[Daverz]

For those not familiar with the Vandersteen M5-HPB highpass filter,  it's used with the Quatro and other speakers in the line to keep low frequencies out of the mid-bass driver by attenuating frequencies at 6 dB per octave (first order filter), with a -3 dB point at 100 Hz.  These are small boxes that are meant to be hooked up between your pre-amp and amp. 

You can see a pic of the guts of one here:

http://www.head-fi.org/t/162792/inside-look-a-high-end-balanced-crossover

And here is John Atkinson's measurement of the response of the filter from Stereophile:



The main takeaway from the graph is that the filter response should be down about 14 dB at 20 Hz.

Unfortunately the accuracy of the Vandersteen highpass filter with my Rogue M-180 tube monoblocks is disappointing.

Here are some measurements I did with my HP400E AC voltmeter.  All measurements are from the 8 Ohm taps of my Rogue M-180 amps.

In the table below, "ideal" is the ideal frequency response for a 1st order highpass filter with -3dB@100Hz knee* (see bottom of page).  The Stereophile graph also shows the response of the filter with an "ideal" load.

"100kOhm" indicates measurements with the 100 kOhm setting on the highpass dipswitches (all dip switches off).  This was the setting that was closest I could find to having the -3dB knee at 100 Hz.

"diff" is the difference between ideal and measured.

Note the large dropoff at the bottom of the range.

Code: [Select]

freq "ideal"  100kOhm  diff
===========================
100   -3.0    -2.2     0.8
 90   -3.5    -2.6     0.9
 80   -4.1    -3.3     0.8
 70   -4.8    -4.0     0.8
 60   -5.8    -5.2     0.6
 50   -7.0    -6.8     0.2
 40   -8.6    -9.0    -0.4
 30  -10.8   -12.4    -1.6
 20  -14.1   -18.4    -4.3

So not only is the -3 dB "knee" in the wrong place, the slope seems to be too steep at the bottom end.   But this is the closest I can find to -3 dB @ 100 Hz with the documented settings for the highpass filter.  (Rogue's specified input impedance on their website of 200 kOhm seems to be way off.)

Here's an undocumented dipswitch setting I found, switches 4, 5, and 10 on.  Better -3 dB point, but still a steeper slope than ideal.

Code: [Select]

freq  4,5,10  diff
================== 
100   -3.2   -0.2
 90   -3.8   -0.3
 80   -4.5   -0.4
 70   -5.2   -0.4
 60   -6.4   -0.6
 50   -7.8   -0.8
 40  -10.0   -1.4
 30  -12.5   -1.7
 20  -17.0   -2.9
 

I have some Rothwell XLR inline 10 dB attenutors, so I placed them between the highpass filter and the amps.  With the attenuators between M5-HPB and the amps the setting closest to -3dB@100Hz was 20kOhm (switches 1, 8, and 10).  So the attenuators effectively lower the input impedance of the amps.  Here the "RW" column is the measurements with the attenuators in place.

Code: [Select]

freq   RW   diff
===============
100   -2.5  .5
 90   -3    .5
 80   -3.5  .6
 70   -4.2  .6
 60   -5    .8
 50   -6.3  .7
 40   -8    .6
 30  -10.5  .3
 20  -14    .1

Now this is pretty good.  The slope is much closer to ideal, and the differences are fairly constant so can mostly be "adjusted away" with the bass level on the speaker. 

I'm not sure I like the "sound" of the Rothwell attenuators, so I'd be interested in hearing about other ways to modify the effective input impedance of the amp to get a better filter response.

For reference, here's a measurement with a Bryston 3B-SST (balanced connection; 2V sensitivity; no attenuation.)  Again the 20 kOhm setting is used since the Bryston has a 10 kOhm input impedance per balanced leg.

