Well, now I got magic smoke from the led string near the DC IN. There was a loose piece of wire underneath the raw psu, it must have shorted something. No more light and the 2W resistor got very hot. Disconnected everything and I'm done for today.
Tomorrow I'll start by replacing the LED's and check if Q2X is still working and after that check Q3X.
Tomorrow I'll start by replacing the LED's and check if Q2X is still working and after that check Q3X.
Last edited:
allowing for a nominal 7Vdrop across the regulator, then a 42Vdc allows upto 35Vdc at the output.
But 42Vdc Input is very high for a 20Vdc regulated output.
The Vdrop will be 22V and that generates a lot of heat.
The CCS mosFET has to dissipate most of that heat.
Have you killed the input mosFET?
One of the tests I carry out is to short the output (0volts output) and using a Variac increase the input voltage all the way to 253Vac (the UK's maximum supply voltage) and check that even though the CCS gets absurdly hot, it survives.
When the output short is removed the circuit returns to normal operation instantly.
For me this is a basic test of survivablity.
But 42Vdc Input is very high for a 20Vdc regulated output.
The Vdrop will be 22V and that generates a lot of heat.
The CCS mosFET has to dissipate most of that heat.
Have you killed the input mosFET?
One of the tests I carry out is to short the output (0volts output) and using a Variac increase the input voltage all the way to 253Vac (the UK's maximum supply voltage) and check that even though the CCS gets absurdly hot, it survives.
When the output short is removed the circuit returns to normal operation instantly.
For me this is a basic test of survivablity.
Yes it is, but something maybe is dragging this raw DC rail a little lower than the 44V other
20V tops is little for 9.1k R3x it would take to chance on a lowest GR range Q3x sample. Do your checks. Is Q3x barely pulling because there is no good Vbe on Q4x for instance? Less than 0.6V? Is Q5x really biasing it? What is the VGS reading on Q6x? If all semis check well then higher R3x will give higher Vout and its just a matter of Q3x IDSS. Also make sure that the voltage drop across R1x indicates about 100mA.
20V tops is little for 9.1k R3x it would take to chance on a lowest GR range Q3x sample. Do your checks. Is Q3x barely pulling because there is no good Vbe on Q4x for instance? Less than 0.6V? Is Q5x really biasing it? What is the VGS reading on Q6x? If all semis check well then higher R3x will give higher Vout and its just a matter of Q3x IDSS. Also make sure that the voltage drop across R1x indicates about 100mA.
I'll need to read that a few times. I'll try to compare readings between the good board and the bad one to see if I can find where is the faulty part.
I'm at work right now, I'll do that in a few hours.
I did some resistance measurements to check for continuity and one thing I noticed was that on the working board I got continuity from Q6x heatsink to ground. The bad board didn't show this continuity.
Q6x drain is indeed supposed to be at ground. Its sink should also show ground when there is conductivity between that MOSFET's tab (drain connected inside) and the sink metal. It does not need insulation for this circuit's purposes and the heat transfer is better without insulation when not needed. On the other hand Q1x's sink should be in continuity with Vout Rail +. If the screws are plastic and the anodizing perfect at the mating area maybe it prevents continuity but its indifferent to have continuity or not to the sinks in this design.
