Any way to limit gain to +/-10.0 dB?

[maxcooper]

Hi,

I am using REW to setup my Onix R-DES EQ. The process is basically that I measure and calculate the EQ settings in REW, and then I copy the settings into the R-DES software and upload it to the unit. It works great.

I do have one issue, however. The R-DES box has different limits on the EQ params than REW has (with any of the Equilisers selected). The R-DES limits are:

Gain: -10.0 - +10.0 dB
Freq: 20-120 Hz
Q: 1.0-15.0

The R-DES unit has a freq resolution of 1 (i.e. the freqs must be entered as integer values), so I usually change the calculated freqs in REW to whole numbers. This work-around works fine.

Sometimes the Q values are greater than 15. The limit for the Generic Equiliser seem to be 20.0. I have found that selecting SMS-1 as the Equiliser will limit the Q values to 10. This work-around works okay, but I suppose the filters could be better if they could be calculated with Q as high as 15.

The main issue is that the Gain value for my biggest peak often has a magnitude greater than 10.0. For instance, my biggest peak in my latest run has a Gain of -10.6. Well, maybe this example wasn't so great, since -10.6 isn't much different from -10.0 , but I've had ones that were larger in magnitude than that one. (Why do I keep measuring? Because I change my equipment/placement/treatments/furniture setup sometimes, and of course, it's fun. )

Question: Is there any way to adjust the limits (and resolution, for the Freq issue) for the EQ settings?

If not, it would be really cool if there was an Options/Preferences dialog where these could be set, so that REW could be used with any parametric EQ. It would also be nice if there was an "Onix R-DES" option in the Equiliser menu that was already setup with the proper limits.

Also: I really love REW! Thanks so much for making it available, and for everyone in the community that answers questions and writes guides, etc.

Thanks!
-Max

[brucek]

Question: Is there any way to adjust the limits (and resolution, for the Freq issue) for the EQ settings?

No, but why not simply enter your own filters into REW once you have taken a raw measurement?

You would then have complete control over the max and min of your entries.

I don't find that REW ever automatically finds the optimum filters. I always have to tweak them and add my own. The "what you see is what you get" is rather good though and after entering filters into my EQ I find the actual results very close to the predicted graph that REW offers.
This means I can take a raw measurement and then go off line and play with filters at another time and be confident that they'll be quite close when entered.

An alternative to filters that exceed your size limits is to simply divide the recommended filter in half and enter them twice into your EQ.

brucek

[JohnM]

maxcooper wrote:

it would be really cool if there was an Options/Preferences dialog where these could be set, so that REW could be used with any parametric EQ. It would also be nice if there was an "Onix R-DES" option in the Equaliser menu that was already setup with the proper limits.

That's on the features list, I've moved it further up

Providing control of the ranges and resolutions is straightforward, the difficult part is dealing with Q/bandwidth. There are a number of ways in which Q can be specified in filter implementations so the same value can give different filter widths. To see what R-DES is using it would be helpful if you could make some measurements of the R-DES alone, just connect its output to your soundcard in place of the SPL meter and make some measurements with a single filter at a mid-range frequency (80Hz for example) and a fixed gain (-10dB, say) but with a few different Q settings, then save the measurements as a .mdat file and email it to me - remember to say what Q was used for each measurement.

[maxcooper]

brucek wrote:

No, but why not simply enter your own filters into REW once you have taken a raw measurement?

You would then have complete control over the max and min of your entries.

I don't find that REW ever automatically finds the optimum filters. I always have to tweak them and add my own. The "what you see is what you get" is rather good though and after entering filters into my EQ I find the actual results very close to the predicted graph that REW offers.
This means I can take a raw measurement and then go off line and play with filters at another time and be confident that they'll be quite close when entered.

An alternative to filters that exceed your size limits is to simply divide the recommended filter in half and enter them twice into your EQ.

brucek

Thanks for responding. I am a relative newbie in sub EQ tuning, so I like the fact that REW will make a very specific recommendation. I like that it just tells you what values to use. In fact, I plan to write an as-simple-as-possible how-to for using REW to setup an R-DES for this reason. From personal experience I know that it can be confusing to choose filter values for the first time, and I suspect there are many people with R-DES boxes (or any other sub EQ) that feel the same way. Also, I think many people aren't very motivated to learn to select filter values themselves, and would appreciate the convenience of REW's receommendations (so long as they are within the ranges supported by their EQ box). Despite the fact that it does take some effort to get an SPL meter and do measurements with REW, it is really nice to have a deterministic (easy to document, easy to perform) process for choosing filter values.

However, after going through a couple of rounds of tuning this weekend, I see what you mean about being able to choose filter values that seem to measure better "by hand". I haven't yet "closed the loop" as tightly as you suggest, though, as I am still doing measurements after each filter tweak.

In any event, it would still be nice to be able to perhaps start with some recommendations from REW that honor, and take full advantage of, the limits of my R-DES box.

Thanks,
-Max

[maxcooper]

maxcooper wrote:

The R-DES unit has a freq resolution of 1 (i.e. the freqs must be entered as integer values), so I usually change the calculated freqs in REW to whole numbers. This work-around works fine.

Update: using the SMS-1 mode in REW solves this, since the SMS-1 also appears to use integral values for frequency.

-Max

[maxcooper]

JohnM wrote:

That's on the features list, I've moved it further up

Excellent!

