Arc between filament and plate

[hartlandrocker]

I completed a Trainwreck Express recently. It works well and I've been running EL34s in it for a few weeks since completion. Tonight I put a pair of new, tested JJ 6V6s in it for the first time. I was setting bias and went to check plate voltage. My probe slipped from lug 3 (plate) to lug 2 (filament). Apparently I shorted them and pulled my probe away immediately but there was a green arc that continued for about 2 seconds with an accompanying hum before I hit the power switch. The lug for the plate is partially burned away, vaporized! And there is a clear burn spot on the lug for pin 2 where it was arcing to. The fuse didn't blow. The filament winding has a grounded center tap.

I've never experienced this before so before I power it up again I would really appreciate any feedback as to where to check for possible damage before I put another set of tubes in as I don't want to cause any further unintentional damage.

[pdf64]

Think about insulating your probe with heat-shrink, or just tape, only leaving the tip exposed.
EDIT - credit where it's due, this was a suggestion from soundguruman http://music-electronics-forum.com/t31469/#post284471
I would replace the tube socket as the insulation may have been degraded by the arc.
Confirm that the heater winding CT to ground is still good.
Consider the fusing arrangements for the amp - is the primary fuse suitable, given that it didn't blow under such an extreme fault - does it have a fuse for the B+ winding?
Prepare for insulation damage on the OT primary, it may seem ok at low volumes but crackle when pushed.
The tubes may be ok, as they weren't the main subject of the fault current, though the screen grid current would have been high with the plate pulled to ground.
Pete

[Steve Conner]

You probably got away without any major damage due to the grounded centre tap on the heater winding.

I'm also surprised that a fuse didn't blow.

[bob p]

Probe slippage happens to everyone sooner or later. There are some things that you can do to protect yourself -- I really like the idea of insulating the length of your test leads with heat-shrink tubing.

Another option is to change the type of leads that you're using. For some applications where I'm working in tight spaces, it's suicide to use pencil-type test leads. In those cases, I prefer to use a mini-clip "grabber" type of lead, like those offered by Pomona. I bought my "Grabbers" with a BNC connection at the instrument end (good for hooking up to my signal generator), but the Grabbers with banana plugs work really well with handheld meters. I also bought a Pomona BNC to Banana Plug converter which helps to cover all the bases. Images below.

[bob p]

I've never experienced this before so before I power it up again I would really appreciate any feedback as to where to check for possible damage before I put another set of tubes in as I don't want to cause any further unintentional damage.

I think you probably already know this, but it bears repeating for anyone else who might be reading along: when you decide to power the amp back up it would be a good idea to power it up the first time without any tubes in place, and run through the amp with your meter.

[bob p]

> I'm also surprised that a fuse didn't blow.

I'm not. That is to say, I recently watched an amazing burn-up where arcing took place and output transistors got burned up and there was a fireball on the PCB lasting for several seconds before the 8A fast-acting fuse on the (+) rail finally protected the house from burning down. The 10A fast-acting mains fuse never blew. It wouldn't surprise me at all to see slow-blow fuses allowing such a fireball to last quite a bit longer -- long enough for you to get to the power switch before the fuse "fails".

This prompted me to review some fuse data sheets to get a better understanding of how long the fuses can last before doing their job. Doing this gave me the opportunity to re-think my fuse selection.

The typical low-amperage "fast-blow" fuse, like a Littelfuse 3AG fast-acting 312, will allow 200% of the rated current for a maximum of 5 seconds before it opens. (In a recent mishap I watched a ball of fire jump around the amp board on a Flame Linear while the 8A rail fuses waited to pass the equivalent of 16A of current for 5 seconds. My output transistors had <16A ratings, and the fuse failed to protect them while the amp board went up in flames.)

In contrast, the typical low-amperage "slow-blow" fuse, like a Littelfuse 3AG Time Lag 313, will allow 200% of the rated current for a minimum of 5 seconds and a maximum of 30 seconds before it opens. Looking at the time-current curves, it will pass a LOT more current for shorter periods of time.

The moral of the story is that fuse ratings are extremely important. The difference between fast-blow and slow-blow is significant (depending upon the application) and what my seem like small differences in current ratings could become very significant if the right rating isn't chosen.

I realize that many people reading this will already know this, but sometimes the numbers are worth reviewing. IME it's pretty hard (near impossible) to get to the power switch before a fast-acting fuse will open, but with a slow-blow fuse there's still a good chance that you can get to the power switch before the fuse activates.

Sometimes even a variac or a current limiter won't help you ... ie: if the mains fuse isn't the one that's going to blow. One thing that I've done to help my response time is to build a dead-man footswitch. I run AC power through a momentary contact footswitch that kills power to the amp if I take my foot off. My natural response to a fireball is to jump backwards, which cuts power to the device under test. You could do the same thing with a dead-man's switch that's attached to a lanyard on your wrist.

[hartlandrocker]

Thanks to all for the input. I really should know better as I've been doing this a long time. I broke one of my own rules and that is to not work on amps when I'm tired. It was late and I should have waited. I was concerned about the filament winding with around 400VDC going through it, so as suggested I will give it a thorough check powered down, then without tubes.

