Slew rate question

Yesterday I opened up a Grace 201 preamp and took a look inside. I saw INA103 followed by AD844 on the output. From looking at their datasheets I see that INA103 has 15V/us slew rate and the AD844 has an insane 2000V/us and is made pretty much for video and high speed stuff.

My question is: electronically thinking, what is the reason to use such a fast IC after a relatively slow preamp? It seems to me that there is nothing to speed up that is left past the 15V/us device. Or am I not getting a bigger picture and a slew rate spec is not that important here for this combination of ICs?

Thanks!

> what is the reason to use such a fast IC after a relatively slow preamp?

Maybe the slew-rate was not important, as long as it didn't limit performance. 

I have a length of 40psi hose draining a sump, max actual operating pressure 0.4psi. Why did I use a hose 100X better than I needed? Because it was the right size, and there was nothing cheaper in the store.

The stand-out spec on the AD844 is the 80mA output current. Also a lovely symmetrical architecture. Ample bandwidth, low-enough noise. The major drawback is a $7 price tag. But a discrete buffer designed and built would cost similar, and $7 is a small dent in overall costs (box, marketing, markup).

I would not call 15V/uS "slow".

If there is gain in the 844, its slew should/could be higher than the 103's by the amount of gain.

However in +/-15V systems, there is little to be gained for slews over 15v/uS. Audio will NOT slew 15V/uS. We don't even want to come close to slewing, and 15V/uS is good margin. There's usually other limits on slew, especially for variable-gain stages. (The 103 and the 844 both side-step this, though to different degree.)

> there is nothing to speed up

I don't want my amps "speeding up" the performance. Yes, 15 or 25V/uS (times gain) in the second stage would be ample. But when you demand large current, low-low distortion, and pre-made solution, market demands make the Video chips very attractive even though the slew specs are more than we need.

Thank you PRR for the excellent reply, this is good info!

Surprising the grace is basically an ina103 based preamp?? Pretty sure I opened my m-audio interface from way back when and it had ina103s in it...

abechap024 said:

Surprising the grace is basically an ina103 based preamp?? Pretty sure I opened my m-audio interface from way back when and it had ina103s in it...

Click to expand...

It's INA103 preceded by multiple high quality WIMA poly caps in parallel (~6.8uF), nice ELMA gold plated switch for gain, separate regs per channel and a nice LM317-based PSU with a lot of filtering. Some TL072's for DC servo I think. There are no electrolytic caps anywhere in the audio signal path.

I was doing a shoot-out between my THAT1570/5171 preamp, the Grace, and Millenia STT-1 yesterday. VERY interesting experience. =)

PS: Regarding the M-Audio thing... I think it's the implementation that matters, not the chip itself. I could probably screw up an INA103 pretty badly if I didn't spend time learning ways to properly design with it.  =)

promixe said:

PS: Regarding the M-Audio thing... I think it's the implementation that matters, not the chip itself. I could probably screw up an INA103 pretty badly if I didn't spend time learning ways to properly design with it.  =)

Click to expand...

Ahh very true

What PRR said...  +1

Slew rate is like money... it's only a problem when you don't have enough. 

Too much is never a problem.  8)

JR

Hi,

I can't be 100% sure without looking at the schematic, but it seems to me that what you are talking about is "the composite amplifier". It is used when you want to add speed.

Generally, a good DC characteristic (input bias current, input voltage offset ...) opamp has "bad" dynamic characteristics, or you can say that it is not that fast. On the other hand you have current feedback opamps which are much faster (up to 8000V/us or maybe more) but have bad dc performance.
The composite opamp is a combination of a good DC opamp as an input amplifier and a high speed current opamp for gaining speed. This way you get an excellent opamp -> fantastic DC and dynamic characteristics.
If the feedback loop is closed around both opamps (the input opamp doesn't have a local feedback) there actually is something to speed up after the 15V/us.
I have build a couple of composite amplifiers and they work great. The problems I encountered were the stability of the whole circuit caused by the lack of experience in using current feedback opamps, and the fact that higher the slew rate you want the higher the gain has to be.

Now the question is why would you need more slew rate than 15V/us?

EDIT: This statement is wrong, I checked and rechecked!

If you have a pure sine wave signal that is lets say 4dBu (pro audio standard for line level), in other words about 1.78V peak at 20kHz, the required slew rate would be 0.219V/us (2 * Pi *f *Voltage)!!! You would say that the 15V/us is an overkill, but...
The sound is a complex signal composed of many many sine waves playing at the same time in which case you have to take into account the sum of each harmonic's slew rate requirement.
Example: if you have a 4dBu sound made of 100 harmonics from 19kHz to 20kHz each 10Hz apart playing at the same time the required slew rate would be 21.32V/us. It is a rare occasion but possible.

