Perhaps you need less efficient speakers. 
Yah, the AD8270/8271 unity gain distortions are clearly a mistake---the 8271 THD+N graph suggests more like -110dB at the stated conditions but it looks like a sim to me and therefore may be optimistic (particularly as it's noticeably better than the +-6dB figures); it's not uncommon for Analog to do that on the assumption customers will apply margin. Linear and TI seem to do the same not infrequently (I much prefer National's practice of using actua measurements). The Burr-Brown parts I mentioned are primarily an audio lineup and Burr-Brown generally makes decent stuff. So it's fairly safe to expect the INA154 to track with the 134/2134's IMD and THD+N versus level behavior---the parts are in a performance class where I'd expect lower signal levels to be noise rather than harmonic limited and the 154 is higher spec than the 134s. (The INA149 is nice too but the high impedances needed for it to handle +-275V make it noisy for audio use.)
The AD8130 could be worth a look; the improved linearity, noise, CMRR, and PSRR over the LT6552 is a good return for the higher price and the wider voltage range makes it easier to fit into an audio application. (If TI has parts using this topology I've not found which parts selector they've hid them in.)
Yah, the AD8270/8271 unity gain distortions are clearly a mistake---the 8271 THD+N graph suggests more like -110dB at the stated conditions but it looks like a sim to me and therefore may be optimistic (particularly as it's noticeably better than the +-6dB figures); it's not uncommon for Analog to do that on the assumption customers will apply margin. Linear and TI seem to do the same not infrequently (I much prefer National's practice of using actua measurements). The Burr-Brown parts I mentioned are primarily an audio lineup and Burr-Brown generally makes decent stuff. So it's fairly safe to expect the INA154 to track with the 134/2134's IMD and THD+N versus level behavior---the parts are in a performance class where I'd expect lower signal levels to be noise rather than harmonic limited and the 154 is higher spec than the 134s. (The INA149 is nice too but the high impedances needed for it to handle +-275V make it noisy for audio use.)
The AD8130 could be worth a look; the improved linearity, noise, CMRR, and PSRR over the LT6552 is a good return for the higher price and the wider voltage range makes it easier to fit into an audio application. (If TI has parts using this topology I've not found which parts selector they've hid them in.)
Last edited:
I have samples of the AD8129 to hand ready to try in my next DAC design - I need the gain so figured I may as well go for that part. I was looking at TI for this topology earlier and also drew a complete blank Seems that the improved audio (as opposed to paper) linearity of the AD830-type parts might just be down to increased emitter degeneration in the LTPs over normal opamps. On NXP's datasheet for the TDA8945 (another obsolete one) they're showing 1k5 degen resistors in the input stage. I seem to recall the LTC6552 has a similar value yet it has individual current sources rather than a single shared one.
Seems that the improved audio (as opposed to paper) linearity of the AD830-type parts might just be down to increased emitter degeneration in the LTPs over normal opamps. On NXP's datasheet for the TDA8945 (another obsolete one) they're showing 1k5 degen resistors in the input stage. I seem to recall the LTC6552 has a similar value yet it has individual current sources rather than a single shared one.Yeah, not many options in the high CMRR line receiver/line driver space if you need more than 6dB of gain. Kind of too bad as the THAT120x series is probably the nicest bootstrapped amplifier implementation around---as far as I can tell NXP is implementing some sort of bootstrap outside of That's InGenius patent and the AD8129/30 are, while not the same circuit topology, similar in boosting common mode input impedance via feedback. AD8476 and THAT1280 are also worth a mention but it's odd how richly populated the market for ADC drivers is while there's almost no parts for precision DAC output buffering. ADC drivers and line receivers can be repurposed for this if one doesn't need much gain. But putting an RCRC or RCRCRC antialias/antiglitch filter in front of them increases both the source impedance and its mismatch in differential out DACs, reducing both gain and CMRR.
As a result I'm in the middle of some MFB sims with the LME49724/OPA1632 and 0.1% resistors but I think I'll have to build a test coupon to verify the designs---the 49724 model neglects about half of the common mode effects and the 1632 model has a suspiciously large output common mode offset. But the topologies offer the ability to easily implement different gains to match whatever rails are in use along with the ability to separate AC and DC feedback paths and tap the summing junctions for filtering. And the implementation cost, complexity, and board area is competitive with most of the alternatives.
You may be right on the emitter degeneration. I haven't done enough with discretes to have a sense of where the optimum tradeoff would lie.
As a result I'm in the middle of some MFB sims with the LME49724/OPA1632 and 0.1% resistors but I think I'll have to build a test coupon to verify the designs---the 49724 model neglects about half of the common mode effects and the 1632 model has a suspiciously large output common mode offset. But the topologies offer the ability to easily implement different gains to match whatever rails are in use along with the ability to separate AC and DC feedback paths and tap the summing junctions for filtering. And the implementation cost, complexity, and board area is competitive with most of the alternatives.
You may be right on the emitter degeneration. I haven't done enough with discretes to have a sense of where the optimum tradeoff would lie.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.