So it is quite clear that on GZ34 datasheets, the maximum listed AC transformer voltage plate to plate is 550V. For a center tapped full wave rectifier, this closely matches the listed peak inverse voltage (550 x 1.41 x 2) = 1550V.
The question is, is the listed plate voltage just a function of hitting the max PIV rating?
I’m interested using the GZ34 in a hybrid bridge rectifier where the PIV is equal to the peak transformer winding (not 2x like a full wave). Therefore, I could theoretically use a plate voltage much higher than 550v and still meet the PIV requirement. Furthermore I could go even further on the plate voltage, and use series diodes with balancing resistors to reduce the PIV. I could also put plates in parallel and use two separate rectifiers so the voltage differential between parallel plates is zero.
This raises the question, how high can you go plate to cathode voltage when the tube is cold?
When the tube is cold, the plate to cathode voltage will be equal to the transformer winding since the tube is not in conduction. This must be the worst case for forward voltage rating. It seems reasonable that the PIV is applicable since this is determined by electrode spacing (I assume).
I am of course assuming that the tube is operating well within spec once hot and in conducting and has appropriate limiting resistance and small input capacitance to limit current.
Curious if anyone has feedback on this.
The question is, is the listed plate voltage just a function of hitting the max PIV rating?
I’m interested using the GZ34 in a hybrid bridge rectifier where the PIV is equal to the peak transformer winding (not 2x like a full wave). Therefore, I could theoretically use a plate voltage much higher than 550v and still meet the PIV requirement. Furthermore I could go even further on the plate voltage, and use series diodes with balancing resistors to reduce the PIV. I could also put plates in parallel and use two separate rectifiers so the voltage differential between parallel plates is zero.
This raises the question, how high can you go plate to cathode voltage when the tube is cold?
When the tube is cold, the plate to cathode voltage will be equal to the transformer winding since the tube is not in conduction. This must be the worst case for forward voltage rating. It seems reasonable that the PIV is applicable since this is determined by electrode spacing (I assume).
I am of course assuming that the tube is operating well within spec once hot and in conducting and has appropriate limiting resistance and small input capacitance to limit current.
Curious if anyone has feedback on this.
"The question is, is the listed plate voltage just a function of hitting the max PIV rating?"
Imho, yes.
"Furthermore I could go even further on the plate voltage, and use series diodes with balancing resistors to reduce the PIV."
Do you mean series GZ34 diodes, or the more familiar use of redundant ss diodes?
"I could also put plates in parallel and use two separate rectifiers so the voltage differential between parallel plates is zero."
I'm not sure what this comment relates to - can you elaborate? It appears to be about having two diode structures in the one valve, and a concern that anode structures (and any associated wiring) may not have enough spacing to support say two x PIV.
In a vacuum, with everything cold, there will be residual leakage currents between electrodes/rods/wires when a voltage differential is applied - as the vacuum is not perfect and stray molecules will carry some charge between sides, or there is some sputtered metal on glass and/or mica surfaces that can cause some current flow. Of course those residual currents are benign in a typical diode, but aging or gassy diodes may have measurable levels.
Imho, yes.
"Furthermore I could go even further on the plate voltage, and use series diodes with balancing resistors to reduce the PIV."
Do you mean series GZ34 diodes, or the more familiar use of redundant ss diodes?
"I could also put plates in parallel and use two separate rectifiers so the voltage differential between parallel plates is zero."
I'm not sure what this comment relates to - can you elaborate? It appears to be about having two diode structures in the one valve, and a concern that anode structures (and any associated wiring) may not have enough spacing to support say two x PIV.
In a vacuum, with everything cold, there will be residual leakage currents between electrodes/rods/wires when a voltage differential is applied - as the vacuum is not perfect and stray molecules will carry some charge between sides, or there is some sputtered metal on glass and/or mica surfaces that can cause some current flow. Of course those residual currents are benign in a typical diode, but aging or gassy diodes may have measurable levels.
Thanks for the input trobbins.
