OP AMP Capabilities in Supporting 24 bit bandwidth

Hello Everybody

Iam in the market in search of a Minimalist Preamp and got pulled towards Morrison ELAD, I saw this piece of message in Don Morrison Website:
(The most important thing about it is that its circuitry is designed around the Analog Devices AD797, a highly advanced op-amp made with a fully complementary IC process. This complex device, consisting of 60 transistors, settles to a full 16-bit resolution in under a microsecond, has a noise spec equivalent to a 50-ohm resistor from 10 Hz to 1MHz, and achieves lower THD + N levels than any discrete audio circuit that has come our way. The preamp consists of two AD797's with associated circuitry - including separate volume controls for each channel)

Looks like it can handle 16 bit resolution at ease, but, how about handling higher bandwidth signals at 24 bit resolution?

Any advice is highly appreciated.

Thank you

I don't know anything about this particular OP amp, but..

1...Every OP amp circuit I ever saw had about 60 transistors.
2... Every OP amp circuit I ever saw had a complementary circuit topography. (That's part of the reason for so many transistors).
3... "16 Bit resolution" is a roundabout way of saying it has a dynamic range of 96 dB.

It may be a good preamp and a good OP amp, but I get a bad feeling when specmanship is evident.
You seem to have locked on a bit of information presented for those digital engineers that have a problem in the analogue realm. Looking at the full spec sheet for the 797 you will find Slew Rate and two listings. 1 for "12.5 volts per microsecond" and one for "Settling Time to 0.0015%--10 volt step --800 nanoseconds" The 0.0015% is the precent value of the LSB for a 16 bit DAC, hence the 16 bit settling time. Analogue folks would have simply divided 1 microsecond by 12.5 volts and multiplied that by 10 and found that 800 nanoseconds was the time that the 797 required to respond to a 10 volt step. Now the difference between a 16 bit audio system and a 24 bit system is dynamic range NOT frequency responce, that depends upon the sampling rate. The full power bandwidth of the 797 at a gain of 10 is 8 MHz, much more than sufficient for any audio system. Back to dynamic range. A standard CD as a maximium dynamic range of 96 db, 6 db per bit, a 24 bit system then should have a dynamic range of 144 db but due to noise limites of the recording and playback loops you will find that the best units specify a range of 110 to 115 db. The 120db noise limit of the 797 exceeds those values by a fair margine and will add no audible noise to any system on the market.
I completely agree with cornelius. 24 bits is a dynamic range issue, not bandwith. The 797 was designed before 24 bit digital audio existed as a regular format, which explains the published information from Analog Devices (and the ELAD spec's). I suspect you may be asking about sampling rate - 96kHz or 192kHz which are also non issues with the 797. It's bandwith exceeds the requirements of 192kHz sample rate by many times.
On a more subjective note, I've had a chance to use an ELAD and it is completely transparent. If you want minimalist, it's a good pick.
I checked out this preamp, and find that it has only 6dB gain, with a switch option for unity gain. It is easy to get distortionless ("transparent") performance from a low or unity gain amp. The trick is to do it with about 30 dB or more gain. I guess you could call this item a "straight wire without gain".
Thank you all, your responses are giving me the most confidence in the ELAD unit. Iam going ahead with the purchase.
Re Eldartford's "30dB or more gain". 30dB of gain is the job of a power amp, not a preamp.
Let's do the math. Most CD players are capable of 1 volt rms output. Most amplifiers have an input sensitivity where 1 to 2 volts rms input will drive them to full output or beyond (and if your speakers are reasonably sensitive, that's gonna blow your head off). So if we put an additional 30dB of gain between your CD player and power amp that means we'll have 31.6 volts rms at the input of your power amp with 1V rms from your CD player. Do you smell smoke?
On another note, if it's so easy to get distortionless performance from a unity gain amp as suggested above, I don't understand why so many people have so much to say about simpler items - like connectors, wire, resistors, capacitors solder, pots, and op-amps. Oops! just built a pre-amp...
Svenss1...Yes, 30 dB would be a bit high for the line stage, but I was thinking in terms of a phono preamp, where considerably more than 30 dB overall is necessary. Also, the line stage usually has excess gain, which is attenuated by the volume control. The circuitry may actually be producing 4 or 5 volts, which is cut back to the 1 or so that it takes to drive the power amp.

