The AM Forum

Thanks again for the help in the last post.

Synopsis:
HB Amp (K7LFE)
One 813 Grounded Grid
3200V (!) on the plate
Plate glows dull red at smallest exciter power


OK, I have added a second 813 because it is easy and I hoped this would drop the plate voltage.  It turns out that my second 813 is a dud.  :-(  The filament lights but it passes no RF.

I pulled apart the power supply.  It has a Sterling transformer, model st-2699 , two secondaries:

1120V-0-1120V   .245A
550 .016A

Currently it uses the full 2240VAC for 3200V on the plates.  Full wave bridge.

Any way to get reduced voltage out of this xfmr?  The other secondary only provides .016A....


Thanks

Rich

How much lower do you want to go? 

A variac would probably be the easiest way. 

I've heard of buck-boost configuration but I'm not sure how that works so I won't be of any help but maybe someone else can explain.

Rich,

You WANT a high voltage on the 813 when in linear to give you the headroom for audio peaks.

What you need to do is lower it's no-signal quiescent idle point.

This is done by adding a string of standard 1A diodes in series with the filament transformer center tap to ground. This will bias the cathode more positive, thus the grids more neagtive. You may need 20 or 30 diodes. At some point the tube will idle down where the plate doesn't glow and you are dissipating about 30 watts or so per tube..   You will then be able to take advantage of the 3200 volts for great audio peaks and a bigger carrier.


All of my linears here use diodes in the cathode for bias.  It would be just a coincidence (or by design)  if the tube happened to idle at the proper current based on your power supply, thus the need to bias and adjust it properly.

If the plates STILL glow when you have resonated the final and tuned C1 and C2 for maximum output, (using normal drive)  then add or reduce tank inductance until you find the best efficiency. This is measured by power in vs: power out.  You can also see it by the glow of the final. Put out a reference 700 watt carrier. (or maximum power you can get out) Then experimentally adjust the coil tap and tune for max power out.  The point where you see minimum plate glow and 700w out is the tank L/C ratio sweet spot for a given voltage.  The higher the voltage, the more L you need. (and opposite)

It goes without saying that the tank coil tap should be changed when the amp and HV are OFF... :-)

T

The very high HV you are seeing might be beneficial to your new amp.
BUT If you gotta lower the B+ then insert a (would this be a good choke, guys???) swinging choke after the diodes in the P.S., then your filter caps. A capcitor input filter will give you the high voltage you are seeing now. A choke input filter will lower the B+ somewhat prolly down to 2700, I forget the amount.

Fred

I bought an 813 at Deerfield from a friend for Bob RKW. Then I noticed an X on the base. I showed it to the seller and he told me he would give my money back if it was bad. I noticed the pins quite dirty so scraped them clean. Bob reported it worked OK.

I pulled apart the power supply.  It has a Sterling transformer, model st-2699 , two secondaries:

1120V-0-1120V   .245A
550 .016A

Currently it uses the full 2240VAC for 3200V on the plates.  Full wave bridge.

this transformer was likely designed to be a full wave center tap utilization.   would give about 1600 V capacitor input.  the secondary center tap spec of 0 volts is the givaway.   MANY TRANSFORMERS MADE THIS WAY HAVE LIMITED CENTER TAP INSULATION TO THE FRAME.  you are running a risk to continue running it as a fullwave bridge.  this mode will raise the center tap to about 1600 V.   time to hunt another way or transformer  ...73 ...John

It never ceases to amaze me, the knowledge to be found on this forum.   Thanks guys for the help here.

OK, assume that I intend to keep the 3200V and the second 813.

I have put aside the power supply and taken a closer look at the amp.  The filament xfmr has a center tap but it is not connected.  The two outputs are put across an adjustable wire wound resistor with the slider centered and tied to ground.  I suppose this is to create a virtual center tap, why do this when there is an actual center tap?

Tom,
    Adding diodes to the center tap seems a clever idea, I have a pile of 1n4001's.  I suppose they would go between the resistor slider and ground. 

John,
    Any way of knowing if this transformer is subject to that problem?  I am not in favor of KV's leaking to the chassis..........


