Hi,
I thought I'd describe a little known method suitable
for adding subwoofers to valve driven speakers.
The basic premise is a 6dB high pass is used for the valve speakers.
This is done via reducing the input coupling cap of the amplifier.
Drive to the SS amplifier is then done via an attenuator from
the output of the valve amplifier that also incorporates an
inverse 6dB low pass matching the valve amps high pass, to
feed the line level inputs of the subwoofer with a flat signal.
You then use the sub adjustments as normal.
Pretty much the best way of adding an active SS sub.
rgds, sreten.
I thought I'd describe a little known method suitable
for adding subwoofers to valve driven speakers.
The basic premise is a 6dB high pass is used for the valve speakers.
This is done via reducing the input coupling cap of the amplifier.
Drive to the SS amplifier is then done via an attenuator from
the output of the valve amplifier that also incorporates an
inverse 6dB low pass matching the valve amps high pass, to
feed the line level inputs of the subwoofer with a flat signal.
You then use the sub adjustments as normal.
Pretty much the best way of adding an active SS sub.
rgds, sreten.
Why do you take a low pass for the Bass? Better take a diff-amp and substract the tube-amp-signal from the source-signal. Then you will get a real mirrow of the high-pass.
Room modes at LF are problematic.
Find it very simple and effective to run the main speakers full range with simple HP filter somewhere in the tube amp and then feed the LF amplifier (<80Hz) through something that has processing in the digital domain and RTA with a measurement microphone for actual in room LF response and then flatten (or not) the response with digital EQ.
L.H
Find it very simple and effective to run the main speakers full range with simple HP filter somewhere in the tube amp and then feed the LF amplifier (<80Hz) through something that has processing in the digital domain and RTA with a measurement microphone for actual in room LF response and then flatten (or not) the response with digital EQ.
L.H
Hi,
It does add some distortion, that of the valve amp running
high passed, compared to a line level straight sub drive.
It is claimed however that this makes the sub sound less
solid state and reduces the change in character of the
bass as it goes from solid state to valve drive.
rgds, sreten.
Hi,
Basically because its easy, simple and effective.
Its just a low pass type shelving circuit to get
the drive to the sub to be flat response.
rgds, sreten.
I found a little mistake in the setup. When you feed the tube amp over the high-pass you will have not the signal for subwoofer on the tube-output...
This is how Vandersteen recommends connecting their subwoofers (whether the main amp is tubes or not) It works well. The main amp doesn't have to work as hard and so distorts less. I found that you want the tube amp to have as clean, extended, low impedance bass output as possible to give a seamless transition.
Hi,
True if you don't use an inverse 1st order filter when
converting the speaker level down to line level.
Not true if you do. The bass will be there flat.*
rgds, sreten.
* Down to a cut off point defined by the shelving filter.
Last edited:
Hi,
Yes the main amplifier doesn't work as hard in the
bass and you can upgrade the input capacitor quality.
The whole point of the arrangement is to make the
transition more seamless, if the valve amplifier
has some bass foibles, they are reduced but
reflected by the derivation of the sub drive.
rgds, sreten.
You can get creative on the passive subwoofer drive,
e.g. include a notch filter for the main room mode,
if the subwoofer itself does not have the feature.
Last edited:
The only potential issue I see with this approach is that it is dependent on the tube (valve) amplifier being relatively hum and buzz free which should be, but is not always the case.
Hi,
Yes, hum (especially) and noise performance is compromised quite a lot,
by the high pass cut, low pass boost arrangement for the bass signal.
If you've got problems in these areas it will make it worse for sure.
If not, its all part of making the sub sound more "tubey".
rgds, sreten.
Yes, hum (especially) and noise performance is compromised quite a lot,
by the high pass cut, low pass boost arrangement for the bass signal.
If you've got problems in these areas it will make it worse for sure.
If not, its all part of making the sub sound more "tubey".
rgds, sreten.
Last edited:
Kevin, may I ask you about ripple levels. I've been working to a rule of thumb where supply noise at any rail should amount to better that -90dB relative to full output. It works well enough but I don't see the sense in referencing it to full output and would rather have a figure referenced to absolute hum/speaker sensitivity.
Hi Allen,
I would probably reference it to 1W rather than full output, and my target depending on speaker sensitivity would be at least -70dBr (rel 2.83V into 8 ohm) or better assuming a speaker system of 100dB efficiency.. (System power is right around 20Wrms per channel, GM70 SE)
My power amp, line stage and parametric EQ combined come in at 750uV over a measurement bandwidth of 20 - 22kHz (IEC filter) or about -73dBr relative to 1W. Relative to full power I achieve about -85dBr. The ripple and hum component is well under 200uV. (and very difficult to measure without some precautions) The power amplifier by the way is not the primary contributor to the output noise..
Yes it is quiet, to the extent that the power transformers in some of my hardware generate all of the buzz ever audible in the room. (Not including the pesky doorbell transformer.. lol)
Bass extension with parametric EQ extends to below 30Hz (Surprisingly all cut to deal with 3 room modes) so 60Hz can be a problem.
