New 4-1000A modulated by 4-1000As AM Project - Fabio II Reborn

[W2NBC]

http://amfone.net/Amforum/index.php?topic=31731.msg247041#msg247041

Bruce's fine sequence with HV primary switched in the mix..

and Heising configs.  I use #2..

[KA2DZT]

Tom

Looking at your line-up, it seems that you should have the drive on ahead of the final.  You have the reverse of this, steps 2 and 3.  Also, no reason to not have the modulator come up at the same exact time as the finals.  With the mod tubes not on there is no load on the mod xfmr when current starts through the mod reactor to the finals.  At a glance, to me that seems like it could be problematic.  In reverse order, you're removing the load off the mod xfmr by opening modulator CT and then after that event you open the final CT.  Big back emf with nothing to absorb it.

To answer your question, antenna on, drive on, HV step-started on to both finals and modulators and then audio drive last.  I'm assuming that you're using some or all self grid bias or maybe some fixed grid bias.  You need drive to limit the plate current, that's why I think drive should be on ahead of the finals.

You're right about step-starting the HV with 100ufd cap.  Is it one cap or number of caps making up the 100ufd??

Fred

[K1JJ]

Jeff,

I am using #3 from the Heising diagram you posted above.  Do you think there is a difference in better bleed off using #2 vs: #3?    Maybe the Heising cap going directly to ground is better than the hot end of the HV supply, I dunno.  Remember that I am using a negative floating HV ground for metering. (one ohm)

Fred, I tried keying the driver first and had fireworks.  The screen arced badly as well as the mod gaps.


After doing even more relay delay, etc tests last night, I've tried about everything that has worked in the past.  No choice but to go to the HV keying idea.  That HAS to work, even if I have to leave the ant relay on for 5 seconds... :-)  With HV keying, the modulator and RF final will always be keyed and sucking power if I have the drive on so eventually will bleed off the EMF.

Using Bruce's method I would also need to key the RF screen. That will require a HV relay of some sort. I may have to key the modulator screen too.


I have a pair of 240V AC contactors I may build up...

T

[KA2DZT]

Tom,

I use #2 Heising.  You're using #4, you may get a better bleed off of the back emf from the reactor with #2.  You'll have to try it to see if it helps.

I mentioned the RF drive only so there is enough bias when you power on the finals.  This would have nothing to do with the arcing.

Not sure if you have tried different sequences on the up-start but,  I think you have the finals and modulators wrong.

IMO I think you need to turn on the modulators before the finals, you're doing the reverse of this.

Try this line-up.

1,  antenna on
2,  modulators,  this puts a load on the mod iron.  At this point with no audio drive you don't need any load on the mod xfmr secondary.  So it doesn't matter that the PA is not loading the mod iron.
3,  RF drive
4,  final PA
5,  audio drive to modulators

Key off, the reverse order.

When you un-key the finals the large back emf from the reactor passes through the mod iron secondary.  The modulators, at this point, are still drawing some current which keeps a load on the mod iron primary which keeps a load on the back emf.  This may reduce the arcing.  It may be worth trying this.

Fred

[KA2DZT]

Tom,

After reading you're comments about the screens.  If you remove the HV from the plates you must remove the screen voltage.

Another scheme to consider,  you can turn off the modulators and the finals by simple removing the screen voltage.  Without screen voltage you could leave the HV on to the plates.  The output from the tubes should drop to near zero without screen voltage.  Although, with 4-1000s you would have try this to see what output may still come from the tubes.

Fred

[WD5JKO]

Tom,

After reading you're comments about the screens.  If you remove the HV from the plates you must remove the screen voltage.

Another scheme to consider,  you can turn off the modulators and the finals by simple removing the screen voltage.  Without screen voltage you could leave the HV on to the plates.  The output from the tubes should drop to near zero without screen voltage.  Although, with 4-1000s you would have try this to see what output may still come the tubes.

Fred

    Just arrived at a San Antonio Hotel as part of a three day business trip.