Rich

Hi Rich,

Get rid of the resistor slider and just use the fil xfmr center tap for the diodes.  This CT series circuit to ground should also have relay contacts to ground for TR keying as well as a plate meter in series for current metering.   (Xfmr CT >  diodes >  Meter>  relay contacts>  chassis ground.)

The resistor slider could have been a form of negative feedback or maybe a method of balancing the fil for hum, dunno - but you already have enuff NFB in the GG configuation - and hum shud not be a problem using the xfmr CT.

Hope this helps.

T

Rich, the one 800 lbs. gorilla that people are missing is what's the idling Ip? You may have, (forgive me Rich Measures) a parasitic or one big oversized oscillator in there. I don't have the operating data for 813's readily available, (but I do know them for Class C operation). Eric WB4VVI (SK) had an 813 amp that he put 4200VDC to it just to prove it would work! So the higher HV can work. From a safety and equipment perspective, it would be good to isolate that transformer and get one in there that will do the job safely.

Mike,
Think about it an 813 running AM at 2500 volts with over 100% modulation hits over 5 KV plate voltage on voice peaks. The higher you go in plate voltage the more critical the layout.
Yup, use the center tap of the transformer and dump the resistive divider. Put some diodes or a power zener in series with it to drop the resting current.
I would think you don't want to see any color in the plate with no drive. No signal I would stay around 30 watts dissipation per tube unless you want to run above class AB. I think the ROT for class AB is resting current 1/5 peak.
A little air on the tube will reward you with long life.

Frank said:
Exactly his premise for running it at that level. However Eric confided in me that it was the first time he ever ruined an 813 tube. He forgot and had the amp mistuned and that is all it took. We thought it was funny when he would tell people he had 4~4.5KV on an 813. Even I thought it was bizzare until he explained it to me using the same reasoning you stated Frank.

Hola

Being a grounded grid/screen amplifier, (triode config) using 3.2kv with no bias will cause the graphite plates to glow during no signal idle conditions - too much heat.  From past experience with GG 813's, if the plate voltage is higher than about 2200 volts, we need to bias it down to idle lower.  The same holds true for triode connected p-p 813 modulators.

3200 volts at even 40 ma of idle each is about 128 watts per tube. (The 813 plate diss rating is only 125 watts)  So it's not hard to show plate color at 3200 V.  I would idle the PAIR (no signal) at about  20 ma TOTAL, (32w, 10 ma per tube) using the correct amount of forward biased diodes in the CT - and call it a day.

Using higher voltage is like a super charger on a car. If it becomes unstable, it can break parts, but once you tame it, the benefits are great and it kicks ass.... :-)

T

OK Tom,
     I removed the adj resistor from across the filament xfmr leads.  I have ten 1n4001's strung out Rube Goldberg style to the center tap and it drops the idle current nicely.  Not enough though, I think you are right, it will take 20-30 but that sure helped.  The plate only glows very dully even when I turned up the exciter for 350W out. 

No sparks fire etc.   The true test will be when I get a second (good) 813.  Then we will see how the power supply holds up etc.

I will mount the diodes on a nice bread board before continuing and will shoot for the 20mA per tube.

Thanks again, you da man.

Rich

PS 

Anyone else get VERY jumpy when doing this sort of thing?  I am jumpy even when I am done.  ;-)  HV should make you jumpy I suppose.  I keep the HV clipped to ground when not in use, and un-plugged of course.

We have a young cat that I had to ban from the basement while doing this work, he keeps trying to use up all his nine lives.....

If you are in general happy with the power supply but want to take care about the center tap issue, one alternative to ditching the transformer is to put a full wave voltage doubler on each leg.

hell yeah you should be jumpy.. as in extremely attentive as to what every move you are thinking about is about 2 steps ahead of what you are doing.
you should also have ur bench set up so that you do not EVER have to reach across what you are working on to reach knobs, test gear, etc. NO wedding rings on ur fingerz. and never stick both hands in a piece of gear. what kills is current across your chest as one point of you is on HV and the other is on ground  or B-.

the most I ever got racked with was offa viking 2 when I was a JN, about 700 volts and full RF.            
                                         

