OPA828 / OPA2828 vs OPA627

Member
Joined 2019
Paid Member
No direct comparisons here but I’m using OPA828 in several projects and listening it every day for over a year.

My conclusion is that it is a fantastic sounding solution. Meaning, it does not have some specific sound signature but is perfectly transparent, providing extremely clean and detailed sound that is very enjoyable end does not build any listening fatigue even after the whole day listening session. Actually, at the end of day, it is very hard to switch off the equipment.

First point of comparison was with amplifier input having Toshiba 2SK170/2SJ74 input, where input JFET buffer pair was replaced with OPA828 in the unity gain configuration. OPA828 was better. It was not only my opinion. Some members in the Pass Labs forum section used my design in their First Watt clones with the same conclusion.

Next, OPA828 was evaluated as a gain or VAS section in a small A class power amplifier with extreme performance (2 MHz power bandwidth, 200 ns rise time, 0.0000x distortion and so on). Again, it does not spoil sound or introduce undesired sound artifacts. Complete bliss.

This is what I’m listening now (even while typing this message) and it will be my reference solution, turned into the proper amplifier. ATM, it is on the breadboards.
 
  • Like
Reactions: 3 users
When you say (in the context of audio), do you mean subjective sound quality?
If so, the data below may not be what you want.
However, it may be the data that supports the sound quality impression of tombo56.

The characteristics of opa828 are also very good for audio.
Especially the distortion characteristic is the best among monolithic OPAMPs I own.
Below is what I always use to evaluate op amps
G=+10 (β=1/10)
The load resistance is the parallel value of the feedback circuit's 2kΩ and the distortion meter's input impedance of about 6kΩ.
The input signal is 20kHz 0dBu (0.775Vrms), the signal source impedance is 50Ω. Distortion at 20dBu (7.75Vrms) output.
Test cir.png


The cyan waveform is the signal input to the distortion meter attenuated to 1/50 (-34dB). Therefore, read the vertical axis as 5V/div.
The yellow waveform is the harmonic waveform with the fundamental removed, amplified 2000 times (+66dB). Therefore, read the vertical axis as 50 μV/div.
Also, since the average of 100 times is taken, the random noise component is compressed to 1/√100=1/10.

OPA828
OPA828_20kHz_20dBu_G=+10_Dist.png

THD is calculated as (8.6965mV / 154.39mV) * 0.001 = 0.0000563%

similarly
OPA827
OPA827_20kHz_20dBu_G=+10_Dist.png

THD=0.00102%

OPA627
The OPA627 has a large distortion component (yellow) and protrudes from the screen of the scope, so the range is changed from 100mV/div to 200mV/div. The yellow waveform is therefore 100 µ/div.
OPA627_20kHz_20dBu_G=+10_Dist.png

THD=0.00222%


Below is the distortion factor measured using the FFT function.
Since the FFT of Analog Discovery 2 has poor resolution, it can only measure distortion of about -70dBc with raw waveforms. Therefore, FFT is performed on the waveform obtained by adding the above two waveforms.

OPA828
OPA828_20kHz_20dBu_G=+10_FFT.png

Since the input signal is -34dB and the harmonics are +66dB, the distortion factor is 100dB emphasized and measured. So THD measurement value of -25.22dBc means -125.22dBc=548%.

With OPA827 and OPA627, the second harmonic (40kHz) was larger than the fundamental wave of 20kHz, and 40kHz was regarded as the fundamental wave, so THD could not be measured.:ROFLMAO:
OPA827
OPA827_20kHz_20dBu_G=+10_FFT.png


OPA627
OPA627_20kHz_20dBu_G=+10_FFT.png




By the way, is this forum suitable for such individual OP-Amp topics? Or Analog Line Level?
 

Attachments

  • OPA828_20kHz_20dBu_FFT.png
    OPA828_20kHz_20dBu_FFT.png
    66.3 KB · Views: 167
Last edited:
  • Like
  • Thank You
Reactions: 10 users
Disabled Account
Joined 2002
First point of comparison was with amplifier input having Toshiba 2SK170/2SJ74 input, where input JFET buffer pair was replaced with OPA828 in the unity gain configuration. OPA828 was better. It was not only my opinion. Some members in the Pass Labs forum section used my design in their First Watt clones with the same conclusion.

