Page 1 of 3 123 LastLast
Results 1 to 25 of 52

Thread: MOTOTRBO radios, poor audio AGC performance - SOLVED

  1. #1
    Join Date
    Dec 21, 2011
    Posts
    4,514
    Thanks
    3,789
    Thanked 6,649 Times in 1,896 Posts
    Country: Canada

    Default MOTOTRBO radios, poor audio AGC performance - SOLVED

    THE AUDIOFIX TOOL IS NO LONGER REQUIRED. SEE BELOW:

    March 2016: Motorola has incorporated my work along with some of their own tweaks, into current firmware. For first-generation TRBO radios, use firmware R01.12.17 or newer. For second-generation, use R02.50.05 or newer.

    June 2015: An audio fix utility is now available. Link here

    It is a well-known fact MOTOTRBO radios have TERRIBLE audio AGC performance on both analog and digital. The DMR-MARC group even goes so far as to require all members on their network kindly disable AGC on subscriber radios, as AGC distorts the audio beyond belief.

    Both original and the newer 2.0 series TRBO line have this problem. To date, there has been no resolution, other than suggestions to muck around with linear audio gain levels until you get it "just right". But this is inconsistent and everyone has different speaking habits. One setting does not fit all.

    One of the projects I undertook this winter was to identify and correct the audio issues which affect the TRBO platform. I firmly believe aside from RF coverage, the most important thing on a system is the audio. If subscriber (or dispatcher) audio is poor, it causes many issues. Some of those issues are:

    - Unintelligibly (having to repeat oneself);
    - User/dispatcher listening fatigue;
    - Disinterest in the system/technology (hams);
    - Profitability (users will go elsewhere if they perceive your system audio sucks);
    - Distraction (users having to adjust volume levels constantly).

    I'm sure there's more. High-quality system audio is critical, especially when working with low-bitrate codecs such as AMBE+2, which is used in DMR.

    Here's what I observed with Motorola's TRBO products:

    - EXTREMELY LOUD, out-of-control audio amplitude; (Remember the ASTRO DSP6 days?);
    - Distortion/clipping;
    - AGC not working properly. It was making loud audio LOUDER, not quieter;
    - Poor dynamic range of gain levels. (AGC doesn't really seem to work at all)

    There are several hidden settings/values within the radio codeplug which control AGC operation and audio amplitude. These settings are NOT accessible via CPS or any "depot" software which is floating around. These are factory values. The settings (and default values in 2.0 radios, including the LITE series) are:

    Code:
    RC_NUMBEROFCLIPPEDSAMPLES 7  (Firmware limit: 0-31)
    RC_MINUMUMHWGAINADJUSTMENT 247  (Firmware limit: 244-250)
    RC_SATURATIONTHRESHOLD 250  (Firmware limit: 0-255)
    RC_DMICREFPL -22  (Firmware limit: 252-33022 [not sure how -22 relates to these values]
    RC_AMICREFPL -24  (Firmware limit: 252-33022 [not sure how -24 relates to these values]
    
    (I have also seen -20 and -20 for both DMICREFPL and AMICREFPL in certain firmware versions)
    There are many other hidden settings, but only these are to blame for the poor TX audio in Motorola's 2.0 series radios. Broken down, they are:

    RC_NUMBEROFCLIPPEDSAMPLES (Number of Clipped Samples): This is the number of samples in a defined period of time, which exceed the "Saturation Threshold" level (amplitude) which will then cause the AGC (a DSP function) to kick in. Think of this as a trigger.

    RC_MINUMUMHWGAINADJUSTMENT (Minimum Hardware Gain Adjustment): This is the minimum amount of gain applied to the audio when the AGC circuit is active.

    RC_SATURATIONTHRESHOLD (Saturation Threshold): This is the predefined value (amplitude) which the number of clipped samples must exceed, before the AGC circuit is triggered.

