I tried composite TPA6120A2 with OPA1602. I need to order some OPA1656, they wasn't widely avaliable back when I was assembling prototypes. Performance so far is good but I'm limited with analyzer performance. Potentially OPA1656 could be even better.
High performance OPA1602 + TPA6120A2 Composite Headphone Amplifier
High performance OPA1602 + TPA6120A2 Composite Headphone Amplifier
Didn't have OPA1656 on my hands, but had OPA2156. While those are obviously two different parts, I can see quite a few similarities. So I thought it might be relevant to post link to some OPA2156 distortion measurement results I made
Microsoft OneDrive - Access files anywhere. Create docs with free Office Online.
Microsoft OneDrive - Access files anywhere. Create docs with free Office Online.
I'm glad our new parts are working out well for you! It might feel like shameless promotion (because it is), but I'll always try to post an update on here when we release a new part. If nothing else, it gives TI quite a bit of (brutally honest) feedback from a very large customer base.
The input stage on the OPA2156 is rail-to-rail, and uses slew boosting to improve the settling time (notice the higher slew rate). Slew boosting is something I don't think most customers are aware of, and we never really do it in audio power op amps. In a slew boosted op amp the tail current of the input differential pair is not a constant current, but rather varies with the differential voltage across the inputs. You can imagine with a large input step, the op amp inputs get briefly pulled apart and the tail current increases to allow the output to drive the inputs back together quicker. It sounds great in theory, but the gm curve of the input stage can get wonky if you're not careful. Also, I typically preferred simplicity in the op amps I developed. OPA1656 had respectable slew rate without slew boosting so I opted to forego that circuitry in the audio part.
OPA1656 also has a pole-zero pair in the open loop gain curve to boost the open loop gain at audio frequencies (hence the 53 MHz gain bandwidth product). OPA2156 does not, it is closer to a single pole response. OPA2156 is also laser trimmed for lower input offset voltage.
The 1 audio op amp I developed which was slew boosted was OPA1692. With only 650uA of supply current we had to throw every trick at it we knew to improve distortion. In that device we actually used slew boosting to reduce the input stage distortion.
The anomaly you mention in your THD vs Frequency graph for 32 ohms is likely increased distortion from the TPA6120A2 at low frequencies. Current-feedback amplifiers often have increased distortion at low frequencies and heavy loads due to thermal feedback on the die to the input stage. On a classic VFB op amp with a LTP input stage, you can mitigate this effect with IC layout techniques. This is also why OPA1622 has a funny pinout by the way. But this doesn't work on CFB amps, hence their distortion gets worse at low frequencies. The loop gain of your composite solution will help a lot, but you can still see a slight rise in distortion at low frequency in the 32 ohm graph.
Great project, and killer performance! Sometimes I feel like the OPA1602 is underappreciated
From all that only GBW would have an affect on measurements I performed. I deliberately stayed away from an input crossover region for G=+1 and a slew rate at 20kHz is still quite low.
Looking at overall current consumption, input noise, short circuit output current etc I can guess that the rest of the circuitry is quite similar. Hence I assume that 1656 would show similar or better results.
I was wondering why the current consumption was so low. A great advantage for me as the available power is limited by the phantom power specification.
Next possible project would be a beam steered microphone that can still be phantom powered.
ES
If a tl071 was used, you probably don't need the extra oomph of the opa1656 and you might be better served by the opa1641. Vastly increased performances over the tl071 but similar power consumption.When is that single version of the OPA1656 coming??? I desperately need two RIGHT AWAY to replace some TL071s!!!
The anomaly you mention in your THD vs Frequency graph for 32 ohms is likely increased distortion from the TPA6120A2 at low frequencies. Current-feedback amplifiers often have increased distortion at low frequencies and heavy loads due to thermal feedback on the die to the input stage. On a classic VFB op amp with a LTP input stage, you can mitigate this effect with IC layout techniques. This is also why OPA1622 has a funny pinout by the way. But this doesn't work on CFB amps, hence their distortion gets worse at low frequencies. The loop gain of your composite solution will help a lot, but you can still see a slight rise in distortion at low frequency in the 32 ohm graph.
Great project, and killer performance! Sometimes I feel like the OPA1602 is underappreciated
Thats very interesting. Thank you for pointing out. Never heard about that distortion mechanism of CFB amplifiers. Actually I thought that composite amplifiers aren't prone to thermal feedback as two chips are separated. I need to perform more test and examine this deeper.
Yes. And this is my only real option at this point. Not so bad, but it would be nice to have that single 1656 op amp to just drop in!
Well, in my application (a mixing console's monitor output), I was hoping to replace the existing, rather-clumsy common-base headphone driver with the 100 mA output capability of the 1656; the 1641 (and most other FET opamps) will only out ~30mA.
You guys (BB / TI) have put a considerable amount of effort into audio opamps, which is fantastic, especially in this day and age, I'm finding lot's of
use for them in pro apps. However it's such a shame BB stopped developing the top tier audio ADC's and DAC's. I'm sure you could now, so many years
after PCM1792 / PCM42222 etc, make some really cutting edge converters.
Or is the market too small / development too costly?
cheers, thanks for the great thread content!
TCD
The converter teams are slowly coming back to the world of audio but I agree, we really need to put out some new "statement" products. The ADC team has started reinvesting in audio products, and their new devices do have fairly good performance, but integrate more channels and features than the pure performance devices. Check out the TLV320ADC6140.
That group has some excellent delta sigma IP, it's just mainly targeting industrial applications. I was hoping they would spin an audio version of the ADS127L01 for example.
On the DAC side, there's less audio development. Paul Frost from the DAC team in Tucson is active on here. I highly recommend you harass him about creating new audio products. Sometimes it helps if its not always coming from me
Last edited:
You're right that a composite amplifier should be more immune to thermal feedback theoretically. However, at a more fundamental level, consider that the distortion of the output amplifier is reduced by the loop gain of the input amplifier. Because the loop gain of the input amplifier is finite, any increase in the distortion of the output amplifier will still appear at the composite's output.
The converter teams are slowly coming back to the world of audio but I agree, we really need to put out some new "statement" products. The ADC team has started reinvesting in audio products, and their new devices do have fairly good performance, but integrate more channels and features than the pure performance devices. Check out the TLV320ADC6140.
That group has some excellent delta sigma IP, it's just mainly targeting industrial applications. I was hoping they would spin an audio version of the ADS127L01 for example.
On the DAC side, there's less audio development. Paul Frost from the DAC team in Tucson is active on here. I highly recommend you harass about creating new audio products. Sometimes it helps if its not always coming from me
If Paul FROST could release a new generation of stereo DAC chips, not especially of a High-End kind but fully integrating the Digital Interface Receiver, the Digital Filter and the Digital-to-Analog Converters with balanced current outputs and clic-free volume control, that would certainly find its market...