PCM1804 bal/unbal adc

Status
Not open for further replies.
Jump to Latest
#2
It seems a bit wasteful. The input takes a balanced signal which is then converted to unbalanced (IC3a IC3b), it is attenuated and then two more op-amps are used to convert it back to a balanced signal to feed into the ADC.
Why not use a PCM1802, which is the same ADC chip which accepts a single-ended input, thus elimintaing the need for two op-amps in the signal path?

Adrian
 
#4
Okay, howzabout this then version then? It uses the OPA1632 (I pretty much followed the application schem in the opa datasheet and filled in the blanks by way of the PCM1804 eval brd schems...

Schematic PDF

PCB Layout PDF

Cheers.

NOTE: The unbal header is for a switch - which when pressed flips the inverting input to gnd for single ended input.
 
#13
It seems a bit wasteful. The input takes a balanced signal which is then converted to unbalanced (IC3a IC3b), it is attenuated and then two more op-amps are used to convert it back to a balanced signal to feed into the ADC.
Why not use a PCM1802, which is the same ADC chip which accepts a single-ended input, thus elimintaing the need for two op-amps in the signal path?
Click to expand...

Because contrary to intution, using the 1802 will not actually reduce the number of opamps in the signal path. The 1802 has an internal single-ended to differential converter prior to its delta-sigma modulator. The 1804 does not have this; the inputs go directly to the modulator. I personally would rather have control over the quality and implementation of this single-ended to differential converter than to leave it to an integrated solution of unknown quality (see pg. 12 of '1802 datasheet). The only real reason to choose the 1802 over the 1804 is to reduce the number of external parts, and therefore cost, at the expence of a little bit of performance. In DIY, the extra opamps and their cost is of far less concern than getting high performance. In mass production, the opposite is true.

Also, again contrary to intution, reducing the differential signal to single-ended and then going back to differential might seem wasteful, but really it is rather necessary. The common-mode rejection of the opamp based differential receiver can be tuned and will almost certainly be superior to that of the differential inputs of the ADC. That's why we don't just buffer each half of the diff input and send it to the ADC directly. Common mode rejection is extremely important in differential transimission.
 
#14
Electronically, does everything look okay to yas tho?
Click to expand...
I'm concerned with the use of 1000 pF caps and 10 K resistors. This combination gives rolloff with a -3dB point of 16 kHz, which I am sure that you do not want. Also, it seems to me that using a 100 pF ca in the negative input side and a 1000 pF cap on the positive input side will make the input unbalanced, screwing up the common mode rejection of the differential inputs at higher frequencies.

The 1804 datasheet uses 1800 pF caps, but 1 K resistors. That gives a rolloff at about 90 kHz. Shoot for that figure. If you are really stuck on using 10 K resistors, then use ~ 180 pF caps. Note that bigger resistors create more noise, but are easier for the opamps to drive. You could try any/many combinations that will give a similar time constant (R*C).
 
Status
Not open for further replies.
Top Bottom