So, In my opinion, there is never enough slew rate!

Regards, Renx.

p.s. http://focus.ti.com.cn/cn/lit/an/sboa002/sboa002.pdf

Awesome info!!! Thank you Renx!

Also, LME49860 Datasheet says +/-20V/uS slew rate. Does that mean 40V/us conventional format slew rate spec?

promixe said:

Awesome info!!! Thank you Renx!

Also, LME49860 Datasheet says +/-20V/uS slew rate. Does that mean 40V/us conventional format slew rate spec?

Click to expand...

No, + "or" - 20V uSec, but that is more than fast enough for audio IMO, do the math. 


JR

Jung's conservative criterion, formulated way back when, was that for audio, 1V/us for every peak output volt pretty much guaranteed no problems with slew-related distortion. I've found this to be an excellent rule-of-thumb over the years.

Peace,
Paul

for a few years i chased chips with fast slew rates...
very rarely was i rewarded with anything i liked.

and most of the old classic gear commanding top dollars
have the slew rates of molasses. (or for the younger generation, corn syrup)

I have always kept an eye on the arm waving audio-phools to not miss the ounce of truth (when present) sometime lurking behind the more extravagant golden ear claims. They usually do hear something, just not always what they think it was. This was part of the thrust behind my "Audio Mythology" column back in the '80s. 

One gentleman who is very visible on another audio forum as an opamp hot rod expert, has been promoting a very fast new IC opamp, but even he admits that the speed is not why it works well and sounds good. High speed is a side effect of high gain bandwidth product, so perhaps just a serendiptious (is that even a word?) consequence of other design improvements.

Back to the topic of slew rate, it is at least as important, if not more so, how the circuit acts in the margin as what the max slew rate is.  Input and output circuitry will often be exposed to RF and edge rates much faster than pedestrian audio. We want such high edge rate stresses to be harmlessly passed or low pass filtered. There are lots of legacy designs with modest slew rates that sound fine.

Slew rate is only important when you don't have enough "and" that lack of speed results in audible artifacts.

Today we have the luxury that several formerly challenging IC opamp performance parameters (noise, slew rate, GBW, linearity etc) are now competitive if not superior to most discrete design.  It's like having an easy button for audio design.  8)

JR

Example: if you have a 4dBu sound made of 100 harmonics from 19kHz to 20kHz each 10Hz apart playing at the same time the required slew rate would be 21.32V/us. It is a rare occasion but possible.

Click to expand...

How did you arive at that figure? A +4 dBu signal which is bandwidth limited to 20 kHz never requires anything near 20 V/µs. You can't just add up the individual slew-rates of composite signals.

I can't be 100% sure without looking at the schematic, but it seems to me that what you are talking about is "the composite amplifier".

Click to expand...

The manufacturer under quest has an informal note on its design concepts online: www.gracedesign.com/about/design.html

I shall say at once though that at least the section on transimpedance/current feedback amplifiers is not entirely correct; e.g. they say that current feedback amplifiers "are not prone to large-signal slew rate limiting". This is simply wrong both in general, and in particular for the INA103.

I recall having read somewhere that they use only the front-end of the INA103, i.e. feed a dedicated balanced line driver from pin 5/12.

Samuel

Samuel Groner said:

Example: if you have a 4dBu sound made of 100 harmonics from 19kHz to 20kHz each 10Hz apart playing at the same time the required slew rate would be 21.32V/us. It is a rare occasion but possible.

Click to expand...

How did you arive at that figure? A +4 dBu signal which is bandwidth limited to 20 kHz never requires anything near 20 V/µs. You can't just add up the individual slew-rates of composite signals.


Samuel

Click to expand...

Pretty obviously he is not band limiting his input signals.

Back when I did my own DIY two-tone IMD  (19:20kHz) analyzer, I found pretty quickly from bench testing that the two-tone results correlated well with simple 39kHz THD (allowing for measurement bandwidth of distortion products, etc).

His hypothetical is questionable since even if he close mic'd a crash cymbal the results would be LPF by the microphone and reality.

Slew rate is swiftly becoming like damping factor... it used to be a something worth worrying about, but hasn't for years (decades in the case of DF unless you're one of those tube guys).

JR

A 4 dBu sound made of 100 harmonics from 19 kHz to 20 kHz each 10 Hz apart.

Click to expand...

Pretty obviously he is not band limiting his input signals.

Click to expand...

Hm, it sounds to me like a signal with a maximum frequency of 20 kHz is band limited to 20 kHz, no?