    RC_DMICREFPL (Digital Mic. AGC Reference Power Level): This is the MASTER mic gain which is referenced while operating in digital mode. Think of this value as a fixed-resistor on the audio input of the DSP. It affects the amplitude of signals before they're processed by the AMBE+2 codec.

    RC_AMICREFPL (Analog Mic. AGC Reference Power Level): This is the MASTER mic gain which is referenced while operating in analog mode. Think of this value as a fixed-resistor on the audio input of the DSP. It affects the amplitude of signals being sent to the modulator of the radio while operating in FM mode. Note: This is NOT deviation control.

    The default settings which Motorola has programmed into the codeplug are sub-optimal. In fact, they're terrible. I don't need to tell the thousands of people who are using TRBO products in a professional or hobbyist capacity, just how loud and distorted the audio can be. We all know. Unfortunately, Motorola doesn't seem to think it's an issue or they would've changed it by now.

    I have modified the default settings so they function properly. Here's what I've done in each area:

    RC_NUMBEROFCLIPPEDSAMPLES: I've reduced this trigger from 7, down to 2. As a result, AGC triggers faster. (Do not confuse this with AGC attack time)

    RC_MINUMUMHWGAINADJUSTMENT: This minimum and maximum gain settings do not have enough dynamic range between them. I have reduced minimum gain from 247 to 244.

    RC_SATURATIONTHRESHOLD: I've lowered the saturation threshold from 250 to 248. This allows the AGC circuit to trigger on lower amplitude. (once clipped samples threshold is reached)

    RC_DMICREFPL: This is the most significant change of all. I have lowered the master digital audio gain from -20 to -30. This dramatically reduces the distortion caused in both linear and AGC audio modes. A night-and-day improvement. This alone, cures the majority of the distortion issue.

    RC_AMICREFPL: The master audio gain on analog is also lowered from -20 to -29. I found analog audio -- especially on the mobile radios, to be too loud with AGC engaged. Linear audio is also somewhat distorted. Changing this gain setting is a big improvement.

    In the original MOTOTRBO radios (XPR 4000 and 6000 series, aka DM/DP 3000 series), there are only two settings which are of interest:

    Code:
    RC_DMICREFPL: -22
    RC_AMICREFPL: -24
    Again, these are the master mic gain levels. I will be experimenting with them to determine the optimal settings and will update this thread. I do not have any 4000/6000 (DM/DP 3000 series) radios here to experiment with at the moment.

    I will be posting some audio samples (hopefully today) of the dramatic before/after changes. I will also submit these findings to Motorola, but I somehow doubt they'll take an interest. If that's the case, I am happy to modify (for a reasonable fee) anyones' radio codeplug with these changes, so they can have better sounding audio on DMR-MARC. All distortion and loud artifacts (including initial SUPER LOUD BLAAAAAAAAAARGH on keyup, which ramps down after a second or two) is resolved. Believe me, I know exactly how bad these issues have been for the last few years. It was total garbage. That's why I got to the bottom of it and completely resolved the issue.

    Codeplug mods are permanent (you can read/write with CPS) until the next firmware update, which will default any changes back to factory settings.


  2. The Following 27 Users Say Thank You to Mars For This Useful Post:

    Alpha (Feb 08, 2015),Astro Spectra (Feb 08, 2015),b16a2 (Feb 10, 2015),com501 (Feb 08, 2015),DC5AJ (Feb 08, 2015),dh6kq (Jul 01, 2015),Dirk Diggler (Feb 08, 2015),hambone (Feb 08, 2015),ka5yth (Aug 27, 2015),kb1bvf (Feb 12, 2015),khw (Feb 08, 2015),kondoros (Feb 08, 2015),Mocom30 (Feb 09, 2015),MotFAN (Feb 08, 2015),N4KVE (Feb 08, 2015),no7rf (Mar 22, 2015),p25smartzone (Feb 10, 2015),PSEhub (Feb 13, 2015),rickak (Feb 09, 2015),SteveC7010 (Feb 08, 2015),SwissMoto (Feb 09, 2015),twojaygee (Feb 08, 2015),uplinker (Feb 12, 2015),wildcat (Feb 09, 2015)