The sockets are NOS American made ceramic sockets I purchased close to 20 years ago that I finally put to use. The plate lug is half gone, so I'm going to attempt to find a socket with similar cinch pins and try and replace that one and the one on the filament pin which is burned a bit but in tact.

I have actually repaired other people's amps that had badly burned sockets from arcing but never witnessed it. The thing I don't understand is why it kept arcing after I removed the probe. I suspect it is something like surface tension of water that flows once it is breached? Maybe someone with a better grasp of physics can shed some light on that.

I'll report back with hopefully good news.

[Enzo]

Air is a great insulator, it takes X volts to breach Y distance. An arc is a plasma, which is conductive and a very low resistance. Once you have struck an arc with your probe, then it is easy for the plasma to continue. Starting an arc is the hardest part, continuing one is easy. That current is not then just flowing through air, it is flowing through the plasma.

Socket pins. ANy chance you left pin 6 on any of the sockets unused? The tube doesn't need it, you could rob it to replace the burnt one.

ANy more it is just me here, but in our shop in the past, we had a rule: no working on tube amps if you were on cold medicine.

[hartlandrocker]

Air is a great insulator, it takes X volts to breach Y distance. An arc is a plasma, which is conductive and a very low resistance. Once you have struck an arc with your probe, then it is easy for the plasma to continue. Starting an arc is the hardest part, continuing one is easy. That current is not then just flowing through air, it is flowing through the plasma.

Socket pins. ANy chance you left pin 6 on any of the sockets unused? The tube doesn't need it, you could rob it to replace the burnt one.

ANy more it is just me here, but in our shop in the past, we had a rule: no working on tube amps if you were on cold medicine.

Plasma, as in arc welding 101. I actually took arc welding in high school, but it never occurred to me. Thanks for making that point.

Thanks also for the tip about pin 6. Between the day job and yard work I won't get a chance to work on it till tomorrow.

[fredcapo]

Another option is to change the type of leads that you're using. For some applications where I'm working in tight spaces, it's suicide to use pencil-type test leads. In those cases, I prefer to use a mini-clip "grabber" type of lead, like those offered by Pomona. I bought my "Grabbers" with a BNC connection at the instrument end (good for hooking up to my signal generator), but the Grabbers with banana plugs work really well with handheld meters. I also bought a Pomona BNC to Banana Plug converter which helps to cover all the bases. Images below.




Bob - I would like to know what is the voltage rating of something like the dip clip...would hesitate to connect to B+

[Steve Conner]

These clip leads made from coax also have a lot of capacitance that can make your circuit go unstable.

[soundguruman]

I completed a Trainwreck Express recently. It works well and I've been running EL34s in it for a few weeks since completion. Tonight I put a pair of new, tested JJ 6V6s in it for the first time. I was setting bias and went to check plate voltage. My probe slipped from lug 3 (plate) to lug 2 (filament). Apparently I shorted them and pulled my probe away immediately but there was a green arc that continued for about 2 seconds with an accompanying hum before I hit the power switch. The lug for the plate is partially burned away, vaporized! And there is a clear burn spot on the lug for pin 2 where it was arcing to. The fuse didn't blow. The filament winding has a grounded center tap.

I've never experienced this before so before I power it up again I would really appreciate any feedback as to where to check for possible damage before I put another set of tubes in as I don't want to cause any further unintentional damage.

You have to replace the sockets usually, the burning between 2 pins forms a carbon trace across the socket, many times.

[bob p]

Bob - I would like to know what is the voltage rating of something like the dip clip...would hesitate to connect to B+

If you want to know the specs then I'd recommend doing a search for spec sheet for the specific model of probe that you're interested in. Pomona keeps all of the data sheets online. The reason that you need to do a search is because there are several different configurations of Grabber leads and attachments, all of which have different specs. There are some rated for higher voltages than others, some use BNC 50-ohm cable, while some use separate cables on banana plugs and jacks. You can have the hook-up made any way that you want, but the specs will change along with the type of cable and the type of connection.

The Grabber lead set that I've got uses a BNC 50-ohm cable and it's only rated to 300 VDC, but that's never been a problem for me, as I bought the leads to use with my signal generator. I don't really use it for taking measurements, and I haven't ever used it for measuring B+. That said, I recently did have to use it to measure voltages in a tightly confined space in a PTP-wired Phase Linear amp, because using the pencil-type leads was just too hazardous for that application. I didn't have any problems using it. If you're worried about capacitance, there are all sorts of Grabber connectors that don't use the 50-ohm coax. You can get them with separate leads to a BNC adapter, separate leads to stacking bananas, even just the connector with a built in banana jack and no wire. The specs are different for each configuration so you'll have to do some homework.

[hartlandrocker]

You have to replace the sockets usually, the burning between 2 pins forms a carbon trace across the socket, many times.

I thought about that, but they are ceramic sockets so I'm hoping they are less prone to arc tracks than phenolic. I could be wrong.

[hartlandrocker]

I replaced the burned pin and cleaned up the soot between the pins that arced going so far as using a dremel bit to grind away at the surface a bit. I then check all voltages without tubes, then with preamp tubes, and finally with power tubes. Everything checked out okay. I played through it just for a minute because it was late at night. This was two days ago and today is the first time I can really give the amp a workout. I'm hoping the arcing issue won't return. I do have some very nice NOS U.S. made Amphenol sockets and if necessary.