So hows the volume boost apart from that?

I think it's doing what it's supposed to do, not 100 percent certain.

When I say "-6" in the ini-file, how much does that translate to in decibels of attenuation? How high or low can I go with it?

My goal was to reduce the AM volume by a significant amount, and boost the FM slightly.

I described it in the earlier post: Simple answer: +6 equates to +3dB, ie 0.5dB per integer.

Long answer:

The player doesn't really know about dB levels when it sets the volume. There is a table of 100 volume levels, and the player just moves to the previous/next entry in the table and applys it in order to decrease/increase the volume. It then reads back the dB value associated with that row in tha table to send to the display so that the user can say "My volumes set to xx dB". So the way that the boost works at the moment by jumping up or down nn rows in the table where nn is the config.ini parameter.So +6 equates to pressing "vol up" 6 times or rotating the knob slowly 6 clicks clockwise.

I also mentioned that I needed feedback on the non-linearity issue. This is because the table is non-linear. Right at the bottom of the scale, we start with approximately 3dB per step, and above -3dB, it is 1dB per step.

But for most of the useful range (-25dB to -3dB), it is approximately 0.5dB per step. Because it is 0.5dB for so much of the range, it doesn't make much sense for me to write code to take a dB value from config.ini and iteratively find the row that is nearest to it. I'd have to get the current volume, read the assosciated dB value, add the config.ini dB value, and then read the table until I found the line. Granted, this could be done without too many coding nightmares, but it doesn't really seem worth it if I can just do integer math on the table index with acceptable results.

Here is the table so you can see what I'm talking about:

	{ 0x000, 0xf80, -9999 }, /* Zero */

	{ 0xfff, 0xf80, -6620 }, /* -66.2 dB intensity 0.023988 */

	{ 0xffe, 0xf80, -6270 }, /* -62.7 dB intensity 0.053703 */

	{ 0xffd, 0xf80, -5820 }, /* -58.2 dB intensity 0.151356 */

	{ 0xffc, 0xf80, -5530 }, /* -55.3 dB intensity 0.295121 */

	{ 0xffb, 0xf80, -5310 }, /* -53.1 dB intensity 0.489779 */

	{ 0xffa, 0xf80, -5140 }, /* -51.4 dB intensity 0.724436 */

	{ 0xff9, 0xf80, -4990 }, /* -49.9 dB intensity 1.023293 */

	{ 0xff8, 0xf80, -4870 }, /* -48.7 dB intensity 1.348963 */

	{ 0xff7, 0xf80, -4760 }, /* -47.6 dB intensity 1.737801 */

	{ 0xff6, 0xf80, -4660 }, /* -46.6 dB intensity 2.187762 */

	{ 0xff5, 0xf80, -4580 }, /* -45.8 dB intensity 2.630268 */

	{ 0xff4, 0xf80, -4500 }, /* -45.0 dB intensity 3.162278 */

	{ 0xff3, 0xf80, -4420 }, /* -44.2 dB intensity 3.801894 */

	{ 0xff2, 0xf80, -4360 }, /* -43.6 dB intensity 4.365158 */

	{ 0xff1, 0xf80, -4290 }, /* -42.9 dB intensity 5.128614 */

	{ 0xff0, 0xf80, -4240 }, /* -42.4 dB intensity 5.754399 */

	{ 0xfef, 0xf80, -4180 }, /* -41.8 dB intensity 6.606934 */

	{ 0xfed, 0xf80, -4080 }, /* -40.8 dB intensity 8.317638 */

	{ 0xfeb, 0xf80, -3990 }, /* -39.9 dB intensity 10.232930 */

	{ 0xfe9, 0xf80, -3910 }, /* -39.1 dB intensity 12.302688 */

	{ 0xfe7, 0xf80, -3840 }, /* -38.4 dB intensity 14.454398 */

	{ 0xfe4, 0xf80, -3740 }, /* -37.4 dB intensity 18.197009 */

	{ 0xfe1, 0xf80, -3650 }, /* -36.5 dB intensity 22.387211 */

	{ 0xfde, 0xf80, -3570 }, /* -35.7 dB intensity 26.915348 */

	{ 0xfdb, 0xf80, -3490 }, /* -34.9 dB intensity 32.359366 */

	{ 0xfd8, 0xf80, -3420 }, /* -34.2 dB intensity 38.018940 */

	{ 0xfd5, 0xf80, -3360 }, /* -33.6 dB intensity 43.651583 */

	{ 0xfd2, 0xf80, -3300 }, /* -33.0 dB intensity 50.118723 */

	{ 0xfcf, 0xf80, -3250 }, /* -32.5 dB intensity 56.234133 */

	{ 0xfcb, 0xf80, -3180 }, /* -31.8 dB intensity 66.069345 */

	{ 0xfc7, 0xf80, -3110 }, /* -31.1 dB intensity 77.624712 */

	{ 0xfc3, 0xf80, -3050 }, /* -30.5 dB intensity 89.125094 */

	{ 0xfbf, 0xf80, -3000 }, /* -30.0 dB intensity 100.000000 */

	{ 0xfbb, 0xf80, -2950 }, /* -29.5 dB intensity 112.201845 */

	{ 0xfb7, 0xf80, -2900 }, /* -29.0 dB intensity 125.892541 */

	{ 0xfb3, 0xf80, -2850 }, /* -28.5 dB intensity 141.253754 */

	{ 0xfaf, 0xf80, -2810 }, /* -28.1 dB intensity 154.881662 */

	{ 0xfab, 0xf80, -2760 }, /* -27.6 dB intensity 173.780083 */

	{ 0xfa7, 0xf80, -2720 }, /* -27.2 dB intensity 190.546072 */

	{ 0xfa2, 0xf80, -2680 }, /* -26.8 dB intensity 208.929613 */

	{ 0xf9e, 0xf80, -2640 }, /* -26.4 dB intensity 229.086765 */

	{ 0xf99, 0xf80, -2600 }, /* -26.0 dB intensity 251.188643 */

	{ 0xf93, 0xf80, -2550 }, /* -25.5 dB intensity 281.838293 */

	{ 0xf8d, 0xf80, -2500 }, /* -25.0 dB intensity 316.227766 */

	{ 0xf86, 0xf80, -2450 }, /* -24.5 dB intensity 354.813389 */

	{ 0xf7f, 0xf80, -2400 }, /* -24.0 dB intensity 398.107171 */

	{ 0xf77, 0xf80, -2350 }, /* -23.5 dB intensity 446.683592 */

	{ 0xf6f, 0xf80, -2300 }, /* -23.0 dB intensity 501.187234 */

	{ 0xf66, 0xf80, -2250 }, /* -22.5 dB intensity 562.341325 */

	{ 0xf5d, 0xf80, -2200 }, /* -22.0 dB intensity 630.957344 */

	{ 0xf54, 0xf80, -2150 }, /* -21.5 dB intensity 707.945784 */

	{ 0xf49, 0xf80, -2100 }, /* -21.0 dB intensity 794.328235 */

	{ 0xf3f, 0xf80, -2050 }, /* -20.5 dB intensity 891.250938 */