Patrick  shows a better way with the doubler .... the C1 caps take the dc off the winding ... you will have to derate current appropriately ... how much does the xfmr weigh ?  ... old buzzard am xmisssions will indicate about 10 lbs weight for 100 W of power ... this may be a bit conservative for some xfmr types .... how replacable is the xfmr ? ...this will help you decide just how hard to push it

now we know what happened to Derb .... ZAAP ... wow ...73 .. John

The penalty is C1. It's microfarads will need to be 4 to 8X C2 depending on cap quality/ESR. Not because of ripple, but because of heat. The current spikes for recharging the capacitor each cycle could be 8 to 10A (if the xfmr will permit) contantlky a 6ms 8A spike, then 10ms later, a discharge of the same magnitude as C2 is charged. over and over... So anyway, if C2 your output filter is 60uF, then C1 should be as much as you can muster, perhaps 240-330uF. At 1600V+ that is a burden, but if the C1 is well chosen, the the transformer is good, the scheme works very well. 

I learned this the hard way with the same circuit built as the single ended (no CT) style used right off the line. It was a doubler for a cluster of 10 strobe lamps. Of course each lamp needed its own capacitor, (they don't like to all arc when triggered together otherwise). So, each one had its own C2 of 20uF/450V. C1, I figured, would be 10X that, because C2(total of ten 20uF ones) was 200uF, therefore each cap was alike. Wrong! the super badass strobe I'd built, true to the style of black lights and hippie days, flashed merrily for about 5 minutes, blinding all within its view, then BOOM! the C1 exploded. How embarassing was that? After cleaning out the mess I replaced C1, and noticed it started getting real hot, then tried a 1000uF one and it was fine since.

When I work on high voltage I sing a little song to myself....Come a little closer, your my kind of man....Be afraid be very afraid. when you stop being afraid you are in big trouble.
Tom, I didn't run the numbers but power adds up fast at high voltage. I wouldn't want to be sitting at max dissipation at idle. Look at a 4-1000A with aroung 3500 V on it. With no bias it will be running about 120 ma with no bias and will swing up to about 700 ma peak.  I think the 813 only wants to do 180ma at high voltage. Sounds like the 813 wants to rest at 15 to 20 ma. 
yup add another tube and increase the bias. RF parts sells power zeners cheap once you know the voltage you can set the bias with 1 diode. 

Frank,

Re-read my post above.  We are saying the same thing...

" I would idle the PAIR (no signal) at about  20 ma TOTAL, (32w, 10 ma per tube) using the correct amount of forward biased diodes in the CT - and call it a day."


BTW, the reason I like separate diodes (instead of zeners) for biasing is the ability to select various classes of operation with a rotary switch by shorting out groups of diodes. In addition, I have a Variac on my HV supply and am always changing HV parameters. So the ability to select any bias is desirable for me. 

Come a little bit closer, you're my kind of man... yep, HV is scary stuff. It always amazes me that there are so few electrocutions in our hobby considering how easy it is to screw up only one time with 3kv++ . (knock on wood)  That and climbing towers makes ham radio as dangerous as any "sport"... :-)

T

I have built a little diode board, the green pcb dangling off the side of the rig in the pics.  Using this cathode bias I was able to nicely drop the idle current to ~10mA.

I got up before the rest of the family to do some testing this morning.  Unfortunately I only got ~20 minutes in before I switched on the HV with my safety jumper still in place and blew my last 5A fuse.

With the new bias it seems to take a lot more power in to put power out.  I am using my ricebox as an exciter for now, it will only put out about 15-20W in CW mode into the amp.  The amp has no input tuning so this is the best it will do.  Into a 50 ohm dummy load it puts out plenty of power.

So before I blew the fuse I couldn't seem to get much power out at all with the new bias, maybe 50W.  This is with one 813 installed. 
With two installed the idle current (same number of bias diodes) was ~18mA.  With two tubes and 15-20W in I still got low power out and the tube with the steel plate was really glowing.   :-(

So here I will show my ignorance in the area of tube fundamentals:

The 813, as I understand it, is rated for 125W of plate dissipation.  How does this number figure into the big picture when I am looking at plate voltage, plate current, power in, and power out?

What kind of power out can I expect with one 813?  With two 813s?  By power, I mean a reading of watts on my inline power meter,
with the exciter feeding and un-modulated carrier.  How do we refer to this type of power?  (It is not PEP, would it be plate input power?)

