Implementing a crossfeed network
Sept 18, 2006 17:06:25 GMT
Post by rickcr42 on Sept 18, 2006 17:06:25 GMT
tricky stuff going on here in any 'stand alone" Passive Crossfeed Box where being an R/C filter network at both input and output relies mightily on what goes before and what come after.
the Meier audio Cross-1 adresses this known weakness by having "user trim" controls to allow adjusting the network to the system and is an answer but not a complete answer unless other requirements are also met.
The Input :
You better hit this jack with a low impedance or you would be better off losing the crossfeed from your system entirely !
Drive a filter with a high imedance and the "driver" becomes part of the filter rather than a bystander as any "add-on" device needs to be.If you look at impedance as a resistor,as you should,a high impedance is a high value resistor and being such becomes part of the overall filter equation and has both audible and structural consequences because it changes the frequency of filter operation making any "specs" on the data sheet moot
Not just that but by being a high impedance implies limited current drive so unless your "high Z" is a big mamma slammer power amplifier with enough muscle to overcme the impedance can not properly drive the resistor/capacitor network on its output well enough to make the stage following the crosfeed se a proper signal.If you want to research this read up on phono stage RIAA EQ nework drive requirements which is in many ways similiar though the crossfeed does not have a "boost" requirement.
Bottom line is if you have a solid state device driving the network you are likely in good shape but driving the crossfeed with a cap coupled triode stage and you may be in deep doo doo unless you either add a matching transformer or an active buffer stage both of which will also have audible consequences (everything added has a sonic imprint).
If the triode stage has good drive,and it should due to it being intended to drive interconnect cables,then the best method of matching the stage to the crossfeed would be to use a 10/600 Ohm matching trasfmormer.This will reduce your gain somewhat (as does the crossfeed being a passive network) so means raising the overall volume but a small price to pay if you care about sonic quality and like some folks need to use a crossfeed network or you can't listen for long periods with headphones strapped on (me )
The transformer option is not cheap and is not exactly without conseuences but it does eliminate the need for another stage which is "option #2"
Option 2 is active buffering of the crossfeed network input.Quick and dirty yet mostly effective would be a simple single ended single JFET Follower stage having a high input impedance and a lowish output impedance,low enough to not interact with the actual crossfeed network.By having a high input Z the preceeding stage "sees" an easier to drive load while being totally isolated from the actual crossfeed since it is now being driven by the buffer and not the former tube stage.
Going form "quick and dirty" (even though fully usable ) to optimised would be t take the single jfet stage up a notch to more of a true "voltage follower" with less consequnece and that leaves one of two choices as I see it ;
1-JFET White/Borbely follower stage
2-National Semiconductor "u-Amp" which is also a two element buffer but instead of using a dual jfet uses a combination of jfet input/bipolar transistor output compound stage.why this is a nice choice is each part is used to strength so for the input impedance/input capacitance part of the equation the jfet is used and for the output drive/output impedance part the bipolar transistor used.JFETs make for shitty current amps but excell as voltage amps while bipolar transistor rule as current amps so in theory the best of both worlds
Where this "answer" can get a bit tricky is in the coupling.Being a single ended stage there will be a large DC potential on the output that may or may not cause havoc to your system depending on what is going on further down the food chain.the crossfeed being totally assive does not give a rats ass what it "see" in the DC rider department and in fact will effedtively block any DC in all but the "bypass" postion.....and that is the danger point right there.
If your "next stage" happens to be a DC coupled amp you will be sending a pure DC voltage of 1/2 the buffer power supply (single ended class-A) DIRECTLY THROUGH TO THE OUTPUTS and that will wipe out any speaker of headphone I know of so use a damn DC blocking cap unless you are shure every potential device after the crossfeeed has them !!!!!!!!
If you build the buffer into te crossfeed the entire shooting match can be bypassed at the same time making any added capacitors unimportant to the overall system context.Being smart is better than wiping out your gear for some etheral "DC Coupled or death" mindset.
OK.So we have a proper input driving source,now what ?
Now we need to look at the output device interface of what is connected to the crossfeed and that includes your interconnect cables ! They have an impedance too and must be considered as part of the input to the next stage (why passive "preamps" have particular requirements).
The combination of the resistive/capacitive output of the crossfeed and the cable resistance/capacitance of your interconnects is large it had better "see" an input impedance on the next stage that sets the filter F3 of this combination well above the audible range or you WILL suffer treble loss !
