The Beast gradually returns to life, thanks to Jan's generosity who replaced my melted AMP2.
The board is populated and as I now have hold of and inductance meter I checked the value of the inductors I had for playing with. Jan’s recommendations for the standard iron powder cores is to wind 29 turns of wire for 11µH, which I did and I measured 12µH, spot on ! Then I wound an MPP core for comparison, and it needed only 11 turns of wire to get the same inductance. So I will install those on my 3rd (in 3 months !) AMP2 : the lesser the resistance the better.
Before installing the board I carried out the first tests of the power supply now equipped with bleeder resistors (50W, 250 ohms) that I installed in order to gently discharge the capacitors at power off and to smooth possible fast voltage variations. The resistors are wired between positive / negative rails and the 0V center tap. Purpose of this setup is multiple : soften power up / power off transients and achieve a permanent full load to the power supply, thus eliminating possible over voltage / over current peaks when the power demand is little.
As you can see, positive and negative rails show perfect matching, 2.0mV of variation is guaranteeing an extremely quiet power block. Voltage is slightly lower than before, it was 59.0 volts in the previous (but defunct) AMP2 due to the consumption of the bleeders. This in turn will allow safer operation for the 2350 chip I guess.
Power supply returns back from 58 volts to 0.5volts within 56 seconds and does it in a perfectly symmetrical way (on my previous amp I checked once the rails after shutting off the power and there was a HUGE difference between + and - rails). Here we have the same decay curve on both rails, which the circuitry it feeds will certainly like.
Taking into account the size of the heat sink and dissipated power in the resistors heat is reasonable. Obviously, good thermal compound is interfacing the resistors with the heat sink.
This time I am taking no risk at all from a calorific point of view as I don’t want to see this again http://www.dcx2496.fr/melt.htm !
You will see on the attached pics that I tried to foresee the worst operating conditions : a BIG heat sink is glued under the chip by the means of a heat conducting kapton foil from Fischer Elektronik (Farnell order code 1211708). The heat slug of the TP2350 chip is interfaced with the board with Dow Corning 340 compound. As an additional safety measure a fan will be blowing cool air taken from the outside of the cabinet onto this heat sink. T
his 12V fan, ex from a Pentium CPU cooler is powered by an external 6.0V power supply - for lower noise - which also unmutes the amplifier by the means of a delayed relay.
I hope this third AMP2 will last forever as I am convinced it is playing amongst the best amps money can buy. I'll keep you posted after first power on, if I'm still alive of course [^]
Negative rail voltage : http://www.dcx2496.fr/3rd/V_m.jpg
Positive rail voltage : http://www.dcx2496.fr/3rd/V_p.jpg
Difference : 2.0 millivolts, nice ! Note the bleeders on the heat sink. Lovely heater for the winter [8D]
Populated board : http://www.dcx2496.fr/3rd/amp2_3.jpg
Heat sink underneath the TP2350 : http://www.dcx2496.fr/3rd/amp2_3_b.jpg
Fan from an old Pentium CPU cooler : http://www.dcx2496.fr/3rd/ventilos.jpg it will be running with 6V DC for quiet operation
It'll cool the chip's belly heat sink located 10mm above it : http://www.dcx2496.fr/3rd/ventilo.jpg
The innards of the beast, ready for rebirth : http://www.dcx2496.fr/3rd/P1000444.jpg