Opinions Wanted on Problem with a Pair of 3-500Z Filaments in Series...

[K1JJ]

Hola,

Last night my 1966 Henry 2K-3 pair of 3-500Z's started acting up. The power dropped to one-half. I looked inside and saw one tube's filament hardly glowing while the other was brighter than normal.

Normally, the Henry design has the tubes in series with 10V split across the two, giving 5V filaments. I pulled the amp apart and measured 2.3V on one tube and 7.6V on the other!

I switched tube positions and the low filament tube still glowed low in the new position, so figgered maybe it was the tube.  I then switched in four more tubes and the same things occurred with all tubes! They would glow equally for one minute and then one would get brighter and the other dimmer.

I disconnected all bypass caps around the socket and the same thing. The fil choke was OK too. Strange.   I always hated series filaments.

So, I said the heck with series filaments and put them in parallel with a new 5V 30A fil transformer and wound up a new bifilar choke. While I was at it, I also added a big squirrell cage to the back with direct air thru the chimneys with a variac to slow it down. Big difference in cooling now. 

I fired the amp up and the filaments of ALL tubes now glowed equally at 4.95V each in parallel. (as you'd expect)  With 4KV on it it did an easy 2KW out. (I use an outboard homebrew power supply)

I'm still baffled as to why the series circuit was acting this way, since this is the Henry design that has worked for years. I even resoldered the pins of the tubes themselves and wiggled the socket pins... they were all fine. As proof, the unit works fine now with all tubes...  What could be happening to make one tube drop much more voltage across it than the other... with six different tubes?

Problem solved, but I'm curious yallo.

T

[WA1GFZ]

Tom,
When I bought my box of 4CX3000As the seller told me that 2 were almost new and two were fairly new and the last pair had a lot of hours on them. At the time I didn't even have a socket. I built a little test fixture to fire up the heaters amd measuer heater resistance.
I ran the tubes at a fixed variac setting putting about 5 volts on the tube because I didn't know they took 9. I figured they took at least 5. Anyway I found no two tubes have the same resistance but the newer they were the lower the resistance.
I put a pair of 100 watt light bulbs in serise under the water meter at beach QTH so it doesn't freeze in the winter. I cover and insulate the area. This past summer I took one of the bulbs out to install in a new fixture. This fall I grabbed a new bulb to replace it. Sure enough one lit brighter than the other.
I think you did the smart thing and eliminated the JS. I'm sure there is not a way to guarantee  two heaters will be or remain the same resistance.
 

[WBear2GCR]

Tom,

How did you get the tank and the HV feed thru areas to handle 4kv??

Whenever the HV gets above 3kv (bleeders go south) the thing arcs like a banshee!

Otoh maybe yours is a different variant, mine seems fairly early.

Also, I measured the Load caps and did not find them to be anywhere near the spec'd value, did you ever check urs??

The differential in the fils, is a bit odd...



             _-_-bear

[K1JJ]

OK Franz,

Yes, series connected filaments are a bad idea.  One thing to be aware of for series fils is if you have one hot tube and then plug in another that is cold (trying different tubes) the cold tube can get its filament popped. The fils have big differences in cold or hot  resistance values. I once popped two series connected 6146 fils in a row not knowing any better. The fil changes resistance quickly from cold to hot.

What I don't understand about this 3-500Z situation is that both tubes started out glowing at the same brightness, then one would drop suddenly (in one second) to low after about 60 seconds. Tube after tube did this.  If I turned the fils off for 30 seconds, they would still be different. But if they cooled down for 5 minutes, they would start out at the same brightness again and then go through the same cycle.

Henry did series to use a light fil choke and wiring... 15A vs: 30A.
I will never use series connected filaments for power tubes again.

73,
T

[K1JJ]

Tom,
How did you get the tank and the HV feed thru areas to handle 4kv??
Whenever the HV gets above 3kv (bleeders go south) the thing arcs like a banshee!
Otoh maybe yours is a different variant, mine seems fairly early.
Also, I measured the Load caps and did not find them to be anywhere near the spec'd value, did you ever check urs??
The differential in the fils, is a bit odd...
             _-_-bear

Hi Bear,

Good question.

At first I had arcing. I notice it did not arc at 4KV into a flat 50 ohm dummy load last night. But when the ant gets reactive, it would arc in one of the tuning breadslicer caps. I try to keep the ant near 50 ohms  j0 to keep the Pi net voltages down.

I also found that the clearance between the top of the tube plate caps (and suppressor network) was too close to the screen enclosure. I put small spacers to lift the top screen cover up about 1/8". This helped a lot.

Loading caps... is yours a PI-L network or a regular pi?  What loading caps do you mean... fixed or the breadslicers.  Yes, I have noticed that the loading cap on 75M is usually near to full in, but I have never run out of range yet.

I have a variable 0-4KV supply. There are days when I cannot get it above 3500V before I hear arcing. But when the tubes are new, the gods smile and I can give it 4KV, that amp really sings. In stock PS config at a sagging 2900V? I usually had trouble getting 1200-1300W out pep. 