	{ 0xf33, 0xf80, -2000 }, /* -20.0 dB intensity 1000.000000 */

	{ 0xf27, 0xf80, -1950 }, /* -19.5 dB intensity 1122.018454 */

	{ 0xf1a, 0xf80, -1900 }, /* -19.0 dB intensity 1258.925412 */

	{ 0xf0d, 0xf80, -1850 }, /* -18.5 dB intensity 1412.537545 */

	{ 0xefe, 0xf80, -1800 }, /* -18.0 dB intensity 1584.893192 */

	{ 0xeef, 0xf80, -1750 }, /* -17.5 dB intensity 1778.279410 */

	{ 0xedf, 0xf80, -1700 }, /* -17.0 dB intensity 1995.262315 */

	{ 0xece, 0xf80, -1650 }, /* -16.5 dB intensity 2238.721139 */

	{ 0xebb, 0xf80, -1600 }, /* -16.0 dB intensity 2511.886432 */

	{ 0xea8, 0xf80, -1550 }, /* -15.5 dB intensity 2818.382931 */

	{ 0xe94, 0xf80, -1500 }, /* -15.0 dB intensity 3162.277660 */

	{ 0xe7e, 0xf80, -1450 }, /* -14.5 dB intensity 3548.133892 */

	{ 0xe67, 0xf80, -1400 }, /* -14.0 dB intensity 3981.071706 */

	{ 0xe4f, 0xf80, -1350 }, /* -13.5 dB intensity 4466.835922 */

	{ 0xe36, 0xf80, -1300 }, /* -13.0 dB intensity 5011.872336 */

	{ 0xe1a, 0xf80, -1250 }, /* -12.5 dB intensity 5623.413252 */

	{ 0xdfe, 0xf80, -1200 }, /* -12.0 dB intensity 6309.573445 */

	{ 0xddf, 0xf80, -1150 }, /* -11.5 dB intensity 7079.457844 */

	{ 0xdbf, 0xf80, -1100 }, /* -11.0 dB intensity 7943.282347 */

	{ 0xd9d, 0xf80, -1050 }, /* -10.5 dB intensity 8912.509381 */

	{ 0xd78, 0xf80, -1000 }, /* -10.0 dB intensity 10000.000000 */

	{ 0xd52, 0xf80, -950 }, /* -9.5 dB intensity 11220.184543 */

	{ 0xd29, 0xf80, -900 }, /* -9.0 dB intensity 12589.254118 */

	{ 0xcfe, 0xf80, -850 }, /* -8.5 dB intensity 14125.375446 */

	{ 0xcd1, 0xf80, -800 }, /* -8.0 dB intensity 15848.931925 */

	{ 0xca0, 0xf80, -750 }, /* -7.5 dB intensity 17782.794100 */

	{ 0xc6d, 0xf80, -700 }, /* -7.0 dB intensity 19952.623150 */

	{ 0xc37, 0xf80, -650 }, /* -6.5 dB intensity 22387.211386 */

	{ 0xbfe, 0xf80, -600 }, /* -6.0 dB intensity 25118.864315 */

	{ 0xbc1, 0xf80, -550 }, /* -5.5 dB intensity 28183.829313 */

	{ 0xb80, 0xf80, -500 }, /* -5.0 dB intensity 31622.776602 */

	{ 0xb3c, 0xf80, -450 }, /* -4.5 dB intensity 35481.338923 */

	{ 0xaf4, 0xf80, -400 }, /* -4.0 dB intensity 39810.717055 */

	{ 0xa56, 0xf80, -300 }, /* -3.0 dB intensity 50118.723363 */

	{ 0x9a5, 0xf80, -200 }, /* -2.0 dB intensity 63095.734448 */

	{ 0x8df, 0xf80, -100 }, /* -1.0 dB intensity 79432.823472 */

	{ 0x800, 0xf80, 0 }, /* 0.0 dB intensity 100000.000000 */

	{ 0x800, 0xf70, 100 }, /* 1.0 dB intensity 125892.541179 */

	{ 0x800, 0xf5f, 200 }, /* 2.0 dB intensity 158489.319246 */

	{ 0x800, 0xf4b, 300 }, /* 3.0 dB intensity 199526.231497 */

	{ 0x800, 0xf35, 400 }, /* 4.0 dB intensity 251188.643151 */

	{ 0x800, 0xf1c, 500 }, /* 5.0 dB intensity 316227.766017 */

	{ 0x800, 0xf01, 600 }, /* 6.0 dB intensity 398107.170554 */

	{ 0x800, 0xee1, 700 }, /* 7.0 dB intensity 501187.233627 */

	{ 0x800, 0xebe, 800 }, /* 8.0 dB intensity 630957.344480 */

	{ 0x800, 0xe97, 900 }, /* 9.0 dB intensity 794328.234724 */

	{ 0x800, 0xe6c, 1000 }, /* 10.0 dB intensity 1000000.000000 */

As to how high or low you can go, at the moment the alpha code will let you go 100 steps in either direction. But, as I said, the volume will max out at row 100, +10dB. If you try to push a +20 boost on top of a -3dB player volume, you're still only going to hear +10dB. (In the same vein, pushing a -50 cut on a player volume of -30dB is going to result in silence!)

