Help me integrate my AI pre. Educate me on impedance.

Mathew_M · « **on:** 6 Feb 2003, 04:33 pm »

Ok I need a little help here. I just bought an Audible Illusions 3A pre that I'm having some problems integrating into my system. My jolida tube cdp overall sounds good with the AI, however it outputs a lot of gain into the already high gain of the AI. This gives me very little headroom for volume control with the AI's stepped attenuators. I have a local buddy who is going to make me up some specially attenuated cables that will hopefully help out.

Another problem is trying to match the high output impedance of the AI (1.2 kohms) to the low input impedance of my Norh Le Amps (22 kohms per IRD's site). The Le amps just squeak by according to AI's 20k - 400k ohms recommendation. This would have to be effecting the sound right?Reading the AI manual, they recommend an amp with a 45k impedance. Then I've also read that one should have a ratio 100:1 when matching pre to amp. This means I would need an amp with an input impedance of about 120 kohms, right? The AI manual also states that one can have a technician alter the amp's input impedance without too much trouble.

any suggestions?

Juan R · « **Reply #1 on:** 6 Feb 2003, 04:46 pm »

I think the ratio is 10 to 1, I have a kora eclipse with high output of 600ohms, low output of 1.5kohms. both output sound good with leamp.

audioengr · « **Reply #2 on:** 6 Feb 2003, 07:33 pm »

First, make sure that your friend puts the voltage divider at the receiving end of the cable, not the driving end. Make sure to use resistors in the 1K ohm range for the divider.

Second, since the output impedance of the driver is 1.2K, this is very high. This is more important than whether the input impedance is 10K or 100K. To make this sound good and avoid HF roll-off, you will need a low-capacitance cable, preferably lower than 10 pF/foot.

Dan Banquer · « **Reply #3 on:** 6 Feb 2003, 07:56 pm »

Mr. Audio Engineer: could you please explain to me how you came to the conclusion that 10 pf per foot was required in this situation?

Beezer · « **Reply #4 on:** 6 Feb 2003, 09:36 pm »

In my AI M3A, you can change lower the overall gain via some internal jumpers. Instructions should be in the manual. Doing this will give you more range with the volume control. Also, I'm using mine with an Odyssey amp, which has a 10kohm input impedance, with no problems at all.

Beez

Mathew_M · « **Reply #5 on:** 7 Feb 2003, 12:20 am »

I've already set the internal switch to low and it's an improvement but not enough for my small room and late night listening habits. At 9 o'clock (on volume dial) it is too loud and the previous increment is too soft with loss of dynamics and imaging. Mind you this is with the tape input and jolida cdp. The other inputs have more headroom and I have no problems with the phono section. Why do I want to stick with the Jolida and tape input? Easy, it sounds by far the best.

I guess amp impedance doesn't mean a whole lot then? I've found so many various answers here and on Audio Asylum. Down the road I'm looking at perhaps tube amplification which most seem have impedances ranging from 100k (vtl) to 450kohms (old dynaco amps). What are some of the abnormalities that I would hear with the a poor impedance match?

So Beezer, you don't have any volume problems? Your model is probably a similar year as mine (with internal dip switches). What are you using as your digital source?

I suppose I should also ask if the cables I am using are causing me problems. Yes, currently I'm using radio shack grade and will upgrade sometime soon hopefully.

audioengr · « **Reply #6 on:** 7 Feb 2003, 02:11 am »

Quote

I've already set the internal switch to low and it's an improvement but not enough for my small room and late night listening habits. At 9 o'clock (on volume dial) it is too loud and the previous increment is too soft with loss of dynamics and imaging. Mind you this is with the tape input and jolida cdp. The other inputs have more headroom and I have no problems with the phono section. Why do I want to stick with the Jolida and tape input? Easy, it sounds by far the best.