Code: [Select]

freq  3BSST diff
================
100   -3.3   -.3
 90   -3.8   -.3
 80   -4.5   -.3
 70   -5.2   -.4
 60   -6.2   -.4
 50   -7.5   -.5
 40   -9     -.4
 30  -11.2   -.4
 20  -14.7   -.6

(The Rogue's stomp all over the Bryston for sound quality, by the way, accurate highpass filtering or not.)

* According to the sox audio program, the "transfer function", H(f),  for a 1st order highpass filter with -3dB@100 Hz knee is

Code: [Select]

Fs=44100
b0=9.929267154013539e-01; b1=-9.929267154013539e-01; b2=0.000000000000000e+00; a1=-9.858534308027079e-01; a2=0.000000000000000e+00
o=2*pi/Fs
H(f)=sqrt((b0*b0+b1*b1+b2*b2+2.*(b0*b1+b1*b2)*cos(f*o)+2.*(b0*b2)*cos(2.*f*o))/(1.+a1*a1+a2*a2+2.*(a1+a1*a2)*cos(f*o)+2.*a2*cos(2.*f*o)))

The response in dB would be 20*log10(H(f)), where f is frequency in Hz.

[jeffreybehr]

Geeky, yes, but I love it.   

First, there's nothing magic about the intended 100Hz filter point.  I read an interview of RV wherein he stated that anywhere from 60Hz to 100Hz was just fine.  The HP filter I installed in my M-60s is at about 40Hz, as I love a bit more warmth than the next person, and setting the FP so low gives me more energy in the upper-bass/lower-MR.  I've never driven the upperbass driver nor the active bass system to distortion that I can hear, so my FP is is just fine for me and my listening habits.

I do not think there's anything wrong with your filter.  The 25Hz level, 2 octaves below the intended FP, in all cases is about 5dB further down from the level at 50Hz, and if you could plot the slope a couple octaves further down, probably it'd get very close to the theoretical 6dB slope.

Set the FP wherever you want it and enjoy the music.   

[Daverz]

First, there's nothing magic about the intended 100Hz filter point.  I read an interview of RV wherein he stated that anywhere from 60Hz to 100Hz was just fine.

Thanks, very good info to know.

I do not think there's anything wrong with your filter.

I didn't think there was anything wrong with the filter per se, it's just that with the best setting I could find (according to the manual) for my amps, I felt I was losing too much bottom.  And who doesn't want to get as much bottom as they can?

Maybe with the info you've given me I can try the 75k Ohm setting without worry.

EDIT: It also occurs to me now that I should try the higher impedance settings with that were lower than -3dB@100 Hz, since if they have a better slope at the bottom I can just adjust the overall bass level up.

[Daverz]

Well, this is embarassing.  I finally had time to try the 75 kOhm setting on the filters today.  Apparently, I just misread the dB scale on my voltmeter back in January and never thought to check settings below 100 kOhm again.  75 kOhm is the obvious setting for the M-180s.  I think I need to cut back on coffee and get glasses. 

This time I read off the volts scale and did the math:

Code: [Select]

freq "ideal"  75kOhm  diff
===========================
100   -3.0    -2.7     0.3
 90   -3.5    -3.2     0.3
 80   -4.1    -3.6     0.4
 70   -4.8    -4.4     0.4
 60   -5.8    -5.4     0.4
 50   -7.0    -6.6     0.4
 40   -8.6    -8.4     0.2
 30  -10.8   -10.9    -0.1
 20  -14.1   -15.0    -0.9

My apologies to everyone here, and to Mr. Vandersteen for impugning his filters.  Even the drop here at 20 Hz may be due to the high output impedance of my BAT 3iX pre-amp.

EDIT: OK, I'm really turning my own post into a farce.  It looks like 50 kOhms might be even better.  How could I be so far off initially?  Well, I originally measured the amp in ultralinear mode, but all the measurements below and today have been in triode mode.  This hadn't even occurred to me before.  Maybe I'll find the best setting for ultralinear mode when I'm not feeling so silly.

UPDATE: I did measure 50k.  75k is still the best match for the M-180s in triode mode.