Bad 4D32 filaments - What is cause?

[K1JJ]

Currently, I'm running three 4D32s in parallel for the RF final in the PDM tube rig. 

Out of seven NOS tubes, four work FB while the other three have problems.

Two of filaments have gone bad.  When I first lit up the filaments of these bad tubes, I saw a very brief bright filament glow, then it settled down to normal. After 30 minutes, the glow came back and stayed this way.  It looked like 12.6 volts on a 6.3V filament... very bright like a light bulb.   The other tubes with their fils in parallel were fine. So I know that the 6.3V source was OK.

The 4D32 has an internal filament center tap. Could it be this CT is shorting to one side of the filament leads, and effectively putting 6.3V on a shorted 3.3V filament?

Again, the other four tubes have been running FB for 2 weeks now, so I doubt it is the rig itself.

Anyone familiar with this problem?

T

[AB2EZ]

Tom

Based on you description of the failure syndrome, and the fact that four tubes were in parallel, but only one tube failed: your suspected cause of the failure makes a lot of sense.

I looked up the 4D32 on line... but I didn't see any mention of the filament center tap in the data sheet I found.

What is the internal filament center tap connected to?

If it is the cathode, then I would expect to see 0 ohms between the cathode and one of the two filament pins of a bad tube. I.e. the other half of the filament would have fused, but the side of the filament that was shorted to the center tap would still be fine.

Stu

[Pete, WA2CWA]

Might be a form of filament sag; touching filament wire with reduced overall length and resistance. I'm not sure when the 4D32 was introduced they had any preventative manufacturing procedures to reduce this problem.

[WA5VGO]

I can't speak specifically to your problem, but this might have something to do with it. The 4D32's that showed up on the surplus market several years ago are of a very late manufacture (sometime in the 1980's, if I recall correctly). I have always suspected that they might not have the same quality as the ones built in the 1940's and 1950's. It's easy to distinguish them from each other. The late ones are labeled in red ink and the early ones have silver ink. Maybe there's no problem with them, but certainly a large percentage of the tubes manufactured during this period had issues.

Darrell

[Jim, W5JO]

I can't explain what the problem is but have a pair of 8643 tubes that I put in a Globe Champion.  I did the conversion using some 5894 tubes and it worked fine.  Then I plugged in the 8643 pair and the fuse went immediately with application of power.

I pulled the tubes to check and that is when I noticed that one of the plate structures was leaning at a very steep angle.  It had shorted against the gird and that is all it took.  Looking at it carefully, it was one that escaped the QC person at the factory for the structure was welded at both the top and bottom supports.

So my guess is you have some that made it out of the factory without checks.

[AB2EZ]

I found an on-line reference for a Raytheon  RK-4D32 that states that pin 3 is the filament center tap.

http://www.r-type.org/exhib/aai0005.htm

But the detailed Raytheon data sheet says that only the 4D22 has a center tap at pin 3.

Stu

[Pete, WA2CWA]

Raytheon's RK 4D32 and RK 4D22 are the same tubes except for filament voltage.
RK4D32 is 6.3 V filament; pins 1 and 7; no center tap to pin 3. Pin 3 is blank.
RK4D22 is 12 V filament; pins 1 and 7 with center tap to pin 3.

http://www.r-type.org/pdfs/4d22.pdf

[K1JJ]

A-ha!

You are right, Stu.... about a short to the cathode.  Correct, no internal fil CT on the 4D32, just the 4D22.

I already sent the other shorted tube back, but this bad one today reads 0.1 ohms from heater side (pin 1) to cathode. The other heater side (pin 7) to cathode is 1 ohm.  

Normally a 4D32 pulls 3.1 A of fil.     6.3V / 2 ohms (normal fil resistance)  = 3.1 A.  So looks like one side is shorted to cathode giving 6.3 V across a 3.3 V fil.   My fil xfmr CT is tied to the cathode to keep the floating cathode to fil voltages low.  This also completes the short circuit in the fil.

Maybe a fil bypass cap is intermitant?  Would that spark an event? I have 0.1 / 100V in there from fils to cathode.   I will replace them all and increase to 1KV.  Is .01 enuff ?

I never had a problem with the 6146Bs. So the basic design is good. But since then I built a whole new plate for the 4D32s.  All new wiring and parts on the sockets.

Yes, they are the RED inked tubes. They are recent govt surplus as found on eBay. Maybe some are marginal and the fils sagged, while others are just good enuff to pass.   The seller is bending over backwards to replace any bad ones.... good guy.

Question:  With an indirectly heated cathode like a 4D32, what is the reason for fil bypass caps?  Since the cathode carries the RF current and is tied directly to ground (floating plate) why is the fil bypassed too? Maybe cuz the fil and cathode are closely coupled and might as well bypass both?