Next, OPA828 was evaluated as a gain or VAS section in a small A class power amplifier with extreme performance (2 MHz power bandwidth, 200 ns rise time, 0.0000x distortion and so on). Again, it does not spoil sound or introduce undesired sound artifacts. Complete bliss.

This is what I’m listening now (even while typing this message) and it will be my reference solution, turned into the proper amplifier. ATM, it is on the breadboards.
Hail to the King! So you found a worthy replacement for the otherwise till eternity used 2SK170/2SJ74 duo that could not be written off. Until now that is. Good news!

It was a matter of time but of course developments in JFET opamps did not stand still. What about OPA1655 (CMOS)?
 
Last edited:
Until the release of the OPA828, ultra-low distortion OP-AMPs only had BJT inputs. Here's one I own with low distortion. The measurement conditions are the same as in #4 post.

AD797
AD797_20kHz_20dBu_-114.5.png

AD797_20kHz_20dBu_FFT.png

THD=-113.157dBc=0.000220%


LME49710
LME49710_20kHz_20dBu_-114.6.png

LME49710_20kHz_20dBu_FFT.png

THD=-114.403dBc=0.000190%
LME49710 has been discontinued, but dual LM4562, LME49720, LME49860 are current products.

OPA1611
OPA1611_20kHz_20dBu_-115.8.png

OPA1611_20kHz_20dBu_FFT.png

THD=-116.309dBc=0.000153%

LME49990 (already discontinue)
LME49990_20kHz_20dBu_-117.5.png

LME49990_20kHz_20dBu_FFT.png

THD=-117.707dBc=0.000130%
Although the price was also cheap, It is a pity that NS was absorbed by TI and was abolished after losing the brand power competition with BB. TI was going to discontinue the LM4562 and others as well.



The JFT input OPA828 has overtaken all these low distortion OP-AMPs.
In post #4 ”-125.22dBc=548%” is incorrect and Correctly 0.0000548%.
Currently, only the OPA828 measures below -120dBc under these conditions.
Also, the main components of distortion are the low-order components of 3rdHD and 2ndHD.
This point is also different from conventional RtoR output such as OPA1611/1612.



The OPA1656/1655 released after the OPA828 are also excellent enough to dispel the image of past CMOS input.

OPA1655
OPA1655_20kHz_20dBu_-114.2.png

OPA1655_20kHz_20dBu_FFT.png

THD=-114.144dBc=0.000196%



I recently learned that an OP-AMP called OPA837 was released. I thought it was a non-compensation type of OPA827, but it was completely different. Please don't give confusing model numbers.:rolleyes:
 
  • Like
Reactions: 13 users
Thank you everyone.
Mason_f8, did you also measure OPA1642!?
I am measuring OPA1641.

OPA1641
Since the distortion is large, the range of the distortion meter is switched to 20 times (+26dB). Since it is picked up at 10mV/div, read the distorted waveform (yellow) as 500μV/div.
OPA1641_20kHz_20dBu_-86.6.png

OPA1641_20kHz_20dBu_FFT.png

Since the fundamental wave is -34dB and the harmonic is +26dB, the THD is emphasized by 60dB. So the -22.878 dBc measurement is -88.878 dBc = 0.00360%.

Since the OPA1642/1641 have improved common-mode input distortion since the initial release, they are very excellent, so I think they are suitable for the first stage buffer. I found out about it a year and a half ago.

Would you be willing to share how you took them with Analog Discovery?
Please give me some time.
 
Is price a relevant parameter when looking for best performance!?
In the case of DIY, it takes a lot of time and money to create the board and process the case. I totally agree with jean-paul's opinion because the price difference of OP-AMP, which is the key part of the performance, is negligible compared to them.

However, I also admire alexcp's design for its cheap and readily available OP-AMPs that deliver excellent performance.
 
  • Like
Reactions: 1 user
Member
Joined 2019
Paid Member
I see OP828 is 3 times more expensive but it is still below 8 Euro a piece.

Besides Mouser, I bought one OPA828 batch from LCSC electronics. They are genuine and only 5.5 $ a piece, below 5$ for ten pcs. Using cheapest postal option at 6$, they arrived after 10 days with no additional import charges, despite new EU import rules.

Worth to check: https://www.lcsc.com/