Hi. I built the Blueglow KT88, pretty much to the design in the drawings, my first ever build. It was working great for a few months, then all of a sudden the left channel cathode bias resistor (500R 10W wirewound, #16 on the diagram) popped with a loud bang, flash, and a puff of acrid smoke. I had noticed some static and distortion in the 10 seconds or so before it blew.
So I replaced it, and after about 5 minutes of it being on it happened again, same static, same resistor, same pop. I was monitoring voltages at the time (B+ and the plate of the driver tube) and they were relatively normal, though the voltage on the driver plate was a little higher than I remember when I built it. I had just earlier checked voltage at the plates of the KT88s and they were normal at the time.
Bad tube? Bad bypass capacitor? Bad wiring? Bad ground? Something else in the chain? I know these components are all related and can affect each other in ways my novice mind can't comprehend just yet.
Before I go out and spend another $200+ on another KT88 matched pair, I wanted to see what you folks with much more knowledge on this have to say. What should my next troubleshooting step be?
Diagrams attached.
Thanks.
So I replaced it, and after about 5 minutes of it being on it happened again, same static, same resistor, same pop. I was monitoring voltages at the time (B+ and the plate of the driver tube) and they were relatively normal, though the voltage on the driver plate was a little higher than I remember when I built it. I had just earlier checked voltage at the plates of the KT88s and they were normal at the time.
Bad tube? Bad bypass capacitor? Bad wiring? Bad ground? Something else in the chain? I know these components are all related and can affect each other in ways my novice mind can't comprehend just yet.
Before I go out and spend another $200+ on another KT88 matched pair, I wanted to see what you folks with much more knowledge on this have to say. What should my next troubleshooting step be?
Diagrams attached.
Thanks.
Check the coupling cap on that channel for leakage at the grid of the output tube, check the grid resistor on that output tube and make sure it isn't open. I'd also check the triode/UL switch and make sure it hasn't failed.
Monitoring the plate voltage of the driver won't tell you much. You need to check the cathode current on the output tube. Sounds like it is slowly rising until it overheats the tube and the cathode resistor. That means a failed grid circuit, positive voltage on the grid or something else causing the output tube to run away.
It also might have been the output tube itself. If it has an internal short or failure, using it again would have caused the same issue. If you have no way of testing it, I would hesitate to use it again.
Check all the passive components around the output tube--grid resistor, triode switch, etc. Remove the output tube on that channel, turn the amp on and see if there is any positive voltage at the grid pin (leaky coupling cap). If not, swap the good tube to that channel, turn the amp on again, and measure DC voltage across a new cathode resistor. If it's stable and nothing overheats, you probably have a bad KT-88.
Monitoring the plate voltage of the driver won't tell you much. You need to check the cathode current on the output tube. Sounds like it is slowly rising until it overheats the tube and the cathode resistor. That means a failed grid circuit, positive voltage on the grid or something else causing the output tube to run away.
It also might have been the output tube itself. If it has an internal short or failure, using it again would have caused the same issue. If you have no way of testing it, I would hesitate to use it again.
Check all the passive components around the output tube--grid resistor, triode switch, etc. Remove the output tube on that channel, turn the amp on and see if there is any positive voltage at the grid pin (leaky coupling cap). If not, swap the good tube to that channel, turn the amp on again, and measure DC voltage across a new cathode resistor. If it's stable and nothing overheats, you probably have a bad KT-88.
The coupling cap is 400V rated.
I always use a coupling cap that is rated for higher voltage than the unloaded B+ Voltage, in your case 600V or 630V.
Monitor the KT88 g1 grid voltage, it should be zero V, or very nearly zero V, when the amplifier is warm for a couple of minutes.
coupling cap leaky?
KT88 gassy and drawing grid current, even with no signal?
You have 41V / 500 Ohms = 82mA. 82 mA x 450V = 36.9 Watts dissipated in the plate + screen.
Some modern manufacturer's tubes might go into thermal run-away with that much dissipation.
I always use a coupling cap that is rated for higher voltage than the unloaded B+ Voltage, in your case 600V or 630V.
