawing about 20 ma.  I came up with 1200 VDC so I would have around 900 volts or better for the 811A's.  With a 100K bleeder, I get 1275.  That should do it for both supplies.  

Screen supply for the 4-65A will be on the RF deck using a SS bridge that will come on with the plate supply.  Bias will come on with the filament. Thanks for the help 73!  :D

Mike - W9ZSL  

You're right about not running the PA current through the first choke.  I do that with my 6146/6550 rig.  I use a separate choke and cap for the 6550s which comes off the same 650V plate supply as the 6146.  The 6550s draw so much current that it was causing some downward modulation due to plate voltage dip through the two chokes (I always use two for PA).  The separate choke solved that problem (total 3 chokes).

I've tested many chokes to see which work the best.  The worse ones are the UTC S-line.  Most Thordarsons due a good job at filtering especially the older ones.  The UTC PA series are also very good.  The UTC CG series are also good.  All the Stancors due a fair to good job at filtering  The Thordarson 19 series chokes are good but not as good as the series that predate them.  The last series that Thordarson made which were the 20 series are again not anywhere as good as the earlier ones.  The Thordarson 19 series plate xfmrs are very light for the ratings. The plate iron that predates the 19s are much better.  The Stancor 80 series plate xfmrs are good and can be used with a FWB rectifier.

Fred

All of my iron for both supplies is Thordarson.  I'm using a T-19P59 which is used for the modulator along with a series 19 4-20 Hy NOS 300 ma swinging choke.  The PA plate iron is a T-16PO3 with another 12 Hy 19 series choke.  Both NOS.  Everything is matching and all rated at 300 ma.  I REALLY got lucky pickin' and except for the modulator plate iron, the other three pieces came in the original boxes.  The 16PO3 arrived in it's original wooden box complete with the Allied Radio sales slip and warranty card!  Sweet!

The 16P03 is a better Thordarson plate xfmr than the 19 series iron.  The 19 series chokes are also good.  I have on the bench a 19P59 and a 19P60 that I'm repainting and improving some of the terminals.  I have no plans for them right now but maybe in the future.  I also have 4 or 5 19 series chokes.  In my HB xmtr I'm using a Thordarson 6411 plate xfmr.  It has the same ratings as the 19P59 but has much more iron in it and heavier copper. The Thordarson 4 digit part numbers predate all the iron that have a letter in the part number.  Through out their history, Thordarson made thousands of different xfmrs.  Thordarson also liked to change their part numbers every few years.  Stancor almost never change their part numbers and they made the same xfmrs year after year.  But, they did changed the construction of them through out their history.  So, some Stancor xfmrs may have the same part number but may be constructed somewhat differently.  You can tell the apprx age of the iron by the way it looks.

The 19P59 is perfect for the 811 modulators.  The class B modulators present a short duty cycle on the xfmr so it won't work up a sweat powering the modulators.

I run both my 811 modulators and the 813 final from the one 6411 xfmr.  It is only rated for 280ma but handles long winded transmissions without getting the slightest bit warm.

You may be able to run your xmtr from the one 16P03 .  Although I forgot the voltage of the 16P03,  I'll have to check my catalog.
You lucked out getting the NOS ones.
Fred

Hi Fred!  The 16P03 is 4280 CT using the high side of the primary or 3480 using the low.  The 19P59 has dual secondaries and is 3120 and 2500 CT.  I bought it specifically for the 811A's and it should give me 1KV. Two supplies will work out fine since I have all the parts I need.  I notice your comments on the UTC iron.  My modulation transformer is a VM-3 which comes in a "G" housing (as I recall) the same as the "S" series.  It is identical to the CVM-3 except for the round can.  Mike

The VM-3 and the CVM series mod iron are not the same as the S series.  Although, the S series mod xfmrs are widely use and should be good.  I only have one S series mod xfmr.  The VM-3 is older than the CVM series and it should be good.  The 16P03 has the tapped primary instead of the dual secondary that the 19P59 has.  The only thing you should check on the 16P03 is the voltage rating of the primary.  The early Thordarson plate xfmrs were 110VAC primaries.  Using a 110 volt primary on 125 volts may saturate the core causing it to run a little hot.  I don't remember what the 16P series plate xfmrs were rated at, 110 or 115.  Check the metal tag, it will show the primary voltage.  That rating is for the high voltage tap.  The low voltage tap can handle higher primary voltages.   As for the S series, I tested a few of the chokes and they did a poor job at filtering than most all other common chokes in that current range.  The S series plate iron are good, somewhat better than the 19 series Thordarsons.