If you want the sound not to degrade, I would recommend that you change the resistors in the stepped attenuator, instead of adding a voltage divider to the cable. This is a real easy thing to do. Just jump across the reisistor that you are most likely to select with another resistor, which will reduce its value by paralleling the two resistors. The combined resistance will be:

(R1*R2)/(R1+R2)

This way, you can determine the exact resistance that you will need to order to get the volumes that you like.

Quote

I guess amp impedance doesn't mean a whole lot then? I've found so many various answers here and on Audio Asylum. Down the road I'm looking at perhaps tube amplification which most seem have impedances ranging from 100k (vtl) to 450kohms (old dynaco amps). What are some of the abnormalities that I would hear with the a poor impedance match?

You are correct, the input impedance is not too much of a concern unless it is perhaps 600 ohms. If your output impedance is too high, however, and you use a highly capacitive cable, you will get HF roll-off. You will lose all of the air and cymbols will not have any sheen etc...

audioengr · « **Reply #7 on:** 7 Feb 2003, 02:27 am »

Dan Wrote:

Quote

Mr. Audio Engineer: could you please explain to me how you came to the conclusion that 10 pf per foot was required in this situation?

Sure. My experience, and particularly the ten or twenty thousand SPICE simulations that I have run to find out what capacitance is a problem at what cable length. Believe me when I tell you that 10 pF.foot can be audible on a 1m IC even with a 10 ohm preamp output impedance, like a Mark Levinson. With a 1K output impedance, this will be even more obvious.

If you do the math, the actual dB drop at 20KhZ is only .04 dB, so this should not be audible. If we were to take a typical 1m Monster cable as an example (50 pF/foot), the dB drop at 20Khz is still only .2 dB, however the phase shift will be significant. But this does not tell the whole story. If you do the SPICE simulations, you will find that the transistors or tubes are not ideal drivers and the result is actually much worse when you simulate it.

Mathew_M · « **Reply #8 on:** 7 Feb 2003, 03:34 am »

Quote

If you want the sound not to degrade, I would recommend that you change the resistors in the stepped attenuator, instead of adding a voltage divider to the cable. This is a real easy thing to do. Just jump across the reisistor that you are most likely to select with another resistor, which will reduce its value by paralleling the two resistors. The combined resistance will be:

(R1*R2)/(R1+R2)

This way, you can determine the exact resistance that you will need to order to get the volumes that you like.

Steve,
An interesting suggestion. I know about zilch (comparatively) when it comes to the internals of electronics; though I am (somewhat) enjoying the trial and error process with the audio hobby. So how would one such as myself with limited skills go about changing the resistor values as you describe? Actually I'm lucky enough to know a guy who could help me here (same guy that is supposed to whip me up the voltage divided cables). He is extremely busy right now though.

Beezer · « **Reply #9 on:** 7 Feb 2003, 03:52 am »

I have a Phillips SACD1000, also use the tape input, and rarely turn the volume up past 10a. The step increments work OK for me, although it sounds like I might have more latitude with my listening volumes. I haven't checked out those internal dips yet. I wonder if they don't affect the tape out for some reason?

Beez

Dan Banquer · « **Reply #10 on:** 7 Feb 2003, 11:19 am »

To Mr. Audio Engineer: I would recheck your math. Given that the pre amp in question has a 1.2 K ohm output impedance and a hypothetical load of 500 picofarads the 3 db point wil be 265.3 kHz. The 1 db point should be roughly half that at 132 kHz. Flat to 20 kHz should not be an issue. Distortion, however could well be another issue, but this is a tube pre amp so I don't think this will be a significant issue for the customer. I would like to further point out that in all cases for distortion this should be measured, not assumed, because I am reading lots of claims these days but no actual numbers to back them up. I would like to futher the point that many units out there with a 100 ohm output impedance can drive 10,000 picofarads in parrallel with 20K ohms of resistance and only be down 0.1db at 20 kHz. How do I know this? I have measured them.
Again; recheck your math, and measurement technique.

audioengr · « **Reply #11 on:** 7 Feb 2003, 07:40 pm »

Dan - these are exactly the numbers that I came up with. 265K ohms reactance. .04 dB down for the 10 pF/foot cable and and .2 dB down for the 50 pF/foot cable. I dont think there was a math error.

