Greetings,
I saw a $$$$ amp that had an optional HV cap bank, no values mentioned but from the looks I'd say it must have been over 2000 µF (2mF). Of course they claimed all sorts of benefits, which made me wonder: how much capacitance is useful for an output stage, say a 100W PP? I've gone up to 500 µF, but is there a point in using more?
Thanks in advance.
I saw a $$$$ amp that had an optional HV cap bank, no values mentioned but from the looks I'd say it must have been over 2000 µF (2mF). Of course they claimed all sorts of benefits, which made me wonder: how much capacitance is useful for an output stage, say a 100W PP? I've gone up to 500 µF, but is there a point in using more?
Thanks in advance.
The only benefit I can think of is the HT will still be high as the heaters go out. There will be no noticeable hum reduction above say 47uF at an HT of 450v.
I bet someone who believes that snake oil is good for you, will buy one!
Push pull output stages are very good at hum rejection as they are.
I bet someone who believes that snake oil is good for you, will buy one!
Push pull output stages are very good at hum rejection as they are.
The capacitance needed doesn't just depend on output power, but also on supply rail voltage (higher V means less C is needed), whether the PSU is conventional full-wave or a doubler, how sensitive the speakers are etc. etc. This is why amps need to be designed.
Caps which are too big add no real value - but may look good in marketing aimed at people who don't understand electronics.
Caps which are too big add no real value - but may look good in marketing aimed at people who don't understand electronics.
Second that. My choice would be small PS capacitors of high quality rather than large capacitors of poor quality. I have antique parallel PP 2A3 monoblocks that have only two 6 uF paper-in-oil capacitors in power supply, plus additional 4 uF for driver stage. Adding more capacitance did not improve the sound; the original supply turned to be the best.
There isn't much more to it than hum reduction. OK, there is also ripple intermodulation but that is proportional to hum. If the mains transformer has highish resistance then there may be some advantage in using a bigger cap, but a bigger transformer may be better.
Some people, of course, have gone the other way and proclaim the 'benefits' of rather small caps. In many cases this is because they are using amps with no feedback, so the gain varies with supply rail voltage so a weedy PSU can add envelope-based gain pumping which can be misinterpreted as better dynamic range.
Some people, of course, have gone the other way and proclaim the 'benefits' of rather small caps. In many cases this is because they are using amps with no feedback, so the gain varies with supply rail voltage so a weedy PSU can add envelope-based gain pumping which can be misinterpreted as better dynamic range.
To add my little bit:
The nature of music (normal music!) is occasional loud notes with relatively long moments of lower required energy in between. That means that a comparatively 'stiff' power supply (large filter end caps) emulates a fixed voltage supply. But intuitively there is a point of limited advantage as said. One needs to remember that any 'deviation' in output of less than 3 dB is not 'worth writing home about' - and that represents a fair variation in signal amplitude. As Kaputt elegantly stated, the disadvantages start rearing their collective heads.
It can be informative to watch the h.t. with a 'scope while a passage of music of different intensity is played (this can also sometimes be a real horror show!) There one can study just how much the rail voltage varies and are restored. Again, in my own 100W amplifier I found that an equivalent of 500µF on a 600V supply was really the limit of any usefull (remember the 3dB difference!) capacitance, and that included organ music. But it certainly depends on the amplifier design.
The nature of music (normal music!) is occasional loud notes with relatively long moments of lower required energy in between. That means that a comparatively 'stiff' power supply (large filter end caps) emulates a fixed voltage supply. But intuitively there is a point of limited advantage as said. One needs to remember that any 'deviation' in output of less than 3 dB is not 'worth writing home about' - and that represents a fair variation in signal amplitude. As Kaputt elegantly stated, the disadvantages start rearing their collective heads.
It can be informative to watch the h.t. with a 'scope while a passage of music of different intensity is played (this can also sometimes be a real horror show!) There one can study just how much the rail voltage varies and are restored. Again, in my own 100W amplifier I found that an equivalent of 500µF on a 600V supply was really the limit of any usefull (remember the 3dB difference!) capacitance, and that included organ music. But it certainly depends on the amplifier design.