Samuel

My bad,  you are correct... Even in my example of 19 and 20 khz tones summed, they each have to be padded down -6dB so the sum doesn't clip, so the combined rate of change is back around the same as single 20 kHz full scale sine wave.

His 100 individual components (not harmonics) would each have to be summed in at something like 40dB below the resultant 4 dBu for the sum to add back up to +4 dBu (peak), so again similar to one 20 kHz signal. It does depend whether the +4dBu is measured peak or average, since the sum will not be coherent if it is measured average such a waveform would clip before ever reaching +4dBu average. 

Another small problem with unrealistic hypotheticals... you can't get +4 dBu without clipping.

JR

JohnRoberts said:

My bad,  you are correct... Even in my example of 19 and 20 khz tones summed, they each have to be padded down -6dB so the sum doesn't clip, so the combined rate of change is back around the same as single 20 kHz full scale sine wave.

Click to expand...

You are right!
My calculations were wrong because I didn't take into account the attenuation of each signal component. The required slew rate for my example is almost the same as a single 20kHz component -> around 0.2V/us which means that a 15V/us opamp really is more than sufficient.

Regards, Renx.

> the sum of each harmonic's slew rate requirement

No, as discussed.....

And if we DID have huge slew-rate on the LP, the needle would break. On CD, big slew may be possible but would HURT the ear if it happened often enough to be audible.

Back in the old days, one high-slew passage was the organ solo on In-A-Gadda-Da-Vida. I can think of some later stronger stuff, but I don't think you EVER want full peak above 6KHz.

> around 0.2V/us which means that a 15V/us opamp really is more than sufficient.

Well, maybe, maybe not.

What happens in the studio is sometimes (though rarely) stronger than you ever want to hear on record. Teeny metalophones can ring past 20KHz, a few small condensers go way past 20KHz, and too-close mike distance can give huge levels. Somewhere down-chain you will low-pass, but that first stage perhaps should tolerate some fast signals, at least not overload rudely. (OTOH, you could move the mike back....)

And the quoted slew-rate is GROSS distortion. It's like saying my car-radio makes 37 Watts.... at 100% THD. It is with the input device SLAMMED past the linear range. We want to stay FAR away from this point for cleanliness.

How far is "far"? Doug Self published some:
Article
graph: http://i.cmpnet.com/audiodesignline/2008/01/self_ch17_fig5t.jpg

"Slew" puts the input devices to their limits, here +/-1.2mA. The transfer flat-tops, gross distortion. But if you squint +/-0.12mA, the transfer is smooth, distortion is small. And by the nature of listenable music, high slew rates should be rare. So even if that's 1%THD at +/-0.12mA, you probably won't hear it.

> lets say 4dBu ... about 1.78V peak

That assumes you have things under control. Sometimes, mike in hot room, your preamp output level is NOT "under control", may be 12V peak. Then we have 1.5V/uS. Maybe more if you are picking up ultrasonics. But still, "1V/uS per Volt of peak signal or supply" gives a lot of margin from slew.

---------
Off-point but perhaps of interest: an IEEE paper on transistor distortion, with some easy points for napkin-analysis:
http://amesp02.tamu.edu/~jsilva/610/Lecture%20notes/Papers/Distortion-Sansen.pdf

"...a bipolar transistor carries a dc current of 1mA and an ac peak current of 100uA. The peak relative current swing is thus 0.1. Without feedback IM2 = 5%, IM(3) = 0.125%, and Vbe = 1.84 mV RMS".

A paper on audio distortion:
http://www.cco.caltech.edu/~musiclab/feedback-paper-acrobat.pdf

Hi,

I have revisited my claims in the last couple of posts, and I did the math again, and these are my conclusions:

There exists such a complex signal that when "passed" through a 1st order butterworth lpf of 20kHz @(-3db), gets minimally linearly distorted (a couple of degrees of phase shift and less than 0.5db in amplitude) but requires larger slew rate than the slew rate required by the 20kHz pure tone signal of the same amplitude peak.


EDIT: This statement is wrong, I checked and rechecked!

As soon as I get my Mathematica graphs uploaded, I will post them here to back up my claims, and would kindly ask all of you to check for any errors in my theory or calculations.
The reason I am doing this is that I remember reading an article from Yamaha, I think, on how the slew rate impacts the sound of power amplifiers. If I recall correctly they were playing the same material through a couple of amplifiers different by their slew rate only, and same speakers of course, and have concluded that larger slew rate is needed than that of a max freq. Unfortunately I can't find it anywhere!
The fact that I have been designing electronics under wrong assumptions for so long is just killing me!
That is why I have to get to the bottom of this.

Regards, Renx.