Monitor the KT88 g1 grid voltage, it should be zero V, or very nearly zero V, when the amplifier is warm for a couple of minutes.
coupling cap leaky?
KT88 gassy and drawing grid current, even with no signal?
You have 41V / 500 Ohms = 82mA. 82 mA x 450V = 36.9 Watts dissipated in the plate + screen.
Some modern manufacturer's tubes might go into thermal run-away with that much dissipation.
He only has 410 across the tube (subtract the cathode voltage) so it's running at 33.6 watts dissipation.
grovergardner,
Good point, I forgot to subtract the cathode voltage.
Hopefully he can fix the amplifier soon, then we can see what is the fix to his problem.
Good point, I forgot to subtract the cathode voltage.
Hopefully he can fix the amplifier soon, then we can see what is the fix to his problem.
I've found that modern KT88s draw enough grid current that a 220k grid resistor isn't enough - the grid drifts up and so does plate current. I've gone to 47k to be safe. 0.22uF into 47k is still fine in my opinion in terms of LF response..
Pete
Pete
I agree that a modern KT88 and 6550A may actually be the same thing. I don't think you will need a matched pair for separate left and right stages. Swap the KT88s over and see if the problem moves across too. A single KT88 replacement will do. Yep dropping the 220k down is good but there may not be enough drive maybe try 100k for 13 and increase 12 a bit to get the drive back. Not sure without simulation. Sounds like a faulty KT88.
Last edited:
When the KT88 cathode resistor opens up, is there more than 100V on the cathode?
If there is more than 100V on the cathode, then the bypass capacitor should also be replaced, along with the 500 Ohm resistor.
If there is more than 100V on the cathode, then the bypass capacitor should also be replaced, along with the 500 Ohm resistor.
Wow, a lot of good information. I'll need some time to digest it (i.e., understand what you;re telling me!) and figure out my next move. Thanks to all so far who've responded. More to follow.
OK, so here's what I did, starting with the guidance in @grovergardner's post. I know I probably should have tested for leakage on the coupling cap before replacing parts willy nilly, but since I had another one I went ahead and replaced it anyway. I also replaced the bypass cap because, as @6A3sUMMER pointed out, I don't know what voltage I had running across that, so better safe than sorry. I also checked that the various resistors weren't open, and they all read within spec.
As for the triode switch, when I tested it I heard some momentary beeps from the continuity tester when I shouldn't have, and although the resistance was high, it wasn't open when it should have been open. I can't say for sure that it was bad--I just don't know enough about these circuits and how all the parts interconnect--I swapped it for a new one anyway since I had one. The new one is completely silent on the ohmmeter when it should be silent, so maybe that was it? If so, great instincts, @grovergardner!
In any case, it all seems to work now (mostly, see next paragraph). I checked for leakage: none. I swapped the known good tube and checked the voltage drop across the cathode bias resistor: as it should be voltage wise and steady (again, see next paragraph).
I then, for giggles, put the original tube back in, and with my finger on the Variac power switch, powered up and watched the voltage across the cathode resistor on the "bad" tube. For 25 minutes it was within specs +/-3v. I tried another 20 minutes in ultra-linear mode, also within specs +/-3v. I monitored both channels for another 20 minutes, and both channels were within 1v of each other and within +/-3v of spec consistently for the entire time. Over an hour, no boom, and a perfectly steady voltage, albeit a little high...see #2 below.
So, here are my questions:
1) Can I assume the "bad" tube is good? That my problem must have been elsewhere, either the coupling cap or the triode switch?
2) The voltage drop across the cathode resistor averages about 43v with my home's A/C voltage--I adjusted my Variac so it mimicked my house power this evening, between 115v-118v. The 43v drop is 2v higher than the drawing's spec, and gives me 35.3 watts dissipation...the KT88's are rated at 35w....Implications? Fixes if necessary?
3) I'll be honest, I didn't understand @pmillett and @baudouin0 and their discussion about changing the value of the grid leak resistor away from 220k and the implications that would have. If you're up for it, I would appreciate a primer, particularly if that would be the fix to my observation in question #2 above.