  3. #2
    Join Date
    Jun 25, 2013
    Posts
    761
    Thanks
    382
    Thanked 173 Times in 116 Posts
    Country: Belarus

    Default

    Now we are in XML & Triple DES days...
    Last edited by MotFAN; Feb 08, 2015 at 07:29 AM.
    [I][FONT=times new roman][COLOR=#696969]Everyone who loves RadioReference get into the hell. [U]Especially those[/U] who also love PL-259 connectors.[/COLOR][/FONT][/I]

  4. #3
    Join Date
    Dec 21, 2011
    Posts
    4,514
    Thanks
    3,789
    Thanked 6,649 Times in 1,896 Posts
    Country: Canada

    Default

    If anyone would like their codeplugs modified, I will do the first 10 members for free, on the condition you post feedback to this thread and comment on the improvements. Thanks.

    marsp25 at gmail.com

    --

    Edit: I received a question about why I intend to charge a small fee for this modification. This is to recover the time I put into researching the problem and coming up with a fix. I don't work for free, and you shouldn't either.

    This is really no different than the microphone mods you may opt to purchase for your ICOMs (http://www.7000mic.com/) which make a massive improvement in on-air audio quality. I'm not affiliated with this site, just listing it as an example.

  5. The Following 2 Users Say Thank You to Mars For This Useful Post:

    kondoros (Feb 08, 2015),MotFAN (Feb 08, 2015)

  6. #4
    Join Date
    May 14, 2012
    Posts
    326
    Thanks
    78
    Thanked 197 Times in 91 Posts
    Country: Canada

    Default

    Excellent work Mars - keep it up sir!

  7. #5
    Join Date
    Feb 28, 2012
    Posts
    1,324
    Thanks
    121
    Thanked 1,499 Times in 453 Posts
    Country: United States

    Default

    Quote Originally Posted by Mars View Post
    I received a question about why I intend to charge a small fee for this modification. This is to recover the time I put into researching the problem and coming up with a fix. I don't work for free, and you shouldn't either.
    Mars prices his work very fairly. A lot of effort goes into this type of research. You would be amazed at the time it takes. Pay the man and enjoy superb audio.

  8. The Following User Says Thank You to Number 6 For This Useful Post:

    CQDX (Feb 08, 2015)

  9. #6
    XPR8300 No Longer Registered

    Default

    I can personally vouch for Mars and his passion for quality digital audio. You can take it to the bank.

  10. The Following User Says Thank You to XPR8300 For This Useful Post:

    CQDX (Feb 08, 2015)

  11. #7
    Join Date
    Dec 21, 2011
    Posts
    4,514
    Thanks
    3,789
    Thanked 6,649 Times in 1,896 Posts
    Country: Canada

    Default

    Attached are:

    - Analog tests
    - Digital tests
    - Images showing time/amplitude of the audio, so you can see the AGC distortion in Motorola's default settings.


    digital-tests.jpg

    analog-tests.jpg
    Attached Files Attached Files

  12. The Following 2 Users Say Thank You to Mars For This Useful Post:

    com501 (Feb 08, 2015)

  13. #8
    Jutod No Longer Registered

    Default

    Can I be one of the ten. Are the slots filled yet? I have several radios and would pay for the others.

    Oh nevermind..... I have only gen 1 radios. I might can send you a 6300 to play with.....

    I'll stay tuned to the channel to see if you have some Gen 1 changes coming.

    Thanks

  14. #9
    Join Date
    Dec 21, 2011
    Posts
    4,514
    Thanks
    3,789
    Thanked 6,649 Times in 1,896 Posts
    Country: Canada

    Default

    There are still freebies available. I can tweak a gen 1 codeplug for you, but no promises it will be optimal. Please email me your existing codeplug. Address is above.

