Fusing cathodes

[demonstratedspatulas]

Assuming a 100 watt 4x6L6 or EL34 push pull amp.

Have you ever built one with fused cathodes?

I have heard this is a good idea to protect against bias failure. (and possibly other issues?)

Seems like a good idea to me, fuses are cheap, tubes are expensive.

What type of fuses did you use? What amperage? Have you ever had this fuse blow, saving your tubes?

Or is this more trouble than it's worth, ie, very rare failure case?

[Stevem]

I have never had a bias circuit fail fully, only a bias adjust pot shift value.

Save yourself the work of installing however many fuses holders per tube and just a 20 turn cermet type pot to make the bias voltage adjustment setting, as they never fail or go intermittent and give far more precise adjustments.

[R.G.]

Let me put this a different way. Tubes are known to short or go overcurrent for an assortment of reasons, and this probably happens more often than massive bias failure. Cathode fuses probably protect the power transformer/rectifiers/filter caps more than anything else.

My slant on fixing the issues surrounding this is to put a current sense resistor and a power MOSFET in each cathode, and have a single IC read the cathode currents through the cathodes, decide whether this is too much current for too long, and turn off one or all of the MOSFETs (and hence the tube it's in series with). For extra points, you can make the IC turn on one or more LEDs to tell you what happened and which tube went crazy.

[xtian]

It's pretty common to see a 500mA fuse in the B+ rail.

[pdf64]

Tube shorts that somehow connect HT current to the heater circuit are a fairly common failure mode; a repair bench often gets presented with an amp where heater balancing resistors have blown and there’s a power tube issue. Unfortunately cathode fuses would offer no protection in that scenario.
A general HT fuse seems a better option; ideally maybe a T type between winding and rectifier, and an F type between reservoir and amp.
A 1/2 watt 1 ohm cathode current sensing resistor will invariably pop if passing fault current

[R.G.]

Mmmmm. You're right - if it doesn't go through the cathode, the fuse can't stop it, nor can any other widget in the cathode.

Just thinking on the heretical side, there are good Hall effect current sensors with thousands of volts of isolation to the current path. Easy enough to stick one in each plate lead.

[pdf64]

Don’t forget about that other common tube failure mode, where a short zaps the screen grid resistor. So need to sense current there too

[trobbins]

You can set up individual cathode fuses, and they should provide better discrimination than a mains side fuse, or even a secondary side HT fuse if the amp has one.

The incentive is if the OPT half-winding failed due to prolonged high current passing through the failed tube. There are certainly many failed OPT's with one half-secondary open-circuit - although that could be from a pin 2-3 short, or some other anode related fault to ground or screen.

The fuse has to withstand the expected peak signal current, and that could be increased due to bias tolerance, and valve mismatch when using a quad etc. So that would set the minimum fuse current rating.

The prospective fault current seen by the fuse would depend on the sag of the B+, and the series resistance of the fault path (ie. including the OPT winding resistance and the plate resistance at the fault operating point). So that would set the likely ratio of fault current to fuse rating, and hence the range of time it could take for a fuse to operate.

If the fuse blows, and it is not the tubes fault, the tube could then be damaged from too high a cathode-heater voltage.

Some discussion on that topic in link:
https://www.dalmura.com.au/static/Valve ... fusing.pdf

[frankdrebin]

i have seen someone,i don't remember where,wiring relay contacts on cathodes,the relays don't close until a predetermined bias voltage is reached,that seems to me a good solution.

[cdemike]

I went looking for this thread in case anyone goes searching for this in the future:

I did some mods on one of my builds today that required me to move the amp back-and-forth from my workbench while I dialed in some values. Like an idiot, I forgot to plug in a speaker cable when I powered it on during one of tests. It was powered on without a load very briefly -- no more than 5s -- and I rapidly shut the amp down, checked around for any signs of damage, and did a voltage check on the OT leads. I plugged into the amp, and it made sound, but it was ok but had lots of cross-over distortion. Thinking I may have messed up my bias settings or shuffled the tubes around into different spots (this amp has dual bias controls) during the mods, I checked the bias and found there was about 20V on each cathode. I actually thought my multimeter's battery was about to die since that was such a weird value, but with a new batter I got the same result. So I visually checked the output section's cathode fuse (0.5A SB shared between 2x KT88s), and it didn't look blown. However, checking it with a multimeter showed that the fuse's resistance had climbed to a few hundred ohms.