Heck, back in the 1980's when I ran a 1 KW DC input final, I use #4 Heising configuration with the cap on the top side. Back then few if any used a sequencer. I just keyed things, and seldom had a problem with flash over. Rigs back then had a distinctive 'Ker-Klunk" which was a give away to who keyed up. Some folks didn't even have to speak. The "Ker-Clunk" followed by 60hz hum during a QSB fade meant a BC-610...

So reading Fred's last post, I think keying the screen might do it. The back EMF spike will be a positive going spike above the B+ level. If the 4-1000 screen is keyed to ground, some plate current will still flow. When the back EMF spike comes, the plate should absorb it. To cut off the 4-1000 this way, I suspect the screen needs to go negative several hundred volts. So the screen would have to go down in two steps, say zero volts, then -500V.

Back to normal programming as I prepare for tomorrows festivities in a semiconductor fab clean room where I am about to rebuild an old machine.

Jim
WD5JKO

[K1JJ]

Ya know, youse guys may have a good point about the screens.  When I first fired up the rig, the screen circuit was arcing like crazy on unkey and took out a disc cap. I even added arc gaps to the screen inductor.   Since changing the sequencing, the screen arcs have disappeared.

But still, the screen may be rearing its head in the form of the HV mod arcs.  Since I cannot solve the problem with any combination of HV relay delay  or sequencing techniques, maybe keying the screen will help.  I ended up keying the screen on my old 4X1  PDM rig and this solved the problem a few years back.

I will need a HV relay since the screen is 500 V and has inductor EMF.  Jim, I presently have a fixed bias on the 4X1 RF final of -90 volts. This cuts the tube off, even when keyed, as long as there is no drive.  Should I simply add a screen relay that takes the screen from +500 volts or ground?

Would I continue to key the CT of both the RF tube and modulators  -  or instead, ground both CT's and key both screens of the RF and modulator?  (The HV wud stay on)

T

[KA2DZT]

Tom,

If the tubes are cut off at -90V bias, does this mean there is little plate current??  If so, then why are you keying anything.  The tubes are already cut off.  Maybe what you should do is to increase the bias to a greater negative bias on receive.  This would allow you to leave the screen and plate voltage on.  This should eliminate all your arcing problems.

For the modulators remove the screen voltage on receive.

You would leave the CT connected.

Fred

[K1JJ]

Tom,

If the tubes are cut off at -90V bias, does this mean there is little plate current??  If so, then why are you keying anything.  The tubes are already cut off.  Maybe what you should do is to increase the bias to a greater negative bias on receive.  This would allow you to leave the screen and plate voltage on.  This should eliminate all your arcing problems.

For the modulators remove the screen voltage on receive.

You would leave the CT connected.

Fred

Fred,

Correct, the finals do not conduct with the -90V on until I key the RF drive.

I have already tried what you suggested, but when the drive gets cut, the arc occurs. The HV and screen stay on. I grounded the finals and mod CT's all the time.   No difference.

I'm starting to think keying the HV with a step start along with the screen might work.  I wud ground the CT's.  Bruce sent me a screen snubber circuit that might help too.


T

[K1JJ]

I'm looking at Bruce's sequence.  He uses a slightly different order for TX vs: RX.  With my sequencer, the keyed order is the same, but reversed when unkeyed.

But what if I used his unkey order for both key and unkey?


It would look like this:

CT of both finals and modulator always tubes grounded.

1) T/R ant relay on
2) HV step start
3) Screen voltage on - for  both finals and modulators
4) RF drive
5) low level audio on

Reverse order to unkey.

I'm still not sure if the -90V fixed needs to be lower so that the final conducts SOME plate current even with no RF drive.  How else can it dump the power?  Right now as soon as the RF drive is cut, all currents stop dead.

** - I ordered two vac relays for the screens. Might as well get the screens switched, no matter what I end up with.

T

[K5IIA]

http://www.youtube.com/watch?v=hIkNY5xjy5k

tom sent me that link. looks like he almost has his spark gap dialed in.

