Simple DIY buffer PSU

[random007]

For a LM3886 chipamp project.

Drops +/-35V from the main regulated power supply (sill in development, nearly there) down to +/-15V and +/-12V. Two of these in series will be used rather than using symmetric regulators (easier to construct this way, for one thing).

15V used by FET buffer.

12V isn't necessary at the moment but maybe used for a relay based source selection switch (if I can be bothered). Prefer to alter the switch circuit to work with +/-15V rather than use this extra regulator but I'm still massaging overall design details.

Any comments, suggestions or corrections welcome.

[AndrewT]

Hi,
r1, 2 & 3 are shown as 1W, why?
Have you calculated the currents and the powers through these resistors?
why have you shown 1k2 & 100r, when 1k3 can do the job?


U1 has 20V across it.
U2 has 3V across it.
What does the datasheet sheet say about this?

What is the range of input voltage?
What is the range of ripple voltage on that unregulated DC supply?

[jitter]

I too wonder why you're meaning to use 1W resistors, since the current running through R1...R3 is low enough to use 0.25 W resistors.

What I also wonder about is why you chose 120R for R1 when 240R would be fine to get stable operation. With R1=240 and R2=2k6 you will meet the minimum load current requirement of 3.5-5 mA (even without load) while cutting the power dissipated in R1...R3 in half.

What is the current you're going to draw from the 15V (including what the 12V circuit would draw if used)? With a drop of 20 V across the LM317T you would need heatsink above 100 mA current at an ambient temp of 25C.

I have no problems with the 12V setup, Vin-Vout is sufficient (usually needs to be at least 2-2.5V), but I would add 100nF capacitors between Vin-gnd and Vout-gnd as close to the pins as possible to supress any noise from the LM7812.

[random007]

Originally Posted by AndrewT

Hi,
r1, 2 & 3 are shown as 1W, why?

Have you calculated the currents and the powers through these resistors?
why have you shown 1k2 & 100r, when 1k3 can do the job?

U1 has 20V across it.
U2 has 3V across it.
What does the datasheet sheet say about this?

What is the range of input voltage?
What is the range of ripple voltage on that unregulated DC supply?

I checked U1 and U2. U2's minimum voltage is 2V so it's close but U1 can have up to 40V differential.

The resistor power ratings: was blindly guessing because I don't know what the power requirements of the buffer are (circuit here STEREO FET BUFFER FOR DAC, PREAMPLIFIER, TAPE REC, ... (eBay item 380167086814 end time 06-Nov-10 03:41:04 AEDST) : Computers Networking). Will look at the maximums of the semiconductors to get an idea - any more accurate assessment is beyond my ability.

Thanks

[random007]

Originally Posted by jitter

I too wonder why you're meaning to use 1W resistors, since the current running through R1...R3 is low enough to use 0.25 W resistors.

What I also wonder about is why you chose 120R for R1 when 240R would be fine to get stable operation. With R1=240 and R2=2k6 you will meet the minimum load current requirement of 3.5-5 mA (even without load) while cutting the power dissipated in R1...R3 in half.

What is the current you're going to draw from the 15V (including what the 12V circuit would draw if used)? With a drop of 20 V across the LM317T you would need heatsink above 100 mA current at an ambient temp of 25C.

I have no problems with the 12V setup, Vin-Vout is sufficient (usually needs to be at least 2-2.5V), but I would add 100nF capacitors between Vin-gnd and Vout-gnd as close to the pins as possible to supress any noise from the LM7812.

See reply above too.

Two resistors were used in series because none of the local electronic retailers stock the correct value at 1W. The problem goes away with smaller rated resistors.

I was reading up on good 317 design and 120R was recommended for better regulation. It was a while ago and I neither recall where I saw it or what reason was given. I'll try to find it. In any case, I'm more than happy to go 240R (as per datasheet, I notice) and use 1/4W resistors.

Thanks for your help...

[AndrewT]

the 120r or 240r option for the upper voltage setting resistor is chosen to guarantee a minimum output current for the regulator if the client circuit draws near zero current during part of it's normal operations.

The "real" lm317 has a minimum draw of 5mA.
The "real" lm337 has a minimum draw of 10mA.

I have no idea what the undocumented copies and fakes have as a minimum current draw.

It seems wise, to me, to always adopt a guaranteed minimum draw of >=10mA for all undocumented 3pin regs. That demands <=120r for the 1.25V that is the output of the 3pin regulator.

Now that the resistor has been chosen one can determine the current flowing to ground through the voltage setting resistor string. Use I^2 * R to find operating power.

[AndrewT]

Originally Posted by jitter

..........With R1=240 and R2=2k6 ......

where did you buy your 2k6 resistors?

[random007]

Originally Posted by AndrewT

where did you buy your 2k6 resistors?

When going shopping for parts today, I assumed it was a typo. 2k7 gets 15.3V while 2k6 gets 14.7V.

---

Revision.

• Resistors changed.
• Read a little about tantalum capacitor reliability issues (like an often closed-circuit mode of failure), so dropped those for low-ESR electrolytics. Larger value than before but I have them so may as well use them as their value isn't critical.

Will probably move the connection pins to have 5mm spacing so screw terminal blocks can be used.

Tested component PCB footprint with polystyrene foam (pictured). Looks fine apart from over-sizing the outer diameter of the 470uF capacitor by a few mm.

[AndrewT]

Hi,
low ESR caps across the output of 3pin regs may not be suitable.
See what the datasheets say about caps at input and output.

[stormsonic]

or HERE