A GFCI will monitor the current balance between hot and neutral on the mains, i.e., the current flowing through the primary winding of the mains transformer. As far as I can tell, a GFCI can't figure out if things go wrong on the secondary side of the transformer. If the rectified B+ voltage accidentally finds a way towards ground (not necessarily the PE wire), how would the GFCI know? What am I missing?
I was not referring to a specific design, but since we're at it: I am working on a tube amp for electrostatic headphones. It's conceptually the same as most other tube estat amps: bipolar output from two tubes working in push-pull. The output to the headphones is taken from tube anodes (or from a mu stage on top of the anodes). As usual with such designs, there is no cap or output transformer. Stax does that, and their amps have a CE mark, so it can't be outrageously wrong. Most newer designs have 5.1k "safety resistors" in their output, which limit the current and pull down the output voltage in case of a fault.
What do you mean by "regular rules"?
Last edited:
Nothing.
I was simply showing that a fuse is never protection enough and that in certain cases other means had to be created.
IF an output capacitor shorts, user is hurt or dead, period.
If you want to protect him from DC, then design a circuit which monitors DC and opens a relay, it is very doable.
I invite you to do that
Can you please show the OTL OCL stage?
So as to analyze it better
Not Turbo Watch but guess he means making a design which is safe, hard to fail, if it fails it won't murder the user, etc.
Personally I would NOT connect earphones which are hanging from my skull, pressing my ears, to any device which can fail and apply hundreds of Volts there.
Never never ever ever.
I couldn't care less even if it had the best sound in the World.
What's the idea behind avoiding galvanic isolation?
Not at that price, for sure.
Don't relay contacts only take a few milliseconds to activate?
This is the current sensing relay I'm referring to:
This is the current sensing relay I'm referring to:
Here are a few examples (they all use the same circuit concept for the output stage):
- Stax tube amp (before they used pull-down "safety" resistors):
https://4.bp.blogspot.com/-IvN0qPJPBc4/VqSLbO71pzI/AAAAAAAAEDU/337Qm8rhOy4/s1600/stax_srm-t1.jpg
- The Blue Hawaii, the grand daddy of CCS loaded tube amps for estat headphones (also no "safety" resistors"):
https://headwizememorial.wordpress....static-amplifier-for-stax-omega-ii-headphones
- Schematic of the "mini T2" (attached), a newer design using the 5.1k "safety resistors" in the output
- aX article on modifying old Stax amps to add a CCS and "safety" resistors (attached, page 48 etc.)
- The OSDEHA amp I am drawing up at the moment https://www.diyaudio.com/community/threads/open-source-dht-estat-headphone-amp-osdeha.407679/
Please read the rules.
Not allowed:
Posts concerning Moderator’s directives or decisions, including reposting of such actions. Please use the Private Message or Report Post button to contact Moderators if you have comments or questions.
Hugo
Moderation team
Not allowed:
Posts concerning Moderator’s directives or decisions, including reposting of such actions. Please use the Private Message or Report Post button to contact Moderators if you have comments or questions.
Hugo
Moderation team
Fuses are there to prevent wires bursting into flames(*), not to protect speakers or amplifier output sections. A speaker coil acts as a fuse for the speaker cable, as can the output transistors of the amplifier (!) - what you want is protection circuits to protect the speaker coil from excessive DC and to protect the amplifier (current or SOA limiting). A fuse cannot discriminate between AC and DC, nor can it blow fast enough for SOA protection.
Fuses can be added to an amp output, but for ultra low distortion they should be within the feedback loop as they show thermal distortion at high output levels. However you really need to compare that to the thermal distortion of the speaker coils to figure out if its worth it(+). There's a chance a fuse can save an amp that locks to one of the rails, so its better than no protection, but its not really a substitute for good protection circuitry - that can make an amp able to withstand frequent output shorting events.
(*) Very dramatic and scary - I've seen a wire blow smoke out of a regular set of holes every few cm along its length before the insulation melted away!
(+) You might want those ultra-low distortion measurements on the datasheet, for instance...
Is your 500 V output voltage peak or the RMS value of a sine wave? Is it the differential voltage or per side?
