Looking for amazing numbers is an other kind of the audiophile disease.
At the beginning of digital (my first work with digital recorders in a studio) we made, of course, a lot of comparisons between the original sounds in the mixing desk and the output of our recorders. If any difference, it was subtle enough to not be an issue, at the opposite of the analog recorders we used in the same time as a security.
One some instruments, while the difference between original and analog copy was obvious (hiss, reduction of micro dynamic etc.) some instruments were sounding more agreeable or easier to work with (nice compression) out from analog tapes.
" That's only Rock'n roll. But I love-it. I love-it " ;-) (© R.N.Marsh)
Now, if some want to use digital for measurements, it is an other issue: he needs performance.
Last edited:
Hi KSTR, what do you use/suggest as a marker signal.
In my testing I am testing for relative changes in throughput according to changes to the loopback cable only, so reference/comparison to the original stimulus file is not required but is useful to measure absolute changes if latency compensation/time alignment issues can be sorted automatically and precisely.
Subtle changes is what I am attempting to measure, and as said in my above answer frequency and level response changes are automatically factored out....time stability and time alignments are the only issues in this case, I think.
Thanks and I have thought that through already and will be one of my strategies to average out timing drift/error.
I have run multiple serial recordings (12) and compared a bunch of permutations to get a feel for system intrinsic timing errors/noise/scatter etc over time, long term and short term.
An airflow proof box/enclosure may be necessary to eliminate thermal confounders or at least to increase precision and null depth.
Yes I am indeed finding that discriminating fine dynamic differences is not so easy.
I will do some mods to my usb interface and see what improvements I can make...or wish for an RME as you advise, or Prism or RTX6001 etc.
Thanks for the DeltaWave link, I will take a close look.
Dan.
Last edited:
If you have changed your 'posts per page' setting for diyaudio, then links to posts will be broken. The post I linked to was #102 in the thread, "[Modding] Topping DX3 Pro."
Dan,
There is no setting in Reaper that can compensate for jitter, noise, distortion, and or clock drift in your sound card. What ASIO does is bypass the Windows sound engine and give Reaper full control of what the sound card ASIO driver allows. All Reaper can do is send blocks to the driver to be played, and retrieve blocks from the driver as they become ready to read in from recording. Latency setting values Reaper uses for compensation are as reported by the ASIO driver.
There is no setting in Reaper that can compensate for jitter, noise, distortion, and or clock drift in your sound card. What ASIO does is bypass the Windows sound engine and give Reaper full control of what the sound card ASIO driver allows. All Reaper can do is send blocks to the driver to be played, and retrieve blocks from the driver as they become ready to read in from recording. Latency setting values Reaper uses for compensation are as reported by the ASIO driver.
Last edited:
I can put source file in track 1 of Reaper.
I then do a loopback recording to track 2 of Reaper.
Because of delays the newly recorded file is time offset (late) wrt the source file.
Edit: I found out how to do this compensation with custom trimmed values, I will try it tomorrow.
I then do a loopback recording to track 2 of Reaper.
Because of delays the newly recorded file is time offset (late) wrt the source file.
Edit: I found out how to do this compensation with custom trimmed values, I will try it tomorrow.
Last edited:
For one thing, it means the sound card's ASIO driver does not correctly report the latency.
I don't know if Reaper can do it or not, but seems to me Cubase had a function to time align an outboard analog signal processing chain. IIRC, it set out one or more test signals, measured the analog loop latency, then saved that value for future use in compensation. All it would do though it perform a linear offset of the ADC data coming back in. It couldn't compensate for real time clock drift without using one channel at all times dedicated to alignment test signals for compensation, which it does not do. You best bet if you have a clock that drifts a lot would be to do as KSTR suggests (including trying to acquire better hardware).
One sample set to 0.9 (normalized) will do. Typically I insert 100ms of silence and set the following sample sequence at 10ms, normalized values : 0.4, 0.5, -0.5, -0.4.
With the marker pulse, alignment (if required at all) is easy with a software like Adobe Audition (Version 3.0 was/is available as a free download). Don't know about Reaper, have limited experience.
