Upupa Epops said:
when the amplifier clips, or goes into a mode in which the thermal protection is cycling on and off quickly, the output harmonic spectrum far exceeds the audio spectrum -- so the 100nF or 330nF starts to do its job. you could view it by measuring the voltage drop across the 1R resistor.
the thermal protection can cycle very quickly between 165 and 155 die temperature -- if you have a big heat sink -- this is for conditions which violate the SOA conditions for the chip -- you can see it on a scope. when the thermal protection kicks in the current drawn by the chip also changes -- it oscillates!
my belief is this --- there is a need for the 100nF decoupling snubber on the V+/V-, as well as on the output of the amplifier. I don't think the minimalist approach works if you are pushing the ampchip to its max.
the situation is probably worse still for the LM3886TF with its insulated case -- the TF can not be pushed as hard as the non-insulated versions. the thermal resistance of the insulation is simply too high. I don't think that there is any reason to use the TF in a DIY amp, but I can think of reasons why JBL or someone would use it in a consumer product.
i guess I prepare for the "worst-case" scenario --
Upupa Epops said:
you aren't trying hard enough: here's a snap of the bottom side of my "dummyload" -- 5 40 ohm, 40 watt precision power resistors in parallel -- and you can fry an egg on it ! -- or melt the scotch tape dispenser as was inadvertently done here.
An externally hosted image should be here but it was not working when we last tested it.
fwiw, I use a 1 watt resistor on the output snubber (I prefer to call it a zobel).
the diode snubbers I use -- really depends on the transformer you rae using.
I will say that I seem to hear improvements at all listening levels, not just when it's loud. In fact I've not taken it very loud at all yet... just on the verge of p;ss;ng other people in the house off, so 'moderate'. Only medium term listening will tell if what I hear is in my mind or not, I don't think it is at this point, the improvement is simply too great.jackinnj said:
Was an ordinary heater not enough for this room?jackinnj said:
Is 1w sensible for the zobel then, or simply what was to hand?
Is 1w sensible for the smoothing cap snubber?
What values and ratings do you use on your bridge snubbers? Can this snubber be done to good effect without measurement?
Do you use a cap across the transformer secondaries, too?
Thanks!
Upupa Epops said:
the one on the power supply is called a decoupling capacitor -- and it will pump a lot of current if the amplifier goes into the extreme conditions I described -- you can demonstrate it by overdriving the amplifier.
the energy stored in the snubber or decoupling capacitor is dissipated by the resistor, whether that resistor is on the power supply decoupling capacitor, or on the output of the amplifier. the energy dissipated by the resistor is:
P = C* f *V
.
peranders said:
I did check the first schematic -- they are all 'snubbing" in a manner --
the points I am making are in respect of the 100nF on the power supply AND the 100nF/2R7 output to ground on the ampchip is under conditions of stress.
when the thermal protection kicks in, or when the amp is driven to clipping it can cycle very quickly -- when I stress tested one of my amplifiers on a big heat sink with forced air cooling you could hear the thing squeal when it was pushed beyond the SOA. -- the heat sink was doing its job unmercifully as the die of the chip went into thermal shock, shut down and then recovered.
as I said before, I replaced some "charred" snubber resistors on the output -- they hadn't gone completely wonky but they clearly showed that they were cooking.
I have also observed a latency period for the protection kicking in -- when the chip is making up its mind the distortion kicks up to over 1% pretty steadily -- I wouldn't recommend this as general practice -- it's like burning out a clutch on purpose.
Don't forget that you don't know really why Mr. Otala has chosen the parts that he has done. Just becuase he has 1 W doesn't imply that it's really needed.
I have attached my snubber simulation which maybe has simplifications. You can see for yourself how small the current is through the snubber.
I have attached my snubber simulation which maybe has simplifications. You can see for yourself how small the current is through the snubber.
My first claim is that this rather small peak in the impedance curve is of minor importance.
My second claim is that this alone C2, C7 should be removed.
So anybody who agrees with me? At the moment we are not talking about how the parts should be chosen when it comes to values and ratings.
It looks like this snubber resistor could be rather small in wattage rating.
My second claim is that this alone C2, C7 should be removed.
So anybody who agrees with me? At the moment we are not talking about how the parts should be chosen when it comes to values and ratings.
It looks like this snubber resistor could be rather small in wattage rating.
PA --
I use a 1 milliohm resistor as a current shunt in Multisim8.
Put a pulse signal between the sine source and the capacitors -- this will simulate the boundary conditions in which the amplifier quickly turns on and off -- this is what you will observe on an oscilloscope -- there is an "envelope" of energy -- if the amplifier is going into clipping the pulse is pretty small, but very fast -- try 1 V at 10kHz. -- if the amplifier is at the border of going into thermal protection the pulse is much larger, but much slower.
i agree that in a steady state, when the amplifier isn't driven beyond its SOA, the snubber resistor doesn't see a lot of current, but in the conditions I described -- where the amplifier is shutting itself down because it is overdriven, thermally stressed etc.
why bother to fuse the power supply if you wont spare the expense of a 1 watt resistor.
and from my experience -- the 1/4 watt resistors were charred.
I use a 1 milliohm resistor as a current shunt in Multisim8.
Put a pulse signal between the sine source and the capacitors -- this will simulate the boundary conditions in which the amplifier quickly turns on and off -- this is what you will observe on an oscilloscope -- there is an "envelope" of energy -- if the amplifier is going into clipping the pulse is pretty small, but very fast -- try 1 V at 10kHz. -- if the amplifier is at the border of going into thermal protection the pulse is much larger, but much slower.
i agree that in a steady state, when the amplifier isn't driven beyond its SOA, the snubber resistor doesn't see a lot of current, but in the conditions I described -- where the amplifier is shutting itself down because it is overdriven, thermally stressed etc.
why bother to fuse the power supply if you wont spare the expense of a 1 watt resistor.
and from my experience -- the 1/4 watt resistors were charred.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.