The only references I could find to clarlosfm snubberised psu's were just what appeared to be normal psu's with snubbers, so I'll assume that's what you mean.
Big caps are designed to deliver longer durations of power when the system demands it. Because of their physical size, they also usually have higher impedances and so lower current delivering capacity - limiting their use for removing or supplying power at higher frequencies. That's significant enough that most analog DIY'ers will say they can hear that difference even at audio frequencies. But chips that feature digital processing will place a requirement on the capacitor to supply power at frequencies many, many times higher.
Luckily, each ripples of power they must supply is small in size, just rapidly repeating. So a faster, smaller capacitance value component will work better.
The problem with smaller capacitance value, high speed caps is that putting them at the output of your PSU and then running tracks and cables to the IC's will create a capacitor / inductor circuit. This usually isn't a serious problem with big caps because they have more impedance. But the small value ones can create ringing oscillations with the inductance.
For that reason, board engineers will always bypass as IC's as close as possible to the pins of the chip - sometimes with the capacitors virtually on the pins. Adding some resistance to the tiny capacitors increases the time constant of the capacitor and so it's ability to form parasitic, resonant circuits with the boards, leads, pins etc of nearby components.
Essentially, the ultimate power supply is one that connects directly to the processing elements it supplies - with zero inductance / resistance between the two.
Big caps are designed to deliver longer durations of power when the system demands it. Because of their physical size, they also usually have higher impedances and so lower current delivering capacity - limiting their use for removing or supplying power at higher frequencies. That's significant enough that most analog DIY'ers will say they can hear that difference even at audio frequencies. But chips that feature digital processing will place a requirement on the capacitor to supply power at frequencies many, many times higher.
Luckily, each ripples of power they must supply is small in size, just rapidly repeating. So a faster, smaller capacitance value component will work better.
The problem with smaller capacitance value, high speed caps is that putting them at the output of your PSU and then running tracks and cables to the IC's will create a capacitor / inductor circuit. This usually isn't a serious problem with big caps because they have more impedance. But the small value ones can create ringing oscillations with the inductance.
For that reason, board engineers will always bypass as IC's as close as possible to the pins of the chip - sometimes with the capacitors virtually on the pins. Adding some resistance to the tiny capacitors increases the time constant of the capacitor and so it's ability to form parasitic, resonant circuits with the boards, leads, pins etc of nearby components.
Essentially, the ultimate power supply is one that connects directly to the processing elements it supplies - with zero inductance / resistance between the two.
shiraz,I used to use 4.7uf/10uf caps.but, I was "experimenting"
and I started trying different values with alligator clips that way I can compare the sound with different caps fast and easily.
I noticed the big difference With the 10,000uf per rail, I got stronger,tighter bass.Just try different values with alligator clips that way you do not have to solder and desoldering caps to test them.(make sure to discharge the caps,specially the big ones before touching them).and after this test if you do not like the sound with the big caps....just put the usual 4,7uf or 10uf caps.
in my case and test! ...I like the sound with 10,000uf per rail better.
and I started trying different values with alligator clips that way I can compare the sound with different caps fast and easily.
I noticed the big difference With the 10,000uf per rail, I got stronger,tighter bass.Just try different values with alligator clips that way you do not have to solder and desoldering caps to test them.(make sure to discharge the caps,specially the big ones before touching them).and after this test if you do not like the sound with the big caps....just put the usual 4,7uf or 10uf caps.
in my case and test! ...I like the sound with 10,000uf per rail better.
eeka chu: thanks for the explanation. That was very informative. So the placement of the caps of different sizes has got a lot to do with the chip getting the power it needs, when it needs it, while still balancing the tradeoffs of small caps vs big caps. So if you're always going to be running at low volumes, doesn't all this become relatively less important? I mean then it should be enough just to have a couple of relatively small caps near the chip.
lanchile07: I think I'll try to do some experiments then. Are you using standard 10.000uf caps or Panasonic TS? Does it make a difference? I haven't been able to find any Panasonic caps where I live.
lanchile07: I think I'll try to do some experiments then. Are you using standard 10.000uf caps or Panasonic TS? Does it make a difference? I haven't been able to find any Panasonic caps where I live.
Well,I am using Black gate 1,000uf 50v x 2 on each "chip board" and on the power supply I am using Elna lp5 10,000uf 63v per rail. I am also using .01uf 400v 5% radial caps on each diode (8) to eliminate some impurity...(These 0.01uF, "103") caps subtract rectifier noise, which reduces heat and radio interference).so far I am happy with the result. I try many caps and capacitance 4,7uf,10uf,220uf,1,000uf,1,200uf,1,200 x 2x4x6 but at the end I liked the sound with "ONE" 10,000uf per rail ,I also try 10,000uf x 2x4...I am keeping this much capacitance(10,000uf per rail) on the power supply!.as I said before...Just give it a try and if you do not like the sound with "juicy" power supply...put the usual 4,7uf/10uf.
PS: let me know if you like the sound with more capacitance (I do).
