HI,
I am looking for a high quality soundcard that is capable of input sample rates of up to 192kHz. It can have a PCI or PCIe interface to the PC motherboard.
However I do NOT want any fancy input effects (or I must be able to disable them).
I want to use it with PC based oscilloscope & audio spectrum analyser software, so any 'compression' or 'equalisation' effects are an absolute no-no!
I know there are all sorts of audio output effects on modern cards (5.1 & 7.1 etc) but am not sure whether all current soundcards mess with the stereo input?
Does anyone have any experience or suggestions please?
Thanks in anticipation
Dave
I am looking for a high quality soundcard that is capable of input sample rates of up to 192kHz. It can have a PCI or PCIe interface to the PC motherboard.
However I do NOT want any fancy input effects (or I must be able to disable them).
I want to use it with PC based oscilloscope & audio spectrum analyser software, so any 'compression' or 'equalisation' effects are an absolute no-no!
I know there are all sorts of audio output effects on modern cards (5.1 & 7.1 etc) but am not sure whether all current soundcards mess with the stereo input?
Does anyone have any experience or suggestions please?
Thanks in anticipation
Dave
Asus xonar D2X seems like a possibility but prettymsire that you will be using WDM drivers and that would kill performance.
In one application I tried to use it as a subwoofer DAC fed from the digital input. There was a 100ms delay introduced by the driver. I eventually gave up. Figured these cards are optimised for output, not input. I use an old e-mu now and though it is limited to 96k, signal analysis is decent and max latency of 9ms. Noise floor is below -120db, which is sufficient for testing amplifiers.
In one application I tried to use it as a subwoofer DAC fed from the digital input. There was a 100ms delay introduced by the driver. I eventually gave up. Figured these cards are optimised for output, not input. I use an old e-mu now and though it is limited to 96k, signal analysis is decent and max latency of 9ms. Noise floor is below -120db, which is sufficient for testing amplifiers.
Is there any noise or distortion requirement that motivates you to look for a good sound card instead of onboard sound? What about input levels and bandwidth?
Most sound cards shouldn't mess with inputs, or even outputs. I would just avoid Creative just to be safe.
Most sound cards shouldn't mess with inputs, or even outputs. I would just avoid Creative just to be safe.
Balanced inputs.
I use E-MU 0202 USB. It works very well for measurements (it sound very bad though). The ADCs are good. It has a few drawbacks :
- It is not UAC2 standard compliant, so it does not work on linux, so I have to use stupid crash-prone windows drivers and fight with windows sound configuration which is a mess. I will never ever buy any other hardware that is not UAC2.
- For instrumentation use, the volume pots would really need to be replaced with rotary switches to get repeatable calibration between volts and sample values. I will probably do this modification.
- The output has lots of switching converter residue and other crap. It has lots of drive current though.
- There is a bit of crosstalk. Muuuuuch less than motherboard audio stereo jack, of course, where both channels share the same GND contact and are unbalanced...
- The preamps are not particularly low noise, but they're ok for general measurements.
For PCI, look toward one of the M-Audio Audiophile cards. They can be had used for great prices. A very, very good soundcard.
Jan suggested the Focusrite, which is the big time favorite among the live sound guys. It's USB, if that doesn't bother you. Not sure if it does 192K. I use a USB M-Audio Fast Track Pro, but would buy a Focusrite without hesitation.
Jan suggested the Focusrite, which is the big time favorite among the live sound guys. It's USB, if that doesn't bother you. Not sure if it does 192K. I use a USB M-Audio Fast Track Pro, but would buy a Focusrite without hesitation.
HI
Thanks everyone so far for your thoughts.
The reason I can't use the on-board sound chipset is it doesn't sample fast enough.
The Spectrum analyser/oscilloscope software I have will allow me to use sample rates of 192kHz, but the on board audio ADC's only do 44/48kHz which is a bit low for full audio spectra.
It's not what the spec for soundcards say it's what they leave out is what concerns me. I am not interested in full surround sound, home theatre multi channel outputs as I only want to input.
I need great S/N ratio, low distortion, low latency but most of all I don't want any added effects, compression, equalization, DOLBY or whatever. I just need the soundcard to do pure unadulterated A/D conversion straight into the PC.
Not much to ask?
Dave
Thanks everyone so far for your thoughts.
The reason I can't use the on-board sound chipset is it doesn't sample fast enough.
The Spectrum analyser/oscilloscope software I have will allow me to use sample rates of 192kHz, but the on board audio ADC's only do 44/48kHz which is a bit low for full audio spectra.
It's not what the spec for soundcards say it's what they leave out is what concerns me. I am not interested in full surround sound, home theatre multi channel outputs as I only want to input.
I need great S/N ratio, low distortion, low latency but most of all I don't want any added effects, compression, equalization, DOLBY or whatever. I just need the soundcard to do pure unadulterated A/D conversion straight into the PC.
Not much to ask?
Dave
What are you going to use to condition the input signal to the Sound Card for Oscilloscope duty ?
You don't need low latency for measurements.
But if you use windows, you really need an ASIO compatible soundcard. ASIO drivers make a direct path from your software to the soundcard and bypass all the crap like windows mixer, resampling, etc.
For example on my E-Mu, it is impossible to do very simple things, like setting the recording levels, without ASIO. When using the standard windows drivers, recording level is controlled by many different software knobs, distributed around several control panels, any of which may have been set to a random value by any software that is running...
I can have one knob set to "recording level 50%" and another to "microphone gain +30" (note the absence of units) and lose a bunch of LSBs. Or just moving the slider by 1 pixel changes the level by 10dB because of some obscure setting somewhere. Or the windows resampler would introduce strange artifacts in the signal. IT DOES NOT WORK.
