I know as little as anyone who hasn't signed their NDA, but you are obviously correct about the fact that it is a much lower power design and will certainly have it's limitations (and the hump of course...). It seems to me Gustard are working very closely with the Chinese ESS representative (I think you voiced a similar impression in the past).
For my own project I will stick with ES9028Pro for now. When I was ready to bite the bullet and spend 50+ USD for a DAC chip, ES9038Pro was out of stock on mouser. Unlucky timing.
For my own project I will stick with ES9028Pro for now. When I was ready to bite the bullet and spend 50+ USD for a DAC chip, ES9038Pro was out of stock on mouser. Unlucky timing.
Similar thing happened to me on Digikey when ordering 2 Raspberry Pi Picos a few days ago. It said they had 2 in stock, so I ordered 2 and my order was shipped but the stock level remains at 2. So maybe the stock level is actually a maximum quantity per customer/order. But I don't suspect there is such a rule on mouser for CS43198 and even if so, why the heck would it be 17?!
Adding a current buffer e.g. LT1010 or LMH6321 might separate the current requirements from the opamp selection and keep the heat and any side effects of it from the opamp.
TPA6120 would work only in a composite amp too, otherwise has only -112dB THD, not good enough for instrumentation, and well under the DAC perfomance. I may try a combo of OPA1612 with TPA6120, as soon as I can get some in the DFN case (they are on order for August), otherwise the SOIC version is too big to fit 4 pcs in the current board real estate.
Last edited:
Here's a direct comparison in the SMPTE IMD distortions between the OPA1656 and the OPA1622. SMPTE signal is 4:1 60Hz:7KHz. 0dBFS=~2.6Veff at the analog output (differential)
- For the OPA1656, note the sharp increase in the IMD distortions for signals over -15dBFS
- OPA1622 behaves much better, SMPTE IMD increases past -6dBFS
- Also note the "hump" residuals, these are because Vcm is a rather high 1V. At Vcm about 0.6V the hump vanishes completely, but then the static current in each I/V opamp is a sinking 21mA, too high for the IMD distortions to behave at high output levels.I decided I can live with this "hump" residual.The "hump" levels scale pretty well, OPA1656 seems lower because most of the hump is buried in noise. This proves that the "hump" is intrinsic to the ESS chip, and is not some external artifact, like detecting some RF crud on the bipolar input of the op amp. OPA1622 is bipolar input, OPA1656 is FET input, essentially no difference in "hump".
- For the OPA1656, note the sharp increase in the IMD distortions for signals over -15dBFS
- OPA1622 behaves much better, SMPTE IMD increases past -6dBFS
- Also note the "hump" residuals, these are because Vcm is a rather high 1V. At Vcm about 0.6V the hump vanishes completely, but then the static current in each I/V opamp is a sinking 21mA, too high for the IMD distortions to behave at high output levels.I decided I can live with this "hump" residual.The "hump" levels scale pretty well, OPA1656 seems lower because most of the hump is buried in noise. This proves that the "hump" is intrinsic to the ESS chip, and is not some external artifact, like detecting some RF crud on the bipolar input of the op amp. OPA1622 is bipolar input, OPA1656 is FET input, essentially no difference in "hump".
Attachments
Last edited:
You might be able to adapt Samuel Groner’s composite with OPA211 (1611) and THS4131. THS4131 has 90 mA max output current, so its drive capability may not be much better than what you’ve tested, but the increased loop gain may help. Would require some experimentation to see if it can be made stable with the right cap values.
A JFET at Idss (Vgs=0) from the negative supply to the op amp output is all that’s in principle needed. But the effect on the overall distortions is hard to estimate other than by trial and error, since there are no perfectly matched JFETs to keep the outputs exactly in balance, these JFETs are not in the feedback loop. And I was not in the mood of experimenting with 4 layers PCBs.
Did someone try a fully differential amp for the first half of the composite amp?
Remember to pay royalties to Self who trademarked and patented this stuff that everyone's been doing for at least 30 years...
Remember to pay royalties to Self who trademarked and patented this stuff that everyone's been doing for at least 30 years...
Can you buy them directly from TI? they seem to have them in stock. With a single unit price of USD 2.42.
I made some tests on a composite amplifier some time ago.
With either OPA1612 or OP1656 + TPA6120A2 I measured distortion below -150dB at 1 kHz with a 50 ohm load. Actually the measured distortion was noise rather than real distortion. This was measured with a low distortion oscillator + notch filter + an RTX6001.
I also measured the distortion with a distortion magnification configuration (increased noise gain) to resolve the distortion, which was otherwise below the measurement limits.
This showed a distortion of -174dB at 1 kHz with a 50 ohm load with the OPA1612 + TPS6120A2. The OPA1656 + TPS6120A2 could "only" achieve -172dB. Of course the noise is higher for the OPA1656.
This was with a non-inverting amplifier configuration, not an I/V.
The gain was 3 and the output level was 5.8 Vrms.
I also measured the distortion with other op-amps (LME49720, NJM2068, NJM8068, NE5532 and TL072), but they were not as good as the OP1612 and OP1656. The LME49720 came relatively close though. I didn't use the distortion magnification for the other op-amps since the notch filter + analyzer was sufficient to see the distortion above the noise floor.
It would be really interesting to see the composite amp used as an I/V.
OPA1612 + TPS6120A2 would probably be ideal, with a much lower noise than the OPA1656. With the TPS6120A2 as output buffer the output current should not be an issue.
If you're in Denmark, Farnell will probably have cheaper shipping. They have some in stock.
tpa6120a2r Stock and Price by Distributor
> I made some tests on a composite amplifier some time ago.
With either OPA1612 or OP1656 + TPA6120A2 I measured distortion below -150dB at 1 kHz with a 50 ohm load.
Nice
Will you share the component values for the compensation?