Dayton OmniMic Precision Measurement System - Page 50

goneten,

Not rude at all.

I pointed to that link since it was only 3 pages (I believe). Basically it started with a difference between two mic's and then into the resolution provided by the software.

I believe Bosso's main complaint was Omni did not proved enough data points to accurately graph the information (when none, 1-6th or below smoothing was used). I was trying to discuss with him if running a trace at 1-24th does show the information as would be needed to cover enough data points but do not believe he responded to this. Maybe he will see this and correct me if I am wrong.

On your most basic question on if Omni will work for your needs I guess I could only say you would be best to read the manual and see if it would.

James

I basically require very high resolution frequency response measurements. Unsmoothed, it must show warts and all. If REW is better at showing it as it is then I'll need to look elsewhere. It's nice to have a graph drawn on screen but if the information depicted differs from reality then it's useless. In a nutshell I need to know that what I'm seeing is accurate.

goneten,

I hope I not going off the deep end here with you, just trying my best to show you information.

Below is a screen shot from Omni with the black trace taken with Omni and the smoothing was set to none. The red trace is from REW taken with no smoothing (this REW frd file was provided to me by someone else for my testing purposes). I just loaded them together to show how they are shown. I stopped the window at 100HZ since that is as far as I had measured that day in testing.

I have also attached the frds that were generated by both programs. I hope this helps in some way.

James

 

copy of rew no smoothing.txt 11.5341796875k . file

 

_5_none.txt 1.0791015625k . file

What is the attached graph supposed to show me ? This isn't a comparison in the same room ? Surely ?

Goneten,

What I was attempting to show is how the data is presented from both Omni and REW when measurements are taken with no smoothing but I think the more important part for you would be the information contained with the frd files that were generated by the programs taking those measurements.

No these are not same room measurements. The REW file was provided to me by another AVS member when I was trying to address (understand) Bosso's discussion of the data points from the thread I linked a few posts ago.

If you have ten minutes I really think reading that thread would be of help to you. Again it is only about 3 pages and Bosso and Bill (Omni creator) get involved in the converstation.

James

Okay, thanks James. I'll take a look at the thread.

goneten,

Sounds good. My apologies for not being able to just give you a flat out yes or no answer to your question.

James

Accuracy is not yes/no, it's a matter of degree.


Yes, and I believe that Bill's response satisfactorily put that complaint to rest.


How much accuracy do you think you need, and why?

Noah,

On the note of data points, I wish Bill would program the software to grab the amount of data that would be used in a 1-96th trace and then smooth the graph according to what is selected. Then also program the software to allow for smoothing to be changed after the frd has been saved. So as an example take a trace at 1-6th, save it load it later and be able to change to 1-24th or vice versa.

I think he may disagree with that but at this time REW does this and as you and I have seen, Omni is being compared to that program. So for a package that does have a cost compared to one that is free (software wise) I just do not think it would be a bad idea.

James

I don't disagree with those as desired features, but I what Bosso was saying was that OM has insufficient resolution.

Without sounding sarcastic, at the end of the day how accurate is REW?
How is software accuracy measured?
I would say it would be more how accurate is the hardware. The Omnimic mic is individually calibrated and comes with it's own txt correction file.
You can read Kal Rubinson's Omnimic review on www.stereophile.com website. He rates it.

Bill please can you elaborate on this.

Hi James,

I'm not very interested in Omnimic, so I missed this post.

First, if no one can see the difference in graphs that have been posted on these boards for a decade done in TrueRTA and REW vs the OM graphs, I truly don't know what to say to that.

The issue here is one of confusing two different things. I used a measurement done by Ilkka in TrueRTA back in 2005 that he used to show Craig Chase the difference between a 1/6 octave graph (6 data points, which he circled for emphasis) that has no smoothing applied and a 1/24 octave graph that has no smoothing applied.

I took his graphs and overlaid the 1/24 trace below the 1/6 trace:



Again, neither of the traces have any smoothing applied.

The 1/6 trace shows a response of (+/-) 5dB and the 1/24 trace shows the exact same measurement having a frequency response of (+/-) 11dB.

My point here is that a measurement hardware/software system is acquired and measurements are taken to show you what your system is doing at your seat. If you take the 1/6 octave trace and stretch it X 4 horizontally and squeeze it X 1/2 vertically (vs the 1/24 trace in a standard REW or TruRTA graph scale), you will be missing a lot of necessary information needed to flatten (or customized in any way) your response.