JohnM wrote:

Providing control of the ranges and resolutions is straightforward, the difficult part is dealing with Q/bandwidth. There are a number of ways in which Q can be specified in filter implementations so the same value can give different filter widths. To see what R-DES is using it would be helpful if you could make some measurements of the R-DES alone, just connect its output to your soundcard in place of the SPL meter and make some measurements with a single filter at a mid-range frequency (80Hz for example) and a fixed gain (-10dB, say) but with a few different Q settings, then save the measurements as a .mdat file and email it to me - remember to say what Q was used for each measurement.

Ah, maybe this explains (at least in part) why my measured results did not seem to match that well with the predicted results. I was using SMS-1 mode to calc filter values for my R-DES. Perhaps the SMS-1 and R-DES handle Q values differently.

I will do some tests as you describe and email them to you, but it might take me a week.

Thanks so much for REW! I love it. And as a science-minded Java programmer myself, I think it is very interesting and very well done. Bravo!

-Max

[maxcooper]

John, I performed some measurements and sent you all the .mdat files, including an explanation and Notes in the data files. Sorry for the large file size (16MB) -- I hope that isn't a problem. I can re-send with less data or spreading the files across several messages if there is a problem with delivery.

Also, here is a graph of a bunch of Q values. I did a soundcard calibration with a plain cable, then inserted the R-DES into the loop. I cleared the meter/mic calibration, and turned C-weighting compensation off. The R-DES has a low pass crossover that can't be defeated, so I set it to 120 Hz, 4th order. I chose 35 Hz as the filter frequency since it seemed to be about the middle of the plateau in the graph. The filter gain was set to -10.0 dB. The graph shows the following values for Q: 1, 2, 3, 5, 7, 9, 11, 13, and 15.

D'oh! I can't post images/links until I have 10 posts. Scroll down...

-Max

[maxcooper]

Sorry, here it is...


-Max

[JohnM]

Got the files Max, thanks.

[JohnM]

I've added an initial emulation of the R-DES parametric filters to the current dev build. Was the subsonic filter bypassed for those measurements? There is a subsonic roll-off which makes it a bit trickier to accurately match the Q. If the subsonic filter was bypassed on R-DES, did you have a cal file loaded for the SB Live?

[maxcooper]

JohnM wrote:

I've added an initial emulation of the R-DES parametric filters to the current dev build. Was the subsonic filter bypassed for those measurements? There is a subsonic roll-off which makes it a bit trickier to accurately match the Q. If the subsonic filter was bypassed on R-DES, did you have a cal file loaded for the SB Live?

I am not sure if there is a subsonic filter or not. That sounds vaguely familar, so perhaps the unit does have one. If there is one, I am not sure if it was on or off. I didn't see any option to enable/disable such a filter; but I also wasn't looking.

I did have a fresh soundcard calibration in effect for the measurements. I performed the calibration right before I wired the R-DES box into the loop. The souncard was pretty flat from ~4.4Hz to (~20KHz? I don't remember). I also cleared the meter/mic calibration, and turned off C-weighting compensation. Thinking back, I may have had C-weighting compensation ON when I did the soundcard calibration. Would that affect the soundcard calibration?

I included a baseline measurement of the R-DES where I had no filters configured. The low-pass filter was present, but that is non-defeatable -- I just mean that I had the gain set to 0.0 dB (which causes them to "gray out" in the UI) for all the tuning filters. Perhaps that could be helpful for determining what effects are from the filter vs. some non-configurable aspect of the R-DES (subsonic filter) or my test configuration (soundcard cal).

If I can help by sending the souncard calibration file, or re-doing the measurements with a different setup, please let me know.

-Max

[Ilkka]

The first version of the R-DES had an adjustable subsonic filter. Here's the manual describing it: http://www.av123.com/manuals/31_R-DE...smanual_v1.pdf

The newer version doesn't have it anymore. But I'm guessing there is still some natural VLF roll-off being present. I haven't seen a single electronic device being dead flat down to DC. It would interesting to see it measured. Maybe you could DL the newest version of the REW (I'm sure John can give you the link) which goes down to 2 Hz.

[maxcooper]

I did a little investigating, and it appears that some version(s) of R-DES had/have a configurable (and defeatable) subsonic filter and 4 EQ filters. I saw it here in the v1 manual (sorry for the non-clickable link, but I don't have enough posts to post a link yet):
www . av123.com/manuals/31_R-DES_ownersmanual_v1.pdf

However, my v1.1 R-DES has no subsonic filter and 5 EQ filters. So I suspect that no subsonic filter was in effect when I performed my measurements.

-Max

[maxcooper]

Ah, I'm too slow.

I am eager to try the latest REW so I can test out the R-DES mods. I can also do a measurement down to 2 Hz in search of a subsonic filter.

-Max

[brucek]

Thinking back, I may have had C-weighting compensation ON when I did the soundcard calibration. Would that affect the soundcard calibration?

No...

[Ilkka]

Max,

Have you been able to measure the low end roll-off of the R-DES?

[maxcooper]

Here's the RDES response measured down to 2Hz, with no filters set. The non-defeatable LP crossover was set to 120 Hz 4th order. The souncard calibration is dashed, and the measurement is in blue:

I just sent an .mdat file to JohnM also, in case that's useful.

-Max

[Ilkka]

Thanks! You have a PM.