[ROBSCIX]

Man, not wikki...such a bad source of information. This can go back and fourth...AFAIK, class "T" refer to class D amplifier using tripath chips. It is not a seperate actual amplifier classification. When you say class "T" some will know what you mean and others will not.
Anyway, call it what you want.
They are great amplifiers...by either name

 

[thoppa]

haha...u love to argue eh ? I remember you wanting a pm argument about whether a 32-bit dsp chip is really 32 bit, even if it says so in the datasheet.

Sorry, but I'm not taking the bait. Talk to the patent office if you still think they're the same.

Anyway, back on topic, here's a blurry photo of the finished Onkyo. 6x LME49720HA and Blackgate NX 47uF 6.3V coupling caps, bypassed with Vishay MKP1837 10nF on the other side.

And another giving a very rough description of the signal path. I've got the yellow on some of the wrong pins but you get the idea...

 

[ROBSCIX]

As I said, call it what you want. That is hardly starting an arguement.

 

[ROBSCIX]

On topic, that card has a very different opamp layout compared to others. I always liked the "faraday" sheilding on that card, but that has been seen on others.

 

[headfinoob]

Does anyone have any experience or knowledge of the POWER-ONE HE12-10.2-A linear power supply? I was considering using it to power the Intel D945GSEJT motherboard. I am assuming that its ripple characteristics are very good, however, I found a PC PSU, the Enermax Liberty Eco 400W, with only 6.5 mV AC ripple on the 12V2 rail for about half the cost.
AnandTech: 300W to 450W: 20 Power Supplies on the Test Bench

Thoppa,
I didn't want to harass you too much, but did you ever get a scope reading on the Intel board with your Sigma 11 power supply? I am curious as the the DC-DC curcuitry quality on the Intel board. The board is still in the mail (as is the Onkyo sound card) and I'm just contemplating my choice of PSU. I picked up a 1F car audio electrolytic capacitor which I might end up not using. Being such large capacitance, I believe it would do well to filter low frequency AC ripple (60Hz, 120Hz, etc.)

In all reality, I expect that choice of power supply for the Intel D945GSEJT/Onkyo SE200-PCI LTD combo won't be of extreme importance because the DC-DC converters in the motherboard and the power filtering of the sound card itself should lower the noise floor to inaudible levels. But I suppose one can't be sure without measuring (I have no access to a scope.)

As always, and info is appreciated. When I get my system pieced together, I'll try to get some pics.

-noob

 

[thoppa]

Hi,

That's a huge cap. I doubt you'll get any ripple after that but there'll still be some noise, and the cap will dominate the ability to supply current, (it'll swallow anything and everything from the psu - a bit like a river feeding a dam; the dam has complete control). So you might want to bypass/parallel it with something that can react faster like a 10uF Solen MKP or something similar.

I just got the ssd yesterday - 32Gb Adata sata - so now I'm ready to finish the build and do some testing. It's all working but not fully assembled etc.

I haven't yet tested the voltages on the board for noise and ripple but I'll give that a go this week. I have a cheapie 50W 12 4A smps psu to compare it against.

What case did you choose ?

Tom

EDIT : Have you checked that the psu's can handle such a large cap ? There's going to be quite a large amount of current drawn as the cap charges up when the power is switched on.

 

[headfinoob]

Hi,
I chose the Apex MJ-16 mico-ATX case. It is about 5.25" wide, 13" high, and 15.75" deep. It is pretty cheap, about $50. It isn't too bad for the price, and comes with a crappy ATX SFX PSU. I was tempted to use it, but I might throw it in an old comp to revive it.

Yeah, it is a pretty huge cap (not as big as my hybrid 50F cap I have hooked up the Audison amplifier.) In order to connect these caps, they have to be charged before switching on the PC PSU. If I don't charge the cap to close to 12 V, then every ATX PSU I have tried will go into over-current protection or under-voltage protection- I am not sure which.

When I had my stereo amp as well as my class-D mono subwoofer 400W amp connected to a car sub in a sealed enclosure, the bass hits would often cause the PSU to go into protection. Adding the cap greatly reduced that problem. It was a PITA, because every time it happened, I would have to power off the PSU, discharge the cap, wait for the protection to go off, charge the cap, and power on the PSU again.

I solved the issue in general by just getting the Corsair HX1000W PSU, which is nearly identical to the Thermaltake Toughpower 1500W PSU. I think the only reason its rated 1000W is to maintain its 80 plus certification. It has respectable AC ripple measurements (about 25mV on the 12V rails.) It is actually two power supplies in one, and I just combined the rails by soldering all the leads on the modular power cables together onto 4 AWG power cable.

