bringing Plate Voltage up on Variac

[W7SOE]

I will (eventually) need to smoke test the linear I am building.  My plan is to initially lower the plate voltage with a variac.  What is the proper way to connect the variac?  A simple two wire connection to the HV transformer primary, three wire with ground to chassis? 

Thanks

Rich
 

[W4AMV]

Hi Rich. I have been monitoring with interest some of your various posts. I am involved in a similar project, a pair of 811A's and my HV supply is separate on a smaller chassis. I posted a view in the thread on HV supply safety. To your question, I brought out my supply, HV xmfr primary fused in series with the LINE voltage and a CONTROL Relay to the LINE port of the variac. The NEUTRAL of my variac goes to the other side of my HV xmfr primary and finally my variac GROUND goes to my supply chassis ground. I enable the relay first with DC 24 V from another supply and turn the variac up slow and monitor the supply DC voltage out, initially at LOW AC input voltage. I know exactly what I am supposed to get at any AC input line voltage and so should you. If the numbers do not make sense, STOP. Something is wrong. Finally I apply full 120 V and monitor DC Vout. Eventually when I am happy nothing is going to flame out, I operate the supply exactly how it will operate when the amp is connected. Anyway, what is key on the setup, is make sure you observe the electrical code on connections of LINE, NEUTRAL and GROUND. There are experts on this forum who I am sure can really educate well on this subject matter.

Alan

[KA2DZT]

I will (eventually) need to smoke test the linear I am building.  My plan is to initially lower the plate voltage with a variac.  What is the proper way to connect the variac?  A simple two wire connection to the HV transformer primary, three wire with ground to chassis? 

Thanks

Rich
 

Do you have a variac that is already enclosed in a housing with an AC outlet or do you have just the variac itself without any wiring or outlet??

Fred

[W7SOE]

Thanks Alan.

I have already brought the supply up on a variac, with no load.  No smoke or fire, behaved as predicted.  It puts out ~3.2KV with no load.  This is rather high and so I want to lower the plate voltage for when I actually apply signal.  This means seperating out the HV primary whereas before I brought everything up together.  As you noted, it is a question of hot, neutral and ground...

Of course I could be lazy and just fire it up as is.....

Rich

[W7SOE]

Do you have a variac that is already enclosed in a housing with an AC outlet or do you have just the variac itself without any wiring or outlet??

Fred

Good question Fred!  It is a separate standalone variac, the kind that plugs into the wall and has it's own outlet.

Rich

[KA2DZT]

Do you have a variac that is already enclosed in a housing with an AC outlet or do you have just the variac itself without any wiring or outlet??

Fred

Good question Fred!  It is a separate standalone variac, the kind that plugs into the wall and has it's own outlet.

Rich

OK FB,  I thought you were asking about wiring the variac itself.  You have a ready to use variac.  To answer your question I would think just connect the xfmr primary with any cord you have using the plug on your variac box.  If you use a plain two wire cord (just for testing) connect a separate ground wire from the xfmr frame to the variac housing.   Assuming the variac has a 3 wire cord.

Fred

[W7SOE]

OK FB,  I thought you were asking about wiring the variac itself.  You have a ready to use variac.  To answer your question I would think just connect the xfmr primary with any cord you have using the plug on your variac box.  If you use a plain two wire cord (just for testing) connect a separate ground wire from the xfmr frame to the variac housing.   Assuming the variac has a 3 wire cord.

Fred

Thanks Fred, will do.

Rich

[W1ITT]

Rich...
You mention that your 3.2kv unloaded is a bit more than you wanted.  What is your bleeder current across the total filter capacitor(s)?  If you are only very lightly bled, you can lower things a bit with more bleed current, at the expense of heat and wasted power.
Depending on what you have for a variac, you can run with it and dial up whatever you want.  I have a back-buster heavy 240 volt variac in the primary feed of one of my HV supplies.  Of course it has some resistance internally, and that negatively affects dynamic regulation of the supply, but that is compensated for by the 240 microfarad oil filled capacitor in the filter.  (It was a gift from my old friend K1RQG many years ago.)  My step start procedure also involves slowly cranking the variac up to fill the capacitor without vaporizing my rectifier stack.
The point of all this is to say that the voltage output and dynamic regulation is a dance between your mains supply line, any variac, the internal resistances of the transformer primary and secondary windings, the filter capacitor and its bleeder stack.  Use the best components you can find without breaking the bank, and remember that perfect is the enemy of good.  Get on the air.  Better components will eventually show up in the flea market. And always use the chicken stick before you go mucking around in there.

