Give it at least an hour, then you can see how big a fuse youll need without the therm. The problem with turn on surge is it varies with the instantanious position of the line sinewave when you switch on, it can be as high as 100A for a fraction of a second. Youll still have trouble in the long run without the CL60s.
Bill
Bill
Babowana said:I think that the content in post #2970 is incorrect regarding the voltage divider because there are other resistors than 1K and 2.6K . . .
Anyhow, the 1K is considered too big because there will be voltage drop across 1K when current flows through it.
Simplified sketch is drawn to help my poor head with a simple picture, and attached here. . .
Cheers,
>><<
I think you have it! Thanks for the input!!! I used jumpers to put a 10R in parallel
with the 1K, tested, and got voltages in the range expected. During steady state,
the voltage drop across the 9.9R pair is in the 150 mV range. This means the
current passing through this part is in the 15 mA range. Which means (again,
for steady state) a 1/4 W resistor will do...
But I have to ask if a resistor is required here at all. I think it may not be required
for the LM317. But I'm concerned about the, "power off" condition, when the coil
wants to disipate its energy, and how the 2200uF cap drains. Could we be
converting the energy in the coil and cap into heat via this resistor? And if so,
would the resistor need a higher wattage value?
I really wish I understood how the coil and 2200uF caps are handled at the
powerdown condition.
I want to thank everyone for their help, and especially Babowana for his
analysis and explanation! Way to go B-man! And a perpetual thanks to
Nelson.
Robert
Bill is right.Bill Fuss said:Nope, it can't!
You have chosen to not read the difference between soft start for the transformer and slow charge for the smoothing caps.
These are very different problems that should receive different solutions.
A large toroid needs a few cycles to start (well all transformers need time to start, but large toroids are worse in this respect than EI). During this very short initial period the toroid appears as the primary RESISTANCE against the mains voltage. Measure the primary resistance of your 500VA transformer.
254Vac (the maximum voltage from the UK mains) the peak voltage can be 254 * sqrt(2) ~ 360Vpk.
If your toroid has a primary with resistance = 3r6, then switching on the toroid at the peak of the mains waveform will instantaneously try to draw 100Apk. Fortunately the voltage drops across all the resistances from the power transformer outside your district to your toroid and back will not be zero. they will be finite and the peak current could be reduced by about half.
As each half cycle alternates between plus and minus the current flow will reduce as the flux builds up in the core of your toroid. I have not tested this (my inability) but I'm told this takes just a few cycles of the mains to achieve. This is why I suggested that the bypass delay be tried as low as 100ms (~5cycles).
Once the toroid is started then it works as a transformer and sends charging current pulses to the smoothing caps. This is the second problem. These very high current pulses can damage both the rectifier and the capacitors. If you limit the current during the charging period then the period is extended from a few tenths of a second to many seconds. That is the purpose of the second delay bypass. If a Thermistor is used in the secondary circuit. It starts out cold (if a bypass is fitted) and reduces the initial charging peaks when the cap to charging voltage is greatest. As the Thermistor warms up the smoothing cap charge has also increased and we have a smaller difference in voltage between the caps and the charging voltage and the reduced thermistor resistance lets through a near constant current. This is why the slow charge is better done with a thermistor. But it must be bypassed to ensure it is cold for the next restart and to prevent it modulating the DC supply rails voltage as current demand changes on each supply rail. Remember that most ClassA power amplifiers do not draw constant current through the supply rails.
audiorob said:
I am a resistor murderer.
It seems that I forgot to consider the startup conditions. The 1/4 W 10R
resistor (but only on the + rail) gave up its smoke. I suspect this is due
to the large inrush to charge the 2200uF caps. Maybe I should have
put the 2200uF in front of the 317. Actually, I tried to do that initially, but
the circuit was not working for me, so I tried putting it behind the 317,
and left it there.
Anyway, I replaced the 10R 1/4 W with 10R 5 W (it was all I had) and
they sustained startup. I can adjust the rails up to about 22 V now. I
could have sworn that I saw 26 V rails when I tested with the parallel
resistor on test leads earlier (again, my unregulated supply is 28V).
Maybe I should decrease the value of R1 (another parallel resistor) in
the voltage divider in the adjustment section. Or maybe I could add
1 or 2K in series to R2. The good news is that the rail voltage seems
to be holding a very constant voltage. But with a 6 V dropout, it better!
Robert
audiorob said:
My mistake (again!).
I start with a 28V unregulated supply. The 317 has about a 2 point
something dropout. That gives me about 26 V to the 2200uF caps.
Then there are the V losses over the source resistors of the output
MOSFETS and the 3.5V loss of the MOSFET. That brings me to
about 22V. I think that might be as good as it is going to get...
Robert
If you use a voltage reference like a TL431, it needs 0.4 mA. That
allows a loss of about 0.5V if you use a 1K resistor from the supply
to the drive for the Mosfet and use the TL431 in parallel with a cap.
Place a diode reverse biased on the resistor that biases the TL431 so
that the Gate of the Mosfet won't see any voltage on turn-off.
There. Now you can have only 4V loss = 24V.
Put a big fat capacitor to ground from the Source of the Mosfet to
reduce the ripple due to audio currents.
allows a loss of about 0.5V if you use a 1K resistor from the supply
to the drive for the Mosfet and use the TL431 in parallel with a cap.
Place a diode reverse biased on the resistor that biases the TL431 so
that the Gate of the Mosfet won't see any voltage on turn-off.
There. Now you can have only 4V loss = 24V.
Put a big fat capacitor to ground from the Source of the Mosfet to
reduce the ripple due to audio currents.
ez soft start
Its pretty simple just use a delay on operate relay and a large wattage resistor I use the gold Dale 50 watt chassis mount types.
the relays tend to be a little pricey unless you can find one surplus.
this has the advantage that it will restart immediately, unlike the thermister that must cool off before you get the soft start effect again. Oh yeah use the normally open contacts on the delay relay , and set time for 10-20 seconds.
korben69 said:amp-guy, do you have links, values or schemas for this ?
Its pretty simple just use a delay on operate relay and a large wattage resistor I use the gold Dale 50 watt chassis mount types.
the relays tend to be a little pricey unless you can find one surplus.
this has the advantage that it will restart immediately, unlike the thermister that must cool off before you get the soft start effect again. Oh yeah use the normally open contacts on the delay relay , and set time for 10-20 seconds.
How about using the following circuit which has slow start, DC protection and even temperature sensing can be added: http://www.diyaudio.com/forums/showthread.php?threadid=6654