Assuming peak and differential (peak because you also want to survive when the amplifier is driven into hard clipping):
500 V/(2 times 5.1 kohm) ~= 49 mA
Dutch residual current devices are usually designed for a guaranteed maximum trip current of 30 mA RMS, because a 30 mA shock is very unpleasant but usually not lethal (at 50 Hz, which is about the worst frequency).
With 10 kohm per side, using resistors with a sufficient voltage handling, and high enough power rating or fusable safety resistors, you can at least keep the current below 30 mA. The cut-off frequency with a 100 pF load is about 80 kHz.
Assuming peak and differential (peak because you also want to survive when the amplifier is driven into hard clipping):
500 V/(2 times 5.1 kohm) ~= 49 mA
Dutch residual current devices are usually designed for a guaranteed maximum trip current of 30 mA RMS, because a 30 mA shock is very unpleasant but usually not lethal (at 50 Hz, which is about the worst frequency).
With 10 kohm per side, using resistors with a sufficient voltage handling, and high enough power rating or fusable safety resistors, you can at least keep the current below 30 mA. The cut-off frequency with a 100 pF load is about 80 kHz.
I quoted the above post because in my so humble opinion it is the best advice given so far about the topic and before anyone else takes priority over it. 🙄
I wanted to highlight it because I guess (but I could be wrong, of course!) that the OP didn't realize what exactly "galvaniv isolation" means in his context request and since I've an isolation transformer that has worked perfectly for this purpose for years I thought that highlighting it was the most useful thing to do do for the OP's needs.
This way he may save both goat (his life) and cabbage (the SQ).
I would like to add that it would be a better choice an isolation transformer (about 1000W) with grain oriented laminations. 😉
Are you talking about galvanic isolation in the mains supply or the amplifier output?
I don't think a mains isolation transformer would protect the user from faults with rectified high voltage DC showing up on the amp output. Correct?
Last edited:
In my own amp I am looking at max. 440 V peak per side. Other estat amps will be different.
I am mostly concerned about failures that cause the B+ or B- to show up on the output as DC. In my amp B+ is 440 VDC and B- is -440 VDC. Again, other amps will be different.
Do I understand correctly that you assumed a zero body resistance for this?
Too sleepy now (5 AM here), later will check those schematics, thanks for posting them.
The idea being to check the *actual" danger posible, before suggesting a fuse to hopefully protect from it.
In any case, I find fuses poor protection against shock (to put it mildly) but hey, let's check those schematics 😄
With an awake brain, of course 🥴
A fuse may not be very helpful here. They are crude things that melt (blow) only in high current situations, when they dissipate much more energy than their rating. They don't protect against over-voltage, only when enough excess current actually flows, for quite some time, which can be far too long for humans to survive. It takes 'much' more energy than their rating, which is given as the current they should NOT routinely melt at (not what they will melt at) and even fast-blow types melt far too slowly.
So the only situation in which a fuse would help is where something else (not the human) transmitted the energy. E.g. a fault when the human wasn't present.
Protection from high voltage should be adequate insulation (to prevent things reaching the human) or earthed conductive cases/shields (to sink the energy instead of the human), or floating the voltage so there is no differential that the human can bridge. That latter might be problematic with earphones though; if failures caused a differential between left and right ears, a reference to ground/earth wouldn't be necessary for current to flow. Possibly a 'very' high series impedence would also work, by becoming a voltage divider should the output see low resistance, though not many audio output applications are suited to high circuit impedence.
So IMO insulation/isolation between humans and any high voltage is the safest answer. And that is important when the objective is safety.
So the only situation in which a fuse would help is where something else (not the human) transmitted the energy. E.g. a fault when the human wasn't present.
Protection from high voltage should be adequate insulation (to prevent things reaching the human) or earthed conductive cases/shields (to sink the energy instead of the human), or floating the voltage so there is no differential that the human can bridge. That latter might be problematic with earphones though; if failures caused a differential between left and right ears, a reference to ground/earth wouldn't be necessary for current to flow. Possibly a 'very' high series impedence would also work, by becoming a voltage divider should the output see low resistance, though not many audio output applications are suited to high circuit impedence.
So IMO insulation/isolation between humans and any high voltage is the safest answer. And that is important when the objective is safety.
Last edited: