Good opamp for that1646

Help Support GroupDIY:

warpie

Well-known member
Joined
Feb 7, 2009
Messages
1,357
I already gave you an answer on that point. You may want to experiment, and find out that PSU rails unbalance has almost nil effect on offset, which is essentially due to input devices and associated resistors matching.

Yes I remember you saying that. I wasn't necessary referring only to the offset but if it's generally OK.
 

FIX

Paul Wolff
Joined
May 5, 2021
Messages
143
Location
Nashville
You must put it in perspective with the fact that, in order to reproduce a 20kHz sinewave at full-blow (+20dBu), one needs only about 1.5V/us.
Datasheet shows that power BW with a 47pF compensation cap exceeds 40kHz, 60kHz with 22pF.
More true than people want to admit... The API 2520 had a slew rate of just under 1.8V/us, and no one complained about the money they received from the hits that were made with consoles that used them. I think people forget that a slew rating is when the amp STARTS to diminish if HF response, it's not a brick wall. Most transformers in those days were all starting to roll off either right before or right after 20KHz. Of course, a lot of that design spec was because they bandwidth limited things because of so many radio and TV stations close by.
 

JohnRoberts

Well-known member
Staff member
Moderator
Joined
Nov 30, 2006
Messages
21,828
Location
Hickory, MS
More true than people want to admit... The API 2520 had a slew rate of just under 1.8V/us, and no one complained about the money they received from the hits that were made with consoles that used them. I think people forget that a slew rating is when the amp STARTS to diminish if HF response, it's not a brick wall. Most transformers in those days were all starting to roll off either right before or right after 20KHz. Of course, a lot of that design spec was because they bandwidth limited things because of so many radio and TV stations close by.
Indeed slew rate is over valued as a performance metric... enough is generally enough. Even power bandwidth over states the slew requirement as we rarely see rail to rail 20kHz signals (from a properly working circuit).

That said slew rate is pretty much a hard limit on output rate of change. At the point where slew limiting occurs the amplifier output is no longer keeping up with the input signal. Depending on the amplifier design, recovery can involve artifacts, delayed recovery, or worse.

Permission to repeat myself. It is possible to design an audio path to be rise time limited. Such a path can not be slew limited while faster signals receive a relatively benign low pass filter. Prof Marshall Leach published an AES paper on the subject back when he was still alive. Some amp designers embraced the improved resistance to slew overload but customers did not. Telling an amp customer that the amp has a rise time of X uSecs didn't result in as many sales as citing large numerical slew rates. At least one amp maker defeated the rise time circuit so they could drive the amp to slew limiting for measurement, and advertising.

JR
 

Ike Zimbel

Well-known member
Joined
Apr 17, 2011
Messages
105
Location
Toronto, Canada
I'm curious about what the consensus is on what constitutes an "adequate" slew rate. It has been my experience that higher slew rates offer sonic benefits, most notably more accurate reproduction of HF transient heavy sources like hi-hats, tambourines, shakers etc.
 

JohnRoberts

Well-known member
Staff member
Moderator
Joined
Nov 30, 2006
Messages
21,828
Location
Hickory, MS
I'm curious about what the consensus is on what constitutes an "adequate" slew rate. It has been my experience that higher slew rates offer sonic benefits, most notably more accurate reproduction of HF transient heavy sources like hi-hats, tambourines, shakers etc.
I stopped worrying about slew rate since the wide availability of op amps like NE553x and TL07x in the 1970s. They were several times faster than needed for audio and didn't break the bank. I don't expect any benefit from faster than those old soldiers, and some very high speed op amps (like for sample and hold applications) may not behave well for audio reproduction.

JR
 

john12ax7

Well-known member
Joined
Oct 14, 2010
Messages
1,975
Location
California, US
I'm curious about what the consensus is on what constitutes an "adequate" slew rate. It has been my experience that higher slew rates offer sonic benefits, most notably more accurate reproduction of HF transient heavy sources like hi-hats, tambourines, shakers etc.
I've found higher definitely helps for accurate audio capture. A good rule of thumb with audio is just use the rail voltage. So 15V aim for 15V/uS. This will put the errors way outside the audio band.

You can hear slew induced distortion with something like a 2520, especially the older ones. But that's also part of its charm, sounds great for rock n roll.
 

JohnRoberts

Well-known member
Staff member
Moderator
Joined
Nov 30, 2006
Messages
21,828
Location
Hickory, MS
I've found higher definitely helps for accurate audio capture. A good rule of thumb with audio is just use the rail voltage. So 15V aim for 15V/uS. This will put the errors way outside the audio band.
perhaps by 10x
You can hear slew induced distortion with something like a 2520, especially the older ones. But that's also part of its charm, sounds great for rock n roll.
I am not aware of that criticism, but I never messed with 2520s.

Slew rate has been well discussed. Back in the 70s there were a handful of attempts to invent new rate of change related audio bench tests (TIM, SID). AKA "transient intermodulation distortion" or "slew induced distortion". I preferred the old conventional distortion tests IMD (intermodulation distortion).

