1 ohm cathode resistors

[Randall]

What is the way to go regarding wattage? I read some folks think bigger is better, some say 1/4 watt is plenty, and can act like a fuse if something runs away with current draw. I am drawn to the 1/4 watt side. School me.

[Enzo]

You want your high voltage fused? Then use a fuse. It sounds great to use a resistor as a fuse, but can you make predictable blowing stats? CAn you tell us what would reliably blow that resistor? I can't. Dissipation is minuscule.

What would it take to blow a 1/4 watt resistor? Assuming 1 ohm, 0.25w = V x I. Since V and I are equal numbers in the 1 ohm case, we solve for the square root of 0.25, which is 0.5. Half a volt across a 1 ohm means half an amp flowing through it. 500ma. Now that gets you to 1/4 watt, which it should do forever. 700ma or so gets me a half watt dissipation. What if we pull a full amp through it? Then it dissipates a watt. Now the resistor won;t be happy at 1 watt, but it sure won;t burn out like a fuse either. And unless you use flame proof resistors of some sort, they may catch fire first. On the other hand, how many of our amps can crank a full amp of B+ into any circuit for more than a moment?

SO a 1/4 watt might work just fine, choosing it for some notion of fuse action I think is misguided.

I use 1% resistors, not so much because I think it matters, but customers have an inflated view of bias criticality, so they expect it. Me, I use 1 watt film resistors because they are physically sturdy and large enough to be reliable test points for my meter probes.

[R.G.]

I agree with Enzo. I put trying to use resistors as fuses in the "penny-wise, pound-foolish" category. Fuse makers spend a lot of their technical efforts on trying to get some kind of predictability on when the fuse element will open up. Resistor makers spend no time trying to figure out when a resistor will open. The resistor makers are trying to figure out what change in resistance they get with various levels of heating below burning out, and how hot the surface gets at various power levels.

I'm back in the contrarian camp: if you want to protect something, sense something that indicates what is going on, and then take some action that prevents as much damage as possible.

In the case of output tube cathode current, sense the current with a small resistor that doesn't change value with increasing current, and turn off the high voltage. Fuses present a semi-clever way to do this, but their performance is marred by being relatively expensive and complex to mount, and in being complex in the thermodynamic and heat flow sense because predicting a phase change of a metal is a complicated process. I prefer to sense the current in a resistor, then use other circuit elements to interrupt the flow of power supply current.

A power MOSFET makes a simply great switch for high voltage DC. These things can turn high voltage DC on and off literally millions of times per second for a large number of years when used properly. They approach a metal contact switch in many respects. I've done two or three iterations of designs that watch cathode current by watching the voltage across a small cathode resistor, and turning a B+ MOSFET off when some combination of current size and time of current flow is reached. It's very good at doing the primary purpose - stopping overcurrents - and costs about as much as two fuses. The biggest design challenge is providing the low voltage DC to power this setup. It also provides a clean way to do a "stoplight" biasing system, giving you a red/green/blue light for cathode current being too high, too low, and just right, and letting you get an accurate bias per tube in seconds.

But this does have the advantage that it takes some skill and understanding to design. Putting in a small resistor or fuse takes very little skill and understanding, but needs a large amount of faith that if and when the sudden-disaster event happens, things will just work out OK.

Could happen.

[Gregg]

Something similar to what R.G. described is going on in the YJM100 only the MOSFETs are located at the cathodes. The KC1-4 lines go to BC847 collectors.



http://www.tonymckenzie.com/Site/ima...ole-chasis.jpg

[R.G.]

There are a couple of different ways to go about it. One MOSFET in the B+ line is a shutoff valve. It can be set up as an inrush surge current limiter too, but you're stuck with one control point. One MOSFET per output tube makes some sense in a four (or six, eight, etc.) setup, as the MOSFETs can then turn off pairs. Depending on how clever you are with sensing and coding of the controller, it can also do things like retrying after a delay, and lighting an indicator saying which tube(s) had an overcurrent fault.

My approach was to run the various cathode resistor voltages into a uC with A-D converters, and do all the watching inside the controller. This let the same IC do both the disaster-preventer control of a MOSFET for B+, but also three LEDs per tube for a stoplight style bias annunciator.

It's like birthday cake: you slice the problem different ways depending on who's eating and who's paying, and perhaps who's second guessing everyone on internet forums.

[bob p]

Old SVT used resistors as fuses. It was't a good plan. Modern SVT use current sense resistors and comparators in the cathode circuits to track cathode current and to determine whether the protection circuit needs to be activated. Check out the current Ampeg schematics.