[W7SOE]

Rich...
You mention that your 3.2kv unloaded is a bit more than you wanted.  What is your bleeder current across the total filter capacitor(s)?  If you are only very lightly bled, you can lower things a bit with more bleed current, at the expense of heat and wasted power.
Depending on what you have for a variac, you can run with it and dial up whatever you want.  I have a back-buster heavy 240 volt variac in the primary feed of one of my HV supplies.  Of course it has some resistance internally, and that negatively affects dynamic regulation of the supply, but that is compensated for by the 240 microfarad oil filled capacitor in the filter.  (It was a gift from my old friend K1RQG many years ago.)  My step start procedure also involves slowly cranking the variac up to fill the capacitor without vaporizing my rectifier stack.
The point of all this is to say that the voltage output and dynamic regulation is a dance between your mains supply line, any variac, the internal resistances of the transformer primary and secondary windings, the filter capacitor and its bleeder stack.  Use the best components you can find without breaking the bank, and remember that perfect is the enemy of good.  Get on the air.  Better components will eventually show up in the flea market. And always use the chicken stick before you go mucking around in there.

240 uF is a lot of capacitance for a HV supply!  I added a second bleeder, other than the balancing resistors in the capacitor bank, and that dropped the voltage a bit.

My latest concern about bringing the voltage up in steps while testing is input from the receiver.  The plate transformer also powers the keying relay in the amp and so I plan on jumpering the relay.  Buuut, this means the amp will be in transmit mode, the amp input will be AC coupled to the cathode/filament of the 813s. 

I am afraid this will damage the connected transceiver...

Rich

[W7SOE]

I applied HV to the tubes for the first time and, as I should have expected, it did not go well.  (Why do I always expect stuff to work on the first try when they never do?)

The setup:
The relays are kept closed by a separate 24VDC supply.  Positive to the relays negative to the chassis ground)
(Idec, like this http://www.coleparmer.com/Product/DIN_Rail_Power_Supply_30W_24_VDC_1_3_A/UX-26901-10?referred_id=778&gclid=CKqxyISE_McCFQNsfgodME8Opw   )

The filament transformer is powered from a two wire cord to an outlet.  No ground.

The HV transformer power from a three wire cord from the variac.  Grounded to chassis.

It all seems ok up until about 1.7KV.  Then the HV transformer draws a ton of current as indicated on the meter of the variac.  The grid current and plate current shoot up.   Up until that point there is just a little grid current and a small plate current.

The max voltage for the HV is 3200 unloaded.  It is interesting that things go awry at 1/2 that.  Just a coincidence perhaps.

Here is a video I may quickly before leaving for work.

http://s1227.photobucket.com/user/rsoennichsen/library/813%20Linear

The first shot of the meter shows the plate voltage (3200 full scale).  Then the grid current (200 mA full scale).  Then the plate current, 500 mA full scale)  You can see the grid and plate current oscillating until the voltage hits a certain point then, as the HV goes higher, it gets steady.  If I keep bringing it up the 10 amp line fuse blows.

I tried running the variac through an isolation transformer, no change.

A lot of heat coming off those 813s.  I am down to two tubes of my four, as one had a bad filament and one, apparently, has it's plate shorted to something as it shows a dead short to the HV.

Rich

[KB5MD]

Since you're short of 813's to swap out, I would try it with just one tube and see if you get the same results.  If one gives the same results, try the other one.  Maybe they are both bad, but I hope not.  Good luck

[W7SOE]

I did try it on one tube, same results.  I will need to double check my wiring...

Rich

[Donnie SWL]

I would shorten up several of your rf an ground paths...How close is that band switch shaft to the plate choke? looks very close or maybe just the angle of the photo....

