ahh you are talking about what is called 'blitter noise' it is caused by your audio card either sharing the power supply (BUS) with a noisy card (probably a graphics card, or possibly network card; does it change when you move the mouse or play graphics? and quite soft when the screen is black?) or just the audio card being in very close proximity to the graphics card or other RFI/EMI noise source. so I am unsure if using the usb to power the unit will fix the problem in this case. do you have other compatible slots that will accept the audio card? something that is further away from the other cards?
if it is (which it sounds very much like it is); I have experienced this at one point with my RME, fixed by moving it to the other side of the case to another PCI slot.
if it is, batteries are meaningless too, as is anything that does not address the card being sprayed with noise through the air.
google it 'blitter noise' I think many apparent grounding issues or power supply issues are actually this lesser known problem
if it is (which it sounds very much like it is); I have experienced this at one point with my RME, fixed by moving it to the other side of the case to another PCI slot.
if it is, batteries are meaningless too, as is anything that does not address the card being sprayed with noise through the air.
google it 'blitter noise' I think many apparent grounding issues or power supply issues are actually this lesser known problem
Last edited:
Not convinced it's on your 12 Volt line. The card is also hooked to the PCI bus, so you could be getting noise down the control lines. In either case that's "Not right".
I have (more precisely had due to lightning causing mayhem) many PCI cards doing audio inside computers and inside an external chassis. None of them have issues remotely close to what you're describing.
I have had the problem in the past though (Different card to above). It happenned to a greater or lesser degree no matter which slot I used. My solution was to junk the computer (it was old and crusty anyway)and the same card then ran flawlessly in the new machine
I have (more precisely had due to lightning causing mayhem) many PCI cards doing audio inside computers and inside an external chassis. None of them have issues remotely close to what you're describing.
I have had the problem in the past though (Different card to above). It happenned to a greater or lesser degree no matter which slot I used. My solution was to junk the computer (it was old and crusty anyway)and the same card then ran flawlessly in the new machine
If Fairlight was happy enough with the situation to move from dedicated hardware to a PCI-e card in a general purpose machine then they must be out of their minds. Their systems start at $48k and go rapidly north from there. They have some of the best sounding software and converters on the planet and certainly could never be accused of being "LoFi"
Another (Albeit weaker) example are the Digidesign ProTools TDM cards. They're used to record most of what you've heard in the last 6-8 years, and large amounts of it for well over a decade.
There maybe questions about their ADDAs, their Mix bus and the ethics of the company but I've never heard anyone complain about unwanted digital noise in a system unless one of the DSPs was fragged.
My point being that if the cards are well designed and are installed in a well built computer then the effect described by Phofman doesn't happen. One last comment about the Fairlight/Digidesign stuff.
Neither company is dumb enough to have ADDAs inside the computer as that WOULD cause problems.
Last edited:
Diginerd, I don't know anything about these products to comment but I stand by my statement - why use a dirty PS to start with & then go to extreme lengths (& charge extreme money) just to clean it up? I understand the commercial logic but not the logic as far as sonics are concerned. BTW, we can do things in DIY that commercial units wouldn't consider so using these as examples is not that convincing
That is why I am saying I am pretty sure it is the amp being supplied by 12V fed from the PC. The amp has no stabilizer, unlike the card itself which has several of them. I do not think it is the card (the chain produced the same noise for several different soundcard models), but the amp power supply.
And I use it to show the PC power lines are extremely noisy, and feeding a USB sound card from an independent clean supply could bring benefits.
Come on, you either know what I am talking about, or you are just playing games with words. Hook your PC power lines to a scope and you will see on your own. Do not forget all decent sound cards use additional regulators, often several of them.
Actually, in my profession I work with communications equipment that often makes use of switching power supplies. One product in particular drives a very high precision 10 MHz reference oscillator with a switching supply. It is not only possible, but it is standard practice.
I think were' talking about the same thing, just from different angles. You're talking dirty power and the need for clean external power, I'm talking beter suppression (and possibly a better design overall) should solve the issue without the need for external power.
This problem has been solved many times over the years without needing to resort to external power. In some cases though if there is a deficiency it could be resolved by adding clean power, maybe the benefits outwiegh the cost of doing the mod (Better supression OR better bpower) + the cost of the device Vs simply getting a better device in the first place.
This is DIY audio though, so improving an existing device is the topic. The more general question is what's causing the issue and what's the best way to fix it.
The same issue with the Hi-Face could well be similar to the potential issue with your soundcard:- not having enough suppression.
Are all powersupplies equally noisy? If so then how do certain manufacturers not have issues and others do?
So far this discussion has drifted from "how to provide clean power to a device", through "is there a benefit to providing clean power to a device?" and now we're discussing "why is there benefit to providing clean power to a device" and "do we always need to provide clean power to a device?"
Without testing no one can say what is the root cause for sure, though previous comments by those more versed in this than me point to an explanation that ties both to this situation and to why your soundcard has issues and none of mine do (despite not having external power).
Nothing personal, and I'm not intending to cause offense.
