I don't have a buffer before the take-off point for my subs but I keep the interconnect to the sub crossovers to less than a metre. It works well but I have one further comment.
I use stereo subs. Previously I tried with a mono sub and couldn't get the thing set up without it affecting the sound of the main speakers quite badly. When the sub was connected, it seemed to suck the life out of the main speakers.
I tried taking the feed for the sub from the volume control and after a buffer but it made no difference.
Nuuk,
that is exactly the kind of behaviour one could expect (among other bad things). Let's look at it:
Say you have a 50k pot in the GC. The input impedance of the sub amp/crossover is probably in the 10-100k range, typically 22k or 47k. This means that if you have the pot in mid setting (GND->25k->OUT->25k->IN) and thus have a -6 dB attenuation and then connect the sub with a 22k input impedance the level suddenly drops to -10 dB!
But this is simply the beginning of the nightmare...
A typical shielded cable for audio has a connector->shield capacitance of about 100 pF/m. So a quite short sub cable of only 3m thus have a capacitance of 300 pF. Together with the pot this will form a textbook lowpass filter thus rolling off the highs. Let's again assume the pot is in mid setting. The corner frequency (-3 dB) will thus with a 50k pot be:
fc= 1/(2*pi*R*C) where R=12.5 kohm and C=300 pF (R will actually be dependant on the input impedance of the sub too)
fc=42 kHz which is out of the audio band but one can still expect a very slight rolloff in the top octave. Using a 100k pot and 5m of cable the corner frequency will be well into the audio band. But things get even worse! We have so far only considered the pot in mid setting. In reality the corner frequency will depend on the pot setting being very high near the extremes and dropping considerably when approaching mid level. We have created an EQ that is dependant on volume setting!
The cable will also be a perfect antenna picking up and injecting hum and all kinds of RFI under the stars into the high-impedance node (input of the GC). None of the GC designs I have seen so far even has an RFI filter on the (low impedance) input and this approach is just begging for trouble!
The conclusion is that this is a very bad way of doing things. Cables for analog audio should always be driven by a low-impedance output (unless 20 km long or so). The best way by far is to use a balanced interconnection as done in all pro sound gear.
On the other hand you might enjoy the "high end" approach where you can tune the response of the system by using different cables, orienting the components differently or reorienting the planets.
Feel free to choose your path!
Cheers
/Magnus
that is exactly the kind of behaviour one could expect (among other bad things). Let's look at it:
Say you have a 50k pot in the GC. The input impedance of the sub amp/crossover is probably in the 10-100k range, typically 22k or 47k. This means that if you have the pot in mid setting (GND->25k->OUT->25k->IN) and thus have a -6 dB attenuation and then connect the sub with a 22k input impedance the level suddenly drops to -10 dB!
But this is simply the beginning of the nightmare...
A typical shielded cable for audio has a connector->shield capacitance of about 100 pF/m. So a quite short sub cable of only 3m thus have a capacitance of 300 pF. Together with the pot this will form a textbook lowpass filter thus rolling off the highs. Let's again assume the pot is in mid setting. The corner frequency (-3 dB) will thus with a 50k pot be:
fc= 1/(2*pi*R*C) where R=12.5 kohm and C=300 pF (R will actually be dependant on the input impedance of the sub too)
fc=42 kHz which is out of the audio band but one can still expect a very slight rolloff in the top octave. Using a 100k pot and 5m of cable the corner frequency will be well into the audio band. But things get even worse! We have so far only considered the pot in mid setting. In reality the corner frequency will depend on the pot setting being very high near the extremes and dropping considerably when approaching mid level. We have created an EQ that is dependant on volume setting!
The cable will also be a perfect antenna picking up and injecting hum and all kinds of RFI under the stars into the high-impedance node (input of the GC). None of the GC designs I have seen so far even has an RFI filter on the (low impedance) input and this approach is just begging for trouble!
The conclusion is that this is a very bad way of doing things. Cables for analog audio should always be driven by a low-impedance output (unless 20 km long or so). The best way by far is to use a balanced interconnection as done in all pro sound gear.
On the other hand you might enjoy the "high end" approach where you can tune the response of the system by using different cables, orienting the components differently or reorienting the planets.
Feel free to choose your path!
Cheers
/Magnus
SC, thanks for that thorough expalanation.
Although I cannot percieve any problems without one, I could try a buffer between the volume control (shunt attenuator) and the active woofer crossover.
As I stated previously, I would only use the very short interconnects without a buffer. In fact, I am planning to move the woofer crossovers/amp right under the main amp and reducing the interconnects to a few inches.
Although I cannot percieve any problems without one, I could try a buffer between the volume control (shunt attenuator) and the active woofer crossover.
As I stated previously, I would only use the very short interconnects without a buffer. In fact, I am planning to move the woofer crossovers/amp right under the main amp and reducing the interconnects to a few inches.
Nuuk,
as you have experienced it mostly works fine over shorter distances. Hope you could make out something from my explanation!
Personally I only use balanced interconnects with XLR connectors because I don't think hum, noise, GSM phone signals or your neighbour turning on the lights have any place in an audio system. You can check out the ESP pages for a starter if you want to head down the balanced road.
On a similar note (if you have the time to try it!), you could check the noise of the amp with respect to the pot setting (inputs shorted or connected to a very low-noise preamp). Noise should be highest with a mid setting and lowest at the extreme ends. This is partly due to the thermal noise of the pot but most of the effect comes from the amp and is related to the "equivalent input noise resistance" of the amp. You get this resistance by dividing the voltage noise (Vn) by the current noise (In) that you can find in the datasheet. It is low for bipolar amps (such as the LM chips) but very high for amps with a FET input and the noise from the latter are thus not very dependant on the pot setting.
/Magnus
as you have experienced it mostly works fine over shorter distances. Hope you could make out something from my explanation!
Personally I only use balanced interconnects with XLR connectors because I don't think hum, noise, GSM phone signals or your neighbour turning on the lights have any place in an audio system. You can check out the ESP pages for a starter if you want to head down the balanced road.
On a similar note (if you have the time to try it!), you could check the noise of the amp with respect to the pot setting (inputs shorted or connected to a very low-noise preamp). Noise should be highest with a mid setting and lowest at the extreme ends. This is partly due to the thermal noise of the pot but most of the effect comes from the amp and is related to the "equivalent input noise resistance" of the amp. You get this resistance by dividing the voltage noise (Vn) by the current noise (In) that you can find in the datasheet. It is low for bipolar amps (such as the LM chips) but very high for amps with a FET input and the noise from the latter are thus not very dependant on the pot setting.
/Magnus
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.