Yes. To elaborate, my goal is to build a hybrid SS diode + GZ34 rectifier that can rectify a 1000VRMS transformer winding. I was going to use series didoes with balancing resistors on all legs of the bridge including in series with the GZ34 plates so the PIV so it is well within spec. The GZ34 cathode connect to the B+ positive output. Theoretically the GZ34 PIV should be within spec with a 1000VRMS bridge, but I know PIV is an issue with modern GZ34s, so why not make it lower with additional didoes and balancing resistors.
I would like to use GZ34 for their soft start characteristics and slow ramp of the output voltage.
This raised the questions above.
When hot, everything is well within spec, but when cold and power is flipped on, the plate to cathode needs to withstand the 1000V RMS.
Even though the PIV is similar for the 550V plate to plate full wave datasheet example, each plate only has to standoff 550V AC when cold in that example.
Good points on the leakage paths when cold!
Yes. To elaborate, my goal is to build a hybrid SS diode + GZ34 rectifier that can rectify a 1000VRMS transformer winding. I was going to use series didoes with balancing resistors on all legs of the bridge including in series with the GZ34 plates so the PIV so it is well within spec. The GZ34 cathode connect to the B+ positive output. Theoretically the GZ34 PIV should be within spec with a 1000VRMS bridge, but I know PIV is an issue with modern GZ34s, so why not make it lower with additional didoes and balancing resistors.
I would like to use GZ34 for their soft start characteristics and slow ramp of the output voltage.
This raised the questions above.
When hot, everything is well within spec, but when cold and power is flipped on, the plate to cathode needs to withstand the 1000V RMS.
Even though the PIV is similar for the 550V plate to plate full wave datasheet example, each plate only has to standoff 550V AC when cold in that example.
Correct, the datasheet lists 550V plate to plate as a means to show maximum PIV, but is unclear what the plate to plate voltage limitation is. I guess I'm just complaining on the DS vagueness 😉 . I was also thinking of using two separate GZ34 didoes in the bridge output legs with plates in parallel for more robustness with the inrush current charging the downstream capacitance (which raised that question on what they could actually stand).
Good points on the leakage paths when cold!
Remember that the PIV rating on vacuum tube rectifiers is plate to cathode, not plate to plate. So the smoothing filter will also play a role.
Just something to keep in mind.
Just something to keep in mind.
The DS is not vague imho - 1550V is the rating you are concerned about for each of the two diode structures in the one valve unit - the 550V relates to an application, which is subtley different.
As with a full-wave application, adding an ss diode in series with the valve diode plate is an excellent way to mitigate risk of valve diode arcing, and allows the combined 'ss plus valve' diode to be used at its 1550V PIV rating. So your bridge rectifier proposal is fine with just a single GZ34, although imho you should mitigate risk with the series ss diode 'yellow sheet mod'. Trying to 'balance' voltages across diodes is not called for, and imho is riskier, in this situation, and certainly far more complicated if trying to achieve that with multiple GZ34's.
Keep in mind that rectifying 1kVrms needs careful attention to the type of ss diode(s) proposed, and also note that the 'yellow sheet' mod may be fine for circa 300Vrms applications, but is quite a risk for 500Vac applications as adjacent socket terminals are being asked to withstand a much higher ac voltage than originally intended (which was why pins 3, 5, 7 were left unconnected or even removed.
As with a full-wave application, adding an ss diode in series with the valve diode plate is an excellent way to mitigate risk of valve diode arcing, and allows the combined 'ss plus valve' diode to be used at its 1550V PIV rating. So your bridge rectifier proposal is fine with just a single GZ34, although imho you should mitigate risk with the series ss diode 'yellow sheet mod'. Trying to 'balance' voltages across diodes is not called for, and imho is riskier, in this situation, and certainly far more complicated if trying to achieve that with multiple GZ34's.
Keep in mind that rectifying 1kVrms needs careful attention to the type of ss diode(s) proposed, and also note that the 'yellow sheet' mod may be fine for circa 300Vrms applications, but is quite a risk for 500Vac applications as adjacent socket terminals are being asked to withstand a much higher ac voltage than originally intended (which was why pins 3, 5, 7 were left unconnected or even removed.