PS: let me know if you like the sound with more capacitance (I do).
Shiraz, I am using .01 uF 400V 5% .4" LS Radial Film Capacitors .you can fin them in many places.try e-bay here http://cgi.ebay.com/01uF-01-uF-400V...oryZ4662QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
and for the resistor (bleeder) I am not using them.they are for discharging the caps.maybe in the future I will put them but,for now I do not see the need to put them.
Make sure if you use big caps to be careful no to touch these "charged" caps,They can give you a real big punch.when you do the "test" you should be in a quiet room (no noise of any kind) and trust your ears no matter what.Get a CD that you know well , do not put attention to the words but to the tones....the lows,mid.highs, The bass should be strong and tight,the mid should be clear and effortless,the highs should be crispy but not fatiguing. You can trust your oscilloscope and instruments to test them but, the final and most important instrument that you have to test the sound is your ears. Trust your ears!.
good luck.
and for the resistor (bleeder) I am not using them.they are for discharging the caps.maybe in the future I will put them but,for now I do not see the need to put them.
Make sure if you use big caps to be careful no to touch these "charged" caps,They can give you a real big punch.when you do the "test" you should be in a quiet room (no noise of any kind) and trust your ears no matter what.Get a CD that you know well , do not put attention to the words but to the tones....the lows,mid.highs, The bass should be strong and tight,the mid should be clear and effortless,the highs should be crispy but not fatiguing. You can trust your oscilloscope and instruments to test them but, the final and most important instrument that you have to test the sound is your ears. Trust your ears!.
good luck.
My speakers are ok but my biggest problem right now is that I just moved into a new appartment, and there's almost no furniture, so obviously there's a lot of reverberation in my living room which makes listening tests really difficult.
I'm using Peter Daniels' lm4780 boards (configured for stereo operation instead of dual mono). So from 1 set I'm getting 2 amps. The first one I've made with the included components. Appart from the weak bass, I think it sounds really good. I'm going to make the second one with big caps, and I've decided to start with that I'll just follow peters recommendations (http://audiosector.com/lm4780 psu.pdf). With 2 amps, 1 with subberized psu and 1 without, I'm excited whether I'll be able to hear the difference.
Once I get the living room furnished I'm going to try different configurations with zobel, caps on the diodes, 3*4700uf in parallel per rail and small caps on the chip.
I'm using Peter Daniels' lm4780 boards (configured for stereo operation instead of dual mono). So from 1 set I'm getting 2 amps. The first one I've made with the included components. Appart from the weak bass, I think it sounds really good. I'm going to make the second one with big caps, and I've decided to start with that I'll just follow peters recommendations (http://audiosector.com/lm4780 psu.pdf). With 2 amps, 1 with subberized psu and 1 without, I'm excited whether I'll be able to hear the difference.
Once I get the living room furnished I'm going to try different configurations with zobel, caps on the diodes, 3*4700uf in parallel per rail and small caps on the chip.
I've used different kinds of approaches for building these things, and currently am running the following configurations:
1. 200VA+10,000uF/rail+LM4766, 6 ohm PSB speakers at work. Coupled with a little USB DAC, and uses relays on the output which may affect the output quality slightly, but unavoidable due to large turn-on thump by the DAC.
2. 220VA/ch+20,000uF/rail+LM3886 parallel, 4 ohm Dynaudio studio Monitors in my project studio
3. 440VA/ch+4,700uF/rail+LM4780 BPA, 8 ohm Sony in the living room
It's like making multiple dishes with rice, they all have it, but they're all a little different. The system with the 4700uF caps subjectively sounds the best to me, with the right mix of mids and lows.
That transformer is pretty heavily oversized for the kind of power output I use (90dB+ speakers), and the others use moderately-sized supplies with much heavier capacitance. The sound isn't as liquid as the main rig, but the speakers aren't of sufficiently high resolution either.
On rigs 1 and 2 I use 1208 ceramic X7R caps soldered directly to the chip pins for bypass, plus Panasonic FM 100uF at the supply pins, a centimeter or so away. Both are P2P (the 4780 is the Audiosector kit) and this helps me make changes quickly. Neither uses snubbers or zobels.
Of snubbers, I tried them in the 4780 amp, and I never tried them after that. They (I don't remember exactly though) did something nasty to the sound.
The next project is a 6-channels amp based on the LM4781 (I have the chips already). That's going to be a very complex P2P, but I'm going to do it anyway.
1. 200VA+10,000uF/rail+LM4766, 6 ohm PSB speakers at work. Coupled with a little USB DAC, and uses relays on the output which may affect the output quality slightly, but unavoidable due to large turn-on thump by the DAC.
2. 220VA/ch+20,000uF/rail+LM3886 parallel, 4 ohm Dynaudio studio Monitors in my project studio
3. 440VA/ch+4,700uF/rail+LM4780 BPA, 8 ohm Sony in the living room
It's like making multiple dishes with rice, they all have it, but they're all a little different. The system with the 4700uF caps subjectively sounds the best to me, with the right mix of mids and lows.