T

* UPDATE:  I connected the ohmmeter to the cathode and heater with clip leads. I shook and tapped the tube and the short went away. I plugged it in and the fil looks normal. Definitely a wimpy element in there that will come back.

[K1JJ]

Problem solved.  Or at least a simple fix that will let me run a tube that is shorted from filament to cathode without burning up the filament.  The problem was created by a tube manufacturing defect.

A friend of mine gave me the fix for the rig. Out of the box thinking. I am waiting for his permission to post his comments.  

The rig works FB with shorted tube(s) in there now....     Works even if they are all shorted fil to cath. The fix will probably prevent it from happening again to ALL other good tubes too, or at least from doing serious damage to the filament structure if it shorts like before. No hum generated - and runs at full power, full voltage.


What is the fix?   Better yet, for fun, can anyone guess this simple solution?

T

[W3RSW]

You switched the filament leads or checked each socket's wiring to make sure that the shorted side of the filament was closest to or at ground.  Then eliminated center tap on fil. Xformer, if any, to be sure that 6.3 volts was fully on the non-shorted side of the filaments then newly grounded or now correctly grounded bad side filament pin.

This works if there is no bias split between cathode and G1.  If there is split bias, say from a cathode resistor, then cathode current metering metering and some other circuit placement items may have to be modified.

Did all the shorted tubes show short closest to same filament pin? This would simplify building up a circuit for all buying the recent red letters tubes.

[AB2EZ]

I may be duplicating what Rich is saying, but here is what the latest information from Tom suggests:

Since the CT of the transformer is connected to the cathode bus (in the unmodified configuration), a short between the center point of the filament and the cathode should have no effect.

A short between the cathode and the filament, located (for example) in the middle of the left half of the filament would result in the following:

A) 3.15V between the left end of the filament and the location of the short.
B) 3.15V between the right end of the location of the short.

Therefore, in this example, one would have 3.15V across 25% of the filament and 3.15V across 75% of the filament

This would cause the left side of the filament to draw 4x its rated current.

The solution, as posted by Rich would be to eliminate the connection between the CT of the transformer and ground, and to connect the end of the filament that is closest to the short to ground (cathode bus)

This will put 6.3V across the portion of the filament whose end is furthest from the short, or across the entire filament if the short clears.

A bypass capacitor from the other side of the filament to ground will keep RF out of the filament.

Stu



Note: in the original configuration, the closer the short is to one end of the filament, the more current would flow through the portion of the filament between that end and the location of the short. In the case that appears here, you would have 3.15V across 0.1 ohms (31.5A) and 3.15V across the remaining 1.9 ohms (assuming the transformer could deliver that much current... which it probably can't).

[K1JJ]

The solution, as posted by Rich would be to eliminate the connection between the CT of the transformer and ground.

Stu

Very close!  Good job.  Though there still needs to be DC continuity from fil to cathode to keep the voltage differential low to prevent breakdown. The datasheet shows fil - cathode rating at 100V max for a 4D32.  

4D32 specs:
http://www.shinjo.info/frank/sheets/138/4/4D22.pdf

It goes to show that by saying there IS a solution, even though there may not be, sometimes opens up the floodgates to good solutions.  


First, let me say that I am not confident about the position of the short or that it is 0.1 ohms.  I discovered that my ohmmeter is not accurate down below 1 ohm. So lets just say there IS an intermittent short from fil to cathode (for sure) and that it makes the tube filament burn VERY brightly when it fuses.  By tapping the tube it will come and go.


Here's the solution as described by Chuck, K1KW, who had the same thing happen to his Ranger and solved his problem: (Chuck is a former Varian-Eimac engineer)


"On the 4d32's, yes, that is a manufacturing problem  since you have no voltage between the filaments and cathode in your set up.  It is a common problem in high power indirectly heated tubes of older design.  The insulation resides on the inner surface of the cathode and the filaments actually are in contact with it and unless the process is correct, the insulation will be broken down and you have a filament to cathode short.  
 
There is a design fix!  Try connecting the fil CT to the cathode through a 1K resistor with a 0.01 cap across it.  Make sure then that the return of tube current is connected to the cathode, not the fil CT.  This way the fil and cathode are at the same potential for DC and RF but if there is a short, negligible current will flow.  You can use a tube with a filament to cathode short this way.  Sometimes you may get a small amount of induced hum if the filament is running on AC but not usually."


