Measurements of roll-off provided by M5-HP at various impedance settings into my power amplifier

[Ross B.]

I received my new (to me) Quatro's yesterday, and first things first, checked and changed the batteries in the M5-HP high-pass filters. As you know, proper function of these filters is essential to proper, and safe, function of the Quatro's. Once I had completed my work, I thought to connect the filters to my measurement rig and get some data!

The attached graphs were measured at the output of my hand-built First Watt BA-3 Class A power amplifier (~45W into 8 ohms), which has a nominal input impedance of 47kohm. The first measurement I ran was with the M5-HP disconnected, and represents the nearly flat top line labeled "Loopback". Then, I inserted the M5-HP with the dip switches set to 50k, the closest value to my amplifier. That measurement is represented by the green line. For comparison, I then ran tests at 75k and 33k targets set on the M5-HP. These are represented by the purple, and orange lines, respectively.

A first-order high-pass filter has a roll-off curve that is down 3dB at the target frequency, in this case 100 Hz. My Loopback test established a baseline level of 61 dB, so the ideal filter configuration should cross 100 Hz at 58 dB, which we see is exactly achieved by the orange curve, or the 33k setting in the M5-HP. The ostensibly "more correct" 50k setting is only 0.5 dB down from the target, which is likely an inaudible difference in practice. The 75k setting meanwhile deviates by a full 2 dB, which starts to reach the point of perception.

Thank you for coming to my Ted Talk. I hope you have found these measurements as interesting and informative as I have!

[Richard Vandersteen]

46 minutes ago, Ross B. said:

I received my new (to me) Quatro's yesterday, and first things first, checked and changed the batteries in the M5-HP high-pass filters. As you know, proper function of these filters is essential to proper, and safe, function of the Quatro's. Once I had completed my work, I thought to connect the filters to my measurement rig and get some data!

The attached graphs were measured at the output of my hand-built First Watt BA-3 Class A power amplifier (~45W into 8 ohms), which has a nominal input impedance of 47kohm. The first measurement I ran was with the M5-HP disconnected, and represents the nearly flat top line labeled "Loopback". Then, I inserted the M5-HP with the dip switches set to 50k, the closest value to my amplifier. That measurement is represented by the green line. For comparison, I then ran tests at 75k and 33k targets set on the M5-HP. These are represented by the purple, and orange lines, respectively.

A first-order high-pass filter has a roll-off curve that is down 3dB at the target frequency, in this case 100 Hz. My Loopback test established a baseline level of 61 dB, so the ideal filter configuration should cross 100 Hz at 58 dB, which we see is exactly achieved by the orange curve, or the 33k setting in the M5-HP. The ostensibly "more correct" 50k setting is only 0.5 dB down from the target, which is likely an inaudible difference in practice. The 75k setting meanwhile deviates by a full 2 dB, which starts to reach the point of perception.

Thank you for coming to my Ted Talk. I hope you have found these measurements as interesting and informative as I have!

Ross B, so the actual input impedance of your amp is closer to 33K not the 47k specified by the manufacture!  RV

[Richard Vandersteen]

Just now, Richard Vandersteen said:

Ross B, so the actual input impedance of your amp is closer to 33K not the 47k specified by the manufacture!  RV

Ross B, the small difference may not be audible, but the point is it can involve a big change in Mid-woofer excursion.  It is critical the high-pass is correct to minimize damage to a very expensive driver.  RV

[Ross B.]

12 minutes ago, Richard Vandersteen said:

Ross B, so the actual input impedance of your amp is closer to 33K not the 47k specified by the manufacture!  RV

Yes! In this case, 47k is the impedance to ground at the input, so presumably the amplification circuit also affects the impedance. I haven’t measured it experimentally, I will have to do so, or inquire. It goes to show that, while specs may be close, with time and some effort, further optimizations may be possible!

[Ross B.]

11 minutes ago, Richard Vandersteen said:

Ross B, the small difference may not be audible, but the point is it can involve a big change in Mid-woofer excursion.  It is critical the high-pass is correct to minimize damage to a very expensive driver.  RV

Right! It’s about what the response looks like when reconstructed inside the speaker. Presumably a value that causes the curve to fall *below* the target curve would be safer than the alternative - simply over-reducing the bass. Of course a full measurement with the interactions from the internal rectification circuitry would be needed to be sure.

[Richard Vandersteen]

12 minutes ago, Ross B. said:

Yes! In this case, 47k is the impedance to ground at the input, so presumably the amplification circuit also affects the impedance. I haven’t measured it experimentally, I will have to do so, or inquire. It goes to show that, while specs may be close, with time and some effort, further optimizations may be possible!

Ross B, this is why every competent Vandersteen Dealer uses the test disc/voltmeter method shown in a video on this site to verify the input impedance as they are rarely correct but very important.  I recommend every owner do this if a dealer has not.  RV

[Ross B.]

I just ran the high-pass calibration from the Vandertones pack. As it turns out, 33k setting that produced the most ideal curve was actually not rolled off enough! Adjusting back to 50k reduced the voltage level to the correct ~0.707v range. Whether my test is slightly off, or the true calibration for the filter is slightly low, the point is, the documentation works! Trust it!

[Holmz]

3 hours ago, Ross B. said:

I just ran the high-pass calibration from the Vandertones pack. As it turns out, 33k setting that produced the most ideal curve was actually not rolled off enough! Adjusting back to 50k reduced the voltage level to the correct ~0.707v range. Whether my test is slightly off, or the true calibration for the filter is slightly low, the point is, the documentation works! Trust it!

Yeah but nothing beats the experience of doing it for yourself.
Now you know.

[GdnrBob]

I really have to do this myself, as well. 

14 hours ago, Holmz said:

Yeah but nothing beats the experience of doing it for yourself.

Knowledge is Power.

B