Slew Rate

koiman · « **on:** 16 Jun 2005, 02:28 am »

Can some one tell me if a lower or higher SLEW RATE in a power amp is better?

warnerwh · « **Reply #1 on:** 16 Jun 2005, 03:20 am »

Good quality modern solid state amps have slew rates that are more than good enough and it's really a non issue.

Levi · « **Reply #2 on:** 16 Jun 2005, 04:29 am »

My BAT VK-200 has a slew rate of 250V/uS. I think it's fast enough.

audioengr · « **Reply #3 on:** 16 Jun 2005, 04:42 pm »

This is often measured with a repetitive signal such as a square-wave, so this spec can be meaningless. What it important is the transient response, or impulse/step response. This is amost never measured or specified. This is what makes an amp responsive to transient signals.

DVV · « **Reply #4 on:** 17 Jun 2005, 06:54 am »

As any specification, so slew rate is relative as well, in more than one sense of the word.

If measured and declared properly, one would assume it applies to the whole amplifier, just as you take it out of the box. Unfortunately, all too often this is NOT so.

For example, Sansui used to declare the slew rate of their input stage only, reasoning that if it doesn't saturate and distort, the rest will manage. Hence, they had wild slew rate specs, 200V/uS and upwards. 25 years ago, this was sicence fiction.

Others will take out of the circuit some components, known to knock down the slew rate, such as input low pass filters and even the output inductor circuit, measure it and declare it - too bad you will NEVER actually have this in real life.

For many years, it was assumed that so long as your amp could manage 0.5V/uS per peak volt of output, you were all right in terms of necessary speed. So, for a say 100W/8 ohms amp, delivering 40V peak to have those 100W/8 ohms, a slew rate of 20V/uS was considered sufficient. When you work out the slew rate you need to deliver those 40V peak into 8 ohm loads at 20 kHz, the figure you get is 7.04V/uS. Consequently, the 20V/uS spec appears to be quite all right.

My own feeling is that the logic is solid, however, I prefer to have 1V/uS per every peak output volt. In other words, twice what is considred as enough. But this must be EFFECTIVELY available, not on paper only, hence it must be the overall amp slew rate, input to output, as delivered to the user.

But the folly can also work backwards. What I routnely do, and I am not the only one, is to have the amp hit 1 MHz and above; consequently, my effective slew rate would be say around 500V/uS. However, I normally insert an input filter, limiting input frequency at say 500 kHz because there's no sense going over that. Thus, my INTERNAL amp slew rate remains the same, but my input signal slew rate is cut to one half of what the amp can do. The amp is therefore always twice as fast as the fastest possible (realistically possible) input signal.

No TIM, no TID, no SID.

Cheers,
DVV

Dan Banquer · « **Reply #5 on:** 17 Jun 2005, 05:07 pm »

I have a general rule of thumb; for every volt out rms you will need 1 volt per microsecond of for a slew rate.
Example: If I have a power amp that can give a maximum voltage into 8 ohms of 25 volts rms, then I will need a 25 volts per micro second slew rate to maintain a near flat reponse to 20 kHz at it's full power level.
Note: this is a general rule of thumb and not an absolute.
d.b.

DVV · « **Reply #6 on:** 17 Jun 2005, 07:22 pm »

Quote from: Dan Banquer

I have a general rule of thumb; for every volt out rms you will need 1 volt per microsecond of for a slew rate.
Example: If I have a power amp that can give a maximum voltage into 8 ohms of 25 volts rms, then I will need a 25 volts per micro second slew rate to maintain a near flat reponse to 20 kHz at it's full power level.
Note: this is a general rule of thumb and not an absolute.
d.b.

So the only difference between us is that you go for 1V RMS of output, while I go for 1V PEAK of output per 1V/uS.

Your amp delivering 25V RMS will have no less than 25 V/uS, while mine will have 35V/uS minimum, the peak value of 25V RMS.

Mine is the still more conservative appraoch, but I'd be hard pressed to say yours is not sufficient; whatever, a bit more surely can't hurt. If I like nervous, even neurotic cars, it stands to reason I'll probably be the same with amps.

