Tubelab SE rect. arcing -looking for more info

[wicked1]

My tubelab se rectifier used to arc. I gave up on the 5ar4 (after killing several in a row) and went w/ a 5u4g. Around that same time I thought the arcing could be caused by trouble w/ the PS caps and replaced them. I didn't have an extra 100uf for c5, which was at the time, 100uf electro, plus 80uf motor run. So, I just took off the electro and left the 80uf motor run as my c5.
So, for a year or two I've been running w/ 5u4g-20uf-choke-80uf.
(for those unfamiliar w/ the design, it goes rectifier-c4-choke-c5-[rest of the amp], so you know what im talking about when I say c5)
my c4 is definitely way below the maximum for the rectifier (my hammond tx runs hot, so I decreased it. It might actually be even lower than 20uf )

Now I've added the 100uf electro back to the c5 position in the amp, just to try it, (so rect-20uf-choke-180uf-[rest of amp])and when I powered it on, the 5u4g had some lightning for a second, and then functioned normally.
I added series diodes to the plates of the rectifier, and powered it back on.. bzz another flash, though smaller than the previous one. A spark jumps across the top mica, from plate to filament.

I've read the several threads here on the topic, and most are people changing things in an old amp, or building a new amp from scratch. This is a solid design, w/ all default value components. The 180uf c5 might be a bit high, but it's not out of the ball park for a tube ps. What could be causing this? All my voltages and bias settings check out after it's powered on, so it's just the initial surge. But, why my amp and not the other many tubelab se's out there?? I've triple checked all of my components and they are all correct.

thanks for any help (again)

[Ty_Bower]

It appears your choke is behaving less like an inductor and more like a piece of wire, at least when the amp is first powered on. What are the specifications on your choke?

[wicked1]

Right, I was thinking that and measured the dcr, and it is still what it should be.. (but that says nothing about what it's doing in the circuit, I guess...)

6H, 200ma, DCR 150ohms

Triad Magnetics - C-14X - Allied Electronics

[Ty_Bower]

It might be worth taking a look at the 20uF cap at C4. If it is leaking DC, that might make life harder on the rectifier.

[tomchr]

I need some clarification here... You say it arcs for a second on start-up. Is this right when you flip the power switch or is it right when the rectifier begins to conduct (filament glowing bright here)?

Have you tried a resistor in series with each plate of the rectifier - or one in the center tap of the transformer to limit the in-rush current? Granted, 20 uF is below the max spec of the 5U4G, but still...

Another option would be to replace the choke with a piece of wire and taking C5 out. Granted, you won't get much ripple filtering, but it would give you a data point. If the rectifier doesn't spark with only C4, then add C5 and use a 150 ohm, 5 W resistor instead of the choke. If you still don't have sparks, I'd say the choke is the culprit and you might need a snubber across it (see Morgan Jones).

If you have access to an oscilloscope and a probe that can handle the B+ voltage, you can probably cut to the chase by measuring the voltage on C4 as the rectifier warms up.

Above is if the arcing happens as the rectifier starts to conduct. If the arcing happens right at turn-on on a dead cold amp, I would blame the mains transformer for having crappy leakage inductance. Somehow the turn-on spike causes the rectifier to arc...

I'm getting curious here. Are you using one transformer for both B+ and filament voltage? Or is the filament transformer separate? Are there any center taps on your filament windings? What are you doing with those?

For reference: I'm currently using an Antek 1T250 in an amp of a design very similar to that of Tubelab Simple SE. I've tried with a 3T350 as well. I'm using a 5AR4 Sovtek rectifier and powering its heater from a 6.3 V winding through a 0.68 ohm resistor to drop the heater voltage to the specified 5.0 V. I'm using a 47 uF reservoir cap (C4 in your circuit) and an MOSFET based ripple filter following the reservoir cap. I've tried RC and LC filtering in addition to the MOSFET filter. I'm using an in-rush limiter (Digikey P/N: KC009L-ND) in series with the transformer primary. I've had no issues with arcing. The B+ is about 300 VDC with the 1T250 and 400 VDC with the 3T350 transformer.

~Tom

[rknize]

It sounds like the inrush surge is a bit much for the rectifier. Note that 5U4xx tubes are directly heated, which means they come up fast. One of the nice things about the 5AR4 is the slow warmup. I would recommend trying an inrush limiter on the center tap of the HT winding. That will slow the charge rate of the caps to keep the peak forward current within limits on the 5U4.

All that said, I did try a 5U4G for a brief time on my TSE and it worked fine. I believe my setup is 47uF - 10H, 80ohm - 220uF. I usually stick with the 5AR4 and have never had one arc in the TSE.

[wicked1]

-- Is this right when you flip the power switch or is it right when the rectifier begins to conduct (filament glowing bright here)?

I always thought it was right when I flip the switch, but I just now did it and paid attention, and it's a couple of seconds.. Right when the filament starts to glow is when it arcs.