[W7SOE]

That shaft is close.  But keep in mind I have no RF applied yet.  The tube(s) are starting to conduct all on their own as I bring up the plate voltage...

Rich

[W1RKW]

oscillation?

[Donnie SWL]

Yes Rich that's why i said what i did about the ground an rf paths.....This is some things i would do....i would get rid of the twisted pair of wires you have connecting tube fil together....Instead run solid say #14 copper wire from socket to socket.....Also get rid of the twisted pair of wires out of the fil choke to the tube an again use a solid wire....remove the cap you have on the output of the fil choke an place it right at the tube socket......
Now you see the the green wires you have for grounding the screen an suppressor grid....why not make your ground their direct to the socket mount bolts this would be a much shorter an better ground path you can use both the socket bolts for these grounds.......
These are just my opinions an you know the ole saying everyone has one ....
 Nice build though Rich keep it up an she will come together...............Don

[W4AMV]

Hello Rich. Just to clarify, are you building the "Junker" amplifier using the 813? There are a few confusion items in the article and not sure if some of those may be lurking in your circuit.
Alan

[W7SOE]

Hello Rich. Just to clarify, are you building the "Junker" amplifier using the 813? There are a few confusion items in the article and not sure if some of those may be lurking in your circuit.
Alan

That is it, except I am running two 813's instead of one.  What confusing items are you refering too?

Thanks

Rich

[W4AMV]

Thanks Rich. The metering and scaling R's you have caught already. The use of Z1, particularly with one tube, perhaps improved with 2 tubes, but why the need for Z1 and what arrangement would you use for 2 tubes? In short, I would add a PI or at least an L match network to the filament connection. I am sure you have T2 connected proper, but just to be sure that center tap connection appears to have a dot on leg of the xmfr. Of course it should not be there. Each grid to ground needs a 0.01 uF and make it as short as possible with the ground connection near to the same ground connection as the plate RFC bypass, 500pF/5kv unit. I measured a 0.01 uF disc and with short leads its ok, about 5 ohms at 3.5 MHz. I can't recall if you posted a pix of your completed layout, but I get a sense that that oscillation is present and that is certainly possible if the layout and ground returns are not proper. All of the other grid connections to ground need to be as short and as direct as possible. The fact that Z1 is present and the filament (cathode) high frequency Z to ground is NOT present in the circuit (as it would be if a PI or L match network were present on the RF input side), makes me concerned. One other comment, question, how are you providing RF decoupling between the grids prior to proving a single cutoff bias supply feed. I wrestled with this a bit and decided to use series chokes, 2 uH, non resonate followed by shunt C, .01uF to ground for each RFC prior to joining the chokes to a single cutoff bias supply. I have seen a variety of approaches to this, highly simplified if you just DC return all the control grids to ground and control cutoff via the cathode (center tap filament) xmfr return.

[W7SOE]

Thanks Rich. The metering and scaling R's you have caught already. The use of Z1, particularly with one tube, perhaps improved with 2 tubes, but why the need for Z1 and what arrangement would you use for 2 tubes? In short, I would add a PI or at least an L match network to the filament connection. I am sure you have T2 connected proper, but just to be sure that center tap connection appears to have a dot on leg of the xmfr. Of course it should not be there. Each grid to ground needs a 0.01 uF and make it as short as possible with the ground connection near to the same ground connection as the plate RFC bypass, 500pF/5kv unit. I measured a 0.01 uF disc and with short leads its ok, about 5 ohms at 3.5 MHz. I can't recall if you posted a pix of your completed layout, but I get a sense that that oscillation is present and that is certainly possible if the layout and ground returns are not proper. All of the other grid connections to ground need to be as short and as direct as possible. The fact that Z1 is present and the filament (cathode) high frequency Z to ground is NOT present in the circuit (as it would be if a PI or L match network were present on the RF input side), makes me concerned. One other comment, question, how are you providing RF decoupling between the grids prior to proving a single cutoff bias supply feed. I wrestled with this a bit and decided to use series chokes, 2 uH, non resonate followed by shunt C, .01uF to ground for each RFC prior to joining the chokes to a single cutoff bias supply. I have seen a variety of approaches to this, highly simplified if you just DC return all the control grids to ground and control cutoff via the cathode (center tap filament) xmfr return.

