Put the 10,000s in the PSU with snubbers then experiment with the caps on the amp board.
Try leaving the 1500 but give some other values a try... 100, 220, 470, 1000. They won't cost much, are bound to come in useful in future and giving them all a listen will give you the experience of having heard the difference they make.
Try leaving the 1500 but give some other values a try... 100, 220, 470, 1000. They won't cost much, are bound to come in useful in future and giving them all a listen will give you the experience of having heard the difference they make.
Interesting cap values for onboard the amplifier include: 150uF, 220uF, 330uF.
Possible variations on sound:
A squeak--This means the cap is either too large or too small
A boom (missing low bass)--This means the cap is too large
Missing bass--This means the cap is too small
The snubber, a zobel, RC dealy is tunable. This item can help allow greater capacitance without loss of driver control. Since a zobel is a masking component, it is also used with a little boost-- a bypass cap. You can find this implementation in the chipamp.com power supply. However, it is necessary for these "effects" to match up with your application.
Several terms for driver control include "driver command" and "bass command" and, well, there's a lot more terms.
However, depending on how well the snubber works, its possible that your 1500uF could do fine. When and if a snubber supply works, it is for the purpose of using large capacitance throughout. So, if its working (at your bandwidth), then cap values of 680uF and higher (up to 4700uF) should work without loss of driver control. This will depend on the bandwidth used versus the tuning of the RC (snubber).
My imagination says that the clear reason for dual mono layouts on chipamps seems to be in more thoroughly controlling the amount of capacitance per each amp chip. However, it also seems that any "more transformer" method could do a similar job in decreasing the reliance on caps.
Also, DC issues can cause loss of driver control. Since its not a bridge amp, then you need absolute zero dc offset for best driver control.
Um, its a good time to mention that a 10w wirewound resistor, like used for speakers, is a nice safe way to drain power caps.
I went to these topics, because its quite unlikely for LM3875 to be bass-shy, although its easy to make too much midrange in a chip amp.
You might also check out some Wima caps (or auricaps sound similar) in especially small values for input filter caps. They have a slightly euphonic sound, and that will include much bass if neither too large nor too small. Interesting electrolytic options include Nichicon KZ, Elna Tonerex (euphonic), Elna Cerefine, Baknor BPT, and Ruby's Blackgate. Those are for audio signal; although Nichicon is an excellent (loud bass) choice for power caps. Inexpensive Mallory (Cornell) can also do a good job (especially higher voltage models), and their reference materials are really good.
Have fun!
Possible variations on sound:
A squeak--This means the cap is either too large or too small
A boom (missing low bass)--This means the cap is too large
Missing bass--This means the cap is too small
The snubber, a zobel, RC dealy is tunable. This item can help allow greater capacitance without loss of driver control. Since a zobel is a masking component, it is also used with a little boost-- a bypass cap. You can find this implementation in the chipamp.com power supply. However, it is necessary for these "effects" to match up with your application.
Several terms for driver control include "driver command" and "bass command" and, well, there's a lot more terms.
However, depending on how well the snubber works, its possible that your 1500uF could do fine. When and if a snubber supply works, it is for the purpose of using large capacitance throughout. So, if its working (at your bandwidth), then cap values of 680uF and higher (up to 4700uF) should work without loss of driver control. This will depend on the bandwidth used versus the tuning of the RC (snubber).
My imagination says that the clear reason for dual mono layouts on chipamps seems to be in more thoroughly controlling the amount of capacitance per each amp chip. However, it also seems that any "more transformer" method could do a similar job in decreasing the reliance on caps.
Also, DC issues can cause loss of driver control. Since its not a bridge amp, then you need absolute zero dc offset for best driver control.
Um, its a good time to mention that a 10w wirewound resistor, like used for speakers, is a nice safe way to drain power caps.
I went to these topics, because its quite unlikely for LM3875 to be bass-shy, although its easy to make too much midrange in a chip amp.