--Have you tried a resistor in series with each plate of the rectifier - or one in the center tap of the transformer to limit the in-rush current?

No, but was just reading about that.. I don't have one on hand that can handle enough power.

--Granted, 20 uF is below the max spec of the 5U4G, but still...

I'm actually at 4.7uf, I just checked! I get a b+ of about 380 this way. I remember when I was using ~40uf I was getting about 430v. (my tx is 250-0-250, hammond 272jx. It runs way hotter (more power output) than it should.


--Another option would be to replace the choke with a piece of wire and taking C5 out.

That is my next step, I suppose.


--I'm getting curious here. Are you using one transformer for both B+ and filament voltage? Or is the filament transformer separate?

One, it's got the 600vct, 6.3vct and 5vct.

--Are there any center taps on your filament windings? What are you doing with those?

I'm guessing you mean for the rect, and yes. It's not connected.. Taped off tied out of the way. (ct on the 6.3v is connected to ground)


-- I'm using an in-rush limiter (Digikey P/N: KC009L-ND) in series with the transformer primary. I've had no issues with arcing. The B+ is about 300 VDC with the 1T250 and 400 VDC with the 3T350 transformer.

I've got a couple of cl-60's on hand. Do you think that will work for this? It's got less resistance, but looks about the same otherwise.
CL-60 datasheet and Application Note, Data Sheet, Circuit, PDF, Pinout | Datasheet Archive


Then I guess you mean I just connect one in series w/ one leg of the tx primary (120v side)? Between the fuse and the power switch should work, right? (It's easy for me to get to that spot)

Or, should I do like rknize says, and place it on the ct of the 2ndary?
(or one in each location since I have 2)

And rknize, I blew through 5ar4's like they were firecrackers! I never knew if I'd turn on my amp if that would be the last of the 5ar4 or not.
I've gotten over a year out of one 5u4g.

I feel like I'm just putting a bandaid on the problem by adding the diodes and now these current limiters, but when I was dealing w/ the 5ar4's blowing the first time, I replaced literally every cap, and tested every resistor in the amp. It made no difference! I've just gone through and re-tested every cap in the amp, and they all test ok. A bandaid might be my only option

one more thing.. It's always the same one 'diode' of the rectifier tube that arcs. But, each are exactly the same in the circuit.. (just tx lead connected to plate)

[rknize]

Adding inrush limiters on the primary will not help your issue as much as if you put one on the center tap. This is because the limiters will already be up to temp (no longer limiting) by the time the rectifier is conducting due to the loads from the tube heaters at initial turn-on.

I don't have limiters on my TSE right now because it is sitting on a breadboard. I will add them once the amp is in a chassis just like I did for the SSE. I put one on the primary to make the transformer and heaters happier and one on the HT center tap to make the rectifier happier.

one more thing.. It's always the same one 'diode' of the rectifier tube that arcs. But, each are exactly the same in the circuit.. (just tx lead connected to plate)

Have you ever put a meter on your transformer output leads? Measure across the entire winding and then between each half and the center tap. Ideally, the last two measurements should each be half of the first measurement. I've seen where someone accidentally used the bias tap as the center tap.

[Ty_Bower]

Originally Posted by wicked1

I've got a couple of cl-60's on hand. Do you think that will work for this?

CL-60 is designed for a max steady state current of 5 amps. You'll never draw enough current through the thing to get it up to full operating temperature. Besides, it's only 10 ohms when cold. That won't do a whole lot of inrush limiting.

I never was a big fan of sticking IRCLs on the high voltage secondary side of the PT. Most of the common CL-xx parts are designed for currents of 1.5 amps or more, and you just don't draw that kind of load off the high voltage winding. Stick them on the primary side, where they have a chance to do their job. Of course you'll also slow down the warm-up of the tube heaters, unless you have a separate filament transformer. Maybe that's not a bad thing.

My favorite is the CL-90. It has one of the highest cold resistances (120 ohms) and is designed for a current draw of 2 amps. That seems to fit well with small push/pull EL84 type amplifiers, or single ended amps with 6L6 or EL34 tubes. The 3 amp rated CL-80 can also be useful if you need more than the CL-90 can handle. I suppose the CL-140 might be interesting: 1.1 amps full load, and 50 ohms cold.

[wicked1]

Ok, I understand the cl's now, thanks.

One more comment, and I need to let this thread rest for a while ...

I noticed the filaments of the 5u4g are kind of flaky/crusty, and when it arcs some of that crusty corrosion (or whatever it is) flakes off or burns up.. the filament is brighter in the places it has arced. Maybe this is all just a coincidence, and my rectifier is going bad. It started exactly when I added this cap, but maybe the little extra pull is pushing it over the edge.. I dont know. I don't know how long one of these should be expected to last, but I might have about 2000 hours on it.
I put a new rectifier in place and it did not arc, but then, it doesn't every time. I'll flip it off and on a few more time over the evening (Im listening to it now tho, so no more testing for a while).

thanks for all the help