I would probably reference it to 1W rather than full output, and my target depending on speaker sensitivity would be at least -70dBr (rel 2.83V into 8 ohm) or better assuming a speaker system of 100dB efficiency.. (System power is right around 20Wrms per channel, GM70 SE)
My power amp, line stage and parametric EQ combined come in at 750uV over a measurement bandwidth of 20 - 22kHz (IEC filter) or about -73dBr relative to 1W. Relative to full power I achieve about -85dBr. The ripple and hum component is well under 200uV. (and very difficult to measure without some precautions) The power amplifier by the way is not the primary contributor to the output noise..
Yes it is quiet, to the extent that the power transformers in some of my hardware generate all of the buzz ever audible in the room. (Not including the pesky doorbell transformer.. lol)
Bass extension with parametric EQ extends to below 30Hz (Surprisingly all cut to deal with 3 room modes) so 60Hz can be a problem.
As for sretens suggestion my pre stage is currently driving 6V6s and Mosfets in parallel. I find the fets easy to drive for just a bass amp, but I'm using a shunt RC low pass filter and it's tricky to get the load light enough at higher frequencies for the driving stage without using very high values in the fet bias divider and losing gain in the filter.
@ Kevin, thanks, yes my biggest noise is my source.. a digital whine, but I need to put my head inside the tweeter horn to hear it. When I rigged up a bass buffer I had the supply under filtered by 10-20dB and still didn't expect hum but there was a little. I guess the discussion on IMD with DHTs and AC filaments has me wanting hum less than just audible. I'm on heaters at the present but a non-linear element still exists.
@ Kevin, thanks, yes my biggest noise is my source.. a digital whine, but I need to put my head inside the tweeter horn to hear it. When I rigged up a bass buffer I had the supply under filtered by 10-20dB and still didn't expect hum but there was a little. I guess the discussion on IMD with DHTs and AC filaments has me wanting hum less than just audible. I'm on heaters at the present but a non-linear element still exists.
Do you not then end up significantly worsening 1/f noise originating in the valve amp, fed to and within the passband of the sub ? Such as flicker noise ?
Yes, and no. The sub is driven by the voltage output of the main amp. When I tried this using a crossover of 80hz. Insufficient damping factor and/or output impedance of the main amp in the 80hz - about 200hz range gave me a subjective "hole" in the sound in the mid-bass. Increasing the tube amps power supply capacitance helped this problem, and if I set this up again, I'll probably use a higher feedback design to begin with. Of course if you have control over the sub's crossover point, you can set that high enough that the problem goes away.
Hi,
Yes you do. Its inevitable. All part of making the sub seem more tubey.
rgds, sreten.
I'm kind of late to the party, but on thinking on this, might this not be one of those cases where having a simple pass-thru amplifier for the bass would be warranted? Gainless. (0 dB). Use PNP/NPN (way, way, way old school) complimentary emitter follower config, in a classic class-A push-pull configuration? Simple as DIRT to make, nearly immune to power supply ripple, moderately high input impedance, 0.05 Ω output impedance or better.
Or, if you want a little amplification, take the 16 Ω OPT winding as input to the amp. Still can use the 8 Ω or 4 Ω for your mid-and-tweeter speaker connection.
Heck, I did exactly this in 1975 for my crazy brother-the-drummer, so that Pa's confiscated stereo system plus an extra HUGE 1950's 8-driver (2 Ω) subwoofer could play loud enough to match crazy brother's drumming. It worked great, but suffered from ridiculous heat dissipation. Again … who cares?
The circuit diagram couldn't be simpler, I think. My power supply was a ± 50 volt job. Transistors were expensive (comparatively, then), so it wasn't an H config. 50 volts (37 RMS) into 2 Ω … P = E²/R = 37²/2 ≈ 500 watts. Probably only ⅓ that in practice … and I used a different kind of low-pass filter: a finite input resistance in series, with a mid-sized inductor to cut everything above about 250 Hz or so.
But it did get really hot. I figure quiescent, it was chucking some 300 watts into the room. Maybe more.
GoatGuy
PS: the nice thing about the R > L > transistor bases approach was that by making R variable, I could change the cutoff frequency, while NOT changing the fundamental passband gain. Remember, it was a 0 dB device IN band. Very convenient for fiddling.
Or, if you want a little amplification, take the 16 Ω OPT winding as input to the amp. Still can use the 8 Ω or 4 Ω for your mid-and-tweeter speaker connection.
Heck, I did exactly this in 1975 for my crazy brother-the-drummer, so that Pa's confiscated stereo system plus an extra HUGE 1950's 8-driver (2 Ω) subwoofer could play loud enough to match crazy brother's drumming. It worked great, but suffered from ridiculous heat dissipation. Again … who cares?
The circuit diagram couldn't be simpler, I think. My power supply was a ± 50 volt job. Transistors were expensive (comparatively, then), so it wasn't an H config. 50 volts (37 RMS) into 2 Ω … P = E²/R = 37²/2 ≈ 500 watts. Probably only ⅓ that in practice … and I used a different kind of low-pass filter: a finite input resistance in series, with a mid-sized inductor to cut everything above about 250 Hz or so.
But it did get really hot. I figure quiescent, it was chucking some 300 watts into the room. Maybe more.
GoatGuy
PS: the nice thing about the R > L > transistor bases approach was that by making R variable, I could change the cutoff frequency, while NOT changing the fundamental passband gain. Remember, it was a 0 dB device IN band. Very convenient for fiddling.
Last edited:
- Status
- Not open for further replies.