Every single R/C equation in an audio system is in fact a filter and when a capacitance "sees" an input impedance (resistance) you are making a low pass filter.This is every single system on the planet and not just with a crossfeed.Because your average audio output is buffered/isolated the only part of the equation that has any audible effect is cable capacitance/next stage input impedance which is agan a simple C/R low pass filter.Due to the low capacitances of even crappy cables and the average high input impedance of of any consumer grade audio gear this filter created usally has its F3 point way up in either the extrmely high Khz range or very low Mhz range so totally transparent for all real world intent.
Add in another passive R/C element however and you need to add it also to the equation because now you are in the BIG CAPACITOR arena (cable capacitance is in the pf range,a crossfeed network has caps in the uF range ) and must be considered as part of the whole.
Under 90% of system possibilities this is a nin event and especially so with short cable lengths and low capacitance interconnects going to a "typical" high impedance amplifier inpt but still a thing to be aware of.High end rolloff may be the culprit at work when I read "the crossfeed dulled everything making the music boring" type reviews which in combination with the intentional high end attenuation would cause just that.
The Fix ?
Only one this time.Using a transformer alone is 100% off the list of "possibles" because any impedance step-down comes also with a voltage step down (attenuation) and since we are already dealing with a "lossy" network any more attenuation will likely result in a very gain lacking sound system,ne you would not like very much
The answer here is a true curent pump style output stage which can be anything from the above compound amp (but with a higher current output) to a "diamond buffer" to a monolithis buffer or in a tube stage a "White follower with coupling transformer (a very expensive but the best "fix" in a tube based system ).since you are already "active" it would be a cakewalk to go the extra mile and make this stage have some actual positive gain to make up for network losses but again THINK or when you are in "bypass" mode you will be sending a serious signal increase through to your amp !
The above is just ptting known potential problems out in the open and is not something you will find is a real worl problem with the average system,a system having high input impedances with low output impedances,but something to be aware of just the same if your subjective testing of a crossfeed network results in reports of "very dull and sluggish sonics.I think I do not like crossfeeds even though I do not like the way headphones image".Matchups are verything in audio even though the least understood part of system design and one rarely even mentioned in any ad copy so you are mostly left on your own up that proverbial creek without a paddle
just another "Public Service announcement" from the dark and scary recesses of the Rickmongo lobed grey area ;D
the Meier audio Cross-1 adresses this known weakness by having "user trim" controls to allow adjusting the network to the system and is an answer but not a complete answer unless other requirements are also met.
The Input :
You better hit this jack with a low impedance or you would be better off losing the crossfeed from your system entirely !
Drive a filter with a high imedance and the "driver" becomes part of the filter rather than a bystander as any "add-on" device needs to be.If you look at impedance as a resistor,as you should,a high impedance is a high value resistor and being such becomes part of the overall filter equation and has both audible and structural consequences because it changes the frequency of filter operation making any "specs" on the data sheet moot
Not just that but by being a high impedance implies limited current drive so unless your "high Z" is a big mamma slammer power amplifier with enough muscle to overcme the impedance can not properly drive the resistor/capacitor network on its output well enough to make the stage following the crosfeed se a proper signal.If you want to research this read up on phono stage RIAA EQ nework drive requirements which is in many ways similiar though the crossfeed does not have a "boost" requirement.
Bottom line is if you have a solid state device driving the network you are likely in good shape but driving the crossfeed with a cap coupled triode stage and you may be in deep doo doo unless you either add a matching transformer or an active buffer stage both of which will also have audible consequences (everything added has a sonic imprint).
If the triode stage has good drive,and it should due to it being intended to drive interconnect cables,then the best method of matching the stage to the crossfeed would be to use a 10/600 Ohm matching trasfmormer.This will reduce your gain somewhat (as does the crossfeed being a passive network) so means raising the overall volume but a small price to pay if you care about sonic quality and like some folks need to use a crossfeed network or you can't listen for long periods with headphones strapped on (me )
The transformer option is not cheap and is not exactly without conseuences but it does eliminate the need for another stage which is "option #2"
Option 2 is active buffering of the crossfeed network input.Quick and dirty yet mostly effective would be a simple single ended single JFET Follower stage having a high input impedance and a lowish output impedance,low enough to not interact with the actual crossfeed network.By having a high input Z the preceeding stage "sees" an easier to drive load while being totally isolated from the actual crossfeed since it is now being driven by the buffer and not the former tube stage.
