Hi Sonnya,
Yes, will keep you informed, and thank you for your input; much appreciated.
Dorkus (primus or secundus??), the Wolfson, made in Edinburgh, seems to have a good reputation and the specs are clearly top quality. However, it requires an output IC since the attenuation is implemented in the feedback, and this brings us back to the sonics of the associated IC. It is also phase inverting, which causes me some concern, even though I realise absolute phase in audio is indetectable.
For personal preferences, I do not wish to use an IC; all circuitry I use is discrete, often hybrid, and use of an IC with its push pull output stage and huge feedback ratio is not for me, at least in this application (line preamp). My experiences with ICs has often been that they collapse the sound stage, and I've spent a lot of time refining SS discrete circuits which do well in this area.
Brett, use of the DS1808 as a shunt element is problematic, as it was not designed for this. The attenuation steps are configured precisely for log output as a divider; used as a shunt, I would expect a jerky and uneven control. Shunt controls work best when operating into known, prescribed impedance; this mandates placing it between two stages, ie buffering it. Strictly, this involves a buffer stage - more electronics, possibly more damage to the music. I'd stick with series.
Thanks to all for the input,
Cheers,
Hugh
Aspen Amps P/L
Melbourne AUSTRALIA
Yes, will keep you informed, and thank you for your input; much appreciated.
Dorkus (primus or secundus??), the Wolfson, made in Edinburgh, seems to have a good reputation and the specs are clearly top quality. However, it requires an output IC since the attenuation is implemented in the feedback, and this brings us back to the sonics of the associated IC. It is also phase inverting, which causes me some concern, even though I realise absolute phase in audio is indetectable.
For personal preferences, I do not wish to use an IC; all circuitry I use is discrete, often hybrid, and use of an IC with its push pull output stage and huge feedback ratio is not for me, at least in this application (line preamp). My experiences with ICs has often been that they collapse the sound stage, and I've spent a lot of time refining SS discrete circuits which do well in this area.
Brett, use of the DS1808 as a shunt element is problematic, as it was not designed for this. The attenuation steps are configured precisely for log output as a divider; used as a shunt, I would expect a jerky and uneven control. Shunt controls work best when operating into known, prescribed impedance; this mandates placing it between two stages, ie buffering it. Strictly, this involves a buffer stage - more electronics, possibly more damage to the music. I'd stick with series.
Thanks to all for the input,
Cheers,
Hugh
Aspen Amps P/L
Melbourne AUSTRALIA
hugh,
true, IC op-amps are not always ideal for audio circuits. however, not all op-amps are bad, and not all ICs are bad. a lot depends on implementation, and i have heard some fantastic sounding circuits that have IC op-amps in them. yes, the majority of IC op-amp circuits can sound pretty icky but you should not rule them out entirely. you can also implement an op-amp with discrete transistors and get excellent sonics (Bryston gear uses discrete opamps, among others). so using an IC is not a requirement of the Wolfson part. it certainly is more flexible than the crystal or BB/TI parts which have internal op-amps. i did not realize the DS108 was entirely passive though, so it looks like a pretty good part.
(oh, and absolute phase IS audible in audio
)
marc
true, IC op-amps are not always ideal for audio circuits. however, not all op-amps are bad, and not all ICs are bad. a lot depends on implementation, and i have heard some fantastic sounding circuits that have IC op-amps in them. yes, the majority of IC op-amp circuits can sound pretty icky but you should not rule them out entirely. you can also implement an op-amp with discrete transistors and get excellent sonics (Bryston gear uses discrete opamps, among others). so using an IC is not a requirement of the Wolfson part. it certainly is more flexible than the crystal or BB/TI parts which have internal op-amps. i did not realize the DS108 was entirely passive though, so it looks like a pretty good part.
(oh, and absolute phase IS audible in audio
)marc
If you want a truly fantastic semiconductor-based attenuator,
back engineer the volume control from the X series preamps,
which uses bipolar switches.
It outperforms the other stuff we've seen by a factor of about
10 in every parameter, and was developed by Wayne, who
has a patent pending on it.
back engineer the volume control from the X series preamps,
which uses bipolar switches.
It outperforms the other stuff we've seen by a factor of about
10 in every parameter, and was developed by Wayne, who
has a patent pending on it.
