Reactor's sparking when carrier goes off

[KA2DZT]

I agree with Don,

I use a single diode limiter with a bias voltage.  I don't use the load resistor, the impedance of the bias supply RC filters creates a load to ground.  Works perfect and it doesn't create any distortion.

As far as the OP's problem,  how are you putting the xmtr into standby??  Switching off the HV plate supply?  Or, leaving the plate supply on and biasing the finals into cutoff by opening the cathode circuit??

Opening the cathode circuit is almost the same as opening the HV B+ line to the mod reactor with a relay.  Doing so will create very high voltages in the reactor with the sudden loss of current.

A single-diode-negative-peak-limiter circuit connected to the Class C terminal of the mod reactor will help to absorb the negative spikes.  Positive spikes should be absorbed the the Class C stage.  Another problem are the spikes (+ or -)that appear at the B+ terminal of the mod reactor.  You might think that the filter caps in the supply would absorb them.  Probably not,  those caps have too much inductance to absorb fast-rise-time spikes.  You would need to have a number of HV disc caps connected to the B+ terminal and ground to help absorb the spikes.  Using a spark gap across the mod iron is another way (as mentioned by others) to absorb the spikes.

I think JMO that it's better to switch off the HV B+ supply on standby and use a step-start to restart the B+ supply on transmit.  The filter caps in the supply allow for a time constant for the current to drop to zero.  This should greatly reduce the HV spikes across the mod reactor.

Just some of my thoughts which may or may not be useful for your transmitter.

Fred

[K1DEU]

My mod reactors need their Q dumped,excessive ringing. Therefore I place some R across it. John, K1DEU

[AB2EZ]

Bruce

Hi!

In response to your question, pasted below.

If the modulated B+ tries to go low enough for the rf tubes to stop conducting, then the average current that is still flowing in the secondary of the modulation transformer (or in the reactor, in the case of Heising modulation) would force the modulated B+ to go back up (positive)... and the rf tubes will continue to conduct.

So, the "hitting the baseline" effect is an issue with distortion on the modulated envelope (splatter, etc.).

The transformer/reactor damage problem occurs if something other than a drop in the modulated B+ causes the rf tubes to stop conducting. As examples:  suddenly forcing the screen voltage to zero, or suddenly disconnecting the grid drive, or something that causes the B+ line to suddenly open up.

Best regards
Stu

Hi Stu,

Thank you for your usual detailed and insightful response.

This would probably explain in large part why Gates, who almost exclusively used triodes as modulator tubes, did not see the need to include spark gaps on the windings of the modulation xfmrs or reactors used in their plate modulated BC rigs.

To carry this one step further, and the answer is probably obvious, but would this also suggest that plate moduated rigs employing triode modulator tubes would be less likely to exhibit modulation transformer or modulation reactor failure in the event of periods of over-modulation, where the carrier is cut-off and the modulation transformer is effectively unloaded?

73,

Bruce

[W2XR]

Thank you very much, Stu.

73,

Bruce

[k4kyv]

I never put the transmitter in stand-by by interrupting the plate current to the final.  That's just asking for trouble from transients and soaring power supply voltage, plus the safety hazard of keeping a transmitter running for hours in stand-by with the full +HV present.  Not only might it be easy to forget that it's there, this also causes unnecessary stress wear on the HV insulation in certain components, particularly the mod xfmr and reactor if their case is grounded.

I prefer to use large "contactor" relays to interrupt the a.c. voltage to the primary of the plate transformer(s). I use choke input filtering, and as much filter capacitance as I can without having to resort to step-start, while using a low enough resistance in the bleeders to maintain critical inductance in the filter choke in the event of loss of load on the power supply, either inadvertently or intentionally, as in the case of operating CW.  Depending on the voltage and current drawn from the power supply, I have found between 25 and 50 mfd to be optimum for the filter capacitance.

If you do interrupt the +HV to the final but leave it on all the time, it would better to break the cathode circuit as long as the insulation in the filament transformer and filament by-pass capacitors can withstand the voltage.  You will not see the full HV appear across the switch or relay contacts.  When the cathode circuit is opened, assuming the grid resistor and bias supply are taken directly to ground, the open contacts on the relay act like an infinitely high cathode resistance.  Current ceases to flow through the tube as soon as the cathode voltage rises above the grid bias cut-off voltage of the tube, so you might experience a couple of hundred volts, depending on the tube, not the full plate voltage that may exceed several kilovolts.  If, OTOH, the switch or relay is placed in the plate side of the circuit, the full HV will appear across the contacts when they open. The contacts in the HV switch or relay are much more likely arc over upon stand-by if it is placed in the plate side than if it is placed in the cathode side.