[KA2DZT]

That sequence would be correct if you use the step-start HV when keying on.  It seems, with all the testing you done, that you covered all the ideas that I suggested.  I guess when you use a big reactor no amount of loading is enough to damp the back emf that occurs when the plate current stops dead.  Also didn't realize that removing the drive cut off all plate current.

The reverse of that order will NOT work when keying off because you're removing the drive before killing the HV.  You need to cut the screens first then the HV then audio drive and last the RF drive.  I'm not sure if the sequencer can handle two different orders, one order keying on and a different order keying off.

Although, cutting the screens first may also cut off plate current and still cause back emf.  You may need to cut the HV first and estimate the time it takes for the HV to drop to a level near the screen voltage and then cut the screens.  The thing you need to worry about is having screen voltage present without plate voltage.  This causes the screens to draw heavy current which could exceed screen dissapation.

In order to keep some plate current you would need to raise the negative grid voltage to something less negative.  In fact you could use some fixed bias and some grid leak bias by using some value of grid resistor.  You could also use all grid leak bias and then you wouldn't have the lack of drive cutting  off all plate current,  you would have the opposite, too much plate current with no drive.

Fred

[KA2DZT]

Looking at the sequence,  you could put the RF drive ahead of the HV then the screens and last the audio.  I think the tubes can handle having drive without plate voltage for a second.

antenna
RF drive
HV
screens
audio

Then you could just use the reverse order to key off.  But, again cutting the RF screen ahead of the HV may be a problem because it may also cut off RF plate current and cause back emf.  You may need to have a slow time decay of the screen voltage.

If you were running the RF screen from a dropping resistor off the plate voltage you wouldn't have  this problem.  I think you using a separate screen supply and screen reactor to modulate the screen.  Instead of cutting the screen voltage between the reactor and screen you could  cut the circuit in the supply before the filter caps and let the caps discharge the screen voltage more slowly.

[K1JJ]

OK, thanks, Fred.

Well, I ordered two 60A used contactors from eBay to build a HV step-start.  I overrated them in case I make a mistake and get a short.

The screen vacuum relays are on the way too.

So this will give some new options to beat this arc problem. 

I like your idea of experimenting with the screen supply by keying the xfmr primary, to get a slower decay.  I will also try switching before and after the screen modulation choke to see what works the best.

So, back to the drawing board with Fabio II for now. 

Been outside on the towers doing catch up antenna work.  I'm finding the pair of 20M  300 pound stacked Yagis are getting too much to maintain.  They are scary to work on. So in the last two days I took them down. Who knows what I'll put in place of them, but I think as I get older I will start shifting back to wire arrays with ropes and pulleys...

T

[w1vtp]

http://www.youtube.com/watch?v=hIkNY5xjy5k

tom sent me that link. looks like he almost has his spark gap dialed in.

COOOL!

[W2NBC]

T,

You've now taken Fabio II to new heights.. Complete control of voltage.. In fact, you have officially become the second "Electron Wrangler".. (can't tell you the first, because he'd have to kill you).. Yes "I can see it now"..

[WA3VJB]

And Jeff I applaud your selection in that image of what appears to be a couple of 50C5 in shove-yank.

[w4bfs]

I've been following this thread since 2009 with some interest as I have several similar projects in mind ... I do not want to seem to be attempting to steal this thread as it is an interesting problem with so far a difficult solution ...

the problem as I see it is what to do with parasitic energy generated by current carrying inductors when the current they are carrying is turned off ... the energy in a 72 Henry inductor carrying 0.5 Amp at turn off can be roughly approximated as   E = L x i x i  or 18 Joules ... this energy manifests itself as a voltage spike of reverse polarity across the inductor as the internal magnetic field collapses