Mathew_M · « **Reply #12 on:** 8 Feb 2003, 04:36 am »

Well my tech buddy came through with a set of voltage divided cables. Didn't get a chance to ask him about the details but he did give me the audiophile grade thumbs up. They work...sort of. I do believe in break in of components and so I'll give them a couple of days. Plenty more volume control; I currently have them going from the jolida to AI and have a huuuge range (almost too much). I don't hear any roll off though they do sound harsh and flat which hopefully will be alleviated with break in.

Dan Banquer · « **Reply #13 on:** 9 Feb 2003, 07:40 pm »

Quote from: audioengr

Dan - these are exactly the numbers that I came up with. 265K ohms reactance. .04 dB down for the 10 pF/foot cable and and .2 dB down for the 50 pF/foot cable. I dont think there was a math error.

The formula that you used is for the -3db point , not reactance. I believe the formula that you should be using is the following:
-3db = 1 divided by 2 times pi times R times C
With that formula the -3 db point using 500 picofarads of capacitiance and 1200 ohms of resistance equals 265.3 kHz.

audioengr · « **Reply #14 on:** 11 Feb 2003, 07:35 pm »

Dan - the formula that I used was the formula for a voltage divider: V2 = V1 ((R2/(R1+R2))) and the reactance of the cable: Xc = 1/(2*[pi*f*C], where f=20kHz. I didn't use any -3dB formula.

Dan Banquer · « **Reply #15 on:** 11 Feb 2003, 07:46 pm »

Now I see where the problem is. You should know that for all Solid State inputs there are low pass RC filters. The choice of R and C are designers choice here. The capacitance of the RC filter is in parrallel with the input impedance setting resistor. When input impedance is typically stated, it is the value of this resistor alone, and does not include the parallel capacitance of the input low pass filter. As you can see form the above the input impedance at 20 kHz is already lower without any cable involved.

audioengr · « **Reply #16 on:** 12 Feb 2003, 02:34 am »

I imagine that most components do have input filters, however, not all of them. Some may have none or limit the B/W in the feedback gain-control for the input stage. Probably a good design practice, however.

When I model an input of 10K ohms in SPICE, I usually put a 20 pF cap in parallel. This is just the capacitance of the input connector and wiring to the circuit board.

Dan Banquer · « **Reply #17 on:** 12 Feb 2003, 11:25 am »

Try using an RC filter with values such as 1 k ohm and 470pf to 680 pf. The 3 db points from the filter above are more indicative of typical input filter responses.

Curt · « **Reply #18 on:** 12 Feb 2003, 02:16 pm »

Quote from: audioengr

I imagine that most components do have input filters, however, not all of them. Some may have none or limit the B/W in the feedback gain-control for the input stage. Probably a good design practice, however.

The IC itself (along with the sources output impedance) forms and becomes a lowpass filter. All cables (more than one wire) cause LP filtering to some degree. Usually, only when it get's too close to the audio range does it become a problem.

The Fc (-3dB) point Dan calculated is correct, the formula is correct, and in this case should not effect the sound.

audioengr · « **Reply #19 on:** 12 Feb 2003, 09:15 pm »

Dan B is correct, and so am I. We are just looking at the problem from a little different perspective. I dont pay too much attention to -3 dB points, because I can hear degradation at -1dB, from both amplitude and phase-shift. -3 dB points are interesting for doing comparison of amplifier responses etc..., but I dont find them particularly useful for cables. I am more interested in the actual attenuation and phase shift at a particular frequency, such as 20kHz.

As for audibility, each of us has our own opinion. I just happen to be able to hear the difference with my system. I must have a highly resolving system in order to effectively design cable. Whether others can hear a difference depends on a lot of variables, including all of their system components, the environment and their own hearing acuity.