NELSON!!!!!!!!!!
i was just about to ask the forum if they knew anything about Wayne's design, but seeing as how it's in the process of being patented i figured there was no hope of ever constructing such a device myself. but if the man himself gives the go-ahead...
btw, i'm in the process of choosing a gain/buffer circuit for my multichannel preamp. was going to try the Walt Yung IC/discrete buffer combo, but after realizing how glorious a cheap passive volume control can sound i'm starting to cringe at the thought of putting so many stages of active circuitry in the way (i.e. ICs). so i'll start looking at some of your simpler designs... i'm also thinking of trying a discrete op-amp using your DIY design tips, but i guess discrete or not you don't like the sound of feedback very much?
p.s. i was meaning to contact you a while ago re: IC opamps and their application in audio for an article i'd like to write for Bound for Sound magazine, did marty dewulf tell you about that?
i was just about to ask the forum if they knew anything about Wayne's design, but seeing as how it's in the process of being patented i figured there was no hope of ever constructing such a device myself. but if the man himself gives the go-ahead...
btw, i'm in the process of choosing a gain/buffer circuit for my multichannel preamp. was going to try the Walt Yung IC/discrete buffer combo, but after realizing how glorious a cheap passive volume control can sound i'm starting to cringe at the thought of putting so many stages of active circuitry in the way (i.e. ICs). so i'll start looking at some of your simpler designs... i'm also thinking of trying a discrete op-amp using your DIY design tips, but i guess discrete or not you don't like the sound of feedback very much?
p.s. i was meaning to contact you a while ago re: IC opamps and their application in audio for an article i'd like to write for Bound for Sound magazine, did marty dewulf tell you about that?
If you want to make a discrete version of the WM8816 attenuator it is possible to get some really good spec's with switches available today. Of course this is costly but iy should be possible to get really high perfomance.
Take a look at the DS1808 datasheet under switch resistance versus voltage with reference to GND you will see that it is not flat but better than most switches. however if you take new switches from Vishay,AD or MAxim you can get a resistance flatness of 1 - 5Ohm instead of 10 - 20Ohm.
And when you are smart you run them in a virtual gnd connection => no voltagechange => no resistance change.. You only will have a small currentchange : 1mA * 10Ohm => 10mV change. Thats something usefull!!!!
Sonny
Take a look at the DS1808 datasheet under switch resistance versus voltage with reference to GND you will see that it is not flat but better than most switches. however if you take new switches from Vishay,AD or MAxim you can get a resistance flatness of 1 - 5Ohm instead of 10 - 20Ohm.
And when you are smart you run them in a virtual gnd connection => no voltagechange => no resistance change.. You only will have a small currentchange : 1mA * 10Ohm => 10mV change. Thats something usefull!!!!
Sonny
The THD specs on the Wolfson part are not as good as one would expect with a well-designed "passive" integrated attenuator. In fact, the PGA2301 seems superior in spite of its integrated buffer amp. It is not very clear to me how Wolfson measured some of the distortion plots. Also, why would the device be inverting?
Sonny, back in the days of 4053 switches, Elektor used to say that the best way to use them was to use the op amp in non-inverting configuration because then its input resistance is several MegOhms so the nonlinear switch resistance would not matter. I guess you can pick your poison whether you want voltage/capacitace modulation in the switch and the op amp or load the switch with a current....
Probably the best way is to drive the attenuator from an extremely low impedance source (but you can do nothing about the internal resistors) and load it with a very high impedance buffer.
Regards,
Eric
Sonny, back in the days of 4053 switches, Elektor used to say that the best way to use them was to use the op amp in non-inverting configuration because then its input resistance is several MegOhms so the nonlinear switch resistance would not matter. I guess you can pick your poison whether you want voltage/capacitace modulation in the switch and the op amp or load the switch with a current....
Probably the best way is to drive the attenuator from an extremely low impedance source (but you can do nothing about the internal resistors) and load it with a very high impedance buffer.
Regards,
Eric
Okay CAPSLOCK.
For the use of analog switches the perfect way. Try look at: http://www.analog.com/productSelection/pdf/ssm2404.pdf
you wrote "Elektor used to say that the best way to use them was to use the op amp in non-inverting configuration because then its input resistance is several MegOhms so the nonlinear switch resistance would not matter"
I do not agree on this one.. When do you have a source impedance of several MegaOhms .. Not in opamp. I would agree with up to ~ 200kOhms.