In the am transmitter under discussion ( maul) what happens to this spike is that it tends to appear across the rf amp anode circuit unless a B+ relay is used .... the value of this voltage can be approximated by conserving the 18 Joules of energy produced and calculating the resultant hv applied to the anode circuit bypass capacitor      E =  .5 x C x v x v   assuming the use of a 500 pF bypass cap and ignoring stray capacities this calculates to be approaching 268 kV ... no wonder things are arcing ... this is why the spark gaps are firing

we need some snubbers ... hv movs can be found but tend to be expensive .... consider the use of a string of say 500 V movs ... I bought a bag of them off Epay for $.25 each .... the data sheets were found on Mousers web site and I am considering the use of a dozen in series to snub a 2kV mod reactor .... when you put this many in series the usual 300 pF capacity decreases to 25 pF and will not mess with the maul's high freq response .... add a series resistor and a neon bulb and you can have a visual indication rather than the crackles of thunder on firing

I think you will need this as well in the screen reactor .... good luck .... 73     John

[K1JJ]

T,

You've now taken Fabio II to new heights.. Complete control of voltage.. In fact, you have officially become the second "Electron Wrangler".. (can't tell you the first, because he'd have to kill you).. Yes "I can see it now"..

Yeah,  COMPLETE CONTROL with the Electron Wrangler.  That's the ticket!


But first, "If I only had a voice."  (Sung to Oz, "If I only had a brain")


"I would talk away for hours, squawk about my towers, with vocal fry that's choice.

I'd sound just like Walter - my chords would never falter... if I only had a voice."


Now all I's needs is senior anchorman  lessons from youse guys, caw mawn.

T

[K1JJ]

we need some snubbers ... hv movs can be found but tend to be expensive .... consider the use of a string of say 500 V movs ... I bought a bag of them off Epay for $.25 each .... the data sheets were found on Mousers web site and I am considering the use of a dozen in series to snub a 2kV mod reactor .... when you put this many in series the usual 300 pF capacity decreases to 25 pF and will not mess with the maul's high freq response .... add a series resistor and a neon bulb and you can have a visual indication rather than the crackles of thunder on firing

I think you will need this as well in the screen reactor .... good luck .... 73     John

You make some good points there, John.

I do intend adding a snubber to the screen circuit. Bruce has suggested a series  .1 - .5 uf  and 500 ohm cap before the screen choke from hot to ground.

The MOVs sound interesting, though I know some who say they fail a lot and might be trouble.

There's a lot of things to try now that I will be keying the HV and screens.  I will look into the HV snubbers if I need to also.

It seems like a difficult task (and juggling act) to dump 70H of energy.  The transmitter plate circuit has to conduct to dump energy into the antenna - but at the same time it needs to slowly die off or this same conducting energy will have no place to go.  Like a Catch 22.

The bottom line is if the arcing gets out of hand, we will be rewinding our modulation transformer - and possibly the Heising reactor - and possible tube and parts damage.


T

[KA2DZT]

Tom,

My PW version of your xmtr also uses a 50hy Heising reactor.  Like I've mentioned,  keying your supplies should solve your problems.

I step-start my HV supply (1500V) and I never had any arcing problems.  I also have no spark gaps on anything.

I don't think you need any snubbers on the screen supply.  All you need to do is turn on and off the screen supply before the filters.  Let the filters discharge the screen voltage.  You could add a few ceramic disc caps (5KV) .01-.05 on the filter caps to ground.  These caps have less inductance than electrolytics and will act faster to short any spikes that might come from the screen reactor.  Also don't use any method to open the screen circuit after the filters.  So forget relays on the screen line to the tube.  Just leave the screen circuit intact.

The same is true for your HV supply, a slow decay of the HV will produce less energy from the reactor, if I remember all that calculus that I mostly forgot.  Don't use any method to open the DC plate circuit.  So forget the cathode relays, leave everything connected.

You just need to use the right sequence for keying on

Antenna on
RF drive on
HV on
Screens on
Audio on

Keying off,  the reverse order.

The timing between the screens and HV should be close.  You don't want the absents of screen voltage to cut the tubes off before most of the HV has decayed.


With the speed at which you build stuff, I'll expect these changes to be completed by the end of today.

Fred

PS to my comment about the reactor producing less energy.  The reactor will have the same stored energy but release it at a slower rate with a slow decay of HV. which should result in less of a voltage spike.

The voltage induced in a coil is directly related to the rate of change in current through it.

