A way to get rid of bipolar zero error

[AD1865]

As PCM56,PCM1700,PCM58,AD1865,AD1862 and so on which all have the bipolar zero error. But if you use two of PCM56, one for high 15 bits data convert and one for low 15bits data convert, it can get rid of bipolar zero error.
For example, if you input data MSB is 1, for example 1101010010001010, the low 15 bits data converter will convert data 1101010010001010, and the high 15bits data converter will convert data 0000000000000000. If input data MSB is 0, for example 0101011010111110, the low 15 bits data is 1111111111111111, the high 15bits data is 0101011010111110.
And we add the two converters current to output. In this way, the MSB data of two PCM56 will not change, so we can get rid of bipolar zero error.
ragards,

[Oli]

By 'bipolar zero error' do you mean DC offset on output?

[AD1865]

when input data change 0000 to 8000 (or 8000 to 0000), it may cause more than 1 LSB error. It's not the DC offset.

[rfbrw]

Originally posted by AD1865
As PCM56,PCM1700,PCM58,AD1865,AD1862 and so on which all have the bipolar zero error. But if you use two of PCM56, one for high 15 bits data convert and one for low 15bits data convert, it can get rid of bipolar zero error.
For example, if you input data MSB is 1, for example 1101010010001010, the low 15 bits data converter will convert data 1101010010001010, and the high 15bits data converter will convert data 0000000000000000. If input data MSB is 0, for example 0101011010111110, the low 15 bits data is 1111111111111111, the high 15bits data is 0101011010111110.
And we add the two converters current to output. In this way, the MSB data of two PCM56 will not change, so we can get rid of bipolar zero error.
ragards,

The PCM63 and PCM1702 do indeed have nice datasheets. Pity they left out the control logic.

[Jocko Homo]

Even if the glue logic isn't shown.......

They explain how "sign-magnitude" works, and there is a nice block diagram showing 2 DACs. Doing just what he describes.

AD1862 has a slightly different approach.

No need to reinvent the wheel when you can buy one the right way..............

Jocko

[Oli]

Cheers AD1865,

I am interfacing AD1865 with OPA627 as IV converter ( I like the sound, although I understand others don't)

I have a 40mV DC offset on the left channel and the suggested 100k trimmer helps a little, but not entirely.

Why am I getting this offset? (The trace to the inverting input runs through the chip legs- I suspect E-field coupling could be a problem)

Is there any quick and dirty method to trim opamp to zero?

(change the value of the pot?)

Help!

[OliFilth]

Sounds like a reason to use two's complement to me.

[AD1865]

May be the opa627 self-oscillate. Power supply connections should be bypassed with good high frequency capacitors positioned close to the op amp pins. In most cases 0.1uF capacitors are adequate.




ad1865,

[Jocko Homo]

They all use 2's compliment............

Jocko

[banana]

I guess the BurrBrown's Colinear (or Sign-Magnuite) approach is now patent free.

Anyone can make their own Sign-Magnitude DAC by designing their glue logic, serial to parallel interface, and banks of discrete R2R network. Just like what MSB Platinum Link and Lavry DA2002 did.