PS, years ago when I found out that article I laughed, it is very entertaining reading.
You guys read it, it's fun.
Some titles:
'I have only one itchin' desire: let me stand next to your wire'
...
'El cajon esta muy cheesy'
...
'Star grounding is alive and well and living in Paris, TX'
Btw I posted the link to page two, here's the link for the beginning of the article:
http://www.stereophile.com/amplificationreviews/54/index.html
You guys read it, it's fun.
Some titles:
'I have only one itchin' desire: let me stand next to your wire'
...
'El cajon esta muy cheesy'
...
'Star grounding is alive and well and living in Paris, TX'
Btw I posted the link to page two, here's the link for the beginning of the article:
http://www.stereophile.com/amplificationreviews/54/index.html
Voltage sources
Since we seem to have some energy for linguistics, lets make sure that everyone understands that a voltage source is called that because the voltage is NOT dependent on the current drawn from it. Voltage sources are in fact designed to supply enough current to keep the voltage constant independent of load.
Of course there is some upper limit as to how much current they can drive but any source component that has trouble driving 10 KOhms is deficient in design (IMHO). In order to drive the cable capacitance for any reasonable cable length at 2V RMS at up to 20Khz requires a source impedance in the few hundred Ohms maximum range.
The device that drives my passive pre has a published source impedance of 6 Ohms, lower than most but not really unreasonable. Most tube stages have source impedances of 100 Ohms or so whereas transistor designs (or Op-Amps) easily reach down to 10 Ohms.
So the whole story about Voltage sources not being able to supply current is not only wrong but exactly backwards. Voltage sources by definition are made to supply current.
Since we seem to have some energy for linguistics, lets make sure that everyone understands that a voltage source is called that because the voltage is NOT dependent on the current drawn from it. Voltage sources are in fact designed to supply enough current to keep the voltage constant independent of load.
Of course there is some upper limit as to how much current they can drive but any source component that has trouble driving 10 KOhms is deficient in design (IMHO). In order to drive the cable capacitance for any reasonable cable length at 2V RMS at up to 20Khz requires a source impedance in the few hundred Ohms maximum range.
The device that drives my passive pre has a published source impedance of 6 Ohms, lower than most but not really unreasonable. Most tube stages have source impedances of 100 Ohms or so whereas transistor designs (or Op-Amps) easily reach down to 10 Ohms.
So the whole story about Voltage sources not being able to supply current is not only wrong but exactly backwards. Voltage sources by definition are made to supply current.
Re: Voltage sources
I agree.
But other things impact the performance.
A 10K pot works much better with a fairly long cable after it than an 100K pot.
And your source should drive better the 100k pot, right?
No, it depends.
The 100K pot should not have more than a very few cm after it.
Talking of pots, I have ALWAYS consistently better results with low impedance pots, even it the pot is directly on a PCB, with short signal path.
My sources are all very low output impedance, I really don't like it any other way.
Sometimes it's not the brand of the pot that makes the difference, it's the impedance of the pot that you choose.
Component values are very important.
hermanv said:
I agree.
But other things impact the performance.
A 10K pot works much better with a fairly long cable after it than an 100K pot.
And your source should drive better the 100k pot, right?
No, it depends.
The 100K pot should not have more than a very few cm after it.
Talking of pots, I have ALWAYS consistently better results with low impedance pots, even it the pot is directly on a PCB, with short signal path.
My sources are all very low output impedance, I really don't like it any other way.
Sometimes it's not the brand of the pot that makes the difference, it's the impedance of the pot that you choose.
Component values are very important.
Hi Hermanv ,
Not really. Many op amps can not drive a 600 ohm load properly, a tube stage may easily be over 2K ohm output impedance. The design standard for audio input impedance (-10 dBu) is 50K.
If op amps could drive a 10 ohm load, there would be no reason for the BUF-03 or the 4556 op amp.
You are correct, a perfect voltage source has 0 ohms output impedance, a current source would have an infinitely high output impedance. We are stuck with various combinations of the two.
Now the big question. What is the impedance of your 6 ohm source after running through a low impedance 10K pot? Hmmm, not 6 ohms anymore.
-Chris
Not really. Many op amps can not drive a 600 ohm load properly, a tube stage may easily be over 2K ohm output impedance. The design standard for audio input impedance (-10 dBu) is 50K.
If op amps could drive a 10 ohm load, there would be no reason for the BUF-03 or the 4556 op amp.
You are correct, a perfect voltage source has 0 ohms output impedance, a current source would have an infinitely high output impedance. We are stuck with various combinations of the two.
Now the big question. What is the impedance of your 6 ohm source after running through a low impedance 10K pot? Hmmm, not 6 ohms anymore.
-Chris
You are confusing drive capability with output impedance, they are not the same. In the same way design standards for input impedance do not equal the drive impedance.
In the first case; applying for example a 600 Ohm load to an op-amp specified for a 600 minimum load will not cut the signal in half. Cutting the signal in half would mean the load resistor is the same as the output impedance. In most op-amps even those specified for 600 Ohms minimum load, the output impedance is milliohms for DC, rising to a few hundred Ohms at very high frequencies. No they will not drive these resistors to full swing but the feedabck loop on a typical inverting Op-Amp circuit makes the output impedance equal to very roughly R * (circuit Gain)/(open loop gain) until the current limit of the device is reached.