I think I will swap the 50F hybrid cap for the 1F electrolytic cap for the amp and just go ahead and get a 10uF Solen MKP for the PSU like you suggested. I'm very illiterate when it comes to small caps, so I wouldn't know which ones to get. Speaking of caps, check out this link to the modded SE90-PCI, I'm impressed.
Jimmy’s Junkyard » Blog Archive » Onkyo SE-90PCI Mod (Finally, It’s Done!)

I am tempted by a Sigma 11, but I don't have the skill level or patience at this point to get a kit. I just want an easy, cheap, and quality PSU. That is why I'm debating a Power-One linear supply for about $150.00.
I am amazed that a computer power supply (the Enermax Liberty Eco 400W) reviewed from a reputable site like Anandtech show measurements of 6.5mv on one of the 12V rails. I will probably pick this up, and I can't fathom getting a better SMPS for the money (I've been known to be wrong.)

-noob

 

[thoppa]

Hi,

Well, I thought it might suck too much juice. I have played about with different caps on smps outputs to smooth away the ripple, using a scope. I found that, as a rule of thumb, about 2000uF to 3000uF per amp is enough to remove most ripple on smps rails.

The Sigma11 is one of the easiest projects to build - you just need a high-heat soldering iron because the earth planes are so large that it can take quite a while for the solder to flow.

I just hooked up the Intel board again to test noise. With the smps, it is about 35mV noise and 5mV ripple unloaded and this increases to about 60mV and 10mV under load, but the noise is all over the place and would be virtually impossible to eliminate.

The sigma11 has no measurable noise unloaded using a fluke 17b dmm (0.000V AC) and the scope, so it's less than 0.5mV. However, I just found that although it works with the board, the voltage across the smoothing caps is 14.3V (needs to be 15v or higher really) and the output voltage has dropped to 11.3V and there is 15mV noise. This is clearly out of regulation (it should stay at 12.00V) so I need a higher voltage toroid. Dang. I checked it and found it is actually 0-12V 50VA, so I'm gonna replace it with a 0-15V 60VA or higher. I have used 0-12V before to create 12V but never for this much current or with a sigma11. Live and learn eh ?

What will interest you is that the noise on the psu output is exactly the same on the scope as the noise on the 12v line to the ssd - which has obviously gone onto and around the mobo - so it looks like this board isn't adding much noise at all. However, to know for sure I'll need that bigger transformer and try again.

I also looked at the noise on the 5v line from a usb port and that had 0.000V AC, i.e. below the level my dmm can measure, and was 5.10V.

So I think this might be a very 'clean' board. EDIT : Unfortunately it's not. Quite a bit of switching noise - see below.

EDIT : I just pulled a 50VA 0-15V toroid trans from another project and hooked it up. Works beautifully - voltage stays in regulation and there is 0.000V AC noise/ripple on the psu output under load, and 1.5mV on the 12V into the ssd. There's no detectable noise on the 5V line either.

And best of all, the sound is very well-defined and the stereo imaging is excellent. I'm very pleased ! I just need to box it all together and I'm done.

 

[ROBSCIX]

Craziness! ...but how does it sound?

 

[headfinoob]

Thoppa,
Thanks very much for the measurements, it is good to know that the Intel board isn't noisy.

The largest Solen MKP capacitor that I found on their website was 330μF for about $97- maybe I'm not looking hard enough.

At zero load, I have measured 0.000V AC with my Fluke 77 on several ATX PSUs. I actually read 0.000V on all of them if I disconnect the cooling fans. Even the ATX SFX PSU that was included with my Micro-ATX case also reads 0.000V. With the fans connected, they read about 0.003V AC. The Scythe fan I had connected to one of the PSUs would show 0.022V when it was connected, and 0.000V when disconnected.

I am curious what they measure at 5-10% load, as that would approximate the current draw from Intel system.

My main PC running heavily overclocked and overvolted (there are also six 120mm fans) measures 0.006V AC when idle. When I run ATI Tool to stress the video card, I get about 0.025V (fluctuates) and the the ammeter on the Furman PM-8 Series II line conditioner shows a whole 1.3A greater. So my system uses over 150W more running ATI tool than idle which adds about 0.020V AC to one of the 12V rails. I estimate that the PSU is running at 70% load with ATI Tool running.

Correct me if I'm wrong, but I thought that a DMM is mostly worthless for AC ripple measurements other than AC ripple at lower frequency. I also don't know the difference between noise and ripple necessarily. I've seen the terms used synonymously and wasn't sure how to differentiate between the terms. I thought it may be how sinusoidal the waveform is.

It looks like your case was specifically built for that toroid! Are you using the ATX power connector and that cable for the final build?

-noob

 

[thoppa]

Hi,

The sound is very good - typical LME49720HA (wide soundstage, amazing details, deep tight bass, clear vocals without being cold). The card has 10ppm xo's with low phase noise and a Wolfson WM8740 DAC - smooth, clear, revealing. I'm running the latest Foobar with kernel streaming to the card. I find that the standout character is the clarity throughout the frequency range - instruments sound very real and are clearly placed. It's very enjoyable !