I use it just for about 150W carrier on AM as a backup and occassionally on ssb loafing at 1200w or so. The headroom is simply for best linearity. To get full output requires about 130W? pep drive, which is overdriving the grids anyway.

73,
T

[Bacon, WA3WDR]

It would be interesting to measure the filament voltage-current curves on the two tubes that did that to you.  Thinking about this, I wonder that filament series connection ever worked at all.  But it was used in radios and TVs for many years.  And every filament is a series circuit of a sort itself.  Imagine sections of a filament deciding to go screwy like that inside of one tube, or in a light bulb.  I've never seen that happen. 

[K1JJ]

It would be interesting to measure the filament voltage-current curves on the two tubes that did that to you.  Thinking about this, I wonder that filament series connection ever worked at all.  But it was used in radios and TVs for many years.  And every filament is a series circuit of a sort itself.  Imagine sections of a filament deciding to go screwy like that inside of one tube, or in a light bulb.  I've never seen that happen. 

Yes, that wud be a worthwhile experiment indeed.

When ya think about it, if the fils acted flakey in a fil parallel circuit, you wud probably never know it cuz the fil voltage wud stay constant across each tube. Even so, how many guys pay close attention to the exact brightness of each tube in their linears when in parallel?

It stills sounds like something was intermitant, but the only thing across each tube is a bypass cap - and I cut it out as a test.  The tube pins were good, since I  clamped them, wiggled them and could not see any indication of change. I resoldered all connections. As further proof, the amp works perfect now in fil parallel... same socket components and tubes...

I still can't believe that the fils of six tubes were flakey and now run FB in parallel...
I must have missed something.

BTW, all these tubes are used pulls I  got at flea markets. Two are full power and the others are fair. Maybe it has something to do with being directly heated filaments of power grid tubes, I dunno.

T

[WBear2GCR]

Tom,

Did the plate and grid meters sit quietly when the fil color changed??
Was a scope nearby to look at the "waveform" or lack thereof?

Thinking oskillashun?

The arc I got was on top of the large left hand cap, I think that's the plate variable tuning cap.

The cap I measured is/was the breadslicers in parallel on the ANT side of the deal.
Measured them in two decks. My cap meter said 250pf both times! The schematic calls for 500pf, iirc.

They're stock, and both decks are the same electrically, one is the floor console and one is the desktop.
The console has good PS and arc'd up and crapped out meters on the deck, the desktop has fried xfmr in the PS and good (maybe) deck. Last time I fired up the "good" deck the loading cap seemed to have NO effect. (into a good dummyload)
Bizzare.

            _-_-WBear2GCR

[WA1GFZ]

Bacon,
I thought the same thing. I bet it is less of a problem when the heater current it lower and voltages are higher. My 2 100 watt bulbs in series one bulb is almost full brightness and the other is kind of dim. When they were both matched out of the box they were equal brightness.  Still I bet the voltage difference is not all that much unless one tube is a lot higher resistance. Heck doesn't take a lot of R at 15 amps.

[Tom WA3KLR]

Tommy Vu,

I'm just reading this for the first time.  It sounds like it is due to the resistance/ positive temperature coefficient of filaments.  Unless they are perfectly matched in the voltage and current curve over a fair range, one will run hot and then the other gets less voltage, cooling down and the resistance goes down, putting more voltage on the other tube.  This is a condition that will continue to create big difference and reach equilibrium at the value you saw.  The see-saw voltage divider. 

I think most tube filaments will tolerate a short period at 2 times the normal voltage.  It would have been interesting for you to short across the tube with the higher filament voltage and see if the tubes would then sit in the opposite state.  I think they would.

When you think about it this is insane, just 2 tubes in series.  Especially directly heated.  Not good for the grid-cathode bias either.  The engineer that designed that should be shot. 

Consider what many muti-tube tube rigs do. They have a series/parallel string and perhaps resistors on the bus too.  With this many other loads, things are force to stay in balance.

What the guy did was not good, but if he had put some power resistor in parallel with each tube it would prevent that from happening; say an additional 50 Watts of wasted power.  What you did was the best thing to do.

[WA1GFZ]

Gee Tom,
You got me thinking. A perverted js like this may self compensate in time. The tube hogging the power would get hot in turn increasing the heater wie resistance making the other tube share more of the load. The see saw action would continue until you get a balance.
Imagine the IMD generated while this dance is going on.
Only an over educated idiot bean counter would think of the money saved on this stupid design idea

[WBear2GCR]

Ya, except for one thing... the one that I have doesn't do that!

At least so far, and as much as I have seen so far...

It has other problems, but not that.

       _-_-bear

[K1JJ]

Interesting comments....

Bear: No oscillations, since I was testing it unkeyed with only filament voltage applied.
The fact that my amp worked OK for years and yours works OK makes this situation strange for sure.

Tom: BTW, I built up your LC network and it matched the 60 ohms, +j22  hardline perfectly.  I used a variable vac cap and a silver plated coil. I was able to tweak it to ~ 52 j3.    After a KW for a minute, the coil and cap are cold, so it works FB.