The same is true for cathode keying a CW transmitter.  Contrary to common belief, the full HV does not appear across  the contacts of a hand key.  Of course, the full voltage will appear across the contacts in the event of a plate-grid short in the tube, so I would be careful when cathode keying a final with kilovolts of DC on the plate.

[Steve - K4HX]

No interruption needed. Just bias the final off and remove screen voltage (if there is a screen). Thousands of linear amps have worked FB for decades this way.

[AMLOVER]

 Hi everybody,

I have read very carefully all your comments about negative peak diode limiter and turning on-off the C class amplifier.
As it has to do with my situation there are some things I can do and some I can't because of the specific construction of mine.
I bet to myself to keep the solid state af amp as modulator even it is very difficult at the moment to fully protect it because :
a)I designed the modulator iron for 4-8ohms primary and
b) this class H mosfet amp has amazingly low dissipation/distortion in comparison with any B class push pull amp I've heard in the neighbourhood.
   
  Steeve (Stu) gave me the best solution to my nr1 problem, how to turn on-off the transmitter without any spark or even any little flash by keeping also the Hv always on because it is complicate to step start it (2 phses with two HV transformers in series etc etc... anyway everything around it is super insulated-no chance for shortings, tested under very stressed situation.
Now I first turn off the audio chain output, then with the same switch I turn off screen and short the reactor's ends by a resistor=Rmod and finally I turn off the driver.As Steve K4HX said, I don't open the cathode as far I use -200v fixed bias and tub cuts off as soon screen and drive turn off.
Don,
spark gaps that I also have got, are fine for irons protection but for every spark they create a trancient travels through windings  back to the solid state, where they shouldn't appear.
Fred,
I just noticed that I don't have any 0.001uf/10kv on the reactor's Hv supply side, I'll put some tomorrow that I'll also add the diodes with zeners in series to the relays' coils.
Are the solid state af amps very sensitive or there is still the possibility to block any trancient from jumping through the modulator's windings in the output of the solid state af amp and kill the mosfets?
 I don't want to quit and go for a B class pp triode amp and let the trancient without punishment laughing at me...  I hope next time to catch it...
Check please the attachment.

Stefano

[WD5JKO]

Stefano,

   Problems like this can be cured at the source, or you can beef up things where susceptibility occurs. Of course you can also do both.

   The audio power amplifier at 3200W into 4 ohms puts out 113 vac rms. I would be tempted to order some varistors, like V130LA10A, and put across the audio lines at the amplifier, and from each audio line to ground,  and also make a common mode choke going to the modulation transformer primary. Could use the choke in one of those 20Amp 250V AC line filters.

   Does that modulation transformer have an electrostatic shield between primary and secondary windings? If so, ground it.

Jim
WD5JKO

[k4kyv]

I disagree with the 3.6k 100 watt resistor connected across the modulation reactor.  That would waste 50% of your audio power.  The whole purpose of the modulation reactor is to present an infinite impedance across the modulation transformer, while simultaneously providing a zero resistance DC path from power supply to final.  A practical reactor won't quite make those ideal characteristics, but will come satisfactorily close.  The modulation impedance of the final (plate voltage/plate current) already simulates a perfect resistor connected across the modulation transformer. A second equal resistance connected in parallel will result in the modulator working into half the nominal modulating impedance.

[WD5JKO]

I disagree with the 3.6k 100 watt resistor connected across the modulation reactor.  That would waste 50% of your audio power. 

  Don, I think the intention was to close the relay just before un-keying the radio to absorb the back EMF spike. Running the modulator with that resistor connected was not the intention, and as you say, would cause problems.

Jim
WD5JKO

[AMLOVER]

Don,
as Jim mentioned, the resistor is not there when transmit, it is entering the sircuit few milisecs before the drive switches off and greatly improves the sparking effect of the reactor. When I turn on the relay disactivate it and the reactor acts freely. It needs of course an extra vacuum relay but believe me worths every penny.
With 5-6kv and 5-7kw power on and off can not be easier than it is now, I believe.