"History:  It showed up in my Ranger after converting it to 8417's.  I was using cathode bias and had a filament to cathode short.  Went to fixed bias, grounding the cathode and returning the fil CT to gnd this way.  Saw the same filament flare up when the CT was directly grounded!  So....that's how this fix came about. I heard a very slight increase in hum but in your setup, I'm not surprised there was none.  Not really sensitive to AC there.  Sure, post the solution with attribution if you wish, perhaps it will help someone. "


Bottom line is that I have the shorted tube in the PDM rig working now at full power and no noticeable increase in hum or filament brightness.  I think this is also a great preventative design idea to protect good tubes IF they ever have a cathode to fil short. The filament structural damage will be nil.  Chuck's comment about the filament normally touching the insulated cathode is an eye opener.  I consider adding the 1K and .01 in series with the fil xfmr CT just like putting a 25 ohm resisitor in series with the HV supply... protecting parts from potential shorts, etc.


T

[Pete, WA2CWA]

Maybe I'm not reading your post correctly Tom, but there is no center-tap filament connection to an external pin on the 4D32 according to the link you provided for the tube.

[AB2EZ]

Tom

Comparing tubes that appear to be working normally, to tubes that appear to have problems (whether working okay in the new configuration or no longer working at all):

What values of total resistance do you see when you measure the resistance between the two filament pins?

I'm wondering if some of these tubes have a low value of resistance, from filament pin to filament pin, because the filament is electrically contacting the cathode in two or more places... producing a low resistance bridge across one or more segments of the filament (even if the tube still works in the new configuration.)

Stu

[K1JJ]

Maybe I'm not reading your post correctly Tom, but there is no center-tap filament connection to an external pin on the 4D32 according to the link you provided for the tube.

Yes, Pete, you are correct - there is no internal tube CT to pin 3. My mistake.   It's only for the 4D22.  I did mention this a few posts back.  The short is from one side of the filament (pin 1 or 7)  to the cathode.... or as Stu pointed out below, maybe contacting the cathode in two or more places.

(Maybe I  confused tube internal CT with xfmr CT a few times in the posts.)



Stu said:

"I'm wondering if some of these tubes have a low value of resistance, from filament pin to filament pin, because the filament is electrically contacting the cathode in two or more places... producing a low resistance bridge across one or more segments of the filament (even if the tube still works in the new configuration.)"

Stu,

That's a very good point.  I did measure the F-F pins of a shorted tube and found the resistance to be somewhat lower than a normal tube, so that may be what happened.  But right now I can't make another measurement cuz I was able to tap/shake the shorted tube until the short cleared... and plugged it in. Maybe I can get a long life out of it as a "normal" tube. It won't matter now if it shorts with the modification, but might as well not wake a sleeping "dog."  

T

[AB2EZ]

Tom

Agreed!

Stu

[K1JJ]

BTW, with the rig producing power, I measured the voltage on BOTH sides of the new 1K resistor just to make sure there was not a voltage drop across the resistor due to "static" current.    The voltmeter showed 0 volts drop. So looks like the fix is doing the job intended and the tube fils are safe from high currents thru the transformer center tap.

T

[Pete, WA2CWA]

What does one do if they're using a transformer without a CT?

[K1JJ]

What does one do if they're using a transformer without a CT?

Good question.


We can make an artificial center tap using resistors.   In fact, I'll bet a pair of 1Ks across the winding, with 0.01 uf caps across each - to form a center tap connection that then runs to the cathode would work FB.

T

[Jim, W5JO]

Is the supplier replacing the defective tubes?  I believe I read in a previous post that they are.  I certainly hope so for I would not want to use defective tubes even though there is a work around.  In my years of ham radio I don't recall any tube supplier urging a work around. 

If the tube in question is the last or a very expensive piece, then that is one thing but I believe the 4D32 is quite common.  I hope you are getting replacements.

[AB2EZ]

Tom

Thinking about the solution in which you connect the CT of the filament transformer to the cathode using a 1Kohm resistor in parallel with a bypass capacitor.

Question

Since each half of the center tapped filament winding has a lot of inductance, what purpose is served by a bypass capacitor in parallel with the 1Kohm resistor?

It seems to me that if you want to ensure that there is no RF between the filament and the cathode, you would be inclined to place the bypass capacitor between one filament pin and the cathode bus, or to use two capacitors to bypass both filament pins to the cathode bus.

The derived center tap method you described would also work since it employs a bypass capacitor from each filament pin to the cathode bus.

Stu

[K1JJ]

Jim,

Yes, the supplier is replacing the tubes. In fact, I sent two back and he insisted that I just throw away any more... no need for him to verify.  He said he would pull the  replacements out of a new case, hoping for a better quality.  Pot luck.   

The fix was suggested by Chuck / K1KW who has no connection with the supplier, except maybe they are both carbon units.


Stu,

I have the filament pins of all 4D32s bypassed to the cathode pins using 0.003 uf 1KV caps.  But I do see your point of why is the 1K resistor bypassed if we are looking for isolation from RF for the fil transfomer ?  Maybe Chuck has a reason and will chime in.  We spent about 1 1/2 hours today chatting on 75M.  Lots of good subjects discussed.


T