Thanks to all who pointed me in the right direction. The amp is still on my bench, I haven't dropped it back into my system yet, because I'd like to know your continued thoughts.
Thanks again.
Mike
As for the triode switch, when I tested it I heard some momentary beeps from the continuity tester when I shouldn't have, and although the resistance was high, it wasn't open when it should have been open. I can't say for sure that it was bad--I just don't know enough about these circuits and how all the parts interconnect--I swapped it for a new one anyway since I had one. The new one is completely silent on the ohmmeter when it should be silent, so maybe that was it? If so, great instincts, @grovergardner!
In any case, it all seems to work now (mostly, see next paragraph). I checked for leakage: none. I swapped the known good tube and checked the voltage drop across the cathode bias resistor: as it should be voltage wise and steady (again, see next paragraph).
I then, for giggles, put the original tube back in, and with my finger on the Variac power switch, powered up and watched the voltage across the cathode resistor on the "bad" tube. For 25 minutes it was within specs +/-3v. I tried another 20 minutes in ultra-linear mode, also within specs +/-3v. I monitored both channels for another 20 minutes, and both channels were within 1v of each other and within +/-3v of spec consistently for the entire time. Over an hour, no boom, and a perfectly steady voltage, albeit a little high...see #2 below.
So, here are my questions:
1) Can I assume the "bad" tube is good? That my problem must have been elsewhere, either the coupling cap or the triode switch?
2) The voltage drop across the cathode resistor averages about 43v with my home's A/C voltage--I adjusted my Variac so it mimicked my house power this evening, between 115v-118v. The 43v drop is 2v higher than the drawing's spec, and gives me 35.3 watts dissipation...the KT88's are rated at 35w....Implications? Fixes if necessary?
3) I'll be honest, I didn't understand @pmillett and @baudouin0 and their discussion about changing the value of the grid leak resistor away from 220k and the implications that would have. If you're up for it, I would appreciate a primer, particularly if that would be the fix to my observation in question #2 above.
Thanks to all who pointed me in the right direction. The amp is still on my bench, I haven't dropped it back into my system yet, because I'd like to know your continued thoughts.
Thanks again.
Mike
LF response might be OK but 47K following grid R requires that a rather lower driving plate R be used.
Then a lower rp driving tube is needed. All to maintain a reasonable looking AC load on the driving tube.
Otherwise D% increases. 🙂
Glad the amp is working properly now. If I were you, I'd get rid of those triode/UL switches and put in some kind of jumper or terminal strip instead. A lot of switches aren't designed to handle that much DC votage. Maybe they're rated for it but I've always thought they were an accident waiting to happen.
I don't see any reason to reduce the grid resistor that drastically. It's really loading down the driver tube and compromising the performance of the amp. 220K should be plenty safe.
Is the old tube okay to use? Well, if the screen was floating due to a failure of the switch, that might have caused the current to rise. It might not have damaged the tube, hard to say. Personally I'd replace it but if you keep and eye on it it's probably okay.
I don't see any reason to reduce the grid resistor that drastically. It's really loading down the driver tube and compromising the performance of the amp. 220K should be plenty safe.
Is the old tube okay to use? Well, if the screen was floating due to a failure of the switch, that might have caused the current to rise. It might not have damaged the tube, hard to say. Personally I'd replace it but if you keep and eye on it it's probably okay.
On the bad tube you could measure across the 220k for grid leakage. You could also check across the 1k2 for screen current and check it's in spec. For cathode bias 220k should be just fine. Not sure about the tube my gut feeling is to replace. The large current will have gone through the KT88. The UL switches agree with grover. If you wish to keep it then a permanent 10k from screen to HT will stop a problem during changeover.
Last edited:
What baudouin0 said, plus I assume you never flip the switch while the amp is on or even warm. Just checking. ;-)
OK, it all makes sense, thanks. I'll get at least one new KT88, maybe a new matched pair.
Re the switch: OK, I'm sold. I think I'll remove the switch from the architecture for now and make it triode-only mode, as I slightly prefer that sound for most music.