When I hear someone say "That is a KW amp", what are they referring to?  Can that amp put out a carrier that will cause my power meter to read one KW?  Or are they referring to a PEP power of 1KW?  If the latter is the case, what would the power meter indicate to allow one to believe that the amp is putting out 1KW PEP?

Thanks

Rich

Looks like a nice amp - and VERY compact, Rich!

Yes, as you add bias and go towards class B, the linear amp will be harder to drive. The efficiency will be higher, however.

You should have the proper drive, but if not, short out some diodes to get a higher idle and the amp will become easier to drive. (and slightly cleaner too with more heat)   It's a compromise situation.

Add some amp tuning L/C input to make it cleaner and easier to drive.

At 3200 volts, the way to determine max power is look at the scope with AM audio tone. Tune up the amp for max power out, then reduce the carrier to 1/5 - and put through a 1khz audio tone into the mic input. Increase the audio level until the waveform starts to flat top on the scope.


You should be able to get out maybe 600-800w pep with 3200v, but this is a guess. Usually it's X2 the tube's dissipation. So 250w X2 = 500w pep for a pair. But the 813 is a very special/strapping tube and may do more for you. You have air blowing so that's even better. Just tune it up as described above and see what you can do.  The scope will tell you. Much depends on loading, HV sag, tube condition, L/C efficiency, drive available and other minor stuff.


After you peak C1 and C2, then DECREASE C2 (loading cap) until power drops about 8% or so. This will make the tube more linear.

A full CW carrier of 1 kw is equal to 1 KW of pep.   AM carrier is 1/5 or about 200 watts for a 1kw pep rig.


Play around with the amp tuning, drive and audio level while watching the wattmeter and scope and you will soon understand the relationship.


BTW, a "kilowatt rig" can mean many things to different people. To me, an AM, 1 KW rig is one KW carrier output.   It used to mean 1KW carrier input in the old days.   On ssb, a 1KW rig means 1KW pep output to me.  Depends on your standards.

I would call a pair of 813's in class C, plated modulated rig a 1KW rig, cuz I've gotten that out with a pair on AM before. (at 3kv lots of air)  On ssb, it is also considered a 1KW rig cuz with 4kv on the plates, (and lots of air)  I've gotten 1KW pep out on ssb. This is really pushing things to the outer limits, though... :-0


Hope this helps, OM.

Tom, K1JJ

Great help Tom, thanks.

When you say "max power out" what does that mean?  The max rating for the 813's?  Just peaking the output at an arbritrary power?  I interpret this as tuning the amp at a given frequency and then increasing the power.

I will have to reduce the cathode bias to get the power out I want.  This will increase the plate disipation at idle.  I am still unclear how the plate disipation changes as I start feeding in more RF...  I guess I don't understand how plate disipation, a DC current, relates to or changes when we are talking about RF amplification.


Isn't the max disipation for the 813 125W?  So x2 would give me 250W PEP?  (More missing fundamentals I suppose.....)


Rich

Rich,

Max power out means how much power the tubes will deliver under the configuration and class they are in. Much has to do with efficiency of the class of operation. 

In class C, max power out may be 4kw pep. In class B, max power out may be 1kw pep.  This is using the same 250w of plate dissiption ratings.  We look to dissipate 250w and get out as much as possible.  Trying to get more power out will results in exceeding the plate dissipation  - or saturating the tube, causing distortion.

Plate dissipation will vary with class as you increase RF drive. In class A, the dissipation stays the same, no matter what the drive is. In class B, more drive causes more plate dissipation.  The same for class C.  On AM, your linear is more an "efficiency" type of amp. It swings from about 30% eff up to 70% when going from no audio with dead carrier to full audio modulation.  You may see little change in plate color (or dissipation) between 0%-100% modulation on AM.  On ssb, the plate color will follow the audio and power output level.

Set up your amp as I described in the last post and play around with drive, amp tuning/loading and audio levels on AM as you watch the power meter and scope - and you will soon understand.

T

Thanks Tom, I will continue to experiment.

Rich

Rich,

To just follow up a little more...