With ASIO there is one level knob, it's the pot on the USB soundcard enclosure. And the clipping LEDs really correspond to clipping. That's a lot more usable.
ASIO is another reason to get a "pro" soundcard (besides balanced inputs)...
I'll say again. M-Audio Audiophile. I have seen it used many times as a very good measurement input. You really need look no further.
If you need to get past the 196K sampling rate, look at Picoscope. You won't get the bit depth of an audio card, but you get more bandwidth.
If you need to get past the 196K sampling rate, look at Picoscope. You won't get the bit depth of an audio card, but you get more bandwidth.
I have 2 or 3 M-Audio "Audiophile 2496" cards and an M-Audio Audiophile 24 - 192. I have used all of them for audio amp measurements. They are just converters, buffer amps, and a PCI buss interface, no DSP. A somewhat old, dated design, but quite effective. Note PCI, not PCIe buss. Many new motherboards do not support PCI anymore.
The loopback (outputs connected directly to inputs) frequency response goes to about 44 KHz on the '96 cards, and to about 90 KHz on the '192. Audio I/O are not balanced so the noise floor depends on the PC. I had the best results with a simple motherboard that supports overclocking, but underclocking it instead. Use on board motherboard video if possible, video cards are noisy. Try the audio card in each buss slot and test the noise floor. There can be 10 db difference between the slots. Usually the slot furthest from the CPU and video card is the quietest. 100 db of dynamic range is possible.....far better than most amps.
They happily sold me a $400+ interface, the Emu 1820M then dropped support for it a year later....do you think I will give Creative any more money EVER? It was a kick a$$ home recording studio system for XP though, even had a phono preamp.
I got it for home recording to replace an old M-audio 1010LT. Emu never released any drivers for the 1820 when Vista came out, and dropped support for all PCI systems. I sold the 1820 fro $100 and put the 1010 back in. It still lives on in my W7 system. The 1010 is basically a 10 channel 2496.
If dynamic range is the ultimate criteria, an external USB or firewire interface will have a lower noise floor since it doesn't live inside the PC.
Hi George, E-Mu 1212M and 0404 are also not officially supported but there are drivers for all versions of Windows up to Win 7. I use both of them and they work fine, DSP and everything.
It's a pity you sold the interface, there was a lot of pressure on them and they did relent.
It's a pity you sold the interface, there was a lot of pressure on them and they did relent.
Thanks George for the follow-up on Audiophile card. It is an oldie, bad a goodie. I have notice the move to PCIe slots, too. I've seen some good measurement work done with this card, it's nice to know it's limitations.
I have M-Audio Audiophile 24192 and 2496, both are great cards for measurement applications and the 2496 is quite decent for simple recording applications as well.
M-Audio Profire 610 24/192
Really flaky firewire operation, sometimes have to reinstall driver when you don't power up or down in a very specific procedure.
But the hardware's resolution and accuracy is very rewarding for measurements.
Differential in/out highly recommended for certain measurements.
The waveform output distortion is also very low, so you can typically avoid having to invest in additional separate oscillators (but I do have one of Victor's).
Really flaky firewire operation, sometimes have to reinstall driver when you don't power up or down in a very specific procedure.
But the hardware's resolution and accuracy is very rewarding for measurements.
Differential in/out highly recommended for certain measurements.
The waveform output distortion is also very low, so you can typically avoid having to invest in additional separate oscillators (but I do have one of Victor's).
The best performance wise IMHO, but as always not the cheapest, are the Lynx sound card. Now also having PCIe brands.
Hp
HI,
I bought also an used E-MU 0202 for measuring purpose from ebay. Looks working but I'm not really satisfied with high frequency noise at output. Is it normal to be more than 100mV PP ?
See the attached picture from oscilloscope screen, channel 1 (top) is connected to 0202 out and channel2 (bottom)to a SigGen. The scope was triggered by channel2 and moved the frequency on SigGen until I see some stability on chanel1. Here you can see is a huge noise (more than 100mV PP) centered to 138khz.
Anyone seeing the same or is my DAC faulty?
I bought also an used E-MU 0202 for measuring purpose from ebay. Looks working but I'm not really satisfied with high frequency noise at output. Is it normal to be more than 100mV PP ?
See the attached picture from oscilloscope screen, channel 1 (top) is connected to 0202 out and channel2 (bottom)to a SigGen. The scope was triggered by channel2 and moved the frequency on SigGen until I see some stability on chanel1. Here you can see is a huge noise (more than 100mV PP) centered to 138khz.
Anyone seeing the same or is my DAC faulty?
That looks about right for a bus-powered unit like the 0202. All the power is being generated off the already poorly regulated 5V line of your laptop/PC.
Unfortunately the 0202 was the weakest link in the entire line of interfaces built by Creative back in the day.
Of course, for speaker measurements it should not be an issue, but electronics is a slight problem.
You can use FFT math available in software to remove the noise, RMAA for example does this automatically. You will need to run one loopback and save it as a result, then do the run with the DUT connected. In the results window, you can subtract the original result from the new one. This should give you a more accurate idea of what the DUT is like.
Unfortunately the 0202 was the weakest link in the entire line of interfaces built by Creative back in the day.
Of course, for speaker measurements it should not be an issue, but electronics is a slight problem.
You can use FFT math available in software to remove the noise, RMAA for example does this automatically. You will need to run one loopback and save it as a result, then do the run with the DUT connected. In the results window, you can subtract the original result from the new one. This should give you a more accurate idea of what the DUT is like.
- Status
- Not open for further replies.