In the same way some tubes might have a plate impedance of 2K, if there is negative feedback the output impedance will be much less than 2K. In many pre-amps the output jack is driven from a cathode not a plate (there are certainly exceptions) so again output impedance is much lower.
Buffer chips exist more to isolate op-amps from instability due to capacitive loads than becasue a typical op-amp can't drive a amplifier input
As far as the results of using a 10K pot, at half rotation, the source impedance is now 2500 Ohms (all othere settings Z is lower), with a typical interconnect cable, capacitance can go as high as 400pF so the 3dB corner works out to 159KHz. Ok by me. I would not recommend 20 foot long cables on a passive pre, of course I wouldn't recommend 20 foot cables on an active pre either.
Hi Hermanv,
When you load an output stage down too far, the distortion goes through the roof. Not very listenable.
What about the input capacitance of the next stage (could be an amplifier or something else)? This will add to the cable capacitance. Also, with a higher source impedance, the S/N ratio degrades due to more noise pickup.
Too many assumptions. Not a reliable way to set up a system. For some , I guess it's good enough for rock & roll ... , nah ... country & western.
If you like it, that's okay, but it flies in the face of known principles. I can't recommend this to anyone.
-Chris
When you load an output stage down too far, the distortion goes through the roof. Not very listenable.
What about the input capacitance of the next stage (could be an amplifier or something else)? This will add to the cable capacitance. Also, with a higher source impedance, the S/N ratio degrades due to more noise pickup.
Too many assumptions. Not a reliable way to set up a system. For some , I guess it's good enough for rock & roll ... , nah ... country & western.
If you like it, that's okay, but it flies in the face of known principles. I can't recommend this to anyone.
-Chris
Re: Re: Voltage sources
Carlos,
I agree (stop looking so surprised, please). Assuming for the sake of discussion that the source Z is lower than the pot impedance, the worst-case situation with driving impedance occurs if the pot is in mid position. For the signal, the two pot halves are effectively in parallel, so in this case the driving impedance (seen downstream from the pot) is 1/4 the pot value. At all other pot possitions it is lower. For a 10k pot, worst-case drive impedance is thus 2.5k. That should be OK for all but the worst and longest cables. A 100k pot has worst-case 25k and that could interact with the cable capacitance for instance.
Anyway, using a 10k pot seems an accepted compromise between cable drive impedance and source load. 50k or 100k pots should not be used except mayby for unbuffered tube preamps with many 10's of k output Z, but those are classifed (in my book) as incompetent designs.
Jan Didden
carlosfm said:
Carlos,
I agree (stop looking so surprised, please). Assuming for the sake of discussion that the source Z is lower than the pot impedance, the worst-case situation with driving impedance occurs if the pot is in mid position. For the signal, the two pot halves are effectively in parallel, so in this case the driving impedance (seen downstream from the pot) is 1/4 the pot value. At all other pot possitions it is lower. For a 10k pot, worst-case drive impedance is thus 2.5k. That should be OK for all but the worst and longest cables. A 100k pot has worst-case 25k and that could interact with the cable capacitance for instance.
Anyway, using a 10k pot seems an accepted compromise between cable drive impedance and source load. 50k or 100k pots should not be used except mayby for unbuffered tube preamps with many 10's of k output Z, but those are classifed (in my book) as incompetent designs.
Jan Didden
Hi Jan,
I can't agree totally with that. In early tube designs, the amplifier input impedance was 50~100K. So a couple K output impedance was reasonable. A rebuild of the same design, driving and amplifier design of the same era is not unreasonable. (as much as I hate high impedance outputs). In today's market I feel your comments have merit. It must be recognized that some SS amplifiers seem to have a non-linear input impedance that can react with higher impedance sources. Therefore, I can not support an attenuator only design that ignores added capacitance by the later equipment and possible non-linear impedance's. Never mind the higher noise levels possible from the lack of loading a low impedance source can give. I feel that a buffer less attenuator is an incompetent design.
Of course, in a way, I am in total agreement with you. It's just a question of degree really.
-Chris
I can't agree totally with that. In early tube designs, the amplifier input impedance was 50~100K. So a couple K output impedance was reasonable. A rebuild of the same design, driving and amplifier design of the same era is not unreasonable. (as much as I hate high impedance outputs). In today's market I feel your comments have merit. It must be recognized that some SS amplifiers seem to have a non-linear input impedance that can react with higher impedance sources. Therefore, I can not support an attenuator only design that ignores added capacitance by the later equipment and possible non-linear impedance's. Never mind the higher noise levels possible from the lack of loading a low impedance source can give. I feel that a buffer less attenuator is an incompetent design.
Of course, in a way, I am in total agreement with you. It's just a question of degree really.
