It's too hot today outside, so I relocated from my barn to a living room.
The space between tube unit and a power supply will be occupied by a control unit that will contain own PS, 3 electromagnetic relays: for transistor PS power control, and for speaker protection, and as well servos and speaker protection networks.
The space between tube unit and a power supply will be occupied by a control unit that will contain own PS, 3 electromagnetic relays: for transistor PS power control, and for speaker protection, and as well servos and speaker protection networks.
Here is what I am going to use as a servo and speaker protection.
Speaker protection is triggered by clipping of servo, i.e. when there is some voltage greater than 0.3V on it's inverting input.
Any comments/suggestions?
Speaker protection is triggered by clipping of servo, i.e. when there is some voltage greater than 0.3V on it's inverting input.
Any comments/suggestions?
Corrected a bit, to protect inputs of opamps from possible overvoltage in case of damaged output stage. Protection must be reliable and self-protected, right?
Are you sure that the servo output as a trigger for protection is fast enough? There is a conflict here: you want the protection to react fast, but the servo to react slow.
jd
jd
janneman said:
Yes, when an output transistor shorts, how long does it take the circuit
to decide it's not the cannon fire in the 1812 and trip; and is that fast
enough to protect the speakers?
Michael
Let's check.
A first, let me repeat for the beginning, the servo output does not trigger protection. A voltage on integrator's inverting input triggers it. In the normal mode this voltage is always close to zero, and changes abruptly only and only when the integrator's output sticks to one of rails, i.e. when the integrator clips a voltage on it's inverting input changes enough to open one of diodes biased by 22 MOhm/750K voltage dividers from 12V sources, i.e. each of diodes is under-biased by 0.39 Volts on each.
In a normal mode the servo works as an integrator, but when something wrong happens an output voltage abruptly goes to +50 or -50. An integrator clips and a voltage on it's inverting input goes up.
Let's check time of the reaction.
1 mOhm, 50V means 50 microam current that in the normal mode goes to an input of an integrator. Time to clip from zero volt to one of opamp's rails would be 12V * 10^-6 F / 50 *10^-6 amp = 12/50 seconds, i.e. 0.24 second. When it clips it is no longer an integrator, and a voltage on it's inverting input starts going up, and 1V shift (let's assume that we need 1V to open a diode and trigger the comparator) happens during 1/50 seconds.
That means, we allow 1/4 seconds for servo to react (it's a cannon fire in Glinka's 1812), and 1/50 seconds only when servo can't correct the situation (it's no longer a cannon fire).
Is it too much?
A first, let me repeat for the beginning, the servo output does not trigger protection. A voltage on integrator's inverting input triggers it. In the normal mode this voltage is always close to zero, and changes abruptly only and only when the integrator's output sticks to one of rails, i.e. when the integrator clips a voltage on it's inverting input changes enough to open one of diodes biased by 22 MOhm/750K voltage dividers from 12V sources, i.e. each of diodes is under-biased by 0.39 Volts on each.
In a normal mode the servo works as an integrator, but when something wrong happens an output voltage abruptly goes to +50 or -50. An integrator clips and a voltage on it's inverting input goes up.
Let's check time of the reaction.
1 mOhm, 50V means 50 microam current that in the normal mode goes to an input of an integrator. Time to clip from zero volt to one of opamp's rails would be 12V * 10^-6 F / 50 *10^-6 amp = 12/50 seconds, i.e. 0.24 second. When it clips it is no longer an integrator, and a voltage on it's inverting input starts going up, and 1V shift (let's assume that we need 1V to open a diode and trigger the comparator) happens during 1/50 seconds.
That means, we allow 1/4 seconds for servo to react (it's a cannon fire in Glinka's 1812), and 1/50 seconds only when servo can't correct the situation (it's no longer a cannon fire).
Is it too much?
Wavebourn said:
I reduced R1 to 240K value. Now 1/16 seconds for normal servo operation, and 1/200 seconds for an emergency case.
Is it better?
I took some time off to make a new front panel for Marantz-7200 since it's DSP board that I used for home theater surround sound died. Let it drive woofers and subwoofer now.
Now I know why I took a time off... Nelson Pass reported today that he started shipping a production with Silicone Carbide output JFETs. I did not know before now that they were already available: http://www.semisouth.com/products/uploads/DS_SJEP120R063_rev1.3.1.pdf
Anyway, it will not be hard to switch from silicone MOSFETs to silicone carbide JFETs; that will mean almost the same design, but with much better sound quality.
Anyway, it will not be hard to switch from silicone MOSFETs to silicone carbide JFETs; that will mean almost the same design, but with much better sound quality.
I promised to continue tonight, but I am not ready yet: the topology I made for the servo+protection PCB I don't like. I am going to redraw it tomorrow more optimally. It's like a poetry: sometimes it goes easy, sometimes does not. I don't like a bad poetry.
Going to order http://www.j-tron.com/product/2_lb._FeCl_Etchant_Powder.html -- mine exausted...
Meanwhile, making wooden body, to match speakers. The same palm tree grill, the same hardwood look...
Meanwhile, making wooden body, to match speakers. The same palm tree grill, the same hardwood look...
Thermal properties of the material that means less of factors of variability of parameters on working temperatures. Dirty currents, effective geometry modulations were significant factors that drove switch from Germanium to Silicon. The same leap means switching from Silicon to Silicon Carbide; the next leap may be switching to Diamonds. Can you imagine devices with 100 GHz Ft, 1,000W dissipation, 100 Amperes/Volt transconductance, and 1 picoampere of dirty currents, in an 1/4 inch bolt screwed into a threaded hole in a heatsink?
Would they actually need a normal sized heatsink? I can imagine them glowing intensely at 600C and not giving up the spirit if pure diamond.
Would they actually need a normal sized heatsink? I can imagine them glowing intensely at 600C and not giving up the spirit if pure diamond.
Yes, I dream of such devices inside of tube cathodes, such a powerful hybrid tube would not need any additional heater supply.
Can you imagine a single tube power amp, very powerful one? Something like zero dB input, 500W output.
To a brave new audio world.
I don't like this hole. May be it is a good idea to cover it by painted Lexan with small windows in front of indicators?