I think the reason why the WOLFSON is perfoming less than perfect in their datasheet is due to the use of OP275 which need a low resistance feedback loop and source to perfom at its optimum (<2K seen by I+ and I-). So another opamp would show of a better perfomance .. i think.
The wolfson also uses the opamp i a way where it perfoms at its maximum... If you compare a opamp in non versus inverting mode you will see that it perform better in inverting mode.
Sonny
For the use of analog switches the perfect way. Try look at: http://www.analog.com/productSelection/pdf/ssm2404.pdf
you wrote "Elektor used to say that the best way to use them was to use the op amp in non-inverting configuration because then its input resistance is several MegOhms so the nonlinear switch resistance would not matter"
I do not agree on this one.. When do you have a source impedance of several MegaOhms .. Not in opamp. I would agree with up to ~ 200kOhms.
I think the reason why the WOLFSON is perfoming less than perfect in their datasheet is due to the use of OP275 which need a low resistance feedback loop and source to perfom at its optimum (<2K seen by I+ and I-). So another opamp would show of a better perfomance .. i think.
The wolfson also uses the opamp i a way where it perfoms at its maximum... If you compare a opamp in non versus inverting mode you will see that it perform better in inverting mode.
Sonny
And look at this Application note from Vishay .. Page 3 under "On resistance Flatness"
http://www.vishay.com/document/70606/70606.pdf
Sonny
http://www.vishay.com/document/70606/70606.pdf
Sonny
Sonnya:
I've wondered about this a lot too. I've seen the papers you've suggested and a few others (old ones by Siliconix, for example). It "seems logical to me" as Mr. Spock would say, that using the switches with an inverting op amp in the configuration you mention, with one of the switches in series with the feedback resisitor for "compensation" would be almost perfect.
But I have NEVER seen a manufactured product using this approach! Are there any? I don't know. But I've seen MANY products using the "Non-inverting amp as switch buffer" approach. I'd like to know why and I don't think it's just a matter of cost.
The last commerical Hafler preamp, the 915 used those old CMOS parts (CD4053/CD4051, etc.) used mention with a very nice JFET buffer (source follower with current sink) for minimal loading of the switches. It was a nice sounding preamp IMHO. I wonderd if fols that had this preamp would sub in one of the Maxim upgraded replacements, would there be a noticeable difference? So far, I haven't found an owner that has done this.
I'm planning on doing something similar with my current preamp - currently I chnages sources by pulling/placing cables
mlloyd1
I've wondered about this a lot too. I've seen the papers you've suggested and a few others (old ones by Siliconix, for example). It "seems logical to me" as Mr. Spock would say, that using the switches with an inverting op amp in the configuration you mention, with one of the switches in series with the feedback resisitor for "compensation" would be almost perfect.
But I have NEVER seen a manufactured product using this approach! Are there any? I don't know. But I've seen MANY products using the "Non-inverting amp as switch buffer" approach. I'd like to know why and I don't think it's just a matter of cost.
The last commerical Hafler preamp, the 915 used those old CMOS parts (CD4053/CD4051, etc.) used mention with a very nice JFET buffer (source follower with current sink) for minimal loading of the switches. It was a nice sounding preamp IMHO. I wonderd if fols that had this preamp would sub in one of the Maxim upgraded replacements, would there be a noticeable difference? So far, I haven't found an owner that has done this.
I'm planning on doing something similar with my current preamp - currently I chnages sources by pulling/placing cables
mlloyd1
I do not have a lot of time right now but check :
http://www.die-klaassens.de/Audio_und_Elektronik/Preamp_L10/preamp_l10.html
He had made a really nice job (preamp) where he uses CS3310 and SSM2404 switches the right way. And the cream on top!! Walt Jung look a like buffer outputstage.
There is not much you can complain him for!!
He says he will make a page when the project is finished.. I haven't seen this page before, so it is a accident that he uses the same ucontroller as i will!! Thats to funny!
Sonny
http://www.die-klaassens.de/Audio_und_Elektronik/Preamp_L10/preamp_l10.html
He had made a really nice job (preamp) where he uses CS3310 and SSM2404 switches the right way. And the cream on top!! Walt Jung look a like buffer outputstage.
