OpenServo.com

[guru]

Hey guys,

Perhaps you can shed light on my audio front end amps to a burr brown PCM3008 PCM codec.

I am trying to decide if I should use single or dual supply op amps for my differential input isolation amp. I would like really good noise cancellation on long cable runs (line level). So I designed a 3 op amp Instrumentation Amp.

I have positive 2.5, 3.3v & 1.2v supplied. I can probably offset both inputs for a single suppy design by using a cap and resistor divider on the input and a fet to offset the signal ground input by .6 v. (Otherwise noise on the signal ground that dips below 0v would get truncated by the input op amp. My spice model reflects this.)

I could make or find a negative supply generator from the positive supply. I only need about 80mA according to my spice model.

Your input is appreciated!

Colin

[Cliff]

Hi Colin,

On your long cable, can you be sure your noise will be less than .6V? I would probably use a dual supply, to insure noise rejection over a wider range.

If you want to try your hand at an inverting DC-DC converter, I have had good results using this guy. For your application, I would use the 2.2MHz. part to build a 3.3V to -3.3V supply, as the L's and C's can be small and cheap, and still give you a nice clean supply. The LT1931A will give you more current than you need, but it won't generate the low frequency noise seen in charge pump designs.

The datasheet has a good application section and LT's SwitcherCAD makes it easy to simulate the design using different components.

Cliff

[guru]

Thanks cliff. I had a feeling you would answer my question.

I read the datasheet. This is exactly what I was looking for in my
negative supply. I simulated a charged cap inverter but I wasnt happy with the output waveforms. I have used one of these things before with good results but I couldnt remember the part#!

No, I couldnt be sure the noise would be below .6v, that's why I was doubting it as a good solution. I got the fet idea from analog devices in-amp ICs for single supplies. This is why I wasnt comfortable with single supply, but I also didnt want to overcomplicate the whole thing if dual is unjustified and thus burn extra power for a DC/DC converter (from efficiency loss).

I notice that the LTC device generates an internal ~1.2v ref voltage regardless of Vin...thus I could connect Vin directly to Vbatt instead of 3V3 and not be added the inefficiency of the 3V3 DC/DC to the inefficiency of the -3V3. Correct? Also, I would not be using current from the positive DC/DC which is powering most of my logic and IO of the fpga. Furthermore, it may add any transients from current draw change of the negative DC/DC input onto the postive supply output. How is my thinking?

I dont have an O-scope so I cant test these things for real....I have to rely on good planning, simulation and lots of rechecking the design. Man I need an O-scope!

This is the last part of my design so I will have some interesting schematics to post soon.

Thanks,
Colin

[Cliff]

Colin,