Why is there a plethora of linestages out there with inordinately large output caps ? If I didn't know better I'd swear there was some kind of 'mines bigger than yours' competition going on.
I have heard the phase shift argument, but simply don't buy having to construct something that's flat to 0.5Hz into a 10k load to get decent bass; especially when most of us are using tube power amps with around 47k minimum input impedance. In my book, smaller caps sound a whole lot better, and also happen to be much cheaper.
The other trend is the use of vanishingly low resistors to load these mega-capacitors, thus apparently defeating their purpose.
Please put me out of my misery. To paraphrase Marvin: what's going on ? Is there a secret society of elephant audiophiles out there with infra-bass hearing down to 3Hz ? Am I the only person with a CD player that cuts out below 20Hz ?
pm
I have heard the phase shift argument, but simply don't buy having to construct something that's flat to 0.5Hz into a 10k load to get decent bass; especially when most of us are using tube power amps with around 47k minimum input impedance. In my book, smaller caps sound a whole lot better, and also happen to be much cheaper.
The other trend is the use of vanishingly low resistors to load these mega-capacitors, thus apparently defeating their purpose.
Please put me out of my misery. To paraphrase Marvin: what's going on ? Is there a secret society of elephant audiophiles out there with infra-bass hearing down to 3Hz ? Am I the only person with a CD player that cuts out below 20Hz ?
pm
IMO, anything bigger than 3.3 muF. is nonsense. 3.3 muF. satisfies the established rule of thumb for 3 dB. down at or below 5 Hz. into the IHF "standard" 10 KOhm load. That's the worst case situation.
Mach1 correctly points out a smaller capacitance is OK when the load is known to be > 10 K. High quality capacitance is EASIER to come by at smaller values.
150 ton Blue Whales must be the owners of the preamps with the gargantuan coupling caps.
Mach1 correctly points out a smaller capacitance is OK when the load is known to be > 10 K. High quality capacitance is EASIER to come by at smaller values.
150 ton Blue Whales must be the owners of the preamps with the gargantuan coupling caps.
Phase shifts are a problem for ~ 10X the 3db frequency, which means a 3db of 5Hz will cause problems to 50Hz which is, theoretically audible. Also, even though the input impedance may be set at 47K, there is also generally a resistor from the output to ground of the preamp itself which is also usually 47K, which means that the load the cap sees is actually more like 24K. Thus, 4.7uF is more appropriate. Once you have spent all that money on building a line stage, why not spend the extra $5 for the bigger cap?
dsavitsk,
Thanks for your input. I have also heard the phase shift argument from Brian Beck, who's opinion I greatly respect:
However, I have not been able to confirm this contention empirically nor find any published work on the subject (if you could point me towards anything however, it would be most appreciated). I have always followed the -3db @ 5Hz rule that Eli quoted, and have never noticed any benefits at the bottom end from going any larger (I do concede that some benefit may be gained if the circuit already has a couple of poles around 5Hz). OTOH I have been able to hear the impact of very large output caps (I have seen 8uF prescribed in some schematics) on the midrange.
I am also under the impression that CD players have a 20Hz -20KHz bandwidth and believe that the lower limit here would be far more likely to smear 'presence and air' than a -3db point at 5Hz induced by a coupling cap. Coming from the age of the gramophone (and still running one) I am also aware of the large amount of power consuming subsonic output that slightly warped records and/or sub-optimal cartridge/tonearm combinations can generate. Many phono amps are deliberately rolled off at the extreme bottom end specifically for this reason, yet still manage to display superb low and mid-bass.
Re the 47k resistor on the output of the linestage - this is related to my second point. Why are such small resistors used ? I have always seen the function of this resistor as giving the tube some kind of load to run into if the linestage is disconnected, with the bulk of the loading coming from the power amp input. My preference is to use 470k minimum.
regards
pm
Thanks for your input. I have also heard the phase shift argument from Brian Beck, who's opinion I greatly respect:
However, I have not been able to confirm this contention empirically nor find any published work on the subject (if you could point me towards anything however, it would be most appreciated). I have always followed the -3db @ 5Hz rule that Eli quoted, and have never noticed any benefits at the bottom end from going any larger (I do concede that some benefit may be gained if the circuit already has a couple of poles around 5Hz). OTOH I have been able to hear the impact of very large output caps (I have seen 8uF prescribed in some schematics) on the midrange.
I am also under the impression that CD players have a 20Hz -20KHz bandwidth and believe that the lower limit here would be far more likely to smear 'presence and air' than a -3db point at 5Hz induced by a coupling cap. Coming from the age of the gramophone (and still running one) I am also aware of the large amount of power consuming subsonic output that slightly warped records and/or sub-optimal cartridge/tonearm combinations can generate. Many phono amps are deliberately rolled off at the extreme bottom end specifically for this reason, yet still manage to display superb low and mid-bass.
Re the 47k resistor on the output of the linestage - this is related to my second point. Why are such small resistors used ? I have always seen the function of this resistor as giving the tube some kind of load to run into if the linestage is disconnected, with the bulk of the loading coming from the power amp input. My preference is to use 470k minimum.
regards
pm
Hi mach1
My view is very similar to yours - use the smallest possible cap you can get away with in order to maximise sonic quality.
And despite what dsavitsk says a big cap does not cost $5 more, at least not a big cap worth having Prices very quickly reach a level where output transformers look like a cheap alternative.
Btw, the purpose of the output resitor is to create a capacitor discharge path to ground. This is mostly an issue if you use a tube pre with a SS power amp. No reason a 470k resitor would not be safe enough.
