Immortal Mods: additional fuses, where to put the fuse holders?

[catalin gramada]

...and not even a fault. Inrush startup curents from bias filter caps can broke a hairwire bias winding. For this reason, think, we often see hundred ohms serial resistor in bias winding -to keep charging startup currents in safe area.The peak of currents can be determined and in respect with excesive big or used caps we can have an vague idea how much the winding toast before it die.

[mikepukmel]

I think what we are dealing with is some blanket regulation in some countries that requires that every transformer winding be fused, not some engineers notion of protecting a transformer. That's why only one version of the amp has the fuse in the bias winding.

Remember that most fuses are there to prevent the appliance from catching on fire. R.G. makes a good point about loss of bias, but explaining that to some bureaucrat in a foreign country is not something that average engineer or company wants to take on. You've got the marketing guys who want to sell in that country and the legal guys telling you that the amp has to pass regulations. So you install the fuse in the version of the amp that has to pass the regulation, and hope for the best. Maybe you even test the amp by removing the bias fuse to see if the other fuses protect the amp from catching on fire. The regulators probably will.

I had the same question: why the export version had each secondary winding fused, but the non export version, only on the primary side.

[trobbins]

There would be a typical minimum winding wire size for practicality in any commercial transformer. Bias windings also aren't high voltage, so C*V*V joule storage is not likely to be huge. The filter time constant versus ripple voltage tradeoff with bias filter capacitance will kick-in to limit a practical value of C. I can appreciate there would be a transient thermal stress from in-rush - but I've not come across winding damage as suggested before.

[R.G.]

My experience with all guitar amps is not encyclopedic by any means, but I believe that having a separate bias winding is rarer than having a tap on one of the high voltage halves. Or the scheme from some Marshalls and others to use a capacitor and high resistance from one of the high voltage halves. That changes the failure mode calculus a bit.

Some rumination on fault protection makes me classify fault protection schemes into ad hoc schemes, and overall sense-and-decide schemes. Ad hoc schemes involve making a separate and local protector for each possible fault to be protected against. I class the idea of putting fuses in each transformer secondary as an example of this class. Sense-and-decide schemes have become practical with the advent of integrated logic and especially microprocessors.

One example of sense-and-decide is the first version of output tube protection I came up with back n the early 2000s. It started out sensing the cathode currents of each power tube and running that into a lashup of CMOS logic gates that eventually turned off the current in the power tubes. By itself that would be and ad hoc, but I realized that with a turnoff available, I could lash other sensed conditions into the flipflop that turned things off. I had to calm myself down and quit adding stuff to the sensed conditions.

Today, I'd put a solid state relay in the incoming AC line and maybe a MOSFET in the power to the output tubes and run the decision making in a $0.50 microcontroller as a true sense-and-decide. The ability to do more complex decisions that you can do with hard logic chains. For instance, there are digital temperature sensors that communicate with uCs with two wires, one of which is ground. These cost about $1.50. You can put one of these on the PT and OT laminations and have the uC pulse an OT warning LED. Or run a current clamp in the main B+ lines for coming out of standby so that standby isn't traumatic to the rectifiers and filter caps any more. And then one of my favorites, tack on one more bell and whistle, have the uC run a red-green-blue LED or two to let you bias the output tubes super easily.

As the cynics will quickly see, a central sense-and-decide block is a super fertile ground for creeping featurism.

The applicable standard for guitar amps is IEC60065. It became my friend and nemesis when I designed the Workhorse series of amps. IEC60950 is for computing and IT equipment if I remember correctly. The requirements are similar but not identical as regards transformers, but I'd have to do a detailed review of them to tell, and I don't have the $300-$400 needed to buy a copy of an IEC standard.

But I need to stop typing. The coffee's ready.

[mikepukmel]

My experience with all guitar amps is not encyclopedic by any means, but I believe that having a separate bias winding is rarer than having a tap on one of the high voltage halves. Or the scheme from some Marshalls and others to use a capacitor and high resistance from one of the high voltage halves. That changes the failure mode calculus a bit.

Some rumination on fault protection makes me classify fault protection schemes into ad hoc schemes, and overall sense-and-decide schemes. Ad hoc schemes involve making a separate and local protector for each possible fault to be protected against. I class the idea of putting fuses in each transformer secondary as an example of this class. Sense-and-decide schemes have become practical with the advent of integrated logic and especially microprocessors.

One example of sense-and-decide is the first version of output tube protection I came up with back n the early 2000s. It started out sensing the cathode currents of each power tube and running that into a lashup of CMOS logic gates that eventually turned off the current in the power tubes. By itself that would be and ad hoc, but I realized that with a turnoff available, I could lash other sensed conditions into the flipflop that turned things off. I had to calm myself down and quit adding stuff to the sensed conditions.

Today, I'd put a solid state relay in the incoming AC line and maybe a MOSFET in the power to the output tubes and run the decision making in a $0.50 microcontroller as a true sense-and-decide. The ability to do more complex decisions that you can do with hard logic chains. For instance, there are digital temperature sensors that communicate with uCs with two wires, one of which is ground. These cost about $1.50. You can put one of these on the PT and OT laminations and have the uC pulse an OT warning LED. Or run a current clamp in the main B+ lines for coming out of standby so that standby isn't traumatic to the rectifiers and filter caps any more. And then one of my favorites, tack on one more bell and whistle, have the uC run a red-green-blue LED or two to let you bias the output tubes super easily.

As the cynics will quickly see, a central sense-and-decide block is a super fertile ground for creeping featurism.

The applicable standard for guitar amps is IEC60065. It became my friend and nemesis when I designed the Workhorse series of amps. IEC60950 is for computing and IT equipment if I remember correctly. The requirements are similar but not identical as regards transformers, but I'd have to do a detailed review of them to tell, and I don't have the $300-$400 needed to buy a copy of an IEC standard.

But I need to stop typing. The coffee's ready.

Thanks for the post. Re your ss relay, what is the sense transistor checking?