There is not much you can complain him for!!
He says he will make a page when the project is finished.. I haven't seen this page before, so it is a accident that he uses the same ucontroller as i will!! Thats to funny!
Sonny
i think any attenuator scheme i come up with would have to be fixed series, variable shunt. the thought of putting solid state switches directly in the signal path makes me cringe. even when used only in the shunt path, the nonlinearity of a solid state switch could be an issue. so i'm trying to devise an attenuator that would have multiple (probably 12-16) shunt resistors individually switchable and controlled by microprocessor. i just need to figure out the optimal resistor values to be able to have usable increments (e.g. 1dB). if practical, i may even end up using small SMT mechanical relays for the shunt switching, so as to avoid CMOS or bipolar switches altogether.
the primary disadvantage of a fixed series attenuator is the variable impedance of course... so most likely i would want to put a buffer before it and a gain stage after it. or, i could use a "distributed gain" setup like Bryston does in their preamps, and use an input buffer with some gain, and an output stage with some gain.
even this scheme is not optimal to me though... having 2 active stages is more than i would like (i don't even want one to be honest, i like the sound of no electronics like my pasive preamp). but i guess everything is a compromise, one way or another... *sigh* i'll just have to continue my quest for the "perfect" preamp.
marc
the primary disadvantage of a fixed series attenuator is the variable impedance of course... so most likely i would want to put a buffer before it and a gain stage after it. or, i could use a "distributed gain" setup like Bryston does in their preamps, and use an input buffer with some gain, and an output stage with some gain.
even this scheme is not optimal to me though... having 2 active stages is more than i would like (i don't even want one to be honest, i like the sound of no electronics like my pasive preamp
). but i guess everything is a compromise, one way or another... *sigh* i'll just have to continue my quest for the "perfect" preamp.marc
When you are thinking about using solid state for volume
controls, you will commonly see applications using analog
gates, and here's a tip:
CMOS analog gates make good shunts to ground. You
attenuate the signal with a resistor and analog gate in series
to ground. One port of the analog gate is grounded.
In this way, the signal voltage seen by the gate in the "ON"
state is very close to ground, and the distortion is very low.
The same effect also works for you if you place the analog gate
in series with the inverting input on a summing junction of
an op amp type circuit because the inverting input is held to virtual ground.
controls, you will commonly see applications using analog
gates, and here's a tip:
CMOS analog gates make good shunts to ground. You
attenuate the signal with a resistor and analog gate in series
to ground. One port of the analog gate is grounded.
In this way, the signal voltage seen by the gate in the "ON"
state is very close to ground, and the distortion is very low.
The same effect also works for you if you place the analog gate
in series with the inverting input on a summing junction of
an op amp type circuit because the inverting input is held to virtual ground.
hi Mr. Pass
maybe CMOS switches won't be so bad here after all. i had a suspicion that they would work ok in a shunt configuration, seeing how app notes usually recommend using them in "virtual ground" mode. so i guess a real ground could only be better. i think i will go with a switched shunt/fixed series scheme using the AD SSM2404 switches. guess the hard part is still gonna be choosing good buffer/gain stages for the whole thing...
i'm assumming the bipolar switches wayne uses in his proprietary attenuator are his own custom design from discrete parts?
cheers,
marc
p.s. i hope you didn't see those posts in the "How long electrolytic caps last?" thread in which i used your name in vain...
maybe CMOS switches won't be so bad here after all. i had a suspicion that they would work ok in a shunt configuration, seeing how app notes usually recommend using them in "virtual ground" mode. so i guess a real ground could only be better.
i think i will go with a switched shunt/fixed series scheme using the AD SSM2404 switches. guess the hard part is still gonna be choosing good buffer/gain stages for the whole thing...i'm assumming the bipolar switches wayne uses in his proprietary attenuator are his own custom design from discrete parts?
cheers,
marc
p.s. i hope you didn't see those posts in the "How long electrolytic caps last?" thread in which i used your name in vain...
If you'd like to know about volume control with bipolar switches, take a look at http://mitglied.lycos.de/MaikHerzog/audio_projects/preamplifier/preamp_text.html.
It has been on the Pass-gallery some time ago, but disappeared from the site for some reason.
Holger
It has been on the Pass-gallery some time ago, but disappeared from the site for some reason.
Holger
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.