Bosso

Then it seems to me if you use room correction like Audyssey you are better off leaving the calibration as is rather than tweaking it based on measurements like these.

Bosso,

Thank you very much for the reply and information. First let me say that yes, for me the information you provided in the screen shot does show a difference. The information showing the 1/6th (6 data points) with no smoothing applied looks as if it was drawn with a ruler along with the difference in the +/- db.

Now here is a basic question that I will ask since I am honestly attempting to find this out and I hope I word it correct.

Below is a very small section of an frd file from a measurement taken with Omni at a setting of 1-24th. Are these data points as you refer to them? Thanks.

James

(Sorry this informaiton posted in the way it did but I am sure you are following it)
10.03365 84.4 0
10.06994 84.4 0
10.10636 84.39 0
10.14291 84.39 0
10.17959 84.38 0
10.21641 84.37 0
10.25336 84.36 0
10.29044 84.34 0
10.32766 84.31 0
10.36501 84.28 0
10.4025 84.22 0
10.44012 84.15 0
10.47788 84.03 0
10.51577 83.9 0
10.55381 83.78 0
10.59198 83.7 0
10.63028 83.64 0
10.66873 83.59 0
10.70732 83.56 0
10.74604 83.54 0
10.7849 83.52 0
10.82391 83.51 0
10.8630 6 83.5 0
10.94177 83.49 0

no, those are not measurement points, those are points along a smoothed curve. The raw, before smoothing, measurement points will be spaced no less than 48,000hz/32768 for OM. Smoothing can give you more points on a curve (or less) but can't increase the resolution. I'd you actually think you will fix a frequency aberration less than 1.5hz wide, then you'll need more resolution. (but don't invite me to hear your system, its gonna suck right out loud if you eq like that!

Bill,

Thank you for your time and information.

So now another basic question, where does the "6 data points" come into play? Is that what the Omni software uses to display the taken measurement?

Thanks again.

James

omnimic starts with points every 1.48 Hz or so. If you use "no smoothing" those are the points you get. Usually that's too wild and hashy when echoes are involved, so you smooth to get something more usable and/or more related to what you can hear. And for saving data it makes no sense to keep 7000 points for the one octave around 10khz, so you take points along the smoothed curve so you the same number of points every octave (higher "hz resolution" at lower frequencies, down to the limit of the raw data).

Omnimic Version 2 on Dayton's site:

http://www.daytonaudio.com/index.php...nt-system.html

Bill,

Thank you for the reply and information.

Based off of this statement so you take points along the smoothed curve so you the same number of points every octave, is this there where the 6 data points being mentioned is coming from? Thanks.

James

I didn't say it, so I can't answer.
The six points are apparently from unsmoothed data, but I don't know if there were more before or if these are interpolated between another set.

I like the full size USB connection, ...I've read individuals complain, I never un-plug mine so once it's in, it's in.

As one of the initial AVS'ers to get one, it's a pleasure to see the product mature.

I fully understand Bosso's concerns and stated preferences, however if an individual cannot adapt and properly discern a graph's characteristics, that's on them. Operator error. There's sufficient resolution granularity, there's very easily configurable vertical and horizontal scales.


Thanks

Apologies for not responding sooner, completely forget about this thread. I think REW has a measuring resolution of 0.36 Hz across the full spectrum. I think that's what some might consider accurate.

I'm not sure if my eyes deceive me, but has the new updated V2 software have upgraded visuals too ?

From 10 Hz to 20 Hz is one octave. If there are data points every 1.48 Hz, then there will be about 6 data points between 10 to 20 Hz. However, 20-40 Hz will get 13 data points, 40-80 Hz will get 27 data points, and 80-160 Hz will have 54 data points. It should then have 1/27 octave resolution and higher from 40 Hz and up with an unsmoothed graph.

DD,

Thank you very much for that information. What you provided matches what I see in a measurement I took with the none setting in Omni.

Does anyone know if the V2 software can be used with a V1 mic? I took a quick look over at the Dayton site but did not see mention of it.

James

Software is the same, spec is the same. Just different construction (and a new manufacturer).

Uh, no, that's not what accuracy is. That's resolution. You can have a very high resolution inaccurate measurement. Or a low resolution accurate answer. Different things.

Looks like a higher list price 399 vs 349?

Thanks Bill.

James