You might consider using damper diodes for your gig. Cheap, plentiful, sturdy, slower heating. Real GZ34s are pretty much all gone.
All good fortune,
Chris
All good fortune,
Chris
It really doesn't matter if you are using it as a full wave rectifier. There can be significant potential between plates because both are cold (i.e. not heated) from a thermionic perspective. But the hold up on the cathode side will push the PIV (plate to cathode voltage) too high. This happen when the cathode is held up by the filter and the plate voltage is at the peak of it's negavive swing.
The Amperex 1954 data sheet shows that with a 500-0-500 secondary and a 100mA load, the DC output voltage is ≈630V at the filter. This means that the PIV when the plate swings negative is 630V+1.414*500 ≈ 1340v. That's the reason for the PIV rating of 1500v. The peak plate to plate delta here is actually 2*1.414*500 ≈ 1400v. The plates can likely stand significantly more potential difference but NOT the cathode to plate circuit.
The GZ34 is not really a high voltage rectifier unit. That's why mercury vapor rectifiers were used in larger equipment.
Thanks for the responses everyone. Yes, I'm not too concerned on the plate to plate voltage, that was more of a thought exercise while I was considering parallel plates and two GZ34s. Yes, the full wave use cases in the datasheets (such as 500-0-500) makes it clear that 1400V plate to plate is fine.
My major concern, was the lack of info on the forward rated positive plate to cathode voltage (not to be confused with the PIV) on startup when the tube is not conducting due to cold filament. Regardless of of the number of diodes to handle the PIV, the positive plate to cathode voltage will be a peak of 1000*1.41 = 1410V when the cathode is cold. This will gradually reduce as the tube begins conducting down to the typical voltage drops listed in the datasheet at a given current. Again for any of the full wave cases listed in the datasheet, the positive plate to cathode when cold is just one half of the PT winding for example 500 x 1.41 = 705V.
This is the big discrepancy between the datasheet full wave applications and a bridge rectifier application (even though the PIV of both are similar).
What is allowable for reasonable reliability with regards to positive plate to cathode voltage on startup is still unclear.
My major concern, was the lack of info on the forward rated positive plate to cathode voltage (not to be confused with the PIV) on startup when the tube is not conducting due to cold filament. Regardless of of the number of diodes to handle the PIV, the positive plate to cathode voltage will be a peak of 1000*1.41 = 1410V when the cathode is cold. This will gradually reduce as the tube begins conducting down to the typical voltage drops listed in the datasheet at a given current. Again for any of the full wave cases listed in the datasheet, the positive plate to cathode when cold is just one half of the PT winding for example 500 x 1.41 = 705V.
This is the big discrepancy between the datasheet full wave applications and a bridge rectifier application (even though the PIV of both are similar).
What is allowable for reasonable reliability with regards to positive plate to cathode voltage on startup is still unclear.
mbeards, I'd suggest that the term 'forward rated' is not valid, but rather 'voltage withstand' may be more appropriate, but also view that as an application dependant concern that is implicity covered by the PIV.
There certainly were a few pre-1950 rectifier tubes that had conditions related to 'start-up' (over and beyond the transient peak anode current rating), and the conditions typically related to using a time delay before applying load, or having a load-constrained time profile.
With a cold cathode there is no forward or reverse, just low level leakage that may be polarity independent. As the cathode heats up, it starts to develop a cloud of electrons around it, and electrons start to migrate to polarity dependent anodes. I don't foresee that the cathode-to-anode conduction during this warm-up time is somehow limited or influenced or damaged by having a plate voltage up to 1550V, but rather is limited by available electrons. If you were keen, then I can suggest the on-line 1962 RCA Electron Tube Design book may provide further insight related to your concern. A close topic may be HV diode design and operation and its quirks.
There certainly were a few pre-1950 rectifier tubes that had conditions related to 'start-up' (over and beyond the transient peak anode current rating), and the conditions typically related to using a time delay before applying load, or having a load-constrained time profile.