You can calculate how deep would be ripples on the cap under the max load and decide, is it acceptable, or not. 120 Hz ripples would be modulated by envelope and low signal frequencies.
One misconception about PP amps is, that does not need so clean power as SE amp, that is not true. You may not hear the hum in pauses when tubes are well biased, but when the music plays, all this dirt adds to the sound.
One misconception about PP amps is, that does not need so clean power as SE amp, that is not true. You may not hear the hum in pauses when tubes are well biased, but when the music plays, all this dirt adds to the sound.
More powerful amplifiers have lower Ra, yielding higher currents from the PS.
How about the PS impedance / amplifier impedance relation? I mean, the reservoir has to replenish after being drained.
The ripple is not the main goal about the p-p, in my opinion.
The great reservoir helps in the dynamic performances.
Mainly if you use a good power tubes that has the capability to deliver a great impulsive current to the load in the region where the L is dropping (=low frequencies in primis).
With the proper instrument it is possible to see, with a Tritim test (by AP), that when you use an impulsive signal the output shape is more clean when the power supply is robust.
Walter
The great reservoir helps in the dynamic performances.
Mainly if you use a good power tubes that has the capability to deliver a great impulsive current to the load in the region where the L is dropping (=low frequencies in primis).
With the proper instrument it is possible to see, with a Tritim test (by AP), that when you use an impulsive signal the output shape is more clean when the power supply is robust.
Walter
In attach some test of a project of mono amp with 4 x KT150each that I have presented on Audio review magazine in Italy ( I sent a thread here also)
On a diagram of the continuos power ( red line) and dynamic power (blue line) is possible to see the differences that comes with the help of a robust power supply , 510 joule each channel.
The test is done with changing the load with a specific sw and AP1 that take a care of distortion, in this case 2% maximum.
On 8 ohms the difference is about 15 watts but on 4 ohms is around 100 watt
Next time I will take a imagine of scope.
Walter
On a diagram of the continuos power ( red line) and dynamic power (blue line) is possible to see the differences that comes with the help of a robust power supply , 510 joule each channel.
The test is done with changing the load with a specific sw and AP1 that take a care of distortion, in this case 2% maximum.
On 8 ohms the difference is about 15 watts but on 4 ohms is around 100 watt
Next time I will take a imagine of scope.
Walter
Last edited:
I am preparing a new amp, stereo with 300B, pushpull with the same concepts on supply.
I will test it on lab to write an article on Audioreview; I can check it both ways.
I have done in the past with other stuff but not recorded.
In every case this is the way to get a good results on dynamic signals as music.
The differences on curves tell this.
Each amp has 8 x 470 uF / 550 vdc so we can't consider it a weak psu.
The test is with continuos signal and impulsive, each 2.5 seconds a 40 mS signal.
This a sw wrote years ago with AP1 by Fabrizio Montanucci in Audioreview.
The Ap1 drive a external hardware ( made inside the Audioreview) that switch the load when the target Thd is reached and record the value that later is translated in a graph.
The load range is written in right corner of the graph.
On 3,2 ohm the power was 290 watts; at 2,7 ohm is 200 watts
This amp as an Ot trafo, double C , potted, with a single sec. set at 5 ohms. The Vdc is at 520 volt, I bias each KT150 at 45-50 mA, not more.
There is a thread where I describe it.
Walter
I will test it on lab to write an article on Audioreview; I can check it both ways.
I have done in the past with other stuff but not recorded.
In every case this is the way to get a good results on dynamic signals as music.
The differences on curves tell this.
Each amp has 8 x 470 uF / 550 vdc so we can't consider it a weak psu.
The test is with continuos signal and impulsive, each 2.5 seconds a 40 mS signal.
This a sw wrote years ago with AP1 by Fabrizio Montanucci in Audioreview.
The Ap1 drive a external hardware ( made inside the Audioreview) that switch the load when the target Thd is reached and record the value that later is translated in a graph.
The load range is written in right corner of the graph.
On 3,2 ohm the power was 290 watts; at 2,7 ohm is 200 watts
This amp as an Ot trafo, double C , potted, with a single sec. set at 5 ohms. The Vdc is at 520 volt, I bias each KT150 at 45-50 mA, not more.
There is a thread where I describe it.
Walter
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.