Announcement

Collapse
No announcement yet.

DTA 2.1BT Amp LP Filter Incursion - The Movie

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • DTA 2.1BT Amp LP Filter Incursion - The Movie

    Note: For full size pictures, right click and open in new window.

    Featuring: Millstonemike
    Executive Producer: Kevin K.
    Screenplay: Shenzhen Fly Dream Digital Technology Co., Ltd
    Distributed By: Dayton Audio

    This film is rated PE.

    Tag Line: "A prod, a mod, a prayer to the solder God ..."

    Intelligence

    PE's new DTA-2.1BT amplifier seems ideal for small to mid-sized 2.1 systems. The one glaring negative is the sub's channel's LP filter. The first issue is the filter is fixed. Worse, the purported LP filter's Fc is 180 Hz (PE's follow up to a customer question). That's just too high for many applications.

    Kevin K, following through on his "Make DIY Great Again" agenda, thought there was a story to be told. So he bought the rights, shipped it to me and production began on a filter modification.

    Diversionary Tactics

    This is a rebranded FX-Audio unit produced by Shenzhen Fly Dream Digital Technology Co., Ltd. The "Dayton Audio" branded unit came with the newer version of DTA's PCB: Ver. V4 2017/06/26. Only this newer version is targeted here as there are significant differences between the two versions. Older stock may exist for FX Audio branded units procured from other vendors.

    The unit is attractively styled in its sleek anodized black aluminum case. But contrary to the marketing photos, the unit is narrower, somewhat stouter than depicted.

    Case Aspect Ratio PCB Versions.png

    A superficial read of PE's web page alludes to a 24 V, 4 A power supply included with the unit. Close inspection of the wording specifies a 2 A supply. However the Dayton unit did come with a 24 V, 4 A supply.

    The PCB uses medium density SMT technology to achieve is small size. Build quality is quite good. Power projection comes from TI's popular TPA3116 class-D amp chip. There is a large internal heatsink spanning both the L-R and sub channel chips. The output filter's inductors are optimized for 8 ohm L & R speakers and a 4 ohm sub; though 4 ohm L-R speakers will work fine.

    The unit is dead quiet with open RCA inputs at full volume; tested with three different switch mode power supplies. The potentiometers are indented - 40 clicks for full 270o rotation.

    The unit is turned on via a momentary push button on the face plate. The same button switches between AUX, RCA and BT inputs. Turn on / off is quiet when using this switch. Switching to BT mode produces a simple three tone signal of medium volume - independent of the amps volume setting. The unit is silent when plugging a powered PS, either into the unit or plugging a connected PS into a wall socket. Removing a powered PS from the units DC connector producers a moderately low crackle. It is quiet when removing AC power from a connected PS. BT reception was not tested.

    AT 24 V, the TPA3116D2 will put out 33 W rms into 8 ohms and 61 W into 4 ohms at 1% THD. Though the PTBL sub channel will likely reach near 65 W into a 4 ohm driver at 1% THD owing to it's reduced output impedance. The PCB has reverse polarity protection diodes on the PS input. These will cause a several hundred millivolt drop to the chip so power out will be slightly less than stated above.

    The onboard PS bypass cap is 1800 uf. Not terribly large especially if running 4 ohm L-R speakers and a 4 ohm (or 2 ohm) sub. That type of configuration will likely require a PS with a higher amp rating to handle dynamic peaks, or, beefier outboard PS capacitors.

    Recon

    Initial measurements confirmed PE's stated LP filter Fc somewhat near 180 hz.

    Reverse engineering determined the LP filter is based on the standard Sallen-Key active LP filter topolgy. The filter's component values create a 2nd order Linkwitz-Riley (LR2) filter with an Fc at 159 Hz, Q = 0.5. Thus the signal is down 6 dB at Fc.

    There is a 1st order LP filter created by the sub channel's L-R summing circuit circa 330 Hz. It's Fc is high enough not to materially affect (e.g., lower) the main filter's Fc. But it does steepen the main filters roll off above Fc.

    There is also a 1st order HP filter created by a decoupling capacitor / DC bias resistors leading in to the sub gain stage. This can be seen in the measurements.

    DTA Stock Measurements and Reverse Engineering.png

    Strategic Options

    In the Sallen-Key filter topology, two resistors and two capacitors determine the filter's Fc and Q. The filter's Fc is dependent on both the resistor and capacitor values. The two resistors are typically kept to equal values; 10K ohms for this PCB. Equal value resistors allow a simplification of the filter's complex transfer function to generic algebraic equations for the filter's Fc and Q. With equal value resistors, the filter's Q is only dependent on the capacitor values.

    The chart below shows possible Fc and Q values for commonly available components used in the filter.


    Filter Options.png


    Tactical Planning


    DIY meets modern production technology. The PCB's SMD components are 2.0 mm x 1.25 mm with similar spacing. A SMT hot-air rework station is the tool of choice to swap components on this type of PCB. Of my three soldering irons, none had a tip that would fit between standoff components surrounding the SMD components. Lacking a hot air station, there was only one recourse.

    PCB Size.jpg

    The filter's two resistors are situated between two op-amp sockets - nearly impossible for an iron to get at. However, the filter's two capacitors offered an option. Removing (then replacing) the heatsink and a can capacitor allowed access to the filter's capacitors with a lite duty iron.

    First Strike

    Kevin's target is a desktop 2.1 system with sealed L-R speakers. Thus an LP Fc circa 110 Hz would be appropriate. Using commonly available capacitors, two 0.15 uf (50 V) were inserted replacing C43 and C47. This modified the LP Fc to 106 Hz. Q was kept the same at 0.5 (LR2). Kemet XR7 50 V capacitors were used. Roughly $0.50 each plus ~$4 shipping from Digikey

    DTA LP Filter - Component Callout.png

    Epilog

    My father taught me to solder as a little tyke so I could wire my Lionel trains and slot car sets some 55 years ago. My first iron was a pyramid "hunk of metal" on a long handle that was heated over the stove burner. And I've been soldering ever since; as a professional, hobbyist and handyman. Modifying this board with a standard soldering iron is difficult - not the right tool for the job. Your soldering skills must be excellent to undertake this solution. Or you need to be experienced / have the proper tools for SMD rework.
    Last edited by Millstonemike; 01-30-2018, 05:22 PM.

  • #2
    That's a really good write up for a nice looking little amp. I really liked the way you narrated it!

    Comment


    • #3
      Mike,
      thank you for putting so much time into digging into this for all of us. I also really appreciated the way you presented the information, the drama, the excitement...

      Can I ask a possibly stupid question... in lieu of soldering in tiny replacement caps for