OTOH, if you just want to know how big of a 1% resistor to put on your cathodes so that they won't fail during operating conditions, 2W should be fine. I bought 3W because they were cheap.

[Randall]

"Me, I use 1 watt film resistors because they are physically sturdy and large enough to be reliable test points for my meter probes."

If that's good enough for Enzo, It's good enough for me!

[bob p]

lead diameter

[mac dillard]

A question re: the 1 ohm 1% cathode resistors used to measure bias....What do you guys do to compensate for the meter leads resistance? I use a Fluke and my leads will read about 2 ohms. Could be higher or lower according to how I make contact. Kinda makes accurate calculations tough.

[Enzo]

No it doesn't. You have a 1 ohm resistor in the circuit, so you are taking voltage readings across it. Your volt meter has an internal resistance of something like 1 meg to 10 meg. Your probes add 2 ohms to that, so 1,000,002 ohm instead of 1,000,000 ohms. That is a very small error added.

And as to upsetting the 1 ohm circuit, your meter is parallel that 1 ohm resistor. SO calculate the result of 1 ohm with 1,000,002 ohms in parallel. So the 1 ohms is reduced by a millionth of an ohm or so, way less than the error margin of the 1% resistor.

So to directly answer, I don't do anything to compensate for my probes. Where probe resistance matters on my bench is when I use them to measure low resistances, like those 0.22 ohm emitter resistors in SS power amps. I always measure my leads by shorting them so I have a base resistance to subtract. Typically I get half an ohm. My meter impedance gets in the way with high impedance circuits, circuits where 1 meg of meter resistance is a substantial part of circuit resistances. A common example is trying to measure grid voltages in long-tail pair phase inverters. But this is like your 1 ohm situation, it is the meter internal resistance that is at issue, not the probes.

If shorting your probes together does not yield consistent resistance, your probe tips are not making good contact, shine them up. If you get inconsistent contact with circuit points, then the probe tips ar enot making well or the surface of the circuit points are oxidized or contaminated. SOmetimes on real old circuits, I wind up refreshing teh solder so I have a fresh surface for my probe points.

I prefer needle sharp probe points, like really needle sharp, as opposed to the dull points on common meters.

[Gregg]

One MOSFET per output tube makes some sense in a four (or six, eight, etc.) setup, as the MOSFETs can then turn off pairs.

AFAIK that is a feature in YJM100 and one of the reasons for those MOSFETs to be there.

[Randall]

My Fluke reads .2 ohms shorted, and I occasionally treat the probes with deOxit. Not worried about it.

[pdf64]

With big power tubes, I've found that 1/2 watt metal film flameproof 1 ohm cathode resistors tend to fuse pretty quickly in response to a typical tube short, usually way quicker than actual fuses
They have the advantage to being specific to the faulty device.

[R.G.]

Not to harp on it, but if what you want is speed of shutdown and/or annunciation of which tube has a problem, you want to go to electronic sensing per tube.

With the use of a comparator watching per-tube current by watching the voltage across a sense resistor, you can achieve sub-microsecond detection of overcurrents, and using a MOSFET for interrupting current flow, you can achieve few-microsecond opening. It's so fast that you'll spend your time figuring out how slow it needs to be to avoid nuisance trips. Fortunately, that's a problem that stays solved when you find the right capacitor value.

With fast fault detection, you can then decide how much overloading time you want to allow, instead of wondering whether XYZ's resistor or fuse will be fast enough to avoid damage to your rectifier tube too. Or wondering if a soft fault in a tube will let it just sit there and cook until it really does cause flames. The soft fault problem is one reason that high voltage fuses are always enclosed in non-flanmnmable cases.

On top of that, I hate going off to search for yet another fuse to replace the one that keeps blowing. Electronic sense means you can simply reset it; and I really don't like having to remove and replace a soldered-in part just because a tube had an "event".

I also don't like the odor of resistors that have just fused. Nasty smell.

Philosophically, we're bridging the gap between an era when it was too expensive to do ANYTHING with electronics that could be done with a single passive part, however poorly the single part did the job, and an era when using a complex IC is so cheap and available that it outperforms the single passive part, and can also do the job faster and with less side effects. In my dotage, I have developed a sense of modest wonder at how very far electronics has come in the last half-century.

[g1]

Just to throw this out as a practical example of excessive current, the Marshall JCM900 uses a 500mA slow blow as a cathode fuse for each push/pull pair.