Going form "quick and dirty" (even though fully usable ) to optimised would be t take the single jfet stage up a notch to more of a true "voltage follower" with less consequnece and that leaves one of two choices as I see it ;
1-JFET White/Borbely follower stage
2-National Semiconductor "u-Amp" which is also a two element buffer but instead of using a dual jfet uses a combination of jfet input/bipolar transistor output compound stage.why this is a nice choice is each part is used to strength so for the input impedance/input capacitance part of the equation the jfet is used and for the output drive/output impedance part the bipolar transistor used.JFETs make for shitty current amps but excell as voltage amps while bipolar transistor rule as current amps so in theory the best of both worlds
Where this "answer" can get a bit tricky is in the coupling.Being a single ended stage there will be a large DC potential on the output that may or may not cause havoc to your system depending on what is going on further down the food chain.the crossfeed being totally assive does not give a rats ass what it "see" in the DC rider department and in fact will effedtively block any DC in all but the "bypass" postion.....and that is the danger point right there.
If your "next stage" happens to be a DC coupled amp you will be sending a pure DC voltage of 1/2 the buffer power supply (single ended class-A) DIRECTLY THROUGH TO THE OUTPUTS and that will wipe out any speaker of headphone I know of so use a damn DC blocking cap unless you are shure every potential device after the crossfeeed has them !!!!!!!!
If you build the buffer into te crossfeed the entire shooting match can be bypassed at the same time making any added capacitors unimportant to the overall system context.Being smart is better than wiping out your gear for some etheral "DC Coupled or death" mindset.
OK.So we have a proper input driving source,now what ?
Now we need to look at the output device interface of what is connected to the crossfeed and that includes your interconnect cables ! They have an impedance too and must be considered as part of the input to the next stage (why passive "preamps" have particular requirements).
The combination of the resistive/capacitive output of the crossfeed and the cable resistance/capacitance of your interconnects is large it had better "see" an input impedance on the next stage that sets the filter F3 of this combination well above the audible range or you WILL suffer treble loss !
Every single R/C equation in an audio system is in fact a filter and when a capacitance "sees" an input impedance (resistance) you are making a low pass filter.This is every single system on the planet and not just with a crossfeed.Because your average audio output is buffered/isolated the only part of the equation that has any audible effect is cable capacitance/next stage input impedance which is agan a simple C/R low pass filter.Due to the low capacitances of even crappy cables and the average high input impedance of of any consumer grade audio gear this filter created usally has its F3 point way up in either the extrmely high Khz range or very low Mhz range so totally transparent for all real world intent.
Add in another passive R/C element however and you need to add it also to the equation because now you are in the BIG CAPACITOR arena (cable capacitance is in the pf range,a crossfeed network has caps in the uF range ) and must be considered as part of the whole.
Under 90% of system possibilities this is a nin event and especially so with short cable lengths and low capacitance interconnects going to a "typical" high impedance amplifier inpt but still a thing to be aware of.High end rolloff may be the culprit at work when I read "the crossfeed dulled everything making the music boring" type reviews which in combination with the intentional high end attenuation would cause just that.
The Fix ?
Only one this time.Using a transformer alone is 100% off the list of "possibles" because any impedance step-down comes also with a voltage step down (attenuation) and since we are already dealing with a "lossy" network any more attenuation will likely result in a very gain lacking sound system,ne you would not like very much
The answer here is a true curent pump style output stage which can be anything from the above compound amp (but with a higher current output) to a "diamond buffer" to a monolithis buffer or in a tube stage a "White follower with coupling transformer (a very expensive but the best "fix" in a tube based system ).since you are already "active" it would be a cakewalk to go the extra mile and make this stage have some actual positive gain to make up for network losses but again THINK or when you are in "bypass" mode you will be sending a serious signal increase through to your amp !
The above is just ptting known potential problems out in the open and is not something you will find is a real worl problem with the average system,a system having high input impedances with low output impedances,but something to be aware of just the same if your subjective testing of a crossfeed network results in reports of "very dull and sluggish sonics.I think I do not like crossfeeds even though I do not like the way headphones image".Matchups are verything in audio even though the least understood part of system design and one rarely even mentioned in any ad copy so you are mostly left on your own up that proverbial creek without a paddle
just another "Public Service announcement" from the dark and scary recesses of the Rickmongo lobed grey area ;D