Alan,
Thank you so much for the advice, a lot here for me to consider.  I cleaned up the filament leads a bit.  I can still add extra grid decoupling caps.  The reason I don't have the grids grounded to the socket mounts is that I used rubber washers to protect the sockets and so would not get a good clamp on the connection.

I am not sure I know what you mean about the dot on the filament transformer CT.  Also what do you mean by high f Z to ground?

I did some more testing this morning.  The grid and plate current shoot up as the plate capacitor is increased.  There will be no problem until you add a certain amout and then the current shoots up.  I have the relay K1 closed for this testing.  The RF input is connected to my transceiver (Not transmitting) and the output goes to a dummy load.  If I do not close K1 the problem goes away.

Rich

[W4AMV]

John,

The link below and the work done by this fellow should be of value. His posted schematic I believe has a small error as the plate parasitic choke is misconnected, but in total an excellent read.  

http://jetpropulsion.co.uk/gw0fzy-amateur-radio/hf-813-linear-amplifier

The dot connection on the filament transformer center tap connects the center tap to one side of the xmfr, I am sure its my copy or my eyes, but suspect you did not make that goof.

The RF chokes need to be checked while they are in place on the chassis. In fact, I would try to check as much of the tuned circuits and chokes as possible before ever applying power. The plate choke is a classic issue, as the surrounding elements affect the total shunt C across the choke and its parallel resonate frequency. Ideally, there are no resonant points and certainly not in band. A GDO is great for checking.

If you look at the trial and tribulations the above article goes through, the 813 as described in the Junker may not be the best configuration, where all grids (super grid configuration) is used. How do you cutoff your tube(s)...? I do not see that arrangement in the Junker and the lack of tuned input matching is another point of concern.

There are experts here on the forum who can guide you through the best arrangement for the 813's and I would be interested in your schematic of a pair of 813s as the Junker just shows one.

I am currently working on a pair of 811a's and I am sure despite checking as hard as I can any faults, will probably run up against some dumb founded oops!!

One thing that jumps out to me in looking at your photo, your bypass C's to ground, the leads to ground are just to long. I see in one case, the C goes to a tie off point and then from the tie off point a lead to ground. You need to take that C from the node to be AC bypassed immediately to ground. May not be you problem at all, but in general, short leads as a rule for RF bypass caps. 

[Donnie SWL]

Nice Rich now i see the rubber your talking about....how about taking the lock washer off the bolt bye pin 3 an 5 could you then put a double nut on them with a thin grounding strap between the nuts an still leave your rubber in place.....
Yes it don't look like they really have a cut off on the schemo i always put a relay between the center tap of the fil tranny and ground so it breaks contact in receive an cuts off the tubes....
the resistor you have off your g1 the 2ohm i would think that should be a non inductive type not sure the one your using is.....

[W7SOE]

Progress!!

I whipped up some parasitic suppressors and that killed the oscillation. Yay!

But still no output and no response from the tune or load caps. On a whim, I removed my power meter and went right to the dummy load. I then received "the scorch of goodness"!  I was making RF!  One of the padding cap leads arced over. Need to use HV wire.
So I had a bad patch cable. Ugh.

So I am getting 300w (40m) out at reduced 2400v. I think one of the 813s is suspect but I have more on the way.

When I had it overloaded for a bit one tubes plate was glowing and the other not. Quickly dived to turn off the HV.

Rich

[Donnie SWL]

Great Rich
I have never had good luck with braided wire on the plate caps on higher bands anyway....An that was one thing i was going to mention when we got to above deck but you beat me to it lol....have fun be careful....Don

[W4AMV]

Congrats Rich. Exactly what was the problem with the parasitic plate suppressor?
Wrong design values, none present, or what else? Also, how are you connecting the control grids for the 2- tubes. Using the grid suppressor provided in the article for each control grid brought to a common point or a different arrangement?

Alan