-Chris
Damn, will someone please tell the folks at Placette Audio, Antique Sound labs, Creek, MSB Technology, Wellbourne Labs, Channel Islands, Monarchy Audio and Audio Synthesis (to name just a few) That their highly regarded well reviewed equipment is incompetent.
Passive designs do have limitations. So do active designs. My home brew passive pre-amp significantly outperformed my Conrad Johnson vacuum tube pre-amp in the only design requirement that actually matters - it sounded better. I chose an input impedance of 6K Ohms (average) with a resultant output impedance of around 1.5K at my most common gain setting. My circuit drives a 240 watt power amp whose input Z is probably 30-50K (it's not specified)
The choice of an active vs a passive design should be a considered choice, but knee jerk psuedo-engineering doesn't help. Non linear, you say? It takes an active stage to really appreaciate non-linear behavior.
Do you really believe that passive volume control components coupling two op-amp stages are somehow magically different from the passive components between two amplifier stages that just happen to be in different boxes? Yes cables do add another dimension to the question, but the cable parametrics you are afraid of almost allways fall well outside the audio band.
For those of you considering the choice of active vs passive I refer you to this review becasue it gets to the heart of the question (how does it sound?).
http://www.10audio.com/placette.htm
Hi Hermanv,
Okay, I read the article. So?
They are very careful to use a high input impedance amplifier, and it's tube. My previous points hold quite well here.
The manufacturers are building product to feed a demand. Right or wrong. They ain't in business to make the best product. They are in business to sell stuff. Don't delude yourself, this is the primary motivation. Everyone else is out of business. They need you to believe they are making the best. So I'll say it, they are making a flawed product for general consumption.
Now, the difference between a "passive preamp" (Arrrgghhh
), and a real live designed product is real simple. When you design a preamp, you generally know the impedance of the driving circuit and the driven circuit. You have control over the linearity of both circuits and the capacitance of all the cabling. You even have total control over the electrical noise in the area. Cool eh?
Now with a "passive preamp" (Arrrgghhh ), there is no control over any of these factors. None. It's a crap shoot. Total guess. Infinite deniability for any problems - hey, the perfect product. Betcha I can charge large for this!
If you stick this attenuator inside a well engineered product, it will sound better (and measure better- sorry) in most if not all of the cases. That is a fact of life. No magic, just poor combinations.
-Chris
Okay, I read the article. So?
They are very careful to use a high input impedance amplifier, and it's tube. My previous points hold quite well here.
The manufacturers are building product to feed a demand. Right or wrong. They ain't in business to make the best product. They are in business to sell stuff. Don't delude yourself, this is the primary motivation. Everyone else is out of business. They need you to believe they are making the best. So I'll say it, they are making a flawed product for general consumption.
Now, the difference between a "passive preamp" (Arrrgghhh
), and a real live designed product is real simple. When you design a preamp, you generally know the impedance of the driving circuit and the driven circuit. You have control over the linearity of both circuits and the capacitance of all the cabling. You even have total control over the electrical noise in the area. Cool eh? Now with a "passive preamp" (Arrrgghhh ), there is no control over any of these factors. None. It's a crap shoot. Total guess. Infinite deniability for any problems - hey, the perfect product. Betcha I can charge large for this!If you stick this attenuator inside a well engineered product, it will sound better (and measure better- sorry) in most if not all of the cases. That is a fact of life. No magic, just poor combinations.
-Chris
anatech said:
Chris, I have started this last discussion here:
http://www.diyaudio.com/forums/showthread.php?postid=758709#post758709
and I meant to say that there's more than the pot impedance (parallel resistance to ground), the impedance on the output of the pot also has a large impact on the audio performance.
I totally agree with you, no high output impedance sources!
No high impedance pots, please, and if the signal path can't be made very short, a buffer may be needed.
There are other considerations.
For me an active pre is an active pre, I use it and I don't discard it, but depending on the design, topology, components used, layout and only after all this if it is an integrated amp, the buffer can be discarded.
Final note: I do have an active pre and the power amp is near the speakers, on a central position, short speaker cables.
I find this much better than centralizing all on a rack and then run long speaker cables.
I have longer interconnects, but that's no problem.
PS: standalone 'passive' pre?
Yuck!
Re: Re: In other words....
True, but most preamps I've encountered have a voltage gain of >1 when it's not really needed, causing deterioration in S/N and distortion. Of course I'm accepting the fact that driving >600pF cables is unnecessarily stupid. If that is necessary then current (power) amplification is needed.Christer said:
Unless you need a gain > 1.
Hi dhaen,
Well that is too easy to fix. No voltage gain, only buffers (X1). Now you have the best of both worlds!
Carlos,
Yes we agree. Basically, it's called using your head. If you design an integrated amp, you design the following stage as your buffer more or less.
One place my experience is a little different maybe than yours. I would much rather run longer speaker leads than signal leads any day.
-Chris
Well that is too easy to fix. No voltage gain, only buffers (X1). Now you have the best of both worlds!
Carlos,
Yes we agree. Basically, it's called using your head. If you design an integrated amp, you design the following stage as your buffer more or less.
One place my experience is a little different maybe than yours. I would much rather run longer speaker leads than signal leads any day.
-Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.