Two questions only.
What is the diode-resistor configuration and the right position on the mod sircuit for better negative peaks cliping?
The irons are ringing-singing my voice back and this produce feedback to my condenser mic which pushes me to keep my audio 10-20% lower than could be. I'll soon move away from transmitter's place but I anyway prefer to improve this effect. Is it right to take the irons out and swim them in varnish or transformer oil or there is another trick?
Jim,
you are right about the varistors, as long they don't annoy audio quality I will put them in place as you suggested. I'll order 150v ones because the final amp to be used is a 5000w one.
Unfortunately there is not an electrostatic shield in the modulator because it is not a single pri to a single sec winding, it is a multiple sandwitch type with 5 primary windings between 5 secondary windings.
Didn't understand about the choke, could you explain me better what you mean "common mode" and where should I put it exactly?
Anything I can add is between the solid state output and modulator pri input, of course without annoying the pure ac audio signal, varistors to ground sound good.
I always thought that all those solid amps have clamping diodes on their outputs for arresting any dc presenting there but....or they don't have or those diodes don't work properly...or this trancient is too fast and huge...
I know that if I used a push pull tube amp I wouldn't realize any of these problems as long the tubes are elephants for such little ents-trancients...now I have the quality, the lowest distortion, the lowest dissipation but the highest sensitivity.
Anyway, I have 3 recomendetations to try and possibly improve the sircuit, some disc caps on the HV supply, diodes on the relays coils and the varistors on the solid af amp's outputs.

Thanks
Stefano

[AB2EZ]

Stefano

I'm trying to think what could be causing the transient at the amplifier's output, but this should help:

During shut down, close the relay that places the resistor across the Heising reactors first. Then proceed with the other steps in the shut down process.

Stu

[w3jn]

Don't neglect a purely acoustical reason for the transient.

My FT-1000MP has a terrible transient due to the "click" from the push to talk switch.

[Bill, KD0HG]

Stephano: You should get a microphone preamp or processor that includes a audio gating feature to reduce background noise from your iron. The DBX 286 is inexpensive. I use one at home for that very reason.

Sent via Blackberry

[W2PFY]

I am confused as usual. Is this a BC transmitter?

[k4kyv]

Don't neglect a purely acoustical reason for the transient.

My FT-1000MP has a terrible transient due to the "click" from the push to talk switch.

One more reason to avoid a microphone with attached PTT switch.

Another problem with the PTT mic switch is that the PTT line may magnetically induce an electrical pulse into the audio line, due to the long mic cord with parallel PTT and audio lines in close vicinity.  I had that problem with my first D-104 with G-stand.  It caused the tuning capacitor in the final to sometimes arc over just when I changed over to transmit. It's a wonder I didn't blow the modulation transformer. As soon as I discovered the cause of the problem I got rid of the chicken-choker once and for all.

I prefer to mount my microphone on a boom, with separate PTT switch attached to the end of a spiral cord, salvaged from some kind of old communications mic.  Been using that system for decades. I don't like the idea of a foot switch, either.  Nor a toggle or rotary T/R switch mounted on the panel of the transmitter or exciter.

My system allows me to walk over to the back of the transmitter and hit the transmit switch, so I can observe what is going on inside the rig just as it goes into transmit.  It has helped me solve numerous flaky type problems and discover where arc-overs occurred.  I can walk away from the operating position while holding the PTT switch, ready to instantly shut down the transmitter in case an intermittent short or arc-over occurs while I am trying to chase down the problem.

I'd leave the combination PTT microphone switch for the Good Buddies.

[K1DEU]

My mod reactors need their Q dumped,excessive ringing. Therefore I place some R across it. John, K1DEU

Like about 10X The Quiescent RF Amp Load.





A common Mental health problem !
A normal problem all hams have is hearing, tinnitus as we must all monitor our own out of phase broadcast transmitted audio.

[WD5JKO]

Didn't understand about the choke, could you explain me better what you mean "common mode" and where should I put it exactly?

   Ok, this could be a large Ferrite clamp on bead with the two wires (could be 12 awg zip cord) run through the core as many times as size permits, or a whole flock of beads over the wires (both in common) like a necklace, or maybe even use a Grounded Grid filament choke (2 wires bifiliar wound over a ferrite rod). Also some commercial power line filters have interesting common mode inductors inside the case.

Hope this helps,
Jim

[AMLOVER]

Jim,
I have a big T200 ferrite ring and few ferrite clamps to add but this would help mostly in hummmms and noises, a trancient will travel due to the cable continuity and reach the af amp output. A short talk with the modulator transformer brought up on the table the use of a capacitance coupling between af amp and modulation transformer. They suggested 400uf-800uf/440vac for 8 ohms and 30hz lowest frequency.
I have some 800uf/450vdc electrolytics and I could use in a combination to create an ac capacitance.
Should they be connected in series +- to -+ or oposite -+ to +- or doesn't matter?

