ST-120 Rectifier Puzzler…

I built my ST-120 in April 2011 and the amp sounds amazing as many of you know. It now has at least 1200 hours and I run Gold Lion KT88’s, GL 12AT7’s and the Tung Sole KT-120’s and a variety of Tungsram and Mullard NOS. They all sound great with slight differences and very musical. I do not use a pre-amp.
Recently I turned on the amp and the Sovtek GZ34 sparked and blew, it may have had 400 hours. I replaced it with another new production Sovtek and that tube and another blew the 5 amp fuse. I decided to try one of my 1958 Mullard GZ34’s. The relay clicked, no sparks and the amp come to life as normal. I played the amp 3-4 days until I received 3 new JJ GZ34’s. I tried a new production JJ and the sparks and fuse blew as with the new Sovtek’s. I know the NOS tubes are much better built with better materials and all that…  The question is why did the amp run on both new and NOS GZ34’s for two years and now only the NOS? Is there a bigger problem in the Amp that I have not found?

The specs to my amp:
https://dynacotubeaudio.forumotion.com/t996-2011-custom-st-120-amp


images

Do you happen to know what the AC line voltage was when you were having problems?

Good question, I do not know the AC volts when the first one blew. The line volts run between 116-123 from the wall. I have a Furman Elite-15 Line Conditioner and it is stable at 118 AC V. I pull the amp off the conditioner and tried all the combinations of rectifiers with AC line power from the wall at 116, no dice. The NOS works and the new production smokes the fuse. Very strange...

At Bob's request I measured the volts off the rectifier with the Mullard running and they are on the high-side of normal.

Bob just recently posted up about issues with line conditioners ...

https://dynacotubeaudio.forumotion.com/t1745-short-rectifier-life-when-vta-amp-is-plugged-into-a-line-conditioner?highlight=conditioner

With or without the Furman E-15 conditioner the NOS GZ34's work and the new production GZ34's blow a fuse. Strange problem....

I recommend installing the diode mod:
https://dynacotubeaudio.forumotion.com/t1006-tube-rectifier-diode-mod

and running without the line conditioner.

"and running without the line conditioner."

The question is why did the amp run on both new and NOS GZ34’s for two years and now only the NOS? Is there a bigger problem in the Amp that I have not found?

IMHO the GZ34 rectifier in the ST-120 is running beyond normal operating parameters especially at start up charging the huge banks of capacitors.  Looking at the datasheet the design does not have enough choke inductance to satisfy 0 ohm plate limiting resistors requirement by a factor of 20.  Two 1.0H chokes in parallel yield 0.5H which is not much considering the large 266uF bank of caps right after the choke pair and even more caps downstream.  The GZ34 datasheet(s) recommends 10H choke if not using plate limiting resistors in our case.  So the B+ inrush current at turn on (relay board kicks in) will be very large and probably well beyond 750mA peak maximum for a brief second.

If you make it past the initial turn on the ST-120 is pulling about 245mA of B+ current just idling and more even playing music.  This at the maximum recommended 250mA steady state current.  So the GZ34 tube rectifiers are worked very hard in this amp at all times.  Only the very strong shall survive.

Other than nostalgia is there any functional reason to run a tube rectifier versus the solid state sub?   Is there sound improvement?  Solid state rectifiers sometimes create switching noise if the wrong diodes are used.  But I hear nothing of the sort with the WZ68 sub.  Plus the tube rectifier chews up 10W of heater power.   The job of a rectifier is to keep the B+ caps charged and solid state can do this more efficiently and at safer operational limits.  What am I missing?
Sovtek 5AR4/GZ34 Datasheet

The GZ34's are definitely run at the edge in the ST120. I have burned through many Sovtek and Ruby rectifiers trying to find one that hangs in there. I have about 200 hours on the current Sovtek.

I've had problems with JJ and Sovtec 5AR4 tubes to the point where I gave up on those brands.

I've had better luck with Ruby 5AR4 tubes.

What Bob said!

And as others have pointed out, the GZ34s are operating awfully close to their limits. At the very least use the diode mod Bob recommends - I much prefer the fast/low noise UF4007s vs. the 1N4007s, but in either case you take a lot of stress off the rectifier with that mod. Me? I'd use SS diodes... today's fast/soft recovery diodes are really good and far more reliable than tube rectifiers.

Are you recommending the "Rectifier diode mod" even with the 15 second start up time delay upgrade used in the ST-120?

