Last edited:
It does seem like a pretty reasonable discrete with a hefty OPS for sure.
I haven't run the sims, but I'd wonder if I could fit a 49860/49720 + class A buffer into a 10 MHz GBW and get that kind of power bandwidth. Probably close. But that also only makes sense if you need it.
Nothing! But...
1. there are a lot of OPs that are totally inappropriate for audio. They are not designed for the audio domain and have extremely high distortions compared to specialized audio ops, e.g. NE5532
2. The maximum voltage range is + -15..20V. With a discrete comps, voltages of + -40V and more are possible. The voltage swing is larger and therefore the SNR.
3. Although you only need 0 to 2 extra components to build an amplifier, there are still people (myself too) who fall into many sources of error here.
Attachments
Last edited:
No opamps please...
For cost the OPA1652 seems to have a good balance of properties. Its design is very symmetrical which can be seen from the PSRR plots. Costlier but also having very good low frequency noise performance is the OPA209/2209. Too much emphasis is placed on slew rates. The OPA209 seems slower, but its unity gain bandwidth is greater than the OPA1652 and NE5534. The OPA1652 would suit high input impedance duties and a natural replacement for the TL072.
Designers of discrete circuits are borrowing techniques introduced into opamps. Discrete circuits have one advantage over chips, in that lower noise is possible. Wide voltage swings aren't needed for most signal applications, so this advantage for discrete is moot. I would use either of the above for audio duties but not as headphone drivers. People who compare opamps driving headphones have lost the plot.
It is difficult to make an objective claim that discrete is better. Discrete combined with opamps allows greater freedom when making opmap selection. The LM6172 is an interesting device, which is a CFA turned into a VFA. It has a class AB input stage but good noise performance for a fast device. A discrete front end for this chip would probably give very good performance.
I can't remember the year when Texas Instruments produced an amplifier project called the Texan to promote the 741 opamp. The power amp section was very simple with a 741, BC182, BC212 and TIP power devices. That design would be great as a beginner build and with modern devices, deliver staggering performance. When people insist on having no opamps in the audio path, they are effectively limiting their horizons.
For cost the OPA1652 seems to have a good balance of properties. Its design is very symmetrical which can be seen from the PSRR plots. Costlier but also having very good low frequency noise performance is the OPA209/2209. Too much emphasis is placed on slew rates. The OPA209 seems slower, but its unity gain bandwidth is greater than the OPA1652 and NE5534. The OPA1652 would suit high input impedance duties and a natural replacement for the TL072.
Designers of discrete circuits are borrowing techniques introduced into opamps. Discrete circuits have one advantage over chips, in that lower noise is possible. Wide voltage swings aren't needed for most signal applications, so this advantage for discrete is moot. I would use either of the above for audio duties but not as headphone drivers. People who compare opamps driving headphones have lost the plot.
It is difficult to make an objective claim that discrete is better. Discrete combined with opamps allows greater freedom when making opmap selection. The LM6172 is an interesting device, which is a CFA turned into a VFA. It has a class AB input stage but good noise performance for a fast device. A discrete front end for this chip would probably give very good performance.
I can't remember the year when Texas Instruments produced an amplifier project called the Texan to promote the 741 opamp. The power amp section was very simple with a 741, BC182, BC212 and TIP power devices. That design would be great as a beginner build and with modern devices, deliver staggering performance. When people insist on having no opamps in the audio path, they are effectively limiting their horizons.
Last edited:
One other thing I have noticed is that the phase response of designs is very variable. The input stage of the OPA1652 seems preferable to the OPA1642. Also the phase reponse is noticeably better than OPA1642, OP275 and OPA2134. People seem to be drawn to distortion and slew rate specs. I evaluate on a broad range of parameters.
Cost sensitivity* and low-voltage applications are perfectly decent reasons; I guess I'm more wary of the noise behavior of a CMOS input opamp than a jfet, even if the '52 hits ultimately lower noise levels.
The 1642 does looks like it has a Bode step and lower GBW. Long and short, both are nice parts.
*The cost difference at Mouser seems pretty small between the 1642 and 1652, at least in DIY quantities.
The 1642 does looks like it has a Bode step and lower GBW. Long and short, both are nice parts.
*The cost difference at Mouser seems pretty small between the 1642 and 1652, at least in DIY quantities.
The PW Texan is a very interesting design that evolved into the Rega Brio. There was the PE Rhondo also which is a development. I built my own take on this using a slightly under-biased class B using simple resistor bias. I played with the output stage gain and liked a gain of 3. I wanted a robust design for a non audio use, I must say is worked very well, I did check it out for distortion and found it much better than I had guessed. Stability seems good. The output stage gain might help that.
A Paul Kemble web page - PW Texan amplifier and the PE Rondo.
A Paul Kemble web page - PW Texan amplifier and the PE Rondo.
While we are on the subject, another console maker chose the RCA4558. This went on to be ubiquitous in recording studios. You know the one, from Oxford, England. Recording engineers thought it was so, so, even crap. Didn't stop thousands of albums being recorded and mixed with them.
My colleague was aghast when he opened a module for the first time, expecting something more exotic. So there are differences. Double blind not needed. The real question is whether it is actually important. I have stood in front of systems built by people worth $600,000 and said they sound great. But not $600,000 worth of great.
Now people are selling 'vintage' 4558 for distortion pedals. I need to to go to sleep and wake up in a different dimension.
My colleague was aghast when he opened a module for the first time, expecting something more exotic. So there are differences. Double blind not needed. The real question is whether it is actually important. I have stood in front of systems built by people worth $600,000 and said they sound great. But not $600,000 worth of great.
Now people are selling 'vintage' 4558 for distortion pedals. I need to to go to sleep and wake up in a different dimension.
Last edited:
I will have a drink with John Deans ex of SSL to know more. NE5534/2 were typical. At one time the were the Worlds largest user of them. I live 2 miles from them.
I don't recall if Douglas Self evaluated the Texan or Rod Elliot of ESP Audio. I have a hunch with a output gain of 3 it's doing interesting things.
I don't recall if Douglas Self evaluated the Texan or Rod Elliot of ESP Audio. I have a hunch with a output gain of 3 it's doing interesting things.
I'm considering scrapping a very nice project due to fails on prototypes using some expensive audio opamps.
Fortunately the inital production batch was not released.
I've read all of the TI/BB app notes and published suggestions on protecting inputs.
This article sheds light on yet another issue with opamps.
They wear out.
When good opamps go bad
Why are the 1-2k input protect resistors not on chip die?
Fortunately the inital production batch was not released.
I've read all of the TI/BB app notes and published suggestions on protecting inputs.
This article sheds light on yet another issue with opamps.
They wear out.
When good opamps go bad
Why are the 1-2k input protect resistors not on chip die?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.