The industry standard SMPTE IMD test mixed a 60Hz signal with 7kHz. Perhaps back when that test was invented 7kHz was considered high frequency. 7kHz has not been considered difficult for several decades. I modified my old Heathkit SMPTE IMD analyzer to operate with 19kHz and 20kHz sine waves mixed 1:1, instead of 60Hz/7kHz. This test very effectively identified slew challenged circuitry, with a distortion product down at 1kHz in the middle of our sensitive hearing band.

To repeat myself by the 70s we had excellent off the shelf op amps with more than adequate slew rate. That said I killed some brain cells after beer o'clock some nights speculating about a dedicated test to isolate and measure slew rate of change linearity. My speculation is that different amplifier topologies will exhibit increasing slew rate nonlinearity before actual slew limiting. Pretty much analogous to how audio circuits begin to get increasingly more distorted before hard amplitude clipping.

This actually supports John12ax7's practice of providing more slew capability than needed, just not ten times :unsure: more slew rate than needed! In my judgement this whole exercise was made academic by the availability of much faster than needed for audio, off the shelf ICs, so I abandoned my quantification pursuit. In addition I couldn't easily explain to other engineers what the numbers meant and how to use them. Perhaps to show how crappy some old (slow) designs were. For new designs just use a modern part and don't worry about slew rate.

JR

PS: to answer the question I raised myself a perhaps useful slew rate headroom of 1.4x is likely adequate, while unnecessary with fast modern components.

PPS: If you think slew limiting is a useful audio effect for R&R it is possible to design a path with slew limiting intentionally, but I wouldn't melt any solder doing that.
 

ccaudle

Well-known member
Joined
Mar 18, 2010
Messages
125
Location
Houston
PS: to answer the question I raised myself a perhaps useful slew rate headroom of 1.4x is likely adequate, while unnecessary with fast modern components.

There is one reason I can think of that might make your life easier using amps with what seems like excessive slew rate: high frequency audio input (e.g. close mic'ed metal percussion like cymbals or chimes), or EMI in that in-between range that is well above audio, but not high enough that the wiring and connectors becomes lossy, such as AM radio in the 500kHz - 1MHz bands. Either one of those getting into the front end of an amp with just enough slew rate for full power audio (i.e. 20kHz bandwidth or just over) would likely produce a lot of intermod distortion. The low-ish frequency RF interference could be helped with a common mode filter on the front end, but a close position small diaphragm in front of a cymbal or other metal percussion could send differential mode signal at 40kHz - 50kHz into the front end amp. Granted, not likely at full amplitude, the higher harmonics should be falling off, but most equipment isn't going to have a low-pass with a cut-off below 40kHz, so whatever is there will get through to the input stage.

Which I think was kind of the point of the Leach papers on "TIM," that you should bandwidth limit the input to whatever the amp could handle with high linearity, and not let signals outside the effective feedback bandwidth hit the amp front end to begin with. Which seems to imply a converse, that if you can't or won't design a low cutoff low pass filter that is low enough noise, distortion, phase shift, etc. for your application, you then have to extend the amplifier bandwidth to cover the natural cut-off of the circuit design.
 

57sputnik

Well-known member
Joined
Jul 8, 2010
Messages
57
but a close position small diaphragm in front of a cymbal or other metal percussion could send differential mode signal at 40kHz - 50kHz into the front end amp. Granted, not likely at full amplitude, the higher harmonics should be falling off, but most equipment isn't going to have a low-pass with a cut-off below 40kHz, so whatever is there will get through to the input stage.

That's an excellent point.

A similar situation occurs in phono preamps where harmonics from tracing distortion and other sources produce signals up to 50 kHz or more. I've done transfers, run Spek on the file and see lots of activity in the octave above 20 kHz. The levels may be low but the lower IM products get magnified by the RIAA curve. This is a RAW (flat) recording.

(Supertramp)-The Logical Song-Vinyl-RAW.wav.png


Returning back to the 1646 op amp driver: The 1646 has 6 dB gain so whatever slew rate you think you need at the output only has to be half that for the op amp.
 

JohnRoberts

Well-known member
Staff member
Moderator
Joined
Nov 30, 2006
Messages
21,828
Location
Hickory, MS
The 19kHz:20kHz two tone IMD test I DIY'd generated a slew rate at peak superposition of effectively 2x or more like 39kHz. BUT this is with full scale top octave signals.

Yes, Marshal Leach (RIP) nailed it back then with properly rise time limited audio paths.

I (over) designed my last phono preamp with an open loop JFET front end that had the drain current feeding into the 75uSec RIAA passive real pole. You could hit that input gate with lightning, and as long as the output was not voltage saturated following RIAA EQ transfer function it remained well behaved.
===
In the real world, like typical recording studios, air and distance make an effective low pass filter so don't look for full scale 50kHz off that cymbal mic.

I used to jangle a key chain right in front of a studio mic to look for IMD from slower audio front ends.

JR
 

FIX

Paul Wolff
Joined
May 5, 2021
Messages
143
Location
Nashville
Look what you started...
 

Latest posts

Top