[K1JJ]

Tom,


Antenna on
RF drive on
HV on
Screens on
Audio on

Keying off,  the reverse order.

Fred

OK Fred -

There certainly are wide varying opinions on the sequencing....

Looking over yours again, it makes sense for my particular use. Plus it can be reversed in the same order.  Think I'll try your sequence above.  Tnx.

I'll be keying the HV primary and the screen primary. This shud soften things considerably. The RF final and modulator CT's will be always grounded.

I like that the RF drive turns off late, just before the antenna relay drops.  This will give the amp every last chance to dump the power into the antenna as the HV and screen supplies are decaying.   I will try it with and without a fixed grid bias to see what I can get away with.

The contactor and vac relays will be here later this week.

Today I added the 1 ohm resistors to the filament primaries. The 7.85 fil voltage dropped almost perfectly to 7.53 vdc at the pins. It shud be 7.50 vdc.   The problem is I under-estimated the resistor power dissipation. At 3.92 volts AC drop and 2.6 A we are talking 10 watts of heat.  The 5 watters I had in there were not happy. Ordered some 100 watt heat sinked wire wounds as overkill since they were cheap enuff..

I'm confident when the new parts arrive the arcing problem will be solved. Evidently linear amps don't mind harsh CT PTT keying while larger plate modulated rigs with their reactive EMF like to stayed keyed and bled off slowly.

I'll post more results as the parts come in and get installed..

T

[w4bfs]

perhaps we should move this to the technical forum as I intend to further muddy the waters ....

relay sequencing is a good idea for many reasons in an am transmitter but I am not sure that it can be done in a fashion that will keep a system under control with regard to unwanted parasitic energies without addressing them directly, so here is a possible solution .... I mentioned movs as they have the advantage being passive until triggered .... a disadvantage mentioned is their reliability .... were they properly applied ? .... there are energy limits as a time function for these devices

for the sake of argument lets assume that movs are not practical for this application .... is there another way to snub ?   yes since we are considering a sequencer, could we not add a sequenced snubber to the modulation reactor ....  this seems doable .... a hv relay sequenced to parallel a high power resistor across the mod reactor at the end-of-transmission just before dropping the carrier could dump and eliminate the parasitic energy .... the advantage of this that it removes the requirement of this action from the rf chain

most interesting, Mr. Spock

[W3RSW]

Well, a diode across the inductance is the traditional way.  ..  You only lose half the audio. 

Can't believe the length of this topic. 

Here's a method of screen protection when using the choke method from the Orr handbook, page 296, 15th addition.  This method might help suppressing back emf across the choke by tube action. Partially Paraphrasing:

"Place a resistor in series before the choke right after the separate screen supply of about 3k to 10k ohms' fig, D, and the screen supply should now be about 1 1/2 times normal with the actual value of the resistor sized that normal screen voltage is supplied at normal screen current."

"When the plate voltage is removed, the screen current will increase greatly and the drop through the resistor will increase to such a value that the screen voltage will be lowered to the point where the screen dissipation of the tube will not be exceeded."

So might this action be approximated through every current cycle, keeping inductance swings "gently" manageable in their charge and discharge?

[K1JJ]

John,

Good idea on using a relay to place a big power resister across the Heising choke during unkey.  I  think this would work.

Frank / GFZ sent me this idea yesterday too.  Wish I had heard it before ordering the new parts, but that's the way it goes.

I suppose it doesn't matter if the energy is dumped into the antenna or into the resistor, same result.

I will give it some more thought as I wait for the parts, but at this point think I will still key the HV and screen circuits via their primary 240VAC transformers and ground the CT of both final and modulator. This is how most big plate modulated rigs do it. Might as well do it right where the EMF spikes will be naturally bled off into the antenna load.   But if the new parts don't do it, I will certainly give this resistor method a try.


Rick - OK on using the resistor in the screen.  I suppose that would work for smaller rigs, but using a 3K resistor would drop 235 watts of heat across it in the 4X1 rig.  A 10K would produce 783 watts...      So solly, Mr. Orr.  

T