[Dan Banquer]

Mr. Audioengr:
         You are entitled to any subjective opinion you like. However; If you make a claim based on improperly done PSpice models or incorrect measurement technique, than some of us are likely to be asking you questions on how you came to such conclusions.

[Dan Banquer]

I would like to proceed on one point in particlular. RC filters to the best of my knowledge do not show phase shift until the frequency response starts to roll off. In the flat area of the bandwith there is no phase shift that I am aware of. Group delay should also be uniform here also.

[audioengr]

Dan - what I have found after thousands of simulations is that the phase starts changing in a significant manner (my definition) well before the amplitude starts to attenuate.  If you select arbitrary limits, like -1dB down and 5 degrees of phase shift, and you simulate various lengths of cables, you will find that you hit the 5 degrees phase shift at much shorter cable lengths than the -1dB attenuation.  You can certainly argue whether humans can hear each of these equally well, but that is my hypothesis.  If you just objectively examine the knees of the amplitude and phase curves, it is obvious that the phase knee starts moving at a lower frequency than the amplitude knee.

[Dan Banquer]

I am going to attempt to be as polite as I possibly can here. It appears that your definition of phase and the rest of the worlds definition of phase are two different things. It also appears that your spice models are not in line with standard engineering definitions. It also appears that you are totally unconcerned about this.
I am concerned about this type of "marketing" and I use the term marketing since it has no basis in standard engineering practice and theory as we presently know it. My concern will continue to be shown at audio circles whenever the need arises.

[Curt]

I should really be working but…

Actually, in an RC filter the phase does start shifting before the signal has any noticeable attenuation. It can be seen with an oscilloscope.

For a low pass filter: at high freq the capacitors reactance is very small and the phase shift is almost nil, at the fc point the caps reactance equals the dc resistance and the phase shift is 45 degrees, and at low frequency the cap reactance is very high compared to the resistance and the phase shift is almost 90 degrees.

Phase shift in cables is a bad thing because it screws up all we do to keep the music in phase at all frequencies and time aligned at the listeners ears. Although, this is not normally a problem with reasonably good quality cables. I like to say away from weird designer flavor cables, simpler is usually better.

If you set up an oscilloscope to show both, a sine wave going into a RC filter (well away from fc) and the same sine wave coming out and then put them at the same scale and on the same baseline on top of each other they look like a single wave form.

Then start to adjust the signal generator in the direction of fc. As the capacitive reactance starts to "kick-in" becoming larger the phase starts to shift and now you can see the sine waves start to separate on the scope, amplitude still about the same but the filtered sine wave is shifting in phase. This starts to happen well before the -3dB fc point and reaches about 45 degrees at -3dB.

Interesting this electronics stuff...


Tips 4 Mathew_M :

1. I have put voltage dividers right on the input RCAs before, inside the preamp box. Sometimes this is easy and cheaper. Just a thought.

2. Your MB-100/Leamp amps are very easy to drive. Listen to them for a while and see what you think.

The load stiffness “rule of thumb” 10:1 or 100:1 originally was for voltage sources that can not drive a load, do not have enough output current. Not really an original audio rule.

Most audio sources easily drive a 22.1K load. If you do change the input impedance of a device and go to high, say 100K plus on the input impedance you can unbalance the input bias current and actually make excessive noise. You can wind up worse off.

I would listen first if all is OK do nothing. But, if your not happy your not stuck.

If you feel the dynamics are missing simply change R4 (22.1K) on the amplifier PCB to 47.5K MF 1%. Then your amplifier input impedance will be 47.5K. I would go no higher than that.

You should be able to change it from the top of the PCB without ripping the entire amp apart.

BTW : This can be done at our service center without voiding the warranty.
(I had to say that)

[JohnR]

Since I'm desperately procrastinating writing a boring report for work, I just ran a Spice simulation. I suppose I should do the math but I'm too lazy With a 1.2k ohm source, 500 pF in parallel with 10k as the load, I get 5 dB of phase shift at around 26 kHz. At that point, the amplitude is down 0.033 dB relative to 100 Hz.

So I don't see anything wrong with Steve's definition of phase either...

JohnR

[Dan Banquer]

As you do your simulation be careful here. Spice has a wondeful attribute of converting calcultions into radians instead of degrees. Please recheck your spice simulations. I also don't understand the term of "5 db of phase shift" please explain. If the figure of .033 db down at 26 Khz is correct than as far as I am concerned the effect is negligible at 20 kHz. Also the 500 pf theoretical figure represents 5 meters of Belden 8422 in the unbalanced configuration that I use. The 500 pf figure was deliberately used to show how much capacitance can be "tolerated" and still be less than than .03 db down at 20 kHz.

[Curt]

I agree with Dan here. The effects (exaggerated capacitance for effect) we're discussing, as interesting as they are, are not large enough to be any problem in the real world. We should all simply understand the basic principles to avoid the possible pitfalls.

Long cable runs should be looked at a little more closely (more cap + inductance[not yet mentioned]) and, as mentioned somewhere earlier in the thread, be aware that some may be more sensitive to these small shifts than others. (i.e. if you hear it fix it; if you can’t hear it why worry too much)

Hi JohnR. For the record phase shift is normally described in linear terms not logarithmic. Was that just a typeO?

BTW, glad to hear everyone likes to use simulations. These can be very good tools but, you have to keep always keep a real world reference in mind to let you know if something (such as setting up the model) goes wrong. If you get weird results you should know you got weird results and question them.

I see many new electronics guys who grew up on computers depend too much on these tools and trust them without question.

I have been using circuit simulation as a design tool since the 80s (Berkley Spice; Intuspice; Pspice; WorkBench; MultiSIM) and can tell you from experience that things can and do go wrong and results can also be misinterpret. All in all wonderful tools that can cut your prototyping to a minimum (do always prototype to be safe) saving big bucks and time.

Nice discussion...

[Dan Banquer]

One of the engineers that I work and I simulated the RC filter in question this morning and found that phase shift at 20 kHz was somewhere in the millidegrees area. I don't get excited about millidegrees of phase shift at 20 kHz. Think about it John R. ; If your 0.03 db down at 26 kHz your probably close to 0.01 db down at 20 kHz and how much phase shift can you have at 0.01 db down using an RC filter?

[Dan Banquer]

One other thing you may wish to double check is your generator output impedance and how much is this adding to your measurement.

[JohnR]

Right, I meant degrees.

JohnR