Power Supply Design

[whitelabrat]

I'm looking to build a power supply for an opamp based phono amp with a 15V+, COM, 15V- input. About 12V would be fine for the job. Right now I'm using two 9V batteries, but it's not a good long term solution.

I've seen a lot of power designs that do 15V, but without the COM, so here's what I'm thinking, borrowing some ideas from RJM's VSPS project:

Amveco TE70072 transformer
dual bridge rectifiers
For each bridge's output:
An LM317 and LM337
Values set with 220ohm and 880ohm resistors
two 480uF caps
two 100uF caps

This should give me dual 6.25V ouputs that can be combined for roughly +-12.5V and a COM.

Perhaps there is a way I can do this minus the dual bridges and such to lower the parts involved?

[theChris]

transformer specs?

[whitelabrat]

The transformer takes 115V or 230V from the wall and is rated for 35Vac to dual 12V outputs (14V no load). 1458 mA which should be plenty for a OPA2134 or dual OPA27 or OPA37 opamps.

I will likely include something in the circuit to help load the power supply as well. I'll draw up a schematic soon.

[sam9]

Look at the Bugle schematic at http://www.hagtech.com/bugle.html. If you like it you could by the kit, very reasonable.

This is very neat package. Using Schottky rectifiers removes one source of PS noise right from the start.

[whitelabrat]

Tangentsoft has a nice kit too.

http://tangentsoft.net/elec/teps/

Thing is, I've already got all the parts. I figure I'll go down the kit route when all hope is lost.

[theChris]

you can connect the 12V secondaries in series, giving a 24V-CT design. addition of a full-wave bridge will give 16V or so unloaded, with a possiblity of 22V with high line conditions, and maybe 14V under low line. use 35V or 50V rated caps. target ripple current of 500mA @ 120hz if possible.

the lm317 and 337 might be set for 10V outputs. use of 2200uF capacitance would allow for suitable ripple performance with a 5W load on the outputs. the LM317 337 should be heatsinked with small heatsinks that keep the temperature rise low enough when the parts are run at 4W. keep in mind that the metal tab is connected to a pin on the device!

this commentary is less applicable if you want lower output voltages, lower output power, ect...

addition of ceramic or film cap of around 100nF to input/output may be useful.

you can use 1/2W bias resistors of smaller then normal values to provide a load. dissipation should be 200mW per device or less. alternately, a parallel combo of 1/4W resistors could be used.

diodes should be rated for 10A or higher non-repetitive surge, with a 3A or better average rating, and with at a 100PIV rating.

keep in mind all this is top of head estimation, and some numbers are doubled or such for margin. This produces a conservative design.


oh, and don't forget the fuse! use something like a slow blow 250mA fuse for US operation, smaller for higher voltage operation.

[whitelabrat]

You read my mind TheChris.

A gentlemen from China suggested that I use a 0.01uF cap between each of the two transformer outputs and four 0.01uF caps between the LM317 and LM337 outputs. Finally two 100Kohm resistors on the output of the power supply to put a small load on things. He also suggested using a choke instead of the resistors, but I'm not sure where that would work.

Thanks ya'll. I think my PS idea will work!

I hope to have a test in place in a week or two!

[theChris]

this form of simple supply is notoriously bad for power-factor. the current is drawn in high-currrent pulses, once per cycle. this makes the RMS current higher then expected. since the transformer is rated in VA, it is useful to oversize the transformer, which should be fine in your case.

this topology also has inrush problems -- if the switch is closed near peak voltage, the current flow can be excessive while it tries to quickly charge the capacitor. this sudden inrush can damage diodes and damage fast-blow fuses.

the addition of series resistance or inductance combats these effects. in your case, i had estimated the series resistance as over 1ohm (refered to secondary), in which case the current peaks would be low.

the inductance can also reduce EMI/RFI noise as the peak current and rate of rise of current are reduced. the low power and low switching rate (120hz) makes these less important than in hard-switched, high-frequency supplies.

[zlast]

http://sound.westhost.com/projects-4.htm