Need low cost ±15 VDC power supply design

If this chip is satisfactory other than the noise, you may try adding some filtering to the output. Put a capacitor, an inductor, and another capacitor in series (T filter). I would probably use 4.7 uF caps with a 220 uH inductor, but you may want to try several values and look at the results on the scope. Also, if you only have one (or a few) components that are sensitive to the noise, but others which are not, just put the filter on the supply for the sensitive component. You may also be getting some noise from other components in the design, such as microcontrollers (if any are present).

Let's get your requirements nailed down a bit.

Do you need isolation between the supply voltage and the output voltages?

How much average variation from +-15 volts can you accept?

Is efficiency important?

Is 25 mA the maximum input current available, or the load current on the +- 15 outputs?

I am looking for a low cost dual output DC power supply design with the following requirements:

-input voltage 4.75 vdc to 30 vdc (must be able to work throughout the range)

-output voltage +15 vdc and -15 vdc

-output current requirement 25 mA maximum (anything over this is overkill)

-relatively noise free output (ie. ideally less than 10 mV

To date we have looked at using a DC to DC converter for this application but the startup current, although brief, is way beyond the 25 mA maximum which makes it unacceptable.

We are also looking at a design that uses a Linear Technology LTC1871 chip. This device shows promise but we are experiencing severe noise problems.

A complete design or any suggestions you might have on using the LTC1871 chip to do this job would be most appreciated.

The ground is common between the supply voltage and the output voltages. Some variation from the 15 volt requirement is acceptable--probably a volt but on the underside. ie. the outputs could as low as +14, -14 vdc. The 25mA is the maximum load current on the +15 and -15 vdc outputs. In practice, it will likely be about 10mA per leg.

Efficiency is important.

If the converter must start at 4.75VDC then the starting current must be greater than: Is= (2*15*0.025)/(0.85*4.75) = 186 mA which is greater than 25mA. Does this design require any laws of physics to be compromised? A flyback topology can completely control it's start up current profile. Harry

I think the LTC 1871 could work, if you have the correct magnetics, and probably a second stage of LC filter on each output (after the feedback point).

But this one might do it with fewer parts and maybe a bit less waste.

Can you point us toward a schematic of what you have built, for us to criticize and make suggestions?

Voltage controlled circuits can also be configured to limit inrush through tailoring of the output rise-time.

As the LTC1871 has only a crude peak input limit, and no slow start, then using the voltage control loop might be the only way to do this.

The OP seems a little confused about why an unspecified DC-DC converter was 'unacceptable'.

He also didn't really specify what the 'noise problems' were with the LTC1871 - emissions, interference, unpredictable regulation or whatever.

Still no word on isolation requirements.

The 4.75-30VDC input requirement suggests that he doesn't have a practical application in mind, as this doesn't describe any typical industrial source. He may simply be reading control chip specs - always a bad sign - because it means you likely can't do it with the chip described in the data sheet he's reading. The 1871 is ruled out -

RL

Hahah-hoo-hoo-haahaa...

Okay, let me clarify:

The application is real and it is industrial. The +15, -15 VDC outputs will be used to drive an instrumentation amplifier. The 25 mA requirement on each leg previously mentioned is likely not necessary. An input current (from the excitation voltage) in the running condition of perhaps 10 to 15 mA is all that will likely be needed. The reason for the wide ranging input (4.75 to 30 vdc) is to accommodate typical industrial requirements when mating the signal conditioner to a PLC, data acquisition unit or meter. These typically run at 5, 10, 12, 15, or 24 VDC.

The design using the LTC 1871 currently has filtering on the input and output including inductors on the output. What we are seeing on the output is a ratty 7kHz signal with peak to peak amplitude of about 600 mV. The frequency increases as input voltage increases from 5 VDC until things eventually go unstable and blow out the 1871 and the accompanying mosfet.

Regarding the earlier design with DC to DC converter, the killer was the huge startup current. We were using a Burr Brown (now TI) DC to DC converter which demanded about a half amp or more of short duration start-up current--an unacceptable condition given that this new design is to replace an earlier design that has a soft start.

I've tried to attach a copy of the schematic to this response but I guess that doesn't work.

As you may have guessed by now, I'm not an electronics engineer. I'm trying to get some suggestions I can give my electronics engineer who seems to be at wit's end on this project.

snip

Hmmmm, why not let the electronics engineer post here?

martin

Tom wrote: (snip)

This implies that the voltage control loop is already unstable.

Yes, badly unstable. (snip)

This is a text only newsgroup. You could post it to alt.binaries.schematics.electronic

Or email it to me and I will post it there, for you.

The schematic (and possibly a photo of the layout) will help a lot. I now think that the filter is not at fault, but the control stability is.

Okay, I've posted a jpg of the schematic for the LTC1871 power supply to the alt.binaries.schematics.electronic newsgroup. I hope it comes thru. Outlook gave me some complaint about using HTML mode.

It showed up just fine. I put my first comment there. You need to get out of the way (no disrespect intended), and get your designer on that group, so we can get him on to a practical approach.