Having said that, your 19P59 will work fine for the 811 modulators.  The 19P59 should get you 1000 volts or 1250 volts.  I would run the 811s on the higher voltage.

My HB 813/811 rig uses a Thordarson 6411 plate xfmr.  It, like the 16P, has a tapped primary.  It is only rating for 110V primary, but I run it on the low voltage tap.  The 6411 is rated like the 19P59 to give either 1000 volts DC or 1250 volts DC.  Running the low voltage tap on 120VAC gives higher than 1000VDC.  It is about 1150VDC.  My rig runs on 1500VDC both the 813 and 811s.

So, how did I get 1500VDC from a xfmr yielding only 1150VDC?  I added another 350 volt xfmr to the supply.  How did I do this?  Using a second xfmr that will yield 350 volts from the diode rectifiers (no choke no caps).  Just the xfmr and it's rectifiers,  connect the output from the diode rectifiers directly to the center tap of the secondary of the plate xfmr and it will bring up the output voltage by the 350 volts.  The current rating of the boost xfmr should be the same as the plate iron.

You can use this method with any xfmr that can handle the CT being above ground.  Your 19P and the 16P can handle it easily.  In fact, most any xfmr can handle the CT tap running above ground by a few hundred volts.  There are others that the CT must be grounded, so can't do it with those.

You can use any type xfmr for the boost.  The rectfiers can be FW or FWB.  With FW, the CT of the boost xfmr is grounded as normal.  With FWB, the negative terminal of the FWB rectifiers is grounded as normal.

If you want to lower the voltage, you can just shut off the boost xfmr and the output voltage will drop by the 350 volts.  No need to disconnect anything, just shut off or turn on the boost xfmr.  In fact, you can leave on the boost xfmr and shut off the plate xfmr and you'll only have 350 output volts.

This boost method will also work with both xfmrs using FWB rectifiers.  If the main xfmr uses FWB rectifiers, then connect the diode output of the boost xfmr to the ungrounded negative terminal of the FWB rectifiers of the main xfmr.

Fred

I'm assuming both the CVM-3 and VM-3 have the same pin-outs?  This VM-3 has the same case as the "S" series.  Both Thordarson plate transformers are 115 volts.  I took measurements with 120 on the primaries and should have about 1900 VDC on the PA and 1 KV on the modulator which will give me a potential of about 175 watts of audio and about 228 watts input RF.

The VM mod xfmrs are part of the same vintage PA series.  They predate the CVM mod xfmrs.  You're right they have the same case style as the S series but they are not the same.  The 125 watt CVM mod xfmr has a round case.  I'm not sure of the pin outs for the CVM mod xfmr but it is probably on line somewhere.  I may have it in my papers, I look for it.

OK on the plate xfmrs, the 115V primary should be no problem at 120 volts.  You need to use the higher plate voltage on the 811s.  With 1900 volts on the PA you'll never make 100% modulation with only 1000 volts on the modulators.

As for the pin outs,  most all these multi-tap mod xfmrs are usually made the same way.  There should be two windings on one side and two windings on the other side.  Each of the four windings usually has a tap.  This gives a total of 12 pins, six on each side.  The taps on each winding are not in the center, they're off center.  The total winding on one side is not the same total to the other side.  Meaning, there is a step-up in impedance from one side to the other.  Or reversing the xfmr there is a step-down.  Because, you're using higher plate voltage on the PA than the modulators, you'll probably need a step-up in the turns ratio.

The best way to connect the mod xfmr is to use the entire winding (no taps) on the modulator side.  The modulators should be connected to the lower impedance side of the xfmr.  You can then use the entire winding on the PA side, this will give a slight step-up in voltage swing.

Fred

DMOD, in regard to the inrush limiter, I take it where the upper relay contact lead to R1 crosses the lower contact's lead form the black "mains" lead, there is no connection?