Cheers,
DVV

koiman · « **Reply #7 on:** 18 Jun 2005, 03:10 pm »

How can i figuare out the slew rate from this information.. Is it possible?
Performance Features:

Power Output:
300 watts per channel continuous @ 8 ohms
450 watts per channel continuous @ 4 ohms
FR (1W @ 8 ohms 10Hz to 20kHz): + 0, - 0.25dB
S/N: "A" Weighted (at rated power @ 8 ohms): >105 dB
THD + Noise at full power @ 8 ohms:
20Hz 0.015%
1 kHz 0.020%
20 kHz 0.015%
THD + Noise at 450 watts @ 4 ohms:
20 Hz 0.025%
1khz 0.025%
20 kHz 0.0150%
Power Bandwidth (-3dB): 3 Hz to 130 kHz
Dynamic Headroom @ 4 ohms: 2.3 dB
IM Distortion (CCIF, full power from 4 kHz to 20 kHz): <0.075%
Gain: 29 dB
Input required for 1-watt output: 0.17 V
Input required for rated output: 1.7 V
Input Impedance:
105 K ohms unbalanced
10 K ohms balanced
Power Consumption (continuous, all channels driven):
Quiescent 540 VA
Maximum 1440 VA
Dimensions: 14" tall x 17" wide x and 8" deep
Weight: 48 lbs
Manufacturer's limited warranty: 2 year parts and labor

John Ashman · « **Reply #8 on:** 18 Jun 2005, 03:27 pm »

You can't.

But, it doesn't too much matter, I don't think there's a speaker made that could come close to mimicking the slew rate of an amplifier. As I understand it, it's simply not an issue. I've never heard of an amplifier that is too slow to reproduce the audible range. If the slew rate wasn't sufficient, the FR would obviously be rolled off because it wouldn't be fast enough to reproduce HF signals.

Dan Banquer · « **Reply #9 on:** 18 Jun 2005, 03:44 pm »

John A is correct. The FR bandwith states -3db down at 130 Khz so the slew rate, whatever it is, should be adequate enough from that spec alone.
d.b.

koiman · « **Reply #10 on:** 18 Jun 2005, 04:25 pm »

Thanks guy's you are fantastic..
Lee

DVV · « **Reply #11 on:** 18 Jun 2005, 04:51 pm »

Quote from: koiman

How can i figuare out the slew rate from this information.. Is it possible?
Performance Features:

Power Output:
300 watts per channel continuous @ 8 ohms
450 watts per channel continuous @ 4 ohms
FR (1W @ 8 ohms 10Hz to 20kHz): + 0, - 0.25dB
S/N: "A" Weighted (at rated power @ 8 ohms): >105 dB
THD + Noise at full power @ 8 ohms:
20Hz 0.015%
1 kHz 0.020%
20 kHz 0.015%
THD + Noise at 450 watts @ 4 ohms:
20 Hz 0.025%
1khz 0.025%
20 kHz 0.0150%
Power Bandwidth (-3dB): 3 Hz to 1 ...

From the power output and power bandwidth specification, calculation show the slew rate to be approximately 80V/uS.

Take this data with a pinch of salt - this is only a ROUGH approximation, not an exact figure. It could be a little less if the input stage is constrained (not very likely, though), and it could be a bit better if the power bandwidth spec was stated for given 20Hz-20kHz distortion figures (if allowed to be higher, the power bandwidth would also be somewhat higher, and consequently, the slew rate could be a bit higher as well). Either way, I think the deviation would be rather small.

You only need about half that to be free of transient distortion caused by slew rate phenomena.

Cheers,
DVV

John Ashman · « **Reply #12 on:** 18 Jun 2005, 05:03 pm »

Oops, yep, I just saw the 20kHz FR figure, not the 130kHz, so, yeah, you could roughly figure out the slew rate it would have to have to reproduce that response.

koiman · « **Reply #13 on:** 18 Jun 2005, 05:19 pm »

John,
So what is your best bet with the powerband width being (-3Db): 3 to 130Khz?
Lee

DVV · « **Reply #14 on:** 18 Jun 2005, 06:53 pm »

Quote from: koiman

John,
So what is your best bet with the powerband width being (-3Db): 3 to 130Khz?
Lee

Lee, why don't you use the following heueristic formula to work it out for yourself whenever you need it:

Sq.root {[P(8 ohms) x 2)]8} = ... x 0.141 x 0.628 x Bandwidth/10

Please note, this is a HEUERISTIC formula, meaning it's not terribly precise, it's meant only to give an indication, not an exact number. In my experience, it's real world error is about +/-5%.

In your case, assuming 300W/8 ohms and a power bandwidth of 130 kHz, by substituting, we have:

Sq. root {(2x300) x 8} = 69,3V peak.