  15. The Following User Says Thank You to Mars For This Useful Post:


  16. #10
    Join Date
    Aug 06, 2012
    Location
    Ontario, Canada
    Posts
    691
    Thanks
    135
    Thanked 228 Times in 114 Posts
    Country: Canada

    Default

    I'd love to have my XPR2500 tweaked if possible.....

  17. #11
    Join Date
    Dec 21, 2011
    Posts
    4,514
    Thanks
    3,789
    Thanked 6,649 Times in 1,896 Posts
    Country: Canada

    Default

    Several users have had their codeplugs tweaked. Looking forward to some feedback over the next few days. (No pun intended, LOL)

  18. The Following User Says Thank You to Mars For This Useful Post:

    MTS2000DES (Feb 08, 2015)

  19. #12
    Join Date
    Mar 14, 2012
    Location
    Atlanta, GA USA
    Posts
    342
    Thanks
    1,345
    Thanked 1,138 Times in 349 Posts
    Country: United States

    Default

    Mars tweaked my 7550 audio and it is a NIGHT and DAY difference. We have 3500's at work, and there is a HUGE difference in the AGC behavior. Of course the ones at work are several releases back because the low rent MSS never bothers to do firmware uplifts, he also made some other tweaks which made the radio more responsive regarding menu lag.

    Mars thanks much! but I have to say it is sad that someone such as yourself has to go so deep within the bowels of the radio codeplug to fix what the maker refuses to.

    I will get on the local DMR-MARC machine tomorrow after work (cannot ping it from inside) and give it some more airtime.

  20. #13
    Join Date
    Dec 21, 2011
    Posts
    4,514
    Thanks
    3,789
    Thanked 6,649 Times in 1,896 Posts
    Country: Canada

    Default

    Thanks, MTS2000DES. I too, am happy with the results. But I am awaiting feedback from you guys and will make additional changes/tweaks if you guys feel it can be better.

    One thing the DMR-MARC group will have to get used to, is slightly louder audio (when AGC is enabled) as finally, AGC will be working properly. But it will NOT go into distortion or be excessively loud, like the default Moto implementation.

    The solution to the botched AGC isn't to make everyone super quiet. It is to eliminate distortion and out-of-control levels. I feel this has been achieved.

  21. The Following User Says Thank You to Mars For This Useful Post:

    com501 (Feb 09, 2015)

  22. #14
    Join Date
    Jan 12, 2014
    Location
    Aachen, Germany
    Posts
    69
    Thanks
    89
    Thanked 76 Times in 27 Posts
    Country: Germany

    Default

    Mars did his magic on my DP4801 as well. I tested it last night an it works excellent, like in the audio samples that are posted above. AGC gives a really good audio now, no distortion or loud audio after keyup witch slowly ramps down like the unmodified AGC. I'm not sure if this is also possible on the XPR or if this is only possible on the EMEA radios, in the DP4801 I can adjust the mic gain settings even after turning AGC on,it's not grayed out. I have set the mic gain to 9, witch I believe is the default. What setting are recommended with the modified codeplug, any experience ?

    The sad part is that while we are now able to transmit very good audio quality even with AGC turned on, there are a lot of hams out there who don't care about there audio quality, they are running ancient firmware for whatever reason, have distorted audio and annoy people on country wide talk groups with EL Cheapo Bluetooth devices connected to their radios, at least this is the situation in my part of the world.

  23. #15
    Join Date
    Dec 21, 2011
    Posts
    4,514
    Thanks
    3,789
    Thanked 6,649 Times in 1,896 Posts
    Country: Canada

    Default

    Quote Originally Posted by DC5AJ View Post
    I'm not sure if this is also possible on the XPR or if this is only possible on the EMEA radios, in the DP4801 I can adjust the mic gain settings even after turning AGC on,it's not grayed out. I have set the mic gain to 9, witch I believe is the default. What setting are recommended with the modified codeplug, any experience ?
    AGC uses its own pre-defined set of values. The gain levels you see under Utilities/Program Radio are for linear audio gain. Motorola recommends 9dB for digital and 5dB for analog. But now that your master audio gain settings are modified, those settings will not be the same. (You'll need to adjust them higher).

