KT88's...max watt rating and where to bias?

Discussion in 'Amps/Cabs Tech Corner: Amplifier, Cab & Speakers' started by DonaldDemon, Dec 31, 2017.


Tags:
  1. DonaldDemon

    DonaldDemon Member

    Messages:
    6,187
    Joined:
    Nov 13, 2007
    Location:
    Fear and Loathing, NJ
    I tried asking this in April on the amp forum but got no responses so I'll try it here. I have yet to get a clear answer on this.

    I installed a quad of Ruby/Shuguang KT88V-STRs. Supposedly these are a return of the Penta Labs/Shuguang KT88SC (or KT88S?), which were my favorite tubes for my Splawn Pro Mod.

    http://www.dougstubes.com/power-tubes/6550-kt88/ruby-kt88-v-str.html

    I've noticed some modern tubes have a different max anode/plate dissipation rating (PaMax), especially for KT88's. Looking at the data sheet for the Penta/Shuguang/Ruby KT88V-STR, the max anode dissipation is 50 watts! Compared to the regular 42 watts or even 35 watts on some more common KT88 data sheets that seems pretty high and really makes a big difference (about 10 mA) for the bias calculations.

    According to calculations for my amp w/ a 495 plate voltage, using 50ma (PaMax) I'd be looking at biasing around 65mA for 65% dissipation, which I wouldn't feel safe doing. Using 42ma (PaMax), it would be 55ma @ 65%, and using 35ma (PaMax), it's 46ma. So, which would it be, because those are some significant differences if I was to get it wrong?

    Here is the data sheet in question:http://www.dougstubes.com/Penta-KT88.pdf
     
    Last edited: Dec 31, 2017
  2. guitarded_1

    guitarded_1 Member

    Messages:
    1,134
    Joined:
    Nov 11, 2003
    Location:
    Hotlanta, GA
    Funny you should ask. I am planning to bias up a pair of these tonight. Unfortunately, I do not have the answer. However, I am going to use 42 Watts when calculating bias. No scientific backing - just seems like a happy medium.
     
  3. DonaldDemon

    DonaldDemon Member

    Messages:
    6,187
    Joined:
    Nov 13, 2007
    Location:
    Fear and Loathing, NJ
    Yep, that's what I usually do as well, or try to find a happy medium between 60-70% dissipation between all of the numbers. I've asked this before and have read similar threads by others and they are all a dead end, no one seems to know how to answer it.
     
  4. HotBluePlates

    HotBluePlates Member

    Messages:
    2,497
    Joined:
    Mar 27, 2017
    The relevant thing to see on the website for these tubes is the phrase, "We've sold these previously as the Penta Labs KT88SC's."

    AFIK, "Penta Labs" was only ever a re-brand of Chinese tubes, so all you really know from the vendor is the tube in question was previously sold under some other designation/brand/type #. If you want the straight dope, you'd have to find out from Ruby, who'd have to find out from Shuguang (unless you can contact Shuguang directly & find out the real limits on that specific tube).

    Your best best is to go conservative, bias based on 42w. Play the amp, see how it sounds, see that the tubes are comfortable. If you're curious after that, inch upward towards the 50w mark. See how it sounds, see that the tubes aren't going into melt-down.

    Two things about that: Tubes last longer if run conservatively, and will the extra plate current do you any good in the amp? Hotter idle bias might not produce any extra power output, though it may make the amp seem easier to drive to full output (a smaller drive signal to the KT88's will get them to their peak plate current). Taking advantage of the higher plate dissipation (if real) is more a matter of design choice with OT primary impedance, screen voltage and bias for maximum power output.


    (I said, "if real" above because who knows if anyone can really believe a data sheet on a modern tube. Current production is inconsistent, and the data sheets published by "Tung Sol"/Sovtek/EH are largely photoshopped versions of the original manufacturers' data sheets with the original old dates removed. Dunno about the Chinese stuff, or even reliable data even exists.)
     
