Same tube keeps going bad

[Alex Kitic]

Hello McEngVic,

 I took a quick look at the schematics and I understand why the coupling caps should be 630V - at startup the ss diodes are much quicker than the tube heaters, and voltages probably reach 500V or more, and the “upper” caps (like C25) probably suffer the blow.

 The first thing that comes to mind is that the bias circuit is a voltage divider. If the values are incorrect (like replacing grid bleeder resistor 270k for a different value) the correct voltage cannot be obtained with too high or too low current. Thus make sure that all values are original - 270k should stay 270k unless you make other changes.
Also, these 2 resistors need to be the same value because they are part of a shared circuit for the 2 tubes (which is definitely a design flaw worthy of modification).

Another question would be about the variable resistors (pots) used to set the bias voltage: I expect those have been replaced? A bad scraper can do awful things in that position, and the issue you are describing looks like a powered tube loosing g1 reference!

Last but repeated concern, not only the cathodes are connected to ground through a shared small resistor, but the g1 bias circuit is parallel to both tubes and thus shared. The le-diodes tell you if the voltage is high or low (current low or high) for the pair - not just one tube. What is perfect for the pair is probably twice the admissible current for one tube with the other tube drawing little or no current? Therefore the problem is not isolated to V5 but includes what is going on with the pair tube.

 The problem must be caused by erratic g1 condition on tube V5 - either lost contact or erratic bias voltage.

You can also contact me directly, I’ll try to help. But we should post the solution here for others to find

[MechEngVic]

8 minutes ago, Alex Kitic said:

Hello McEngVic,

 I took a quick look at the schematics and I understand why the coupling caps should be 630V - at startup the ss diodes are much quicker than the tube heaters, and voltages probably reach 500V or more, and the “upper” caps (like C25) probably suffer the blow.

 The first thing that comes to mind is that the bias circuit is a voltage divider. If the values are incorrect (like replacing grid bleeder resistor 270k for a different value) the correct voltage cannot be obtained with too high or too low current. Thus make sure that all values are original - 270k should stay 270k unless you make other changes.
Also, these 2 resistors need to be the same value because they are part of a shared circuit for the 2 tubes (which is definitely a design flaw worthy of modification).

Another question would be about the variable resistors (pots) used to set the bias voltage: I expect those have been replaced? A bad scraper can do awful things in that position, and the issue you are describing looks like a powered tube loosing g1 reference!

Last but repeated concern, not only the cathodes are connected to ground through a shared small resistor, but the g1 bias circuit is parallel to both tubes and thus shared. The le-diodes tell you if the voltage is high or low (current low or high) for the pair - not just one tube. What is perfect for the pair is probably twice the admissible current for one tube with the other tube drawing little or no current? Therefore the problem is not isolated to V5 but includes what is going on with the pair tube.

 The problem must be caused by erratic g1 condition on tube V5 - either lost contact or erratic bias voltage.

You can also contact me directly, I’ll try to help. But we should post the solution here for others to find

This is fantastic info, I'll go over it with a fine tooth comb asap. 

[Alex Kitic]

On 4/6/2020 at 9:31 AM, MechEngVic said:

This is fantastic info, I'll go over it with a fine tooth comb asap. 

Hello MechEngVic -- any news? Did you find the time to test some more, in particular those bias setting variable resistors?

[MechEngVic]

4 hours ago, Alex Kitic said:

Hello MechEngVic -- any news? Did you find the time to test some more, in particular those bias setting variable resistors?

No, I've been waiting for some parts to come in before I opened the amp up. Delivery is slower now a days.

 

The variable resistors were replaced and they seem to be working good, no rough spots while turning or jumps in resistance.

 

What you mention about one tube drawing twice the current and the other drawing little or none, do you mean during warmup? And does that condition go back to normal once the amp is warm? I imagine if this condition existed while I was using the amp it would be noticeable, no? Poor SQ or arcing during use?

 

Is been a few weeks now and the amp has been turning on, warning up, and running all day with no problems on a daily basis. V5 has only ever arced during warmup. So ever since the last arcing event, I have been making sure I have a music signal going to the amp before I turn it on (preamp already automatically turns on several seconds before amp), and I don't know if it's helping with arcing, but it hasn't happened again in weeks. Do you think having a signal running during amp warmup can have an effect on this issue? 

[MechEngVic]

On 4/6/2020 at 12:21 AM, Alex Kitic said:

Last but repeated concern, not only the cathodes are connected to ground through a shared small resistor, but the g1 bias circuit is parallel to both tubes and thus shared. The le-diodes tell you if the voltage is high or low (current low or high) for the pair - not just one tube. What is perfect for the pair is probably twice the admissible current for one tube with the other tube drawing little or no current? Therefore the problem is not isolated to V5 but includes what is going on with the pair tube.