69,3 x 0.141 = 9,77 x 0.628 = 6,14 x 13 = 79,77 V/uS

Cheers,
DVV

Dan Banquer · « **Reply #15 on:** 18 Jun 2005, 08:05 pm »

That sounds pretty close, to me.
Just as a note; when I do amp and pre amp designs I generally look at slew rate last. It's the full power bandwidth at full power with worst case load that I look at first. If the frequency response is flat or very near flat to 20 kHz then the slew rate will be adequate.
d.b.

John Ashman · « **Reply #16 on:** 18 Jun 2005, 11:24 pm »

Quote from: koiman

John,
So what is your best bet with the powerband width being (-3Db): 3 to 130Khz?
Lee

Hey, I said it could be done, I didn't say *I* could do it!

DVV · « **Reply #17 on:** 19 Jun 2005, 07:24 am »

Quote from: Dan Banquer

That sounds pretty close, to me.
Just as a note; when I do amp and pre amp designs I generally look at slew rate last. It's the full power bandwidth at full power with worst case load that I look at first. If the frequency response is flat or very near flat to 20 kHz then the slew rate will be adequate.
d.b.

Right on, Dan!

However, these days, you do have to add some spice, in view of 192/24 technology, so I do what you do, but at 100 kHz. If I'm at -1dB or less at 100 kHz, I'm home and dry (and so free to have a beer).

I'd love to read some comments here about how a slew rate can become potentially dangerous ... not in the sense of anything exploding, but in the sense of how it can begin to potentially negate its own benefits. THAT side, The Dark Side Of The Force if you like, is all too often never even mentioned.

Cheers,
DVV

Dan Banquer · « **Reply #18 on:** 19 Jun 2005, 12:12 pm »

Luke: you must come to the Dark Side!
Actually if you look at what goes on in the recording chain there is no need to extend the analog bandwidth of pre amps and amps. The bulk of the microphones used for recording roll off pretty steeply after 20 kHz, the analog filters in front of the A/D converters start the roll off after 20 kHz, and the latest studies show that there is nothing useable to the human ear after 22 KHz. The ultrasonics that some instruments create are literally unuseable. Tweeters tend to have some real problems when too much ultrasonic energy is introduced and can cause IM distortion.
There are some benefits for the recording studio at higher sampling rates. statisically less chance of intersample peaking, and slightly better transient response. For all the crap that I have read and observed I have found that from moitoring CD's at the analog output of a player there is little evidence that better transient response is a major issue, just a very minor issue.
During the mid 90's I was at a local AES meeting where a panel had recommended the following to the industry:
One sample rate for DVD Audio of 96 kHz. ( this is the same sample rate for DVD video) 24 bit pro, and 20 bit consumer. I or someone else recommended that the digital filter be flat to 20 kHz, have a transition slope of 6db to 12 db down at 40 kHz and the "brick wall" would start at 40KHz and be greater than 100 db down at 48 kHz.
This was obviously ignored by the industry., who created the "cluster f**k called DVD A.
In any case this is really all a moot point given the hyper compression used in too many recordings today. I am coming to the conclusion that fidelity will die with my generation, becasue the youth are being brainwashed so that hyper compression now sounds normal. I am hoping against hope for some miracle here.
d.b.

DVV · « **Reply #19 on:** 19 Jun 2005, 06:48 pm »

Actually Dan, I was referring to far more mundane issues, so we'll discuss your views as posted elsewhere.

What is often totally ignored is that slew rate is a function of the bandwidth. So is RF interefernce - go high enough, and you will start to have problems with it. And it may not be the worst - how about parasitic capacitance in many circuits?

They are not an issue when one uses general purpose op amps (i.e. dirt cheap), as found in many audio products, but if one substitutes a say TLO71 with a say National Seminconductor 6171, chances are one will have oscillation and likely RF problems as well. And the ONLY reason why wnybody would try the change is because the audio community has discovered ultra fast video op amps, read that they have a selw rate of 3,000V/uS or more, and somehow decided that this, and only this, was true fidelity.

Not to even mention that high speed, high slew rate op amps can, and I think most often do, show up circuit shortcomings elsewhere. For example, swap the input garden variety op amp with a fast, low noise op amp such as for example AD829, and the following garden variety op amp will have problems for sure. If not in passing on the signal, then in being exposed as what it is - a garden variety op amp.

Oddly enough, most don't even know any of this is even possible ...

Cheers,
DVV