What's weird is the lack of noise - I keep expecting to hear a HDD spin up or something, but there's nothing. If it wasn't for the Sigma11 led, and the led on the mobo, I wouldn't be able to tell if it was on or off (except, of course, for the music !)

I looked at the noise on the scope and there is clearly some switching noise pollution from the board - shame. It's adding lots of noise spikes up to 25mV in size and generally the dense wideband noise is around 3-5mV on the scope, hence the dmm's rms (?) reading of 1.5mV. I can't see any ripple. I hooked up the smps instead and looked at that - almost 100mV noise spikes and noticeable ripple too. So the sigma11 isn't a complete cure but it certainly helps keep the electrical noise level down.

The sigma11 is now giving out 12.13V and the mosfets are running at 37.2 degrees. It seems my heatsinks are a bit of overkill...but it means it'll run forever and the PC won't get too hot either, so that's cool (sic). Plus there's plenty of juice in reserve if I want to add usb stuff and/or another HDD/dvd-rom. I think I've got up to another 1 amp and that's well within the safe limits for everything. Toroid is 3.4A x 0.8 = 2.7A. Peak use on the board is 1.6A during boot in it's present state, and it uses 1.4A steady state.

The processor is running at less than 5% and the mobo heatsinks are barely warm to the touch.

I've hooked up a cheap XP remote from Ebay - very easy to do with Foobar's global hotkeys.

The music is streaming from a Buffalo Livestation connected to a wifi n router. That's getting about 120Mbps and the streaming is about 1% of the bandwidth. No issues there. I can play dvds from the livestation too so I don't need a noisy dvd drive unless I want to watch one I haven't ripped yet. I might look around for an external case one with it's own psu....and replace it with a silent one ! I wonder if this baby will handle blu-ray....

All in all, I'm really pleased with this.

 

[thoppa]

Quote:

Originally Posted by headfinoob /img/forum/go_quote.gif Thoppa, Thanks very much for the measurements, it is good to know that the Intel board isn't noisy. The largest Solen MKP capacitor that I found on their website was 330μF for about $97- maybe I'm not looking hard enough. -noob

If you want to smooth any ripple for this board, four Nichicon HD or Panasonic FM 2000F 25V caps together with four Solen MKP caps - but get the Solens in decades = 100uF, 10uF, 1uF, 0.1uF. This will help with filtering noise at different frequencies too. Add a couple of Sanyo Oscon 47uF 16V because these have very low resistance at high frequencies. IMO, This'll be as good as you can do without an active filter circuit. If you want to passive filter too, then get a toroid transformer and just use one side of it as a choke. It'll pass the DC but filter the AC. Get one that can handle the current ! You could even make a pi-style filter with two toroids and a cap in the middle but I think this'll be overkill.


Quote:

Originally Posted by headfinoob /img/forum/go_quote.gif At zero load, I have measured 0.000V AC with my Fluke 77 on several ATX PSUs. I actually read 0.000V on all of them if I disconnect the cooling fans. Even the ATX SFX PSU that was included with my Micro-ATX case also reads 0.000V. With the fans connected, they read about 0.003V AC. The Scythe fan I had connected to one of the PSUs would show 0.022V when it was connected, and 0.000V when disconnected. I am curious what they measure at 5-10% load, as that would approximate the current draw from Intel system. My main PC running heavily overclocked and overvolted (there are also six 120mm fans) measures 0.006V AC when idle. When I run ATI Tool to stress the video card, I get about 0.025V (fluctuates) and the the ammeter on the Furman PM-8 Series II line conditioner shows a whole 1.3A greater. So my system uses over 150W more running ATI tool than idle which adds about 0.020V AC to one of the 12V rails. I estimate that the PSU is running at 70% load with ATI Tool running. Correct me if I'm wrong, but I thought that a DMM is mostly worthless for AC ripple measurements other than AC ripple at lower frequency. I also don't know the difference between noise and ripple necessarily. I've seen the terms used synonymously and wasn't sure how to differentiate between the terms. I thought it may be how sinusoidal the waveform is. -noob

Yep, about DMMs, that's what I understand too but I don't know cos I've never actually confirmed this. I think they also give rms readings so that's why a scope is sooo useful.

As I understand it, ripple is a repeated waveform and noise is spikes riding on the waveform.

Quote:

Originally Posted by headfinoob /img/forum/go_quote.gif It looks like your case was specifically built for that toroid! Are you using the ATX power connector and that cable for the final build? -noob

I had to cut the case to fit the toroid - the top tray is designed for a HDD - so I cut and bent the metal to accommodate the toroid so it would be as far from the soundcard as possible. It's very nice of you to say it looks like it was specifically designed - actually, there are quite a few rough edges and overcuts....I'm not so good with power tools ! I was concerned to make sure I don't create a shorted turn - so the toroid top must never touch the case. It's recessed enough to do this.