The following examples are just rough examples not related to your 813 amp:

Let's say you have an input power of 1 KW = 3200v X  312 ma.

Using a class A amp at 30% efficiency, you would dissipate 700 watts in the tube's plate and put out 300w to the dummy load.

Using a class B amp at 65% efficiency that becomes 350 watts in the tube and 650 watts output

Using a class C amp at 80% eff = 200 watts dissipation in the tube  and 800w to the dummy load.

So, you can see how your output power is limited strictly by the class of operation and the tube's ability to absorb the wasted power as heat.  Each class has its own advantages and disadvantages.


For AM, in your case, the two tubes' dissipation is pretty much set up when you key the carrier on. If your carrier output is 125 watts, then you may be dissipating  250 watts in the tubes at 33% efficiency. When you modulate 100%, the efficiency swings up to 65%, and the power output to 500 watts. This means 175 watts of dissipation and 500 watts out pep. So, your tube color (plate dissipation) should change very little under modulation and even go down slightly depending on your loading.

Think about it for a while and then experiment.

Later -

Tom, K1JJ

Check the voltage rating of your scope probe. It sucks to blow them out.
A pair of 813s may do more than 500 W PEP but I think you want to blow a little air past them. 200 watts carrier is pushing them a bit but might work if the tubes are good and you don't let them get too hot.

My scope hooks to my TR switch box which has a connection for the scope.  Simply 2" of stiff wire that runs parallel to to one carrying RF.  So there is no direct connection but the readings are only relative, not absolute.

Rich

Yep!  When I was  JN, I had my cat jump down from an unfinished ceiling right into 400 VDC + RF on my operating bench.  He jumped down, yowled and straight up from whence he came quicker than an eyeblink!  Oh yeah, in spite of this death-defying jump, he still lived to over 20 years old!!

speaking of filament sag, you need to remount your 813's to put the plane of the filaments to vertical when you use them mounted hortizontally ... the amp shows a lot of promise ...73 ...John

Yep, you're correct, John - I didn't notice it myself.

Rotating the tubes 90 degrees will also put the plate structure in a vertical plane. Right now it's horizontal.  I understand it can sag too and create shorts for the screen, etc.

Haven't thought about this for years cuz all my amps have the tubes standing up.

T

I saw the note about orientation in the spec a few days ago but was in Deeeeeniiial.

 ;)

Rich

Play it safe and mount that transformer above the chassis. 3/4" plywood and nylon hardware will hold it in place. You could cobble up a little alarm circuit to light a bulb to tell you if the case becomes voltage active.

Decades ago I repaired an AL-80 for a customer who had a hot transformer case. That transformer is still running fine today in a bench test PS mounted as I mentioned above. It even has a warning label on it :D

Carl
KM1H

Unfortunately the power supply is mounted in a Heathkit speaker chassis and the transformer is an "interference" fit.  This precludes insulating it from the current chassis.  So I will have to replace the transformer,  covert it to a voltage doubler circuit as Patrick suggested, or remount it in a bigger chassis.
The last option would be the easiest.  I also want to add a better bleed resistor setup.  The current one does not work well, it takes several minutes to bleed down.

In the mean time I will make sure the chassis is grounded well.


Rich

I have a better understanding of the operating point of the 813's in this amp but I an not as clear as to what I should do from practical point of view as far as where to set the cathode bias.

When I got the amp there was no bias and the plate glowed a bit.  Can I assume that it was operating as a class A amp?  (and exceeding max plate dissipation)  Maybe this was ok for the builder as in class A the dissipation is constant, I think...

As I add bias I approach class B operation, I am in the AB class now correct?  Is this type of amp ever operated in class B or is there too much distortion?

Ok so anytime I set the bias so that the tube idles at some plate current that is lower than in class A I am in class AB, correct?   I suppose the exciter signal strength would determine where class A becomes class AB, correct?

 I still don't know where to set the bias.  Is there an optimum bias where this amp could be used for both AM and SSB?  Or do I need to add a bias control to switch in more/less diodes?

Rich
 

Rich,

Set the diode bias (by adding or shorting out diodes in the CT string) where the amp keys up (with no signal, idle) and draws about 10ma per tube. That's it, done.   The amp will be set at a compromise between efficiency, cleanliness and ability to drive easily. This is for BOTH AM and ssb.