On your "other note" I don't understand it either.
When i first read about the ELAD, i was reminded of the old AGI 511A, which i'm still very fond of today. Both products took a very straight-foward and scientific approach to audio circuit design and utilized IC's as the main source of gain. In effect, i'm quite certain that the design of the AGI was influential in how Morrison approached the ELAD.

Quite honestly, i think that David Spiegel of AGI is the Father of high speed, low noise, wide bandwith, high linearity audio circuitry i.e. the "straight wire with gain" school of audio design. Many before him set out to build products of that nature, but i think that Spiegel was the first to achieve the mass majority of those goals within the confines of one simple product. I just wish Spiegel would have actually produced the prototype amplifiers that i read about way back then. I'm certain that they too would have been very forward-thinking in terms of their design, implimentation of circuitry and sonic performance.

As far as the ELAD goes, it would probably be a nice piece for a very basic system. It is limited in terms of quantity and type of inputs and is very small in physical size. While it uses "good" quality parts, those parts can be bettered. Personally, i think that the sonics would improve along with upgrading those parts, but this is strictly my subjective opinion.

Other than that, Cornelius did a very nice job of summing things up. Svens1 also pointed out that the IC used has the bandwidth capacity and noise level below that of most commercially available gear, BUT, that doesn't mean that all of the support circuitry is up to the same capacity of the IC itself.

As Steve at Empirical will tell you, one can take the same circuit using the same exact parts and lay it out in various manners. Even though you've got the same circuit and the same parts, how it measures, performs and sounds could be very different due to the varying impedances involved. That's because impedances affect power transfer characteristics, transient response, crosstalk between channels, the residual noise floor of the circuit, etc...

As a side note, Stereophile had an article comparing the actual measured performance of several different IC's from various manufacturers a few years ago. I'm not certain, but i think that Ben Duncan wrote the article and provided the test results. Ben has done quite a bit of interesting audio related research and had stumbled across FM distortions within cabling. That is another matter for another thread though : )

Even though the IC's shared the same part number and should have been identically interchangeable in a circuit due to sharing industry spec's, some brands tested FAR inferior to others. Going out of memory, i think that some models had a noise floor of -145 dB's ( PHENOMENALLY good ) to where other so-called "identical" IC's of the same part number measured in at -85 dB's. This is a HUGE difference of 60 dB's While the "bookworm" type of engineers will look at those spec's and say that both are well beyond the audible limits, i have to resort back to the earlier comments made about circuit design and lay-out.

If you start off with a "noisier" part, and then install that into an inferior circuit design and / or poorer circuit lay-out, you won't even be able to reach the full potential of that "junky" part. As such, the difference between an optimally configured -145 dB part in a really good circuit and the much noisier -85 dB part configured in a poor circuit with a bad lay-out can become even wider than the aforementioned 60 dB's. Whether or not this is audible is still subjective, but to my way of thinking based on past experiments and experiences, i personally think that it is.

With all of that in mind, i think that The Audio Critic actually tested and measured an ELAD. From what i can recall, it did quite well on the test bench, demonstrating excellent spec's in most every category. Obviously, whether or not this unit will deliver what they want sonically is a matter of personal tastes and system synergy. Cosmetically, the unit doesn't have much going for it, but then again, neither did the AGI. Both units were built with form following function, so looks were a secondary consideration. For as much of a "straight wire with gain" design as Spiegel set out to make the AGI, it was a tremendously versatile unit. The ELAD is a much more bare-bones approach and lacks much of that versatility / switching arrangements.