    Last edited: Jan 1, 2018
    DonaldDemon and pdf64 like this.
  5. pdf64

    pdf64 Member

    Messages:
    4,739
    Joined:
    Aug 9, 2008
    Location:
    Staffordshire, UK.
    The above dilemma illustrates the issue at the heart of the 'set bias to idle @ xx% of the plate dissipation limit' biasing method.
    Apologies if I'm incorrect, but I think it leads people into the perception that tubes have to be idling at a certain % for them to work correctly / sound good; really it's not like that, the tube's heaters do the necessary heating, plate dissipation is just a limit, the tube doesn't require yet more heat from the plate in order to achieve its characteristics.

    The amp was presumably designed to sound good with KT88; all functional KT88 have the same characteristics, that's what makes them KT88; whether they have 35 watt, 42 watt, 50 watt etc limits on their plates is immaterial to those characteristics. Hence for a given amp, there's no reason to idle a 50 watt KT88 hotter than would be appropriate for a 35 watt KT88. Changing the bias setting primarily changes gain and (in class AB amps) conduction angle; for a given amp design, why would one KT88 need more gain / a greater conduction angle than another? If a 70 watt KT88 was introduced, would it really make sense to idle it twice as hot as a vintage KT88? As @HotBluePlates has demonstrated, doing so would likely reduce the amp's power output significantly, and the method clearly becomes counterproductive.

    Regarding that Penta KT88 info, I think it's unfortunate, and perhaps ignorant, that it conflates absolute max with design max:mob
     
    Last edited: Jan 1, 2018
    HotBluePlates and xtian like this.
  6. DonaldDemon

    DonaldDemon Member

    Messages:
    6,187
    Joined:
    Nov 13, 2007
    Location:
    Fear and Loathing, NJ
    This is the most reasonable explanation I've read yet and agree 100%. The Penta and Ruby's are just select Shuguang factory tubes as you noted. Good point on the data sheets, I can't say I entirely trust them and you made a clear point as to why. Next time I bias I will do it like you stated,v which I believe is how I did it last time. In fact I have my notes from last time and my plate voltage was around 492 with the tubes biased around 52 ma.
     
  7. DonaldDemon

    DonaldDemon Member

    Messages:
    6,187
    Joined:
    Nov 13, 2007
    Location:
    Fear and Loathing, NJ
    I agree with that too and it does become a dilemma since we use those design maxes as a crucial part of the bias calculation. I can't claim to know much about tube and amp design but at least for the layman diy'er it becomes confusing and frustrating, probably even more so than if I was a tech and could just a scope to sort it out.
     
  8. guitarded_1

    guitarded_1 Member

    Messages:
    1,134
    Joined:
    Nov 11, 2003
    Location:
    Hotlanta, GA
    So I put these in my VHT Deliverance 60 last night. Biased them up to a relatively conservative 52mA or 60% dissipation and assuming 42 Watts max. After 10 minutes at low volume, one tube started red plating. I switched tube positions today and the red plating follows the tube. I feel like it’s probably not an amp issue. Is the red plating likely to indicate a defective tube, or is it more likely that the tube is reasonably healthy and the bias should be lowered?
     
  9. pdf64

    pdf64 Member

    Messages:
    4,739
    Joined:
    Aug 9, 2008
    Location:
    Staffordshire, UK.
    A tube that’s red plated should be assumed to be damaged / unfit for service.
    You may have a valid warranty claim.
    What is the control grid circuit resistance of the amp, ie from pin 5 to chassis 0V?
    But really, 52mA seems a very high idle current! Most such circuits are well out of crossover by 10 to 15mA. Why so high?
    What do VHT suggest as a suitable idle current?
     
  10. guitarded_1

    guitarded_1 Member

    Messages:
    1,134
    Joined:
    Nov 11, 2003
    Location:
    Hotlanta, GA
    VHT isn’t much help. I’m a noob, honestly. Simply using a bias probe and a bias calculator to arrive at 60% of max dissipation. The tubes should be covered under warranty from Ruby. I’ll also try backing down the bias and see what happens.
     