After replacing the fuse and re-biasing, the amp sounds great. Although it's not a substitute for fusing HT and mains, I was inspired by this thread to to idiot-proof my amp, and I think it possibly saved me at least a new set of output tubes and possibly an output transformer.

[trobbins]

Thanks for taking the time to describe your experience

Did your recent mods change the output stage in anyway that may have influenced the cathode fuse to nearly blow ?

Did you have an input signal applied to the amp when you turned it on without a speaker load ?

Normally the cathodes wouldn't experience a current surge during start up, even with no speaker connected. When you subsequently rebiased the output stage, was the bias significantly out of whack?

The 0.5A fuse element may have physically sagged, but not been blown asunder, given that a short duration fault of some kind may have only had a lowish 'over-current' within the circuit path.

20V across the fuse (given a common cathode), with a few hundred ohm presented by the fuse, would indicate say less than 100mA but more than 20mA, so maybe 0.4 to 1W dissipation in the fuse. The extra cathode bias voltage from the degraded fuse, along with the fixed grid bias voltage would certainly suppress the cathode current below nominal idle, and may have reached a compromise scenario in the few seconds you made your measurements.

[cdemike]

Thanks for taking the time to describe your experience

Did your recent mods change the output stage in anyway that may have influenced the cathode fuse to nearly blow ?

Did you have an input signal applied to the amp when you turned it on without a speaker load ?

Normally the cathodes wouldn't experience a current surge during start up, even with no speaker connected. When you subsequently rebiased the output stage, was the bias significantly out of whack?

The 0.5A fuse element may have physically sagged, but not been blown asunder, given that a short duration fault of some kind may have only had a lowish 'over-current' within the circuit path.

20V across the fuse (given a common cathode), with a few hundred ohm presented by the fuse, would indicate say less than 100mA but more than 20mA, so maybe 0.4 to 1W dissipation in the fuse. The extra cathode bias voltage from the degraded fuse, along with the fixed grid bias voltage would certainly suppress the cathode current below nominal idle, and may have reached a compromise scenario in the few seconds you made your measurements.

Interesting! I didn't consider quantifying the bias condition that existed with the semi-blown cathodes fuse after I found the cause of the cross-over distortion. The recent mods shouldn't have impacted idle conditions in the amp. The background of the story is that I was experimenting with lowering filtering on the amp (it's a 50w take of a 100w Marshall; ideas for how best to scale the power section's response came from a thread a few months ago in the Dumble section of the forum). In other words, I just tweaked the filter capacitor values (uniformly lower across the output section vs previous), so I shouldn't have made any changes that would impact the idle output conditions. Dynamic loads should also be closer to a colder bias than previous since B+ should sag more while the bias supply remained the same.

The way this happened was that I unplugged the amp without flipping the power or standby switches to make the adjustments before the accident. I then drained the caps and bent a resistor around two turrets so that it was in parallel with one of the cathode resistors to try a lower value, and then went to test it. So when I plugged the power cord into the IEC jack the amp was on since I was being dumb not being methodical about properly powering down the amp. In the intervening time, before realizing my mistake, I plugged a guitar in and then connect a speaker. When I went to power the amp on and saw the switches were both set to "on," I realized my mistake (side note: there is not indicator light at this time since the amp doesn't yet have a faceplate, and I didn't want to disconnect and reconnect the light during that installation). Although the amp was only on for a brief period of time, the total time off was probably about 2 minutes (amp takes about 30s for the caps to drain below 10V by the time I reach my bench from where I had the amp set up), so my thought is that the heaters didn't have much time to cool off and the warm-up time was shortened. If that were true, it might have been enough time for some signal to flow.

I also should mention that part of what I discovered yesterday during this process is that I had a microphonic tube in V2 that would also make rare, intermittent crackling and buzzing sounds without being physically moved. This tube is set up as a cathode follower, though, and without standby, I suppose it may be possible that it experienced a heater-cathode arc. The tube was bad before this issue with the fuse began, and it sounded the same afterward, so I'm doubtful that this was the source of the offending signal.

When re-biasing the amp, it was very close to where I left it when I first installed these tubes, so no major drift after the incident. These KT88s have approx 30 hours of play time since I first installed and biased them. at 440V at the plates, I biased them to 44mA and 46mA, and when I checked on them again after the fuse swap, they idled at ~42mA each. So I think it'd be hard to know if the drift was regular wear-and-tear of the tubes, margin of error viz different wall voltage between when I installed the tubes and yesterday, or if they sustained some damage during the accident.