As a side note, Morrison also produces some interesting speakers too. I've never heard them, but they make use of a relatively novel and complex to build bass alignment that was developed by Stuart Hegemann many years ago. I've always wanted to experiment with this type of a bass alignment, but never gotten around to it. I can see it having multiple benefits, especially if using a woofer with a high impedance peak at resonance. Then again, i would avoid ever choosing a driver like that to begin with, so i don't know how beneficial this approach would be using a higher quality driver to begin with : ) Sean
My first reaction to Sean's comment about the benefit of a noise floor much lower than other ICs and far below audibility is "so what". But then I remembered my own experience in designing tests of complex electromechanical equipment, and my insistance, (against strong resistance from bosses and even the customer, the US Navy) on implementing procedures that were significantly more accurate than the test accuracy spec we had to meet. My argument was that there will be sources of test error that will come to light later, and so you do each part of the procedure as well as you know how so as to leave room for these circumstances. More than once there was near panic when a new error source was discovered, and I was a hero when I could stand up and say "it's OK. We still meet requirements".

But, this (overly?) conservative approach did cost you poor taxpayers some money :-)
El: That's why i said "bookworm" engineers i.e. those that design things so that they look good on paper according to the theories and formulas in their textbooks, but really have no idea as to how things will work under various operating conditions in the field. These are the kind of idiots that should be made to use their own products that they design. Then they could see how we as end users have to deal with all the flaws, glitches, quirks during normal use. After that bit of education, then they should be forced to repair these items once they fail. Just going through the process of having to figure out why it broke, how to repair it and how / where to get the necessary parts would blow their mind. I think that both their design philosophies and design prowess would both change for the better i.e. less "design revisions" and more getting it right the first time.

Other than that, i've always thought that one should design and build for the "worst case" scenario. If you do that, any other type of situation is simply easier on the gear. Then again, something that has to strain to keep up or get the job done obviously won't work as reliably in the long term and / or may provide less consistent results in the short term. At the same time, keeping things simple also helps as there is less to go wrong. More unnecessary features, functions and gadgets only means a longer signal path with more connections and things to go wrong. In that respect, the ELAD is based upon several different tried and true design premises. Sean
The dean of the mechanical engineering school where I went told us that the final exam would be to design a ten ton crane hook, and stand under it as full load was applied.

For most things, worst case analysis is not conservative enough...a "safety factor" is added. However, sometimes you must go with something less than worst case, so that a failure is not impossible. The best example of this that I know is the plumbing of a multifloor apartment house. You simply cannot design for the situation where every toilet in the building is flushed at the same moment. Few airplanes have the structural strength to withstand aerodynamic forces that the pilot can easily cause by manipulation of the controls. (And I fly to Arazona tomorrow).
Sean...And to lower the tone a bit further, consider the design of residential sewage disposal fields. For ages everyone designed for "worst case" and then made it larger for "good measure".

Turns out that an oversized field is a bad idea. Parts of it will dry out, and the microbes that break down the sewage will die.

In audio (remember that) I once thought to improve a regulated power supply by adding capacitance. Bad idea. The regulator became unstable.

So I guess that every situation must be understood and accomodated in the design.
I agree that everything has to be factored in. Then again, if EVERYTHING is built "gonzo", nothing is left to go wrong : )

As to regulation circuits, most are piss poor. This is why there is a lack of dynamic headroom in regulated amps. That's because the regulation circuit is both too small and too slow to pass the necessary amount of current that the circuit requires on a dynamic basis.

Since the entire amp would be pulling juice through the regulation circuit, the regulation circuit itself should be sturdier with more capacity than both channels of the driver and output stages combined. I've never seen a regulated amp built to anywhere near this level, let alone heard anyone ever discuss such things. Sean

If you are not happy with AD797 opamps, try LME49990.