Having said that, what would you recommend if I wanted to switch between UL and triode? I don't even know where to start...I took a look at Mouser for jumpers and shunts and was overwhelmed. Any recommendations? Can I just use standard electrical male/female quick disconnects (insulated and properly rated of course)?
I've got bleeder resistors on my filter caps which I confirmed are working fine when I had it on the bench last week, so I'm fairly certain I didn't flip the switch with any significant voltage running to the tubes. In any case, I hadn't flipped the switch for weeks prior to the first boom, so I don't think that's what caused this.
Thanks again.
-Mike
Re the switch: OK, I'm sold. I think I'll remove the switch from the architecture for now and make it triode-only mode, as I slightly prefer that sound for most music.
Having said that, what would you recommend if I wanted to switch between UL and triode? I don't even know where to start...I took a look at Mouser for jumpers and shunts and was overwhelmed. Any recommendations? Can I just use standard electrical male/female quick disconnects (insulated and properly rated of course)?
I've got bleeder resistors on my filter caps which I confirmed are working fine when I had it on the bench last week, so I'm fairly certain I didn't flip the switch with any significant voltage running to the tubes. In any case, I hadn't flipped the switch for weeks prior to the first boom, so I don't think that's what caused this.
Thanks again.
-Mike
My understanding is that if the screen gets disconnected it should drop down to 0V very quicky and the KT88 goes out of conduction. However, the sudden collapse together with the OPT leakage inductance may cause the plate voltage to go very positive (a few Kvs) and trigger a breakdown. The amount of current you describe is probably internal plasma/arc in the valve. Can you see any pitting on any part of the plate. Me I would add a 10K from screen to HT and use relays to do the switching.
Last edited:
Adding a 10k series resistor to the screen would do the following:
Minus's
1. Increase the harmonic distortion (the negative feedback would hide some of that).
2. Reduce the maximum output power
3. Reduce the output stage gain; and so reduce the open loop gain; reducing the effect of the negative feedback (less negative feedback).
'Plus's', (or not).
4. Reduce: plate current, screen current, and cathode current, which would make for a cooler output tube.
But you could decrease the self bias resistor's resistance, to increase the current.
Beam Power mode is the only mode that allows for a bypass capacitor to ground, the other end of the bypass cap to the junction of the 10k resistor and screen. The same applies to Pentode mode (EL34).
Generally, with the amplifier powered, switching from triode mode to UL mode, or back, may be bad for the output transformer, the tweeter, and your ears too.
Tradeoffs . . .
Minus's
1. Increase the harmonic distortion (the negative feedback would hide some of that).
2. Reduce the maximum output power
3. Reduce the output stage gain; and so reduce the open loop gain; reducing the effect of the negative feedback (less negative feedback).
'Plus's', (or not).
4. Reduce: plate current, screen current, and cathode current, which would make for a cooler output tube.
But you could decrease the self bias resistor's resistance, to increase the current.
Beam Power mode is the only mode that allows for a bypass capacitor to ground, the other end of the bypass cap to the junction of the 10k resistor and screen. The same applies to Pentode mode (EL34).
Generally, with the amplifier powered, switching from triode mode to UL mode, or back, may be bad for the output transformer, the tweeter, and your ears too.
Tradeoffs . . .
Last edited:
I don't know how much room you have under the amp, but you could use a terminal block to reconfigure the amp from triode to UL. Anything that a) forces the amp to be off and turned over and b) provides a hard connection.
Suppose the power cord is removed, and there is no B+ Bleeder Resistor, from B+ to ground.
Then you move the screen connections with your fingers.
Ouch!
The schematic does not have any B+ bleeder resistor.
Do not rely on the tubes to discharge the B+ voltage all the way.
Even with a bleeder resistor, it takes time for the B+ capacitors to discharge.
Safety First!
Prevent the "Surviving Spouse Syndrome".
Then you move the screen connections with your fingers.
Ouch!
The schematic does not have any B+ bleeder resistor.
Do not rely on the tubes to discharge the B+ voltage all the way.
Even with a bleeder resistor, it takes time for the B+ capacitors to discharge.
Safety First!
Prevent the "Surviving Spouse Syndrome".