    DMR-MARC recommended settings are here: http://www.dmr-marc.net/FAQ/audio.html

    The sad part is that while we are now able to transmit very good audio quality even with AGC turned on, there are a lot of hams out there who don't care about there audio quality, they are running ancient firmware for whatever reason, have distorted audio and annoy people on country wide talk groups with EL Cheapo Bluetooth devices connected to their radios, at least this is the situation in my part of the world.
    It's the same on all ham systems. Most VHF/UHF hams are appliance operators. Not all, but quite a few. They just don't understand, care or comprehend the impact these settings/firmware have on performance.

    A group of us on the board started a private IP Site Connect system. We use RAS and AES encryption. Only 2.0 (2nd gen) TRBO radios are permitted on the system. Everyone runs latest firmware and has audio set up properly. No mouth breathing. Lots of technical chat. We coordinate HF radio contacts/testing with each other. Friendships have formed. That's the way a hobbyist network should be!

    I'm really hoping TRBO-using hobbyists make enough noise to get Motorola committed to implementing these or similar changes into their products.

  24. The Following 2 Users Say Thank You to Mars For This Useful Post:

    CQDX (Feb 09, 2015),DC5AJ (Feb 09, 2015)

  25. #16
    Join Date
    Jan 31, 2015
    Location
    Portland OR
    Posts
    649
    Thanks
    907
    Thanked 880 Times in 427 Posts
    Country: United States

    Default

    Impressive work Mars. I distinctly heard the shift in volume. I would say the modified settings were "more natural". The harshness around the unmodified AGC=ON sample was gone in the modified version. Yet, to my ear, all four digital samples were daq 3.4 or better, which just highlights the subjective nature of trying to quantify digital audio quality. It would be interesting to hear samples with high background noise (ie: traffic, crowded hall, train platform, industrial machinery) to see if the AGC reduction helped the local voice punch through.

  26. #17
    Join Date
    Aug 19, 2012
    Location
    FLORIDA/QUEBEC
    Posts
    500
    Thanks
    87
    Thanked 238 Times in 144 Posts
    Country: Canada

    Default

    Quote Originally Posted by DC5AJ View Post
    there are a lot of hams out there who don't care about there audio quality, they are running ancient firmware for whatever reason, have distorted audio and annoy people
    We have users with similar problems here. They have radios, but do not have the ability [CPS, or cables] to program them. Every year, at the Orlando Hamfest [which is this weekend] Rick, the Trbo rptr admin asks them to bring him their radios for a FW upgrade, & he makes sure all radios have the same mike gain. Trouble is, many of these people are using 450-512 radios, & since he does not have a saved codeplug for these radios, he can't do anything to them w/o losing their ham freq's. But those with Gen 2 radios all have CPS, & cables, & are up to date. GARY

  27. #18
    Join Date
    Dec 12, 2012
    Posts
    176
    Thanks
    3
    Thanked 16 Times in 13 Posts
    Country: Canada

    Default

    Still open spots?
    Id like to test this on the ham system and see how well it works out between my 4550 and 5550 is possible?

  28. #19
    Join Date
    Dec 21, 2011
    Posts
    4,514
    Thanks
    3,789
    Thanked 6,649 Times in 1,896 Posts
    Country: Canada

    Default

    Sorry, all of the freebie slots have been taken. Please email me to coordinate a modification.

  29. #20
    Join Date
    Feb 28, 2014
    Posts
    57
    Thanks
    44
    Thanked 9 Times in 7 Posts

    Default

    This format plays nicer with some voices than others, and unfortunately, I've been told mine is kind of rough. Disabling AGC on both my 7550 and 5550 with version 2.4 improved my voice enough that the complaining stopped.

    As soon as I turned on AGC the complaining would start. Mars modified my plugs and doing some testing myself, I could hear a big improvement. I have been running AGC for about two days now with zero complaints.