This problem has been solved many times over the years without needing to resort to external power. In some cases though if there is a deficiency it could be resolved by adding clean power, maybe the benefits outwiegh the cost of doing the mod (Better supression OR better bpower) + the cost of the device Vs simply getting a better device in the first place.
This is DIY audio though, so improving an existing device is the topic. The more general question is what's causing the issue and what's the best way to fix it.
The same issue with the Hi-Face could well be similar to the potential issue with your soundcard:- not having enough suppression.
Are all powersupplies equally noisy? If so then how do certain manufacturers not have issues and others do?
So far this discussion has drifted from "how to provide clean power to a device", through "is there a benefit to providing clean power to a device?" and now we're discussing "why is there benefit to providing clean power to a device" and "do we always need to provide clean power to a device?"
Without testing no one can say what is the root cause for sure, though previous comments by those more versed in this than me point to an explanation that ties both to this situation and to why your soundcard has issues and none of mine do (despite not having external power).
Nothing personal, and I'm not intending to cause offense.
+12V and +5V lines from the same PC PSU, both feeding the same busy and power-hungry motherboard, look like a similar fruit to me
+12V and +5V lines from the same PC PSU, both feeding the same busy and power-hungry motherboard, look like a similar fruit to me
You original comparison between an analog amplifier section driven off the 12 volt line, and the hiFace situation, is not a valid comparison.
I'm outta here - goof luck phofman
O/P's question asked & answered!
Long ago, but we're still in a related and equally valid discussion. Sorry to see you leave.
So to summarise your argument that the mod is 'not necessary', you're saying its because digital has a high immunity to noise.
However this is a red herring - no-one is suggesting the bits are being changed here. Your claim therefore falls flat.
Any more claims?
Here is something of interest, from the manufacturer's site:
hiFace white paper
Toward the bottom, the manufacturer reports the sampling frequency oscillator, while running as intended (USB power and all!), to exhibit a precision of 2.5 parts per million. So less than 0.5 Hz error at 192 kHz sampling frequency.
Consider we are trying to reconstruct a 3 kHz tone - a tone that falls in a sensitive part of human hearing. Considering this worst case positive offset, the output would be 3000.0075 Hz... instead of 3000 Hz.
Assuming the power supply is entirely to blame for this frequency offset (which is incorrect) and we can 'fix' the oscillator to now have a precision of 0 ppm (which we can not), the resultant improvement would not be heard. That error is just too small to be detected by human hearing.
Again - the manufacturer is stating that when used as intended, the output jitter is 2.5 parts per million. It is simply a non-issue.
hiFace white paper
Toward the bottom, the manufacturer reports the sampling frequency oscillator, while running as intended (USB power and all!), to exhibit a precision of 2.5 parts per million. So less than 0.5 Hz error at 192 kHz sampling frequency.
Consider we are trying to reconstruct a 3 kHz tone - a tone that falls in a sensitive part of human hearing. Considering this worst case positive offset, the output would be 3000.0075 Hz... instead of 3000 Hz.
Assuming the power supply is entirely to blame for this frequency offset (which is incorrect) and we can 'fix' the oscillator to now have a precision of 0 ppm (which we can not), the resultant improvement would not be heard. That error is just too small to be detected by human hearing.
Again - the manufacturer is stating that when used as intended, the output jitter is 2.5 parts per million. It is simply a non-issue.
I have read them, but those pages were subsequent to your response to my request for your reasoning.
So now it seems you're claiming the mod is unnecessary based on different arguments from the ones you gave at first. Which is fair enough, but I was only examining the argument you gave at that stage, which was an obviously flawed one. So why offer the completely broken argument when you already had a better one?
No, that's not what the manufacturer is saying here, you have misread it. I'm not blaming you for this btw as their English expression does have considerable room for improvement
This 2.5ppm is the longer term accuracy (they say precision) of the frequency setting, not the jitter.Here are some of their words:
We think we’re listening to the right sampling frequency, but actually we listen to a different sampling frequency and all instruments are “out of tune”.
This by the way demonstrates a complete misunderstanding of what 'out of tune' really means. Even a person with absolute musical pitch wouldn't pick up on a 2.5ppm error in a note's pitch. They're also wrong about PLLs having a frequency error, because PLLs lock to the incoming frequency, their long term accuracy is totally determined by what the long term accuracy of the incoming reference is. If not, then the PLL isn't locked. They are right about PLLs having poorer short-term accuracy though (they're calling it phase noise) as against crystals, but that's another matter.
So to sum up, jitter is concerned with the short term accuracy and their 2.5ppm figure is about the longer term accuracy. Even ignoring this important fact, this is another of your red herrings. You already correctly note that changing the power supply won't change this accuracy.
So you continue to claim. But its just a straw man.
You omitted remarking on the details about the phase noise of the oscillator. Do you know anything about how an oscillator's phase noise varies with the noise on its power supply? Do you have any data on how much phase noise (and at what frequency offsets) might be audible?
Last edited:
- Status
- Not open for further replies.