SS rectifier?

Discussion in 'Tube Audio' started by cyto, Jun 20, 2018.

  1. cyto

    cyto Well-Known Member

    Messages:
    546
    Location:
    Birmingham, AL
    I am in the process of making a separate power supply for my Cornet 2 phonostage (schematic attached), the reasons are that I want to put my Cinemags in there and need room, I don't want the Cinemags close to the tranny, every audio device that I have ever built sounded better with an external power supply, and mainly just because I want to :)

    I got to thinking about changing over from tube rectification to SS. It simplifies the build, reduces the current draw on the tranny and reduces heat. I know about the voltage drop differences and I should be able to drop that to what I need.

    I know there will not be a slow startup with a SS full wave bridge but honestly I am not sure about the cathode stripping issue that is often brought up. I went to electronics school in the 70s and I don't remember that ever being brought up and I also remember a lot of televisions with SS bridge rectifiers and tubes, they did not have slow start ups that I can recall (but that was the 70s and my memory of that decade is sketchy).

    What do you guys think?
     

    Attached Files:

     

    Please register to disable this ad.

  2. Retrovert

    Retrovert AK Subscriber Subscriber

    Messages:
    3,295
    Location:
    New York City, NY
    This topic is endlessly debated with no resolution.

    In summary, when B+ is applied before the cathode is hot, the space cloud has consequently not yet formed and emission occurs from islands of charge instead of a contiguous surface. The cathode will be stressed.

    Emission occurs from the hottest point (best emission point) and that increases the temperature at that point. (All that cathode material which ends up on the grid originates from a cause. This is how it is ablated.) If you read the papers on the subject the problem is clear, no matter what anyone claims. Transmitting tubes will be instantly destroyed because of the higher voltage. Signal tubes survive such abuse because cathodes are robust. But a tube which lasts a few thousand hours may last longer if not abused.

    The solution is to delay B+ until the cathodes are all hot. You can do that with a MOSFET and an RC delay. Circuits have been posted.

    I suggest you research the issue and take your own counsel on this instead of believing the dogma which endlessly circulates until it becomes true.
     
    darkblue94, bberkom, tcdriver and 2 others like this.
  3. kward

    kward AK Subscriber Subscriber

    Messages:
    3,224
    Location:
    Utah
    Concerning how to drop the extra voltage needed when converting from tube to silicon rectification, a simple method is to place a small value resistor after the rectifiers but in front of the first reservoir cap. This resistor causes voltage drop from the DC current and from the ripple current so it needs to be a little larger wattage type than if just used as a DC voltage dropper. I use Vishay wire wound types for this. I like this approach because it requires one single passive device (the resistor) and therefore minimum real estate, doesn't add a lot of series resistance to the circuit overall, and reduces ripple current.

    But you could also put series resistance in with the diodes, more or less as you have shown, or you could use a choke if you have the real estate. There are a few ways to skin that cat.
     
  4. peterh

    peterh AK Subscriber Subscriber

    Messages:
    1,792
    Location:
    Gothenburg,Sweden
    A grand lot of experience has failed to find any negative sideeffects of Si diodes and rapid application of B+
    before the cathodes are fully warm. In fact the slow and gradual buildup of plate currents might be benefical in
    contrast to the harsh puls that occurs if, lets say a relay, suddenly applies full B+ This will also avoid
    the possible output pulse caused by charging an output coupling cap, instead the output cap will slowly
    charged.

    Thus, if the caps can stand the peak and no grids <> cathode will have hight voltages ( easily fixed with a 1n4148 diode)
    then by all means go for Si diodes.

    The alternative is not a delay relay, the alternative is what you want to avoid: a tube rectifier.
     
    triode17 likes this.
  5. Retrovert

    Retrovert AK Subscriber Subscriber

    Messages:
    3,295
    Location:
    New York City, NY
    That's just internet dogma. Whose experience? Who looked for negative effects?

    If the space cloud has not yet formed where does the emission happen? Answer: from point sources on the cathode. This is why transmitters blow up cathodes when B+ is pulled before the cathode is hot. The voltage is a lot lower in most signal tubes so it takes longer for the tube to die, but it is still damaged.

    A similar phenomenon occurs when too much capacitance is added after a rectifier tube. The hottest spot carries the peak current and this gradually blows the cathode apart. The capacitance rating is really a maximum instantaneous current rating.

    Where do you think all of the cathode material on the grid originates? Answer: it is ablated from the cathode each time the tube is powered on. Because the cathode has lots of emission material it survives this for a considerable time, but it eventually fails and the tube dies.

    The early tube computers used slightly lower heater voltages, delayed heaters, and delayed B+ to extend tube lifespan. Once this was done the tube failure rate precipitously dropped.

