Quote:
Do any of you think that there may be certain products that are more or less apt to be

Krell made themselves "cable insensitive" by creating a proprietary link between components, which use proprietary cables that Krell sells for about $500/meter pair. They say something about the proprietary interconnections being "current driven" rather than "voltage driven". I'm not sure what that means and I'm not sure if it makes any difference.

As far as components using regular cables, I do find that certain components are less sensitive to cables than others. I think McIntosh components are less sensitive and VTL components tending to be more sensitive.

Quote:
They say something about the proprietary interconnections being "current driven" rather than "voltage driven". I'm not sure what that means and I'm not sure if it makes any difference.

With the Krell "current driven" topology, the line-level drivers (e.g. preamps) are constant current sources and the sinks (e.g. amps) are low-impedance current sinks. As the name implies, a constant current source will attempt to drive the same current regardless of the load impedance. This load impedance includes, of course, the interconnecting cable as well as the input impedance of the receiving sink. An ideal current source will have infinite source impedance and and ideal current sink will have zero input impedance. If the source and sink are correctly designed, the cable impedance will have inconsequential effect on the current arriving at the sink.

All other components I know of are either voltage sources or voltage sinks. A voltage source, of course, will attempt to drive the same voltage regardless of the load impedance. An ideal voltage source has zero impedance and an ideal voltage sink has infinite impedance. Once again, if the source and sink are properly designed, the interconnecting cable has inconsequential effect on the voltage arriving at the sink.

Bottom line, either way (current mode or voltage mode) will work just fine (minimal effect from interconnecting cable) if correctly designed, and I have no idea why Krell chose this unique "current mode" design.

Boulder claims that if you use their preamp and amps together because of specifically chosen output and input impedance as long as cables meet a given lever of quality there is no difference above that quality level in cable. They also believe that with respect to connecting their sources to preamp.

It makes SOME sense, but it sounds more like a marketing way of "encouraging" customers to have an all Boulder system. $14,000 phono stage anyone? I bet most manufacturers gear works A LITTLE better if paired together.

Hi Buddha,

If I may add some more info, I would say some solid state amps can be more sensitive, esp if they use FETs input devices. At least some FETs have fairly high, to high junction capacitance. Some one might have more info on this subject. This capacitance fluctuates with Vds voltage, the drain to source voltage until a minimum Vds is reached. The IRD-820, for instance is 25 volts.

As far as capacitance and FR in tube preamplifiers, if the output impedance (Z) of the preamplifier is changed from, say 2,000 ohms to 100 ohms, and we use a high capacitance 250pf IC, the high frequency response changes approx 0.4db at 100,000hz (100khz). If the IC is a low 50pf of capacitance, the high frequency response is reduced less than 0.02db at 100khz, so virtually none. (Using cirmaker program.)

However, some tubes like the 6SN7 and especially the 12AX7/12AT7, with their high plate resistance can be reduced 2db+ (6SN7) to 17db at 100khz (12AX7) if precautions are not followed.

This generally means that some sort of high feedback output buffer stage or stepdown output transformer is necessary to lower the output impedance.

Hope this helps.

Quote:
The output impedance of a given component will dramatically affect how sensitive that component is to the interconnect cable's capacitance. A high output impedance (such as some tube preamps) combined with a high capacitance cable will audibly roll off the high frequencies. A low output impedance component will have flatter high frequency response with the same cables.

This is the correct answer. Related, this means preamps should be voltage sources, not current sources. A voltage source has an infinitely low output impedance. A current source is infinitely high, and thus is affected directly by a capacitive load.

--Ethan

Quote:
This is the correct answer. Related, this means preamps should be voltage sources, not current sources. A voltage source has an infinitely low output impedance. A current source is infinitely high, and thus is affected directly by a capacitive load.

But the input impedance of the receiving end in a current mode topology should be as close to zero as possible. As I described in my previous post, either properly designed voltage mode or current mode systems should work well to minimize any effects of the interconnecting cable.

Searching for any technical details on Krell's CAST (Current Audio Signal Transmission) only found technobabble marketing descriptions. A glaring error was found on their web page describing their Evolution Two preamp where the main regulators of the output stage are described as:

Quote:
Operating in current mode, with fully complementary gain stages and output drivers, wide bandwidth, low output impedance, and high current capability, the main regulators easily exceed the requirements of the analog stages.

I'd guess that the power regulators are constant voltage sources, not current sources. I also found a posting from an engineer (Victor) on the HTGuide forum that included the lines:

Quote:
The CAST signal transmission is a fancy word for a simple and cheap circuit that connects the current output trans-conductance output stage to another stage that expects the current input. This is great as it is esoteric, but an engineer in me asks, - why do it? I tell you this; - it would surprise me to no end if a sensible reason for such techno-acrobatics is actually found.

Quote:
Searching for any technical details on Krell's CAST (Current Audio Signal Transmission) only found technobabble marketing descriptions.

Always frustrating. Either really tell me how it works or don't bother. And don't mislead me!

I love the phrase "techno-acrobatics" in the quoted post.

"I'd guess that the power regulators are constant voltage sources, not current sources. I also found a posting from an engineer (Victor) on the HTGuide forum that included the lines:
The CAST signal transmission is a fancy word for a simple and cheap circuit that connects the current output trans-conductance output stage to another stage that expects the current input. This is great as it is esoteric, but an engineer in me asks, - why do it? I tell you this; - it would surprise me to no end if a sensible reason for such techno-acrobatics is actually found."

Maybe...but in my experience Krell rarely, if ever, just does somehting for how it sounds in a press release. Those guys seriously do not mess around! Ever hear a reviewer give any Krell a bad review? Crazy expensive- yes, in general. Over built equipment that is often best in class- yes that too. I'd have to hear it, but by experience and reputation I'd be willing to give them the benefit of the doubt.

I completely agree that Krell designs and manufactures top-tier audio equipment. I do have diffuculty, however, with the marketing approach used by Krell and many other vendors who attempt to distinguish their products by throwing out these very unrigorous technical descriptions of their products. Very few in their target audience have sufficient engineering background to evaluate that the technologies employed by product A would actually produce a more pleasing sound than the technologies employed in product B. It would be much better I think to describe the overall system goals and benefits to the customer.

I've been reviewing current source theory since reading this thread. It's said that the theoretically perfect current source has infinite output impedance. But no current can flow through an infinite impedance, even if the load impedance is zero. How then, can it be a source of anything? Can any of our engineering-minded colleagues unravel this apparent paradox I've created in my mind?

Quote:
But the input impedance of the receiving end in a current mode topology should be as close to zero as possible.

I can't address specific issues with Krell. But I can tell you that every other piece of audio gear in the world is voltage-based.

--Ethan

Thinking about this a little differently; if there is a two-terminal current source in a circuit that may include other voltage or current sources, and the current that source is attempting to drive is zero amps, that source will appear as an open circuit (infinite impedance) regardless of the voltage across its terminals. Have a look at the Wikipedia page for current source.

Ah, the internal resistance is infinite with zero current. It makes sense now. I had seen the Wikipedia entry before, but somehow missed that point. Thank you very much for your reply, Charly.