The amp will be idling  and running in class AB.

When you drive it, the class does not change except for the amount of grid/screen current flowing. If grid current flows, then you are in AB2. If not, then it's called AB1.  Since this is a GG amplifier, it will always draw grid current with drive, so your amp is in class AB2.

The plates shud show no color cuz you are keying up with only 3200v X 10 ma = 32 w per tube. They are rated to dissipate 125 watts.

When you drive it hard in ssb, you may see some plate color at times . In AM, the plates shud not glow during dead carrier conditions or with full audio. A little bit from time to time is OK if you try to squeeze ever last watt out of it before saturation, but keep the air flowing well, if so.

Yes, if the bias is set so the tube idles high and the plate current doesn't change during full ssb drive, it is class A.  Idle it lower and the plate current changes under drive and you are in class AB. Class B is when the tube theoretically draws no current at idle - it's exactly at cutoff. Most amps are run somewhere in AB for best cleanliness and least crossover distortion.

Once you start running tests with tones, watching the scope, you will see the subtle differences of class operation if you add or subtract diodes.  But use class AB2 as set by the 10ma idle per tube setting for this GG linear rig.

Hope this helps.

Tom

Tom,
  I have the diode board installed, set for 20mA for both 813s. 
I applied an AM carrier (using the ricebox) and set the power in for 125W of carrier out.  Unfortunately the power can not be turned down enough to get 1/5 of that.  ( I will double check this, perhaps I was looking for 1/5 on the scope (voltage) instead of the wattmeter(power) )

With the exciter turned all the way down I sent in the 1KHz signal and increased the audio gain.  I never saw the signal flat top even way beyond 100% modulation.  Only when I turned the power up did I see the tops flatten.

This morning I tried using the Viking II to drive it.  First I tried driving it directly, using the "drive" control to attenuate the output of the Viking.  This did not work.  I can see the signal Viking carrier on the scope but there is no increase in the amp plate current and the the roller inductor in the amp tank arcs over!  No power at all on the wattmeter.

I quickly cobbled together an attenuator as you described in another post.  Two turns on the primary of a binoc core in series with a dummy load.  Two turns on the secondary going to the amp.

This worked as far as tuning up the VII and the signal through the amp was clean but still nothing on the plate current meter.  (No arcing this time)

Rich

 :P

All the problems with the JVII in the previous post were due to a simple wiring error.  (Brainfart)

When I have time I will try Tom's test for maximum power using the VII which gives me better control over the input power.


Rich

yep. "pins 2 and 6 in vertical plane" Amp looks great!

Olde Kitty sometimes would sit on the dot matrix printer, printing or not. Once the tip of her tail go into the cog belt that ran the head shuttle, and caused a "print error". From that day on, the sound of that printer would send her under the bed! Kitty lived to 22 years. Half of my life at the time. You get attached to them. 1983 to 2005 RIP

Hi Rich,

Tune the amp up for as much as it will put out. (like 500 -700w?)

THEN reduce the carrier down to 1/5.   When you modulate 120%, you will hit the max pep power of whatever the limit of the amp is. (500-700w in this case)  Watch the scope to see when it flat tops and then back the audio and or drive down to just before that point. You want to be able to modulate 125% with whatever carrier you set it at. It's a juggling act - you have to go back and forth with your adjustments until you get it right.

Sounds like you drove it only to 125W out and reduced it to a carrier of 25w.   It will do much better than that.  With only a 25w carrier, you SHOULD see huge headroom.. :-)

But sounds like you're getting real close to final operations, OM.  Good show!

Tom, K1JJ

OK, I guess I am being timid.  I cranked up the power.  I got to about 350W, I can't remember for sure what was on the plates, something like 150-200mA.

At that point the output power did not go higher even as I increased the input.   After about 30 seconds I got fireworks from inside the roller inductor tube.  Tried it again, same result.

I am not sure why the power stopped going up and it looks like I should take that inductor apart.  Maybe it just can't hang with two 813s.....

Rich

Here are some pictures of the inductor zorch.  It looks like the arcing is between elements of the center shaft (at ground potential) and the far end of the coil.  Blew the solder ball right off the "spoke" attachment.