Without block-averaging, clock and reference drifts effect on the diff in a null baseline test with a good ADC/DAC soundcard (RME class) is usually close or even below the analog noise floor, once the device has fully wamed up. Hum/buzz, popcorn noise and the occasional power line transient or ESD glitch is way more likely to stick out in the residual.
OTOH, with heavy block-averaging (10.000 or more blocks, 40dB noise reduction), which also needs a long time to record the initial response, clock and ref drifts can have significant impact on the diff. One way to mitigate this is the interleaving of (mutiple) A and B blocks, using a single record-while-playback process twice as long. This most often requires a hardware switching box to select A and B (like a cable being connected "forward" vs. "backward" in a directionality test) which must be controlled synchronously as well... this complicates things big time, you need to construct and build the hardware, blocks containing the switching event must be discarded, etc...
Last edited:
Several years ago, I tried a few sound cards and even did the suggested mods to lower noise pickup within. Eventually, I had to give up on them and go to better hardware at higher cost to get stable and accurate readings at low levels of distortion. No latency issues were involved in my app but because of the over-all poor results for my app, I would not use ''sound cards'' for anything but general purpose apps.
THx-RNMarsh
Last edited:
This previous Saturday, I attended a San Francisco Audio Society meeting that featured a guy who sets up hi fi systems for a living. And is he detailed and expensive! He takes literally days to get a first class system (like a WAMM) running its best. He freely showed us how to do it ourselves, if we had the time and patience to do what he does. He would take 1 speaker and position it carefully by listening, then raising it up and down for the best listening, then do a tilt: forward and side. He did not use much measurement equipment, except for a self-leveling laser line generator ($500 or so) to make sure that the speaker drivers were exactly the same height.
One important thing is that he did not recommend damping rooms too much. In one case, he removed $40,000 in sound absorbing material and in the end, put none of it back. He stated that the room has to work with the speakers, and that is what careful positioning of the speakers does. He also DeOxits every piece of electronics, inside and out, and then comes phono turntable set-up (probably an extra day), and so forth. He gets paid real bucks for this, and he flies all around the world to do system set-ups, next week Chicago, then Singapore, then Copenhagen. What a cool gig!
One important thing is that he did not recommend damping rooms too much. In one case, he removed $40,000 in sound absorbing material and in the end, put none of it back. He stated that the room has to work with the speakers, and that is what careful positioning of the speakers does. He also DeOxits every piece of electronics, inside and out, and then comes phono turntable set-up (probably an extra day), and so forth. He gets paid real bucks for this, and he flies all around the world to do system set-ups, next week Chicago, then Singapore, then Copenhagen. What a cool gig!
Last edited:
Richard, I agree with you about sound cards, but then we are pros and they are not. Just so they know the limitations of these accessories.
RME interfaces measure about as well as any high end gear, including the Benchmark. Seems you don't know the limitations?
I doubt you've even owned one, from your previous posts.
My Tek only goes to 200 MHz. And my Audio Assylum 401 only resolves down to about 1 ppm (if I dress all the cabling carefully). The noise floor (50 averages) on loop back only goes to about -140 dB.
We are rank amatuers and do not deserve the company and wisdom of our elders who frequent this thread.
We are rank amatuers and do not deserve the company and wisdom of our elders who frequent this thread.
I found the method in Reaper to manually set latency compensation and will try to set it today ...FYI Preferences>Recording>Use Audio Driver Reported Latency tick box with windows for manually inserting correct/custom values.
Dan.
I do have a sound card plug in on one of my computers, but then I got a Stanford Research FR-1 and have never looked back. I trust that Richard is more familiar with the potential problems with a plug-in.
Do-you try to make some jealous ?
...Who have not always been so well equipped ;-)
Last edited:
It's true, this where, you, Scott Wurcer, truly excel: AD797's in test instrumentation. I put them in my own equipment, long ago.
Since it was designed based upon a special request from a big customer named Teradyne, who builds and sells very high end test equipment, I am not shocked that it works extremely well in test equipment. I'm sure Teradyne's "must have" spec requirements, were expertly conceived.. I'm sure AD's delivery vehicle, the AD797, was also expertly conceived.
Is your 797 still the best OPA on the market in concern with distortion level in the audio range ?
- Status
- Not open for further replies.