OK, cool, I think that matches my observed behavior. Personally, I don't mind the integer/curve issue, I think it's fine the way it is.

So what do you think about the pop issue? Do you think you know where the problem lies and can find a solution?

I have 2 possible avenues of attack for solving the pop issue.

First idea is to mute the mixer whenever we apply the eq. This is possibly a good idea anyway - the way that the player sets the eq values is to write the 20 sections straight to the DSP registers. I don't know how the dsp is supposed to handle this - whether it is supposed to read them all synchronously and apply them in one step or whether it just starts using the new values as they are written in...I suspect the 2nd, in which case the frequency of a section gets written before the dB/Q, which means that changing eqs could lead to transients, even without the extreme Q factor that we are using in the tone controls.

The second approach would be to apply 0 tone controls and then ramp the tone controls up to the set values.

Both of these have potential issues that I'm going to have to investigate.

The issue with the first is that when modifying an eq band completely unrelated to the tone controls, we'd be muting and uumuting the mixer. This'd happen a lot - every small frequency change would do this. It might or might not be audibly noticeable.

The issue with the second is that the tone controls would take some time to ramp. Admittedly, a very short time, but again - it'd happen every time an eq was adjusted.

So I'm between a rock and a hard place at the moment.

Oh, I've just seen something else that I can try first..

I think I know what the problem is.

I didn't stop the original eq from getting applied in the ioctl, I initialise the tone controls after this has happened. This is fine for bootup, but it means that setting a new eq has the same effect as the old "Off" problem.

Give me a couple of miuntes and I'll have a new binary.

Try this.

hijack246-volboost.zimage


volboost246.patch

Pop seems to have been fixed.

There is occasionally a slight bit of noise when switching between sources, but I don't think it's any more than there was before your changes were implemented. In other words, I think it's behaving exactly as desired.

I'm now working on balancing my inputs to the way I like them. At the moment, I think I've got a happy medium right here:

[hijack]
disable_bassboost_FM=1
volume_boost_FM=4
volume_boost_AM=-10

Since I don't have anything plugged into the Aux, I don't need an alteration to that input.

I'm still not certain about whether or not the disable_bassboost_FM is working or not. Anyone else want to take a stab at testing that?

Finally... :rolleyes:

Tony, are you testing on a MK2/MK2A ? Hardware revisions up to 6 (MK1 I'm guessing) don't appear to have a bass boost to start with, and the disable_bassboost_FM won't do anything.

Yes, this is on my mark 2. And I know that the FM bass got boosted in the original software because there were certain announcers on NPR that would make my car shake when they popped a "P". The thing is that I haven't listened to the tuner long enough with your code to have heard those announcers yet.

I haven't done any side-by-side comparison, but I'd venture to say that the bass boost isn't working. My FM seems to have a heck of a lot more bass than my MP3 but I also know that many FM stations are given a lot of bass at the source to sound more "full." I can however say that the volume boost/cuts seem to be working very well. However, I couldn't get the v246 patch to work right, I had to flash the binary instead. The patch is trying to create files that already exist and such, I don't think it was diff'ed from a clean Hijack source tree or something.

Anyway, thanks for all of your hard work, genixia... My initial test of this functionality seems to be going pretty well. Now I just need to dial in the right values... Any chance you could work this into the Hijack menu itself so we can adjust on the fly? Or is that not possible?

Err, the patch isn't creating any files. It is modifying two files that should already be present. I just checked the patch and it looks good. Maybe you weren't supplying the -p1 parameter?

cd kernel

ls

>hijack-246

>volboost246.patch

cd hijack-246

patch -p1 <../volboost246.patch

The bassboost has got me stumped. Just for kicks, try supplying 0 as a parameter. I could'a got the logic reversed! And I've been out for a coupla beeeeeeeeers, so I'm not going to check now.