Pretty darn sure that the tank was "dipped" properly.  I am going to clean up the interior roller contacts and give it another whirl.

Perhaps a better solution is to build my own coil with fixed taps.  Are there large benefits to the adjustable inductor vs fixed?  I suppose you can always get a perfect dip for any frequency....

Rich

Rich,

I didn't want to ctiticize the rig, but, yes, I've also had very poor luck running those small roller inductors in amplifiers.  Usually the small roller itself starts to make poor contact and starts pitting and goes intermittant.  It's all a matter of the power you run thru it.

Remember that a resonant tank circuit (L/C)  has circulating currents within the loop that can be extremely high - in the tens of RF amps.  I think the formula is Q times the plate current?

I use these type of rollers in my mini ant tuners for inputs to amps, but not as outputs.

Anyway, consider doing what you suggested - build up a fixed coil with taps and you'll be fine and bulletproof to move on to the next weak part. It's all a shakedown run.

T

When it is all shaken out and running FB then the fun and learning are over.

:-(

Rich


PS Brace yourselves for 2134 postings regarding the proper construction of a multi-tap tank coil.   ;-)

I was taught, if you solder your taps you are not QRO.

As opposed to alligator clips?

Nuts and bolts with straps.
If you tune up and the output stops going up as input increases look at the plate current. If the plate current is still increasing there is a good chance the pi network L is too high meaning Q too low. A small inductor will run out of steam at some point as the series resistance becomes an issue.

Straps = strapping?  ;-)

I put the roller inductor back in after cleaning it up.  It still arcs over...

I was able to get it up to 500W.  After that the output power did not increase and the plate current did not rise.  I would guess that the B+ transformer was out of poop.  200mA X 3200V = 640 W

The transformer is rated for 249mA at 2260 VAC

Looks like I need to get an old fashioned tapped coil in there.

Maybe next on the list will be a new power supply.


Rich

The roller inductor probably has some carbon tracks on it now. In any case, if it is arcing, you need to lower the voltage  or get a bigger one.

Rich,

If you look carefully where the "roller tap" lead comes out to its connection, you may find it passes very close to one end of the coil. If so, there can be huge voltage from one end to the other and you're getting that arcing.  Maybe you can find a way to isolate the roller wire, dope it with goo or something else.

Though, the best way will probably be to wind a new coil using 1/4" copper tubing and tap it.   Use some kind of 3"-4" form to wind it on and then slip it off and support it as an air coil.  Don't worry about taps until you install it and look for the sweet spots by trial and error. Use the tank calculator to get close and go from there.

Do a search on Google for "Pi Network Calculator."   Plug in 50 ohms for the output, the band frequency, a Q=12 and the output impedance of the pair of 813s. It will tell you the inductance for L and capacitance for C1 and C2.  Then search for an inductance calculator and plug in your coil dimensions and find how many turns you need and tap it.  Tune C1 and C2 for max using this L. 

The best coils for highest Q are made with the physical length to diameter ratio held between 1:1 and 2:1 over the various bands. This will require more than one coil for 5 bands, but don't sweat it if you want to use one coil. 

There are calculators on this BB also.



T

Hi Tom,
     I like the 1/4" tubing idea but that would make the coil too big/long.  I think I can squeeze in a 3" coil but will have to use a smaller diameter inductor to get the turns/inch I need.

This is a nice doc showing some homebrew inductors:

http://www.arrl.org/tis/info/pdf/9708033.pdf

I am researching how to calculate the plate resistance of the 813's.  There is enough mixed messages out there that I have no found a clear answer.  Maybe it is as simple as the plate voltage/current but then I have to make an assumption about the current....

Rich

Rich,

For operating points other than class C, the load impedance isn't the simple resistive calculation it is for Class C due to the dynamic impedance variation throughout the conduction cycle, even for Class A where the observed DC plate current is constant.  The plate impedance is usually much less, for example, for Class AB single ended RF service, it is typically half the average resistive value.  For Class A RF service, it can be about a third. 

I'd go with about 4K for two 813's as the class AB linear load impedance for your pi-network calculations.  If the actual impedance is a bit higher, this just means the Q of the tank will be higher which is fine.