That transformer is pretty heavily oversized for the kind of power output I use (90dB+ speakers), and the others use moderately-sized supplies with much heavier capacitance. The sound isn't as liquid as the main rig, but the speakers aren't of sufficiently high resolution either.
On rigs 1 and 2 I use 1208 ceramic X7R caps soldered directly to the chip pins for bypass, plus Panasonic FM 100uF at the supply pins, a centimeter or so away. Both are P2P (the 4780 is the Audiosector kit) and this helps me make changes quickly. Neither uses snubbers or zobels.
Of snubbers, I tried them in the 4780 amp, and I never tried them after that. They (I don't remember exactly though) did something nasty to the sound.
The next project is a 6-channels amp based on the LM4781 (I have the chips already). That's going to be a very complex P2P, but I'm going to do it anyway.
Which brings me to my first question, which I think was never really answered:
What's the difference between having different sized (from big to small) capacitors down each rail, as opposed to a snubberized PSU?
snagram:
On the 4780 rig, are the 4700 uf caps the only caps you've got between the PSU and the chip? And what do you have on the PSU? 10uf?
What's the difference between having different sized (from big to small) capacitors down each rail, as opposed to a snubberized PSU?
snagram:
On the 4780 rig, are the 4700 uf caps the only caps you've got between the PSU and the chip? And what do you have on the PSU? 10uf?
Sorry should've been clearer. It's the Audiosector kit with 1500/0.1 on the amp board and 10uF on the supply board. The stock configuration was excellent, if slightly raspy in the midrange. 10,000uF (supply board only, the amp board is in stock configuration with 1500/0.1) killed the midrange. It didn't help much in the bass region either. Snubbers exaggerated the problem. I switched these to 4700uF and am now very, very happy. The amp board is still stock.
But my favorite colour is blue, yours may be red. So experimentation is the key to achieving your desired results. In general I see:
1. High-cap supplies improve bass heft at the expense of the midrange. My transformers are heavily oversized for my application and listening levels, so I never experienced a lack of bass, not even with the stock configuration. But in my other rigs, where the transformers are much weaker, caps help chips stand bass transients.
2. Snubbers are supposed to bring back some of the midrange magic into a high-cap supply, or some such thing. In my experience they made the midrange very grainy and raspy. I didn't like it at all. Anything else I say would be very subjective. I did experiment extensively over a month with values and locations, but never could find a happy medium.
3. It really, really helps to have sensitive speakers. Chips (unless stacked) can't take low-impedance loads well. And once you put an insensitive beast in front of a chipamp, you have to use large caps or the supply will buckle. A single LM3886 can't drive a 85dB speaker to reference level. Maybe to a normal listening level, but at (my) reference level SPiKE kicks in. Speakers are a big factor in your results, as is your taste and preferences.
But my favorite colour is blue, yours may be red. So experimentation is the key to achieving your desired results. In general I see:
1. High-cap supplies improve bass heft at the expense of the midrange. My transformers are heavily oversized for my application and listening levels, so I never experienced a lack of bass, not even with the stock configuration. But in my other rigs, where the transformers are much weaker, caps help chips stand bass transients.
2. Snubbers are supposed to bring back some of the midrange magic into a high-cap supply, or some such thing. In my experience they made the midrange very grainy and raspy. I didn't like it at all. Anything else I say would be very subjective. I did experiment extensively over a month with values and locations, but never could find a happy medium.
3. It really, really helps to have sensitive speakers. Chips (unless stacked) can't take low-impedance loads well. And once you put an insensitive beast in front of a chipamp, you have to use large caps or the supply will buckle. A single LM3886 can't drive a 85dB speaker to reference level. Maybe to a normal listening level, but at (my) reference level SPiKE kicks in. Speakers are a big factor in your results, as is your taste and preferences.
Can someone measure the ripple on each rail of a plain Big cap supply and the so-called snubberized supply? It would be great if we can not only hear the difference, but also see the differences.
To measure the ripple, access to a scope is a must. However, test tone can be generated by a PC sound card. A high power resistor as dummy load is also needed.
To measure the ripple, access to a scope is a must. However, test tone can be generated by a PC sound card. A high power resistor as dummy load is also needed.
sangram:
Thanks for this. This is very informative for a greenhorn like myself.
My speakers are supposed to be 91db/w, so I'm hoping they're ok (despite not being so expensive).
I was just wondering about the 1500/0.1: Where did you put the 0.1 uf cap? On the pins of the 1500uf cap? Because as far as I can see, on the audiosector kit, there aren't any dedicated pads for an extra 100 nf cap there.
Thanks for this. This is very informative for a greenhorn like myself.
My speakers are supposed to be 91db/w, so I'm hoping they're ok (despite not being so expensive).
I was just wondering about the 1500/0.1: Where did you put the 0.1 uf cap? On the pins of the 1500uf cap? Because as far as I can see, on the audiosector kit, there aren't any dedicated pads for an extra 100 nf cap there.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.