Tnx again.

Oh yeah... while I was testing various cables in the shack, I came across a male to female right angle coax barrel that measured 140 ohms through it. I couldn't understand this, so I sawed it apart to take a look. The bastards build it with a tiny coiled spring that connects the two pins inside. Dave Boot warned me of this shoddy type plug years ago. He said at UHF it had a high swr and he sawed one apart too.  Evidently it could not take the 200W from my FT-1000D.   I've got about 10 of them in service here.. cheezz..

73,

T

[Tom WA3KLR]

I've  seen at least 3 bad elbow connectors in my career.  That dry fit is the only way to make them unless someone wants to get into welding.  But everyone expects to pay no more than a couple dollars for the stuff.

I was given a coax rotary switch a few years ago.  Every SO-239 on it had an elbow.  I found 2 bad elbows.  I bet the previous owner thought it was a bad switch.

We had a bad one on our repeater receiver antenna input jack - gave us an intermittant receiver.

Obviously for a high reliability situation you want to avoid using them if possible.

[KB2WIG]

               "I've  seen at least 3 bad elbow connectors in my career. "

Years ago, a friend of mine used an 'elbow' and the little screw on cap & chain to make what he called a  " coax-a-bowl " .   klc

[WA1GFZ]

The first thing I do with anything SO239 is convert to N.

[K1JJ]

What bugs me about 'N' connectors is that tiny, thin center conductor... compared to a PL-259.   The N certainly has a better pin mating fit and connector seal, no doubt.

T

[WA1GFZ]

Tom,
I have put 5 KW through N but bigger then go to HN. PL259 elbow it is common to have the cheezie spring thing.
C connectors is also good as the center pin is bigger but crappy for moisture. They are hard to find.

I can't remember the last time I saw anything mil or quality test equipment with an SO239 because they are junk. The only thing they are useful for is to bash out the guts and use them as a strain relief for a coax shield in a HV lead. Then run the center conductor through the hole.
I have never had a n connector fail if installed properly. Do not use 75 ohm N with 50 ohm N
75 center pin is smaller.

[kf6pqt]

Years ago, a friend of mine used an 'elbow' and the little screw on cap & chain to make what he called a  " coax-a-bowl " .   klc

aka nose-hair burner.

[W3SLK]

Frank said:

The first thing I do with anything SO239 is convert to N.

My old buddy and psuedo Elmer, Eric, WB4VVI (SK), said they were good but that out doors, he had problems with the pin due to expansion and contraction of the contacts in inclement weather. I've never had any problems with the SO-239's.

[K1JJ]

Frank said:

The first thing I do with anything SO239 is convert to N.

My old buddy and psuedo Elmer, Eric, WB4VVI (SK), said they were good but that out doors, he had problems with the pin due to expansion and contraction of the contacts in inclement weather. I've never had any problems with the SO-239's.

Yes, I've never had repeatable problems with PL-259's and SO-239's outside as well... other than when the waterproofing failed.  I have maybe 100 connectors out in the field, some for 10 years. I use them for interfacing hardline to coax, etc.

Sealing them well from the WX is the key. I've pulled some apart after many years and they are still bright and shiny inside.   But there have been a few that got condensation inside and turned green...  the N connector would have done better with its better seal.

But there is always a potential problem when any connector is sealed and used outside. It always traps some air, and this air can easily condense from temperature variations away from the time the air was trapped, no matter what kind of connector is used.  Is this correct?

 73,
T

[WA1GFZ]

The taller type n connectors are great. The connectors on my quad were good after 20 years. Yes the seal is the key.  The ground must be tight on a pl259 for it to work while the n has 4 or 6 prongs to make gnd and it doesn't have to be tight.

[K1JJ]

  The ground must be tight on a pl259 for it to work while the n has 4 or 6 prongs to make gnd and it doesn't have to be tight.

Do you tin the braid and then solder it to the inside of the PL-259 connector? I use a torch to sweat the ground shield for full contact to the housing.   Then tape it with at least six layers of Scotch 33. Then I cover it with a thick layer of RTV. The RTV is good for about five years and then needs to be re-applied, but works FB.

T

[WA1GFZ]

I take a PL259 adapter to BNC or N and move on myself. I have not used a pl259 since jsertrola days. I haven't used on in theshack since about 1974. Yes I used to tin the shield so solder would flow easier. The trick is a lot of heat and move fast.

[The Slab Bacon]

I take a PL259 adapter to BNC or N and move on myself. I have not used a pl259 since jsertrola days. I haven't used on in theshack since about 1974. Yes I used to tin the shield so solder would flow easier. The trick is a lot of heat and move fast.

I never tin the braid, that makes it too difficult to get the connector on. Like Frank says:
"use a lot of heat and move quickly". this is a good app for one of those big, manly
Weller 240 / 325w big soldering guns. If you try to use a little girlie soldering iron you just cant get emough heat to make the solder flow.

I hate to see those wimpy S.O.B.'s who are afraid to solder the braid on their pl-259s. This is just asking for intermittant problems when you are running high powered equipment.
                                                     The Slab Bacon