Don,
I also use the independent ptt key with 3 teflon rg cables and the audio chain is totaly separated. Now that I am located next to the transmitter I have the problem of irons talking back, later the audio chain will be relocated in a separate room and of course it will be no problem at all. I can also move around with the ptt on hand and control the amp by being next to it, it is really very helpful.

Stu,
I didn't realize any problem by shuting down first with the same order the screen ac voltage & enter the resistor and shuting down second the drive but I can change it to as you suggest if this can make things even safer.

W2PFY,
I am also confused, it is a mixed style transmitter...power and modulation like broadcast, ptt like communication, use and problems like both of those types...

Thanks
Stefano

[W2PFY]

I think you need to hook up the caps so they look at the circuit as non polarized.

In other words, the caps should be hooked up in parallel with one end both pos and neg tied together and likewise on the other end and the other pair in the other feeder would be wired the same.  If you were to look at this as two flashlight D cells, it would look like a short but for capacitors of this type wired this way, they would be non polarized.

W2PFY,
I am also confused, it is a mixed style transmitter...power and modulation like broadcast, ptt like communication, use and problems like both of those types... Undecided

I guess my question is what frequencies do you operate this transmitter on? This has been an on going thread and perhaps you mentioned it before?

[AMLOVER]

This is a monobander low band transmitter . But we 90% use the 1650khz-1720khz freq for full power and local "races".

I become more and more confused about the cap so I made a schematic with A, B, C and D solutions.
Should I use the AC poor man's cap to each lead of the AF amp or in one lead would be enough?
What is the right configuration???

Stefano

[WD5JKO]

Stefano,

    I once worked for a company that used a pair of Crowm M600's bridged to drive a HV step up transformer. They drove 7.5 Khz PWM (bipolar signal) to make 40,000 volts DC at 50 ma (2KW). They used a series cap setup like B or C as you draw, and the caps were 1600uf @ 450v computer grade with the big screw terminals. I got upset at the idea, "how could you guys be so dumb as to...".

   I added some big diodes across each cap so each capacitor could not be reverse biased. It made NO difference, in fact without the diodes, and measuring with a hand held DVM, I could measure no Net DC voltage across either capacitor, even when the PWM drive had a DC component to it (< 50% duty cycle). I ate crow on that one.

Jim
WD5JKO

[W2PFY]

Hi Stefano, Figure A is the one I have seen in the past where a 400 watt  solid state amplifier was hooked up to a transformer to provide power to a public address system. They only used one set of caps as in figure A.  Now that was a long time ago so I may be wrong and if I am wrong and there is someone to point out why it's wrong, I would like to know

Thanks for the information on the races. I don't know where your located but it would be great to hear the system. Perhaps when finally all the bugs are worked out of it, you could make an MP3?

[Gito]

Hi

THe way I see it,when You used Heizing modulation,You used  a cap ( 1.5uf 15 Kv ) to couple the secondary winding of the Mod.trans to uper connector of the reactor.

What happened at this C ,since the secondary winding is grounded via the winding to ground.and the other end of the C is connected to B+  above the reactor ,it is charged to to the B+ .

SO  it is Ground ---- Secondary winding T.modulator----C (1.5 uF) ---Reactor ----B+---power supplu(HT).

If the B+ is 1000 VDC for instance, than The Coupling C is charged to +/- 1000 V


Now when You  Switch off (Unkey) the transmitter ,What happen  to  capacitive load C1.5 uf (energy stored in it) ---- 1000v DC---,
It begin to discharge the load trough Secondary Mod transformer -----C --- Reactor --B+ supply (that is also dicharging) and caused a big EMF across the Reactor.

It is behaves like a CDI (capacitor discharge ignition),since the energi(EMF) has no way to go.Than Swap there's a spark from the reacktor to ground.


Gito

[Gito]

Hi

Another though  comes to Me ,I have the conclusion after reading all the comments.in this article, is A storage energy in the reactor ,when the current is suddenly is gone, and there's  noway to go ,it can cause a flash ,or spark occur in the reactor.

So Why don't we used a bias  for the PA,that makes the PA tube conducting when idle ,a small standing current like 25 ma.

We still work at class C when this bias Voltage +(grid leak R X grid current) voltage = Class C operating voltage,

The difference when idle,no drive ,there's a small idle current, that makes the PA tube on conducting state

When You unkey the transmitter ,the current trough the Reactor is gone ,and make a high voltage across the reactor

, but since the tube is in conducting state, It will conduct this current,and limit it to 25 ma,so the  positive peak voltage is shunt across the tube .

And we used also a negative peak limiter(3 diode negative  peak limiter),if it is a negative peak voltage that can cause the spike to occur. It limits the negative spikes.


Just A though

Gito