  11. pdf64

    pdf64 Member

    Messages:
    4,739
    Joined:
    Aug 9, 2008
    Location:
    Staffordshire, UK.
    As I’ve tried to show above, with such high capacity tubes in a regular amp, setting bias to a % is pretty much a nonsense.
    Listen carefully to how the amp responds; apart from the gain increase, is >50mA really any better than say 35mA?
    Just set the bias to the lowest current that sounds good.
     
    guitarded_1 likes this.
  12. guitarded_1

    guitarded_1 Member

    Messages:
    1,134
    Joined:
    Nov 11, 2003
    Location:
    Hotlanta, GA
    Will do. Appreciate the feedback.
     
  13. guitarded_1

    guitarded_1 Member

    Messages:
    1,134
    Joined:
    Nov 11, 2003
    Location:
    Hotlanta, GA
    Here’s what is printed on the board. 1.15vdc = 60%. Honestly, I have no clue how that converts to mA. Looks like it’s time to google.

    [​IMG]
     
  14. HotBluePlates

    HotBluePlates Member

    Messages:
    2,497
    Joined:
    Mar 27, 2017
    We have a 10Ω resistor from TP1 (appears to be cathodes of both output tubes) to TP2 (ground). 1.15v = 60%, 1.25v = 65%, 1.35v = 70% (Factory).

    1.15v / 10Ω = 115mA, but this is for both tubes, so 57.5mA per tube. Repeat to get 62.5mA per tube for 65% and 67.5mA per tube for 70%.

    I guessed a max dissipation of 42w, so 42w * 0.6 = 25.2w, and 25.2w / 0.0575A = 438v (plate to cathode, not ground). Repeating with 50w leads to a plate-to-cathode voltage of 521v at 1.15v from TP1 to TP2 (cathodes to ground).

    You can estimate the dissipation rating of the tubes used originally by adjusting the bias pot to yield the correct voltage across TP1 & TP2, and measuring voltage from plate to TP1.

    NOTE: All of the above ignores the fact that there's some screen current in the measurement across the 10Ω resistor; rounding of results should land on a reasonable guess. Note the manual only recommends Russian 6550/KT88 from Sovtek, EH or Svetlana.

    I agree it's confusing & frustrating. I'm an "advanced" self-taught hack... It took me an embarrassing number of years to make sense of how the output stage really works because the modern books targeted towards players/hobbyist-techs breeze past critical fundamentals and focus on the wrong things (though I'll admit I haven't read every book out there; could be some good ones I haven't seen).

    What matters for output tubes is current swing.

    The output tubes swing from a peak plate current to some minimum; for class AB amps, that minimum is zero-current. The peak plate current possible happens when the control grid (G1, pin 5 of the common octal/8-pin output tubes in guitar amps) is driven to the same voltage as the cathode (for most common guitar amps). The peak current possible at that moment is dictated by screen (G2) voltage.

    So if the peak plate current is set by what happens at G1 & G2, and (in a class AB amp) the minimum current is a fixed value of zero, then bias doesn't factor in for output power. Instead, the idle bias keeps the tube from overheating during the signal cycle. There's quite a broad range of acceptable bias in many amps, and as @pdf64 said biasing hotter/cooler can adjust the apparent gain of the output tubes; hotter bias (without overheating) will make the output stage seems easier to drive, which is the "gain" we're talking about (not talking about distortion here).
    ________________________
     
    pdf64 likes this.
  15. guitarded_1

    guitarded_1 Member

    Messages:
    1,134
    Joined:
    Nov 11, 2003
    Location:
    Hotlanta, GA
    Good stuff! Thank you!
     
  16. HotBluePlates

    HotBluePlates Member

    Messages:
    2,497
    Joined:
    Mar 27, 2017
    You're welcome!

    Just another note: I forgot to say my calculations assume both tubes drew the exact same current. If you're using a bias probe, you can determine whether that's really the case or not.
     
  17. pdf64

    pdf64 Member

    Messages:
    4,739
    Joined:
    Aug 9, 2008
    Location:
    Staffordshire, UK.
    It looks like that particular amp was intended for KT88 with a hot-ish bias; the grid leak resistors look to be 100k, which complies with the Penta limit, but not other makes, eg EH, but it isn't usually a problem unless the tubes are run very hot.
    It's still valid to try a cooler bias and so benefit from a longer tube life etc.
     
    Last edited: Jan 2, 2018

Share This Page