My view is very similar to yours - use the smallest possible cap you can get away with in order to maximise sonic quality.
And despite what dsavitsk says a big cap does not cost $5 more, at least not a big cap worth having
Prices very quickly reach a level where output transformers look like a cheap alternative.Btw, the purpose of the output resitor is to create a capacitor discharge path to ground. This is mostly an issue if you use a tube pre with a SS power amp. No reason a 470k resitor would not be safe enough.
One can save a lot of money if one can avoid paying the audio tax. If you must go huge, low impedance oil motor run capacitors are a good option. I will agree that its best not to oversize, cutoff a decade or so from any frequency of interest seems a common rule of thumb.
Non ferrous 500VAC 80uF polypropylene in oil capacitor. $12, ships internationally.
Non ferrous 500VAC 80uF polypropylene in oil capacitor. $12, ships internationally.
All of this is good reason to direct couple the output and use a servo to get rid of offset. That way, the cap coupling to the last stage can be small without a bass phase shift penalty.
Total cost of servo (assuming you've got power supplies in the preamp already): less than $1.50 for two channels.
Total cost of servo (assuming you've got power supplies in the preamp already): less than $1.50 for two channels.
analog_sa said:
Off the top of my head:
1.) Use high resistances around the opamp to avoid loading the tube circuit in question. Use good MF resistors.
2.) This means a FET input opamp, low-offset of course
3.) Use the best small cap around the opamp that you can find. You want it to behave perfectly up to ultrasonics to keep the opamp from “chattering”. Since this will be smaller than a coupling cap, knock yourself out with a Teflon or Unobtanium cap. Or maybe just an Auricap.
4.) Use a far faster, lower distortion opamp than you think would be needed for a mere servo. An OPA627 might be overkill, but...maybe not, if you’re obsessive like moi.
5.) Pull current out of the opamp's NPN output with either a resistor to -Vcc or a simple CCS (sink) to -Vcc, to keep the output in Class A (the rest of the servo circuit may accomplish this already without this help).
6.) Choose a very low servo frequency because the opamp will respond at a 6dB/oct declining rate across the audio band, and we want to start that decline at a very low frequency to get more opamp attenuation in the audio band. If this low frequency creates a settling time problem at start-up, then consider output shorting relays or sequential power-up. I once included a timer-relay-switched pair of servo caps - one small, one large. The smaller cap is used at first to settle very quickly (make sure that you’re charging the larger cap simultaneously) then it would switch to the bigger cap for a deeper response. Hopefully, you won’t need the switched caps because that is a big complication.
7.) Add a second RC low-pass filter after the opamp at a pole frequency at least tens times, preferably more, than the servo frequency to filter any residual opamp "chatter" before it hits the tube circuit. Use a decent cap here, too. This also filters any signal going backwards into the opamp’s output. This added low-pass creates a second-order filter with a potential bump in the response, so you want the two poles spread far apart.
I am NOT saying that you must do all of these things to make a servo that steps out of the way - but this is my complete laundry list as I recall it right now.
EDIT: I think I've exceeded SY's $1.50 budget, but again, this is a list of ideas and you don't need to use them all. Also, please consider what you might have to pay for the best 4.7uF coupling cap (or cap combo) that you could find.
EDIT 2: I think I used a switched RESISTOR servo speed-up circuit before rather than caps, but the principle is the same.
analog_sa said:And despite what dsavitsk says a big cap does not cost $5 more, at least not a big cap worth having
"Cap worth having" is the operative phrase here. A 3.3uF 450V Auricap costs $16.50 at soniccraft.com. A 4.7uF 450V Auricap costs $18.70. I have used a lot of different caps, and while the Auricap is not the best I have tried, it is pretty good as a baseline. Beyond it is entering into the land of esoterica where reason seems to break down and you enter at your own risk
Yes, although you may need negative level shifters and/or negative supplies if you want a ground referenced output, which does add complication. The best prior art on DC amplifiers comes from the old vacuum tube operational amplifier work. If you can find a book by Korn and Korn titled "Electronic Analog Computers" that would be a good buy, since it shows the many ways to do this. What we need is a complementary P-channel-type tube ("P" for "proton"). Has anyone got one?
Try the The Los Alamos Neutron Science Center, Brian. I know you like using big tubes at high voltage and power for fidelity. Electrode life is a bit problematic though.
1ma average proton beam source, 16ma peak.
Burle 7835 Electron Tube Parameters:
Type: Cathode-Driven Triode
Gain: 13.6 dB
Designed: 1960
Filament: Thoriated Tungsten
Filament V: 4 V
Filament I: 6800 A
Plate Dissipation: 300 kW*
Plate Voltage Max: 40 kV
Water Flow: 200 GPM
*Administrative controls limit the operation to plate dissipations of less than 250 kW. Catastrophic failures have been significantly reduced.
Electrode life is a bit problematic though.1ma average proton beam source, 16ma peak.
An externally hosted image should be here but it was not working when we last tested it.
Burle 7835 Electron Tube Parameters:
Type: Cathode-Driven Triode
Gain: 13.6 dB
Designed: 1960
Filament: Thoriated Tungsten
Filament V: 4 V
Filament I: 6800 A
Plate Dissipation: 300 kW*
Plate Voltage Max: 40 kV
Water Flow: 200 GPM
*Administrative controls limit the operation to plate dissipations of less than 250 kW. Catastrophic failures have been significantly reduced.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.