    I don't know if it is necessary for everyone, because I talk to some with AGC on and they don't sound bad. But for those of us with the type of voice that encodes badly, this is a great mod. Now if he could just mod my wife's voice so it wasn't so grating live.

  30. The Following 2 Users Say Thank You to Msingewald For This Useful Post:

    Forts (Feb 11, 2015),Mars (Feb 11, 2015)

  31. #21
    Join Date
    Apr 01, 2014
    Location
    South East USA
    Posts
    10
    Thanks
    4
    Thanked 25 Times in 6 Posts
    Country: United States

    Default

    Noticeable diffrence in the audio on my SL7550. Nice work.

  32. The Following User Says Thank You to Aegis For This Useful Post:

    Mars (Feb 12, 2015)

  33. #22
    Join Date
    Dec 21, 2011
    Posts
    4,514
    Thanks
    3,789
    Thanked 6,649 Times in 1,896 Posts
    Country: Canada

    Default

    For the benefit of Moto who may be reading this thread, the menu lag some of the members have commented on (which I resolved) is nothing more than the Temporary Message Display Timer. By default, it's 1500ms. That's TOO LONG. 500ms is more realistic.

    Code:
    RDIS_TMPMSGDISTMR 1500

  34. The Following 2 Users Say Thank You to Mars For This Useful Post:

    ka5yth (Jun 28, 2015),kondoros (Feb 12, 2015)

  35. #23
    Join Date
    Dec 21, 2011
    Posts
    4,514
    Thanks
    3,789
    Thanked 6,649 Times in 1,896 Posts
    Country: Canada

    Default

    Alpha and I worked overnight to improve the AGC customizations even more. I believe we improved the original results in the following areas:

    - Analog and digital audio is now slightly louder. The first "fix" padded the audio down just a tiny bit too much. That has been fixed.

    - AGC attack and decay (recovery) times have been sped up. This makes the AGC functionality even more responsive.

    Linear audio tests between factory and modified settings sound almost identical. This was the goal from the beginning; it was AGC functionality which needed improvement. The second goal was to match digital and analog audio levels so you don't have to adjust your volume when switching between analog/digital channels. Many have commented digital is much louder than analog with Motorola's default settings.

    I've attached some audio samples of what the enhancements sound like.

    For the Moto engineers, here are our new optimizations for your consideration/testing:

    Code:
    RC_NUMBEROFCLIPPEDSAMPLES 2
    RC_MINUMUMHWGAINADJUSTMENT 244
    RC_SATURATIONTHRESHOLD 248
    RC_DMICREFPL -29
    RC_AMICREFPL -29
    RC_MICATCKRATE 6
    RC_MICDECAYRATE 4
    To those who sent me codeplugs for modification over the last five days, I will be emailing you codeplugs with these new settings. All new modifications from this point forward will include these new settings.

    Update: Enhanced codeplugs have been emailed to all involved parties.
    Attached Files Attached Files

  36. The Following 5 Users Say Thank You to Mars For This Useful Post:

    com501 (Feb 12, 2015),DC5AJ (Feb 12, 2015),kondoros (Feb 12, 2015),Msingewald (Feb 12, 2015),PSEhub (Feb 13, 2015)

  37. #24
    Join Date
    May 14, 2012
    Posts
    326
    Thanks
    78
    Thanked 197 Times in 91 Posts
    Country: Canada

    Default

    Very nicely done Mars and Alpha!

  38. #25
    Join Date
    Nov 05, 2012
    Posts
    329
    Thanks
    277
    Thanked 234 Times in 145 Posts
    Country: United States

    Default

    The improvement is really significant. Will be doing some testing in the coming weeks in a crowd-noise application. It seems the values Mars uses are really an all around sweet spot so far.

    Good test noise (not as accurate as real thing) https://www.youtube.com/watch?v=Ash7YBU7qLU

    Do these values sound good on RSM's too?