    The RC delay on the MOSFET gradually ramps up the B+ to avoid jackhammering.

    Tube rectifiers are awful things and we are well rid of them. Unless one is building a guitar amplifier and wants sag, and even then that's easily accomplished through other means.
     
  6. gadget73

    gadget73 junk junkie Subscriber

    Messages:
    35,883
    Location:
    Southern NJ
    Well, we can all agree that this has been debated to death with no resolution :)

    Honestly for a phono stage I don't see where it matters a lot. We're not talking large voltage, nor are we talking anything that isn't straight class A so the impedance of the vacuum rectifier doesn't matter here.
     
    Retrovert, cyto and Lavane like this.

     

    Please register to disable this ad.

  7. cyto

    cyto Well-Known Member

    Messages:
    546
    Location:
    Birmingham, AL
    Honestly, I did not intend for this to be a debate about cathode stripping. If I do this and think I need a slow start up I will most likely install a switch. I really just wondered if switching to SS from tube rectification would be a good idea while I am rebuilding a power supply anyway.
     
  8. Retrovert

    Retrovert AK Subscriber Subscriber

    Messages:
    3,295
    Location:
    New York City, NY
    Adding a B+ switch can be a prescription for disaster.

    When a tube is operated without any B+ potential from cathode-to-plate the residual silicon in the cathode (used to activate the alkali metals when the tube was made) reacts with the barium (and strontium, but the cathode is mostly barium) to form orthosilicates which functions much like an insulator. This is called "cathode poisoning" or, in the vernacular, "sleeping sickness". The chemical reaction is irreversible. The tube is ruined.

    If you leave the switch in the off position and forget about it you can destroy every tube in short order.
     
  9. kward

    kward AK Subscriber Subscriber

    Messages:
    3,224
    Location:
    Utah
    You'll be fine in moving to SS rectification. Use the techniques discussed to reduce the rail voltage down to what you would have gotten with tube rectification.

    It may not be that much of a concern with a preamp since the currents are much lower than say a power amp, but one additional thing to be aware of when converting to SS rectification is diodes switching off abruptly may cause the power transformer to ring. You don't see this ringing when using vacuum rectification because the tube rectifier doesn't switch off as abruptly. One way to address this is to add an R/C snubber across the secondary. But I doubt you will even need to worry about that with a preamp since the circuit doesn't draw that much current.
     
    tcdriver likes this.
  10. Brice

    Brice AK Subscriber Subscriber

    Messages:
    1,448
    Location:
    Hoboken NJ
    Kward, what snubber R/C values do you use in that case?

    Thank you.
     
  11. Retrovert

    Retrovert AK Subscriber Subscriber

    Messages:
    3,295
    Location:
    New York City, NY
    tcdriver likes this.

     

    Please register to disable this ad.

  12. kward

    kward AK Subscriber Subscriber

    Messages:
    3,224
    Location:
    Utah
    To get the values exactly right I think is best determined by experimentation. It's a whole lot easier anyway for me vs. trying to model the whole system mathematically. The last power amp I did this on I ended up using a 0.1 uF cap which was fixed, and I eventually settled on a 220Ω resistor. Anything from 100Ω to 300Ω got it pretty close though, but 220Ω seemed to get it closest to a "critical" damping. I even posted the scope shots before vs. after here on the tube forum a while back. I'd have to look to see if I can find that thread.
     
  13. Brice

    Brice AK Subscriber Subscriber

    Messages:
    1,448
    Location:
    Hoboken NJ
    Thank you Kward,
    I would be interested in finding how you do it by experimentation to get the best results.
     
  14. triode17

    triode17 Well-Known Member

    Messages:
    973
    Cyto - I don't worry about cathode stripping and have been using SS rects,now for years without any incident. The whole stripping started when computers used tubes for on and off only. 1's and 0's. They found that if a tube is left in cutoff mode for a long time, that it couldn't be switched on anymore. but is takes time like days or weeks. In fact, rect. tubes are turned on with higher than full transformer voltage until they warm up but they don't seem to strip, ever. Explain that.
     
    peterh likes this.
  15. Brice

    Brice AK Subscriber Subscriber

    Messages:
    1,448
    Location:
    Hoboken NJ
    I agree, McIntosh, Acrosound and many others did it for decades and their amps were/are very reliable.
    It is a physical fact, but becomes academic for audiophile.
     
  16. kward

    kward AK Subscriber Subscriber

    Messages:
    3,224
    Location:
    Utah
    For a power amp, here's how I did it. Set the scope to read on the lowest voltage setting, mine is 2 mV. Hook a 8 ohm resistor load across the 8 ohm speaker connections (or 4 or 16 ohms, for that respective speaker connection). Hook the scope probe across the resistor load. Turn the amp on and let it idle. (no input signal).