 The problem must be caused by erratic g1 condition on tube V5 - either lost contact or erratic bias voltage.

So should I measure current between the plates a & k for each tube in the pair?

 

[MechEngVic]

33 minutes ago, MechEngVic said:

So should I measure current between the plates a & k for each tube in the pair?

Actually, I guess you measure voltage between the plates.

 

On 4/6/2020 at 12:21 AM, Alex Kitic said:

What is perfect for the pair is probably twice the admissible current for one tube with the other tube drawing little or no current? Therefore the problem is not isolated to V5 but includes what is going on with the pair tube.

How would I measure this?

[Alex Kitic]

14 hours ago, MechEngVic said:

No, I've been waiting for some parts to come in before I opened the amp up. Delivery is slower now a days.

 

The variable resistors were replaced and they seem to be working good, no rough spots while turning or jumps in resistance.

 

What you mention about one tube drawing twice the current and the other drawing little or none, do you mean during warmup? And does that condition go back to normal once the amp is warm? I imagine if this condition existed while I was using the amp it would be noticeable, no? Poor SQ or arcing during use?

 

Is been a few weeks now and the amp has been turning on, warning up, and running all day with no problems on a daily basis. V5 has only ever arced during warmup. So ever since the last arcing event, I have been making sure I have a music signal going to the amp before I turn it on (preamp already automatically turns on several seconds before amp), and I don't know if it's helping with arcing, but it hasn't happened again in weeks. Do you think having a signal running during amp warmup can have an effect on this issue? 

Whether there is a signal (music) or not during start-up should be totally irrelevant for the proper start-up of the amplifier.

 

This is obviously an erratic issue that does not happen every time and is not repeatable. It’s not a sequence that can be reproduced.

 

The issue must be caused by a component that is malfunctionig occasionally, and is related to g1 on V5 - either the connection socket, or the grid negative voltage (bias) being wrong for some reason.

This is made more complicated by the design where bias setting components are common to the pair of tubes, and the cause may be with the other tube in the pair - a lost anode connection means that all the current is drawn, at least for a short moment, by V5.

 

The condition is not necessarily detectable by looking at the bias leds, nor listening to the music signal, and probably happens all the time but not necessarily long enough as to cause the arcing as an evident event. Probably corrects itself once the tubes heat-up and expand.

 

I just cannot shake off the possibility of a bad socket. I have seen socket issues happening with new sockets - never used and new production - with tubes with seemingly perfect pins...

[Alex Kitic]

9 hours ago, MechEngVic said:

Actually, I guess you measure voltage between the plates.

 

How would I measure this?

 

I guess you need first to measure the DCR of the 2 sections of the primary for each tube, write it down, and than measure the voltage simultaneously with 2 multimeters between B+ and each anode. That would show inbalance as well as give an exact measure of the anode current.

 

What I would also measure would be the g1 voltage of both tubes, with 2 meters, simultaneously.

Edited April 11, 2020 by Alex Kitic

[Alex Kitic]

Thank you, Jeff, for putting this in a simpler way.

I am always striving to use as little words as I can and sometimes my interlocutors don’t get it right.

 

While this is obviously an intermittent issue that occurs on startup when tubes are cold, by checking all these details we can narrow it to the most likely cause. The main hindrance here, and if I may, the point where this design was either not taught through properly is the bias common to both output tubes - you know if the pair is drawing the correct current but not if this current is correctly distributed between the two tubes.

 

[MechEngVic]

6 hours ago, Jeffrey D. Medwin said:

 Hi, 

 

Vic tells me, he has not recently experienced the problem, so he is not rushing to do the measurements you suggested.  He is just happy, listening.

 

Alex, I wonder if something like a CL-90 thermistor, or maybe a CL-80, placed in series in one of the power transformer's primary leads, might be longer-term helpful with this mixture of tube and solid state parts.  I don't understand solid state stuff, but you do !!  Just a thought.  Care to comment??  Pros and cons.

 

Jeff

Yup. Still no second event with V5, the same tube that arced is warming up and behaving like it should. It's been several weeks. The amp is on every day from morning till night, playing music at sometimes loud volumes as well as you tube, shows, and movies. 

[Alex Kitic]

14 hours ago, Jeffrey D. Medwin said:

 Hi, 

 

Vic tells me, he has not recently experienced the problem, so he is not rushing to do the measurements you suggested.  He is just happy, listening.