It's highly unlikely that it'd be added to the hijack menu. The main reason is the extremely limited flash resources available. I virtually had to beg Mark for 12 bits of flash in total for storing bass and treble settings for both AC and DC modes! (Yes, each setting is stored in 3 bits...) *
Secondly, although this is a little awkward to configure, it should only need to be done once, and maybe tweaked a bit later if you decide you really wanted another +1 on the FM mode after all. ie, it shouldn't be a dynamic setting (unlike the bass or treble which can vary by song).

If Mark finds 100 bytes of flash stashed away somewhere then obviously there will quite a few candidates trying to use it in preference to config.ini values, but until then the flash will be reserved for the most deserving cases

* Well, maybe 'begging' is a bit strong. I asked for 24 bits and got 12....

I had the p1 parameter. The patch is *definitely* doing more than just modifying two files that "should already be present." At least the one you pointed me to is, at http://h0000e108b21e.ne.client2.attbi.com/genixia/empeg/volboost246.patch

Take a look at it, it's got no less than 20 files that it's trying to create, a whole bunch of stuff in include/asm-arm/arch, etc....

Aaargh. Sorry, shift reload helped. I kept viewing the cached version.

Yeah, ok I screwed up. The 'extra' files aren't really. The kernel source tree comes with a load of undefined symlinks, that get set by 'make dep'. So once you've built a kernel, there appears to be a whole load of new files compared to the same tree previously. My virgin 246 tree hasn't been built, but the volboost tree obviously has.

Try again now.

I gotcha. I'll give that patch a try tonight.

BTW I agree with your earlier points on why not to allow hijack-menu setting of the volume offsets. They should be "set it and forget it." Unfortunately, I still haven't found the right settings just yet so I have to keep dragging my Empeg in and out to find them.

I haven't done any side-by-side comparison, but I'd venture to say that the bass boost isn't working.

Funny, I'd come to the opposite conclusion last night.

True, FM stations always sound like their bass is boosted compared to MP3s, because they do over-EQ their signals before sending them out. But that's always been around, there's nothing unusual about that.

What I had noticed is that, over and above the bass-boosting the stations already did, the player would boost the bass in radio mode yet again, making it sound truly unnatural, especially when announcers were speaking.

Gen's 246 binary he released yesterday seems to have (for me) removed the extra bass boost as intended. However, I did not A/B test this feature by changing the INI file and running back and forth to my car. Someone should do this to be sure.

Okay then I officially change my position to "I have no idea if bass boost is working or not."

So, what's the verdict so far?

Is this ready for me to send to Mark for inclusion in Hijack as beta?

The main thing I need to know is that no previous functionality got broken and that the feature appears to work as advertised - which from the posts above, I take to be a yes. ( the bonus disable_bassboost_FM feature aside - if the jury's still out about that, then fine - that code is clean and simple and I can't see any reason why it shouldn't work and it is 'bonus' to the original feature request.)

So basically, that means that the boost/cut appears to work, and that changing volume/input or eq doesn't result in any new nasty noise artifacts...

??

Yeah, I think it's ready for prime time. Make absolutely sure there's no PrintK's left over in the code...

Yeah, I'm glad you spotted that one - I have my serial port car_rate=115200 in my config.ini, which is neccessary to see any debug messages when in DC mode (got added when I was chasing down the DC tone initialisation issue), but a side effect is that the stacatto effect doesn't happen either...

Thank you both your help in testing this code! Couldn't'a done it without you.

Mark may well optimise my code, so sanity checking the first Hijack build with it in is probably a good idea, although I think that he sprays his keyboard with bug-repellant...

>Is this ready for me to send to Mark for inclusion in Hijack as beta?

Gimme!

I'll have a look and probably stick it in.

Cheers

volboost246.patch

(No change since the previous post 13 up that had this link for those wondering.)

Now maybe someone (hint hint) would be willing to take a look at implementing volume adjustments together with L/R delay feature for a more accurate shift.

Then all we need is three quick presets (passenger, driver and OFF) with some button code method to cycle them and the feature is more than prime-time.

Bruno