    Turn the time window down (or is it up?) on your scope so you can see the 120 Hz ripple. Without the snubber and with "regular" diodes, you should be able to see the ringing very easily. Add the snubber. The ringing goes almost away totally. You can see it change in real time by changing the value of the R in the snubber. If you use soft recovery diodes it isn't as pronounced, but it's still there.

    I found the scope shots I mentioned previously. See post #1 on this thread: http://audiokarma.org/forums/index.php?threads/the-1625-amp-mk-ii.778418/
     

     

    Please register to disable this ad.

  17. Brice

    Brice AK Subscriber Subscriber

    Messages:
    1,448
    Location:
    Hoboken NJ
    Very Interesting. In fact you look at the amp's noise on the scope and reduce it in real time by changing the snubber network.

    I'll try that. I busted one channel on my scope doing noise measurement on the B+ despite that the VDC was lower than input rated, since then I am careful with that type of inspection :)

    Thank you for this.
     
  18. Retrovert

    Retrovert AK Subscriber Subscriber

    Messages:
    3,295
    Location:
    New York City, NY
    Having slogged through a number of application notes on snubbers, I can say that some uncertainty exists about the optimal selection of values for damping factor, called ζ.

    Cornell-Dubilier suggests 0.5< ζ < 0.9, but this is best for switched-mode supplies, while a ζ of 0.5 appears to be better for traditional linear power supplies, with ζ = 1/sqrt(2) = 0.707 being nearly optimal and nicely between the two endpoints. That larger value for ζ reduces the size of the snubber resistor to dissipate the ringing, because it's not as heavily damped.

    Barbie says, "Analog Design is Hard!".

    This application note from Cornell-Dubilier explains the calculations for the snubber capacitor and resistor:
    http://www.cde.com/resources/technical-papers/design.pdf
    Design of Snubbers For Power Circuits
    By Rudy Severns
    28 pages​

    It is an excellent tutorial and application note.
     
  19. Retrovert

    Retrovert AK Subscriber Subscriber

    Messages:
    3,295
    Location:
    New York City, NY
    Not exactly. You've merged the two concepts.

    What is called "cathode stripping" has nothing to do with a tube being in cutoff, i.e. no current flowing through the tube. It has to do with pulling current through a tube before the space cloud is fully formed, which is a function of the cathode being insufficiently heated. This causes point emission from the cathode and damage to the tube. In rectifiers it results in rapid deterioration and failure because of the current involved.

    The cutoff issue, where B+ is not flowing across the tube, controls cathode poisoning from orthosilicates. This is electrochemistry. It has nothing to do with excessive current flow through the tube, it concerns no current flow at all when the reactions in the cathode occur between silicon and alkali metals may occur. It is not limited to flip-flops and other digital logic, although this was a common source. Tek had to source special low-silicon tubes for its scope front ends which maintained tubes in cutoff.

    The statement that no damage occurs from premature current flow is internet lore and it defies logic as a basic understanding of how the cathode works explains why this is a bad idea. I need only point to the datasheets from tube companies for transmitting tubes and rectifiers; signal tubes are also damaged, but at a far slower rate.

    Rectifiers are commonly damaged by pulling excessive current in a short time interval, called the conduction angle. This is why a maximum permissible capacitance is specified for a capacitor-input supply. This value really specifies the maximum instantaneous current, but engineers would not have understood how increasing capacitance narrows the conduction angle and greatly increases dI/dt to the point of causing damage. The fact is that the rectifier tube has a finite lifespan, and the larger the capacitance being driven the shorter that lifespan. The rectifier eventually fails with an arc. This is why an NTC thermistor, which delays passage of current, extends the lifespan of the rectifier.
     
  20. Retrovert

    Retrovert AK Subscriber Subscriber

    Messages:
    3,295
    Location:
    New York City, NY
    Ok, but that won't show the high-frequency signal produced by the Qrr. It is in the AM band, hundreds of kHz. That is modulated by mains harmonics, typically twice mains, but higher orders also exist.

    What you're mostly seeing there is transformer ringing because it is being stimulated by Qrr. The other stimulation for transformers is short conduction angle because once the switch is made to a solid-state rectifier the input-capacitance limit of the tube is removed. (The rectifier still has one, however, it is just larger.) That creates a giant squarewave charging pulse which causes the transformer to ring.

    The transformer should be snubbed as well. The Quasimodo or Cheapomodo empirical snubber selectors are the best way to go to snub the transformer and prevent ringing.
     

Share This Page