 

Alex, I wonder if something like a CL-90 thermistor, or maybe a CL-80, placed in series in one of the power transformer's primary leads, might be longer-term helpful with this mixture of tube and solid state parts.  I don't understand solid state stuff, but you do !!  Just a thought.  Care to comment??  Pros and cons.

 

Jeff

While adding an NTC helps with current inrush for whole amplifiers, power supplies, or just heater supplies - the thermistors you mentioned are for a minimum current flow of 500mA which is way above the normal current draw of one side of this PP pair. It would possibly not make any difference (not really slowing startup) but would represent an additional resistance between anode and primary of the OPT with adverse effect (I don’t think it would heat enough to become the .7 ohm resistance mentioned in the datasheet - CL90 is 120 ohm at 25 degrees C and maximum current is 3A - good for 2A3 heaters but not anode current of a tube that will not draw more than 100mA).

 

In addition to that, if there’s something wrong with the connection of g1 in the socket, sooner or later this will strike and the g1 will remain without ground reference, rising in potential and causing the problem. If that’s the cause, which is my first impression. That’s why I suggested measuring the g1 voltage at startup for both tubes in the pair with 2 multimeters (obviously tied to g1 and ground so there is no touching and full attention to the readings).

[MechEngVic]

23 hours ago, Alex Kitic said:

While adding an NTC helps with current inrush for whole amplifiers, power supplies, or just heater supplies - the thermistors you mentioned are for a minimum current flow of 500mA which is way above the normal current draw of one side of this PP pair. It would possibly not make any difference (not really slowing startup) but would represent an additional resistance between anode and primary of the OPT with adverse effect (I don’t think it would heat enough to become the .7 ohm resistance mentioned in the datasheet - CL90 is 120 ohm at 25 degrees C and maximum current is 3A - good for 2A3 heaters but not anode current of a tube that will not draw more than 100mA).

 

In addition to that, if there’s something wrong with the connection of g1 in the socket, sooner or later this will strike and the g1 will remain without ground reference, rising in potential and causing the problem. If that’s the cause, which is my first impression. That’s why I suggested measuring the g1 voltage at startup for both tubes in the pair with 2 multimeters (obviously tied to g1 and ground so there is no touching and full attention to the readings).

I read a guy on another forum installed a pair of CL-80's on a Dynaco ST-70 power transformer's primary. I believe he installed one on each hot leg of the primary, but I'm not sure. Is this a viable application of these NTC thermistors? Jeffery mentioned using them but bypassing them once warmed up. Not that I am considering this as a fix, nor am I really interested in changing the basic circuitry of this amp, but I have read that a solid state power supply can beat up the tube side of the amp.

[Alex Kitic]

1 hour ago, MechEngVic said:

I read a guy on another forum installed a pair of CL-80's on a Dynaco ST-70 power transformer's primary. I believe he installed one on each hot leg of the primary, but I'm not sure. Is this a viable application of these NTC thermistors? Jeffery mentioned using them but bypassing them once warmed up. Not that I am considering this as a fix, nor am I really interested in changing the basic circuitry of this amp, but I have read that a solid state power supply can beat up the tube side of the amp.

Just as I wrote above, installing an NTC similar to the CL80 or CL90 will not solve because:

1) these are rated for much higher current (so will not limit anything in particular because all the current a tube can draw will pass thru anyway),

2) not being adequate these will not heat enough to significantly change their resistance (becoming a very small resistance that literally disappears in the circuit).

 

The fact that he bypasses the CL80 with a switch confirms 2), and also confirms 1) which is obvious from the datasheet anyway. Instead of the CL80 you could install a 120 ohm resistor (resistance of CL80 or CL90 at 25 degrees C) and a switch to bypass it once the amp is warmed-up.

Just like the  L80, the 120 ohm resistor will not limit the current relevantly, as the anode voltage will just be slightly lower as per Ohms Law. The NTC would need to be rated for maximum 200mA, constant 100mA and have a considerably higher resistance at 25 degrees C x i.e. at idle.

 

If this problem is caused by cold tubes receiving high voltage too soon (which I believe is not the root cause), a viable solution would be to install a switch that would allow you to power-up the amp without the B+ hitting the tubes, and to switch on the B+ once the tubes have been properly heated (let’s say after 1 minute). The switch would need to be able to withstand the B+ voltage, let’s say good for at least 500V DC.

 

Edit

The switch should be placed between the CT of the HV secondary and ground. I believe the schematics calls it red/yellow, but you should check which one is really the CT of the HV secondary.

The switch must be well insulated and able to withstand at least 500V DC (better go for higher and keep in mind you need the DC not the AC rating usually found on switches).

Edited April 22, 2020 by Alex Kitic