You might also check out some Wima caps (or auricaps sound similar) in especially small values for input filter caps. They have a slightly euphonic sound, and that will include much bass if neither too large nor too small. Interesting electrolytic options include Nichicon KZ, Elna Tonerex (euphonic), Elna Cerefine, Baknor BPT, and Ruby's Blackgate. Those are for audio signal; although Nichicon is an excellent (loud bass) choice for power caps.
Inexpensive Mallory (Cornell) can also do a good job (especially higher voltage models), and their reference materials are really good. Have fun!
Ok, sounds good. I was reading CarlosFM's thread on his snubberized PS, so I think that I'm going to do this. Keep in mind that what I have set up is a 'stereo' implementation - two amp boards on a single PS board.
2x 10000uF (ish) caps on the PS board
100nF film cap and 1R resistor in series on each rail of PS board
220uF cap || 100nF ceramic cap on amp board
Now why do I need such huge wattage on the capacitor discharge resistors? CarlosFM's design has 2.2k 2W resistors for that purpose. The caps won't reach a voltage higher than 50V, I'm guessing, which means that they'll have to dissipate somewhere around 1.34W. My rails are +/- 25V, so that's more like a quarter watt. Shouldn't 2w resistors be fine?
2x 10000uF (ish) caps on the PS board
100nF film cap and 1R resistor in series on each rail of PS board
220uF cap || 100nF ceramic cap on amp board
Now why do I need such huge wattage on the capacitor discharge resistors? CarlosFM's design has 2.2k 2W resistors for that purpose. The caps won't reach a voltage higher than 50V, I'm guessing, which means that they'll have to dissipate somewhere around 1.34W. My rails are +/- 25V, so that's more like a quarter watt. Shouldn't 2w resistors be fine?
That's funny. Um, I had recommended a sandcast resistor because of the nice big ceramic "handle" as a safe spot to hold onto it if you're manually discharging caps or in a hurry. Its even nicer with some test leads on it, for a dedicated discharging tool.
That's a safety in addition to your bleeder resistors aboard the power supply.
That's a safety in addition to your bleeder resistors aboard the power supply.
How sensitive are your speakers, and what impedance?
I've built a bunch of these guys from different chips and different capacitance in the PS and different configs. I generally find that for insensitive speakers you need high current capability, but for very sensitive speakers low capacitance is fine.
My main speakers are specced at 90 dB/w, 8 ohms, so a relatively easy load driven by even one LM4766 (that's just 30 watts) sounds fine, with sufficient bass control. At high volumes the chip struggles a bit, but it's anyway got a severely restricted current output. The main listening happens with a single LM3886 for each channel, though. 10,000uF at PSU and 100uF FM at chips (P2P wiring).
I also have a pair of Dynaudio BM6 passive studio monitors. Those are specced at 86dB, and 4 ohms. And they are difficult to drive. I use a BPA LM4780 with 4 chips (thats = 8 LM3886) driven by a balanced soundcard source. there's just 4700uF per side/per channel (total 8x4700uF) in addition to the stock 1500 uF FC provided by Peter. It drives the Dynes quite happily, but the transformer is also pretty heavily specced - 10 amps, 22 volt windings, with oversized core and winding wire.
With either config I can reach volume levels sufficient to bust my eardrums, and with tactile, quality bass.
I've built a bunch of these guys from different chips and different capacitance in the PS and different configs. I generally find that for insensitive speakers you need high current capability, but for very sensitive speakers low capacitance is fine.
My main speakers are specced at 90 dB/w, 8 ohms, so a relatively easy load driven by even one LM4766 (that's just 30 watts) sounds fine, with sufficient bass control. At high volumes the chip struggles a bit, but it's anyway got a severely restricted current output. The main listening happens with a single LM3886 for each channel, though. 10,000uF at PSU and 100uF FM at chips (P2P wiring).
I also have a pair of Dynaudio BM6 passive studio monitors. Those are specced at 86dB, and 4 ohms. And they are difficult to drive. I use a BPA LM4780 with 4 chips (thats = 8 LM3886) driven by a balanced soundcard source. there's just 4700uF per side/per channel (total 8x4700uF) in addition to the stock 1500 uF FC provided by Peter. It drives the Dynes quite happily, but the transformer is also pretty heavily specced - 10 amps, 22 volt windings, with oversized core and winding wire.