I used some speaker cable to make an ATX connector into the board. It should be a twisted pair so I'll separate and twist it before the final installation. I got an atx power extender cable from a local shop and cut it to get the atx connector.

I bet you can't wait to start your build too eh ? I can tell you love doing this kind of stuff too, and like me, learn as you go along. How many electrical shocks have you had over the years ? I'm so bad at remembering to disconnect stuff...I even gave myself one yesterday....

 

[headfinoob]

hmm... its interesting how the noise spikes go from 35mV unloaded to 100mV when using the SMPS and measured through the loaded board.
Assuming your Sigma 11 has virutally no noise spikes, 25mV doesn't seem too bad for the board to be adding. I'm curious how much can be attributed to the wireless card. I went with ethernet because I didn't want to deal with added RFI/EMI from the antenna as well as electrical noise and load on the PCI Express bus. Who knows, the ethernet controller probably is no saint either.
I'm glad that your happy with the final product, I'm going to be totally stoked because this is going to be superior to anything else I've owned. Think about the possiblities with a high-quality computer for a source: remote access, remote controls, wireless keyboards and mice, user restrictions, software equalization, application flexibility (if only I can get kernel streaming to work with a DVD player program), etc.
Thanks again for all the information and help.
-noob
P.S. I'm pretty Blu-ray performance will not be acceptable. The CPU is pretty slow (my old Athlon XP 3200+ could barely play 720p without dropping frames.)

 

[thoppa]

The smps psu is pretty poor tbh - it cost US$15 from a local shop and is designed for flat-screen displays etc. Any atx psu would be better than this, but it is just to help me set things up and give a relative measure.

Although I'm pleased with the Sigma11, the board is obviously using switching power systems, so its advantages are being limited. It's impossible to say how much, but I am content I have done the best I can, and it sounds very good which is what really matters.

Yeah, I am concerned about wifi noise too, but overall, it's the best solution as I want to send the files to different rooms. It's way up in the Ghz too so I don't expect it to be a problem, just an imperfection. I think with a cable, there is a signal loss, so neither are ideal, but if the cable is short, then ethernet does seem better.

Yeah, blu-ray does seem to be a dream too far....hehe

I'm looking forward to seeing your build too !

 

[ROBSCIX]

Quote:

Originally Posted by thoppa /img/forum/go_quote.gif Hi, The sound is very good - typical LME49720HA (wide soundstage, amazing details, deep tight bass, clear vocals without being cold). The card has 10ppm xo's with low phase noise and a Wolfson WM8740 DAC - smooth, clear, revealing. I'm running the latest Foobar with kernel streaming to the card. I find that the standout character is the clarity throughout the frequency range - instruments sound very real and are clearly placed. It's very enjoyable ! What's weird is the lack of noise - I keep expecting to hear a HDD spin up or something, but there's nothing. If it wasn't for the Sigma11 led, and the led on the mobo, I wouldn't be able to tell if it was on or off (except, of course, for the music !) I looked at the noise on the scope and there is clearly some switching noise pollution from the board - shame. It's adding lots of noise spikes up to 25mV in size and generally the dense wideband noise is around 3-5mV on the scope, hence the dmm's rms (?) reading of 1.5mV. I can't see any ripple. I hooked up the smps instead and looked at that - almost 100mV noise spikes and noticeable ripple too. So the sigma11 isn't a complete cure but it certainly helps keep the electrical noise level down. The sigma11 is now giving out 12.13V and the mosfets are running at 37.2 degrees. It seems my heatsinks are a bit of overkill...but it means it'll run forever and the PC won't get too hot either, so that's cool (sic). Plus there's plenty of juice in reserve if I want to add usb stuff and/or another HDD/dvd-rom. I think I've got up to another 1 amp and that's well within the safe limits for everything. Toroid is 3.4A x 0.8 = 2.7A. Peak use on the board is 1.6A during boot in it's present state, and it uses 1.4A steady state. I've hooked up a cheap XP remote from Ebay - very easy to do with Foobar's global hotkeys. The music is streaming from a Buffalo Livestation connected to a wifi n router. That's getting about 120Mbps and the streaming is about 1% of the bandwidth. No issues there. I can play dvds from the livestation too so I don't need a noisy dvd drive unless I want to watch one I haven't ripped yet. I might look around for an external case one with it's own psu....and replace it with a silent one ! I wonder if this baby will handle blu-ray.... All in all, I'm really pleased with this.

Can't you build a filtering network to remove the last of the noise? -I know you can only get it so clean and if it is not a problem, why worry?
I have a similar setup for Foobar with a remote etc..
Nice job, enjoy.