Maintarget wrote:Are you recommending the "Rectifier diode mod" even with the 15 second start up time delay upgrade used in the ST-120?

Yes. It's VERY cheap insurance!

Jim Thanks,
Does this mod prevent back flow current to the tube? or what exactly is happening?
Thanks for your Help!

Do check out the thread "Tube Rectifier Diode Mod" - but it essentially prevents the tube from having to withstand large reverse voltage situations which reduces the chance of breakdown. So you were real close in your speculation!

I add the diode mod to every ST-120 I build whether it has the TDR (time delay relay) board or not. As Jim McShane wrote, cheap insurance.

On another note, if you have a AC amp meter or a Kill-A-Watt that will measure AC current, with the NOS GZ34 tube and the amp biased at 50ma per tube, the amp will pull about 2 amps AC, if it is pulling more than that, you might have an issue with your amp, possibly a bad section in your quad capacitor.

Sal

Adding an in rush current limiter couldn't hurt either even if the fuse doesn't blow at start up. If memory serves, the ST-120 is fused at 5 amps?  If so the CL-60 (or CL-50) is the one rated at 5 amps, but there are many other values available.  Goes between the AC Line in and the PT's 120 volt (or 240) input.  I use them on everything I build as they add another layer of protection easily and inexpensively without any loss of performance.

Maybe 2 CL-120's one for each leg on the High voltage AC side would ease the instantaneous initial current charging the capacitors.

Sal

I have only ever used one current limiter per build, but now that I think about it the call to use two: one per each side of the tranny's input windings, could be better as it is AC.  Those CL-120 values should work with a balanced input, or 60vac on both the L and N.  

In the USA though the AC line voltage normally isn't balanced as there is usually + or - 120 vac on the Line side and N is usually held to zero.  There would be flow at start up that would pop a CL-120 on the "L" side if my thinking is correct, and it could be all wrong! 

Just thinking out loud here, not disputing.  I appreciate the thoughts, suggestions and any addtional feedback.

I too have lost JJs and Sovteks and Grove Tube 5AR4s, which is why I spend the money on Mullards.  Mullards seem to plow through the ripples that fry other inferior rectifiers.  I usually look for a GE or RCA re-badged Mullard as they're usually less expensive.

I have learned never to power off and power back on the amp without giving it time to bleed off.  A Mullard can take that kind of abuse to a point, but the JJs cannot.  I live in an area that has less than reliable electrical service, so anything to improve and prevent rectifier failures I'm open to.

I've seen where people put a second thermister on the B+ line ... and that a CL-90 should be able to handle 5 amps nicely, and also gives a bit more warmup time. I don't imagine average draw on these amps comes close to that anyway. Another bonus using one of these is that it will never go fully open. Line voltage should drop a couple notches at least. My power averages 122+, so that's not a bad thing. Main thing is to make sure you have plenty of air space around them, as these suckers get HOT!

I also invested in a "non-latching" GFCI adapter that sits between the wall and amp. That requires a manual reset if there's a power blip and gives you time to let the caps bleed out properly ...



Cheap insurance also at $18 ...

Whoa!! A CL-90 is rated for 2 amps maximum steady state current, so it won't live very long. I would not recommend the CL-90. As well, the listed maximum current flow is rated at 25°C, so derating is required if the ambient temp is above 25°C (77 degrees F). Yes, they get hot - but too much heat shortens their life and lowers their current ratings.

A CL-150 would be a better choice by far.

CL-90's have been used at higher amp ratings for years ... the important part is "maximum steady state current".

I've used a CL-90 in a ham amplifier running sideband and rated at 10 amps for years. I mean, what's the worse that could happen? I can get away with it because it's not a steady load.



Here's one mounted in an amp rated and fused for 4A. Several years after mounting.



Easiest way to tell if a thermister can't handle the load is when it cracks and fails open. That happens, go a step up. Would that all such problems were so easy to solve.

Can't stress enough the importance of good spacing and air flow though. Mostly to protect the rest of the components and circuitry.

Understood. But regardless of the anecdotal evidence you provided my prior post still stands. BTW, these devices are NOT standard thermistors, they have a much different time constant.

This is all very good information, for a person that has very limited knowledge of the design. I still am not sure why the new GZ34 tubes worked for 2 years and now only the NOS works. I did not realize the Sovtek was operating at it's design limit. Would changing the chokes or other parts be a better solution? Is the problem current back flow or total current demand? I can make the mods but need a good map and part numbers.