With either config I can reach volume levels sufficient to bust my eardrums, and with tactile, quality bass.
It's the speaker in this thread: http://htguide.com/forum/showthread.php4?t=13154 Should be very easy to drive...
Here's what I'm wondering. I've been reading through the snubberized PS thread, and everyone in there was using dual mono setups with something like 4x 10000uF per channel. I would be using 2x10000uF for BOTH channels. Is it worth it to make the switch?
peranders, I think I mis-read what you said. It's late. I don't think the harshness on the high end is new clarity or lack of distortion. It doesn't sound clear and bright, it sounds grating and painful. Norah Jones' voice shouldn't make me wince.
peranders, I think I mis-read what you said. It's late.
I don't think the harshness on the high end is new clarity or lack of distortion. It doesn't sound clear and bright, it sounds grating and painful. Norah Jones' voice shouldn't make me wince.
Certainly 10,000uF is better than nothing and I wouldn't blindly follow 27.2mF "rule" In fact, I know about person who uses my premium kit with 10,000uF BG caps and with his speakers it was the best solution.
I presented my views on snubberized supply here: http://www.diyaudio.com/forums/showthread.php?postid=838981#post838981
I'm presently testing 12" Electro Magnet Speaker drivers: http://cc.msnscache.com/cache.aspx?q=72918736801214&mkt=en-CA&lang=en-CA&w=e1b40950&FORM=CVRE and the bass is more than I've ever dreamed of. And I'm only using 100uF at the chips.
I presented my views on snubberized supply here: http://www.diyaudio.com/forums/showthread.php?postid=838981#post838981
Peter Daniel said:
I'm presently testing 12" Electro Magnet Speaker drivers: http://cc.msnscache.com/cache.aspx?q=72918736801214&mkt=en-CA&lang=en-CA&w=e1b40950&FORM=CVRE and the bass is more than I've ever dreamed of. And I'm only using 100uF at the chips.
Attachments
Peter,Peter Daniel said:
Is that 100uF the TOTAL rail capacitance, or is that just what you have at the chips? Also, I didn't see a sensitivity for those particular speakers, though the max power was 20W, so I am thinking that these are pretty efficient.
That got me wondering "out loud". . . . .
My speakers aren't the most efficient beasts in the world, and though I don't have a bunch of other speakers to test against, I am betting that the "lack of bass" that people observe for the gainclones is related to the efficiency of their speakers.
I have noticed that most of the people (like me) who complain about the gaincard circuit being anemic at the low end need more power to drive their systems, while the people driving efficient (often full range) speakers tend to get more bass out of this circuit. As a corollary statement, I would posit that most of the SET (and other low power tube amp) fans that rave about the Gaincard are using pretty efficient speakers too.
Since you (Peter) have obviously seen many more gainclone based systems that I, would you care to comment as to the validity of this observation? I could be WAY off, but it seems to make sense to me insofar as bass reproduction in less efficient speakers would be taxing the rail capacitance much more than bass reproduction in "efficient" speakers.
Just a theory. . . .
Those drivers are 98dB efficient and most speakers I was using with chipamps had efficiency of 90dB or more.
There are additional 1000uF in the PS and nothing more.
In fact, some people use my amps just for the bass drivers, the best example is that system, where GC monoblocks (1000uF supply caps) power the 30-foot long straight bass horns: http://www.6moons.com/audioreviews/westernelectric/300b.html
These are the actual amps used in that system:
Certainly the chipamps may work better with efficient speakers, but I also have a theory that they allow you to finally hear what your source equipment is all about and if it lacks the bass, that what the amps will show
There are additional 1000uF in the PS and nothing more.
In fact, some people use my amps just for the bass drivers, the best example is that system, where GC monoblocks (1000uF supply caps) power the 30-foot long straight bass horns: http://www.6moons.com/audioreviews/westernelectric/300b.html
These are the actual amps used in that system:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Certainly the chipamps may work better with efficient speakers, but I also have a theory that they allow you to finally hear what your source equipment is all about and if it lacks the bass, that what the amps will show
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.