help me understand differences between THAT line receivers for ADC

[Dimitree]

Hi everyone,
I'm trying to understand how those two approaches compare when used as ADC front ends.
I extracted the schematic from the datasheets and adapted/modified them (so they may be incorrect after I edited them).
The goal is to provide a balanced line receiver to a 5.6Vpp differential input ADC.

The first approach employs a THAT 1206 line receiver with -6db gain. R8/R9/R20/C4 need to be calculated to drop the signal a little bit more, so that 5.6Vpp output is achieved.
The second approach is using a THAT 1283 (+3dB gain) followed by a resistor attenuator that drops the signal to 2V RMS (about 5.66Vpp).

As far as I understand, THAT 1206 is a better IC than THAT 1283, since CMRR is as high as it can be under impedance mismatch on the differential input.

Other than that, I'm wondering if both are good approaches and if the schematics look correct.
Do I need buffers in front of the THAT 1206? If not, why do I need them in front of the THAT 1283?

thank you!

 

[JohnRoberts]

My understanding is that you want to drive A/D convertors from a very low impedance so the input sample and hold circuits don't cause artifacts.

JR

 

[ccaudle]

Other than that, I'm wondering if both are good approaches and if the schematics look correct.

Both are good approaches. At first glance the schematics look correct. I'm trusting your calculator on the attenuation values, I didn't double check those, but the approach is right in general.

Have you downloaded all the tech docs from ThatCorp? Demystifying Analog Circuits, Analog Secrets Your Mother Never Told You, More Analog Secrets.
The Digital Mic Pre seminar and Interfacing Digtial Mic Preamps to AD Converters are both really good and have a lot of good advice for ADC front ends even if you aren't planning on putting a mic amp as your front end.

Do I need buffers in front of the THAT 1206? If not, why do I need them in front of the THAT 1283?

Looks at the block diagrams in the datasheets for both parts.
The diagram makes it clear that the 1206 includes buffers in front of a 4-resistor diff-amp, and 1283 is just the 4-resistor diff-amp alone.
If that was all the 1206 had it would be roughly like the TI INA1650/1651 devices, but the 1206 has a common mode bootstrap circuit that increases the common mode impedance a lot in the audio range. That is what makes it less affected by imbalance between the impedance of the input signals.

 

[Bo Deadly]

The second approach is using a THAT 1283 (+3dB gain) followed by a resistor attenuator that drops the signal to 2V RMS (about 5.66Vpp).

Looks like -3dB orientation to me. But I suppose that's what you want.

Also, I'm not sure a 4580 is good for Ref. Something with lower noise might be better.

 

[Dimitree]

Thanks guys, is much clearer now.
I’m wondering which one of the two approaches adapts better to the addition of an input sensitivity control (like a pad, or a variable potentiometer) since drawn like that, both circuits are not ideal for real world input signals, without wasting ADC resolution

 

[arthurdb]

It is very easy to add a potentiometer to the first circuit, at the output of the 1206, although beware of loading effects of R12 (although that resistor can be used to your advantage, working like a 'tapering resistor', allowing you to give logarithmic response to a linear pot).

In the second circuit, the signal stays balanced throughout, so you need a balanced attenuator which will increase the complexity a little.

The circuits you show have very high input impedance. If that is not a requirement, you don't really need the buffers.

Do note that if you use a symmetrical bipolar supply to power these circuits, you will not have symmetrical headroom, since your reference is not 0V but the ADC's Vref (2.5V I'm guessing?). Not the biggest deal with the first circuit, since you are already attenuating the signal at the first stage, but with the second, that initial gain is already giving you a 3dB headroom penalty, to which you'll add that 2.5V headroom drop for positive peaks, and so, you might start clipping much sooner than you would like.

 

[jp8]

My understanding is that you want to drive A/D convertors from a very low impedance so the input sample and hold circuits don't cause artifacts.

JR

That was my understanding too. An ADC driver with in-the-loop compensated op-amps driving the ADC input LPF capacitor does this. Don't know how much distortion improvement can be achieved with such a circuit, but a.o. Cirrus recommend this circuit in their CS4272 datasheet.

CS4272 | Cirrus Logic (Figure 13 in datasheet)

https://www.analog.com/en/analog-dialogue/articles/ask-the-applications-engineer-25.html

 

[jp8]

Looks like -3dB orientation to me. But I suppose that's what you want.

Also, I'm not sure a 4580 is good for Ref. Something with lower noise might be better.

Intuitively, one would indeed expect a lower noise level. But the noise from the 4580 will appear as a common mode signal to the ADC inputs. So I don't think much improvement can be expected from using a low noise op-amp.

Cheers, Jan