With a cold cathode there is no forward or reverse, just low level leakage that may be polarity independent. As the cathode heats up, it starts to develop a cloud of electrons around it, and electrons start to migrate to polarity dependent anodes. I don't foresee that the cathode-to-anode conduction during this warm-up time is somehow limited or influenced or damaged by having a plate voltage up to 1550V, but rather is limited by available electrons. If you were keen, then I can suggest the on-line 1962 RCA Electron Tube Design book may provide further insight related to your concern. A close topic may be HV diode design and operation and its quirks.
Trobbins, thanks for the insight.
I suppose you’re right that PIV would cover either voltage polarity between the plate and cathode. If there is no electron flow from the space charge, then element spacing, residual leakage paths, and any gas would determine current flow or arcing.
You made a comment on selecting diodes carefully and that balancing may be risky.
Care to elaborate?
I was looking at using 2KV 1A rated diodes and using 2 per bridge rectifier leg except for the GZ34s which should only require one in series with each plate.
My understanding is that tube rectifier reverse leakage current can be much lower than SS diodes, and therefore would take majority of the PIV. Wouldn’t swamping the SS diode and tube rectifier leakage with a large value resistor be advisable?
Appreciate the input!
I suppose you’re right that PIV would cover either voltage polarity between the plate and cathode. If there is no electron flow from the space charge, then element spacing, residual leakage paths, and any gas would determine current flow or arcing.
You made a comment on selecting diodes carefully and that balancing may be risky.
Care to elaborate?
I was looking at using 2KV 1A rated diodes and using 2 per bridge rectifier leg except for the GZ34s which should only require one in series with each plate.
My understanding is that tube rectifier reverse leakage current can be much lower than SS diodes, and therefore would take majority of the PIV. Wouldn’t swamping the SS diode and tube rectifier leakage with a large value resistor be advisable?
Appreciate the input!
mbeards, I tried to elaborate on that topic in chapters 2 and 7 of linked discussion doc.
https://dalmura.com.au/static/Power supply issues for tube amps.pdf
https://dalmura.com.au/static/Power supply issues for tube amps.pdf
trobbins,
Excellent reading! I have come across your paper before and I appreciate you linking in this thread. Great compendium of rectifier knowledge!
I was interested in the part in 7.3 where you discuss checking leakage using an insulation tester. I have had many new production GZ34s arc in very “kind” full wave rectifier circuits with plenty of resistance and low capacitance at 11uf.
Have you had any luck correlating measured plate to cathode leakage to an arc failure in circuit?
When you discuss leakage changes after running the heaters, I assume you are measuring the tubes without the heaters energized, but after a “burn in” period?
I have had some GZ34s arc and then “clear up” after a second or so. And other that would continue to arc until the fuse blows or I shut the power off. Screening GZ34s, especially for the 1000v application above could be very useful.
Excellent reading! I have come across your paper before and I appreciate you linking in this thread. Great compendium of rectifier knowledge!
I was interested in the part in 7.3 where you discuss checking leakage using an insulation tester. I have had many new production GZ34s arc in very “kind” full wave rectifier circuits with plenty of resistance and low capacitance at 11uf.
Have you had any luck correlating measured plate to cathode leakage to an arc failure in circuit?
When you discuss leakage changes after running the heaters, I assume you are measuring the tubes without the heaters energized, but after a “burn in” period?
I have had some GZ34s arc and then “clear up” after a second or so. And other that would continue to arc until the fuse blows or I shut the power off. Screening GZ34s, especially for the 1000v application above could be very useful.
The batch of 5U4 I tested were of unknown history, but we're likely 'pulled' from equipment by a tech. Each of those valves was IR tested with no pre-conditioning, and then retested with their heater powered and perhaps a minute of warm up time, but no plate current (apart from the resulting IR test current), so yes the heater was powered at the time of IR testing. The opportunistic aim was to get a large sample size as an indicator of what age/use can lead to.
I haven't had the opportunity to do 'arc' testing on that batch, as a way of confirming conditions that would cause an arc on a tube with known IR.
I haven't had the opportunity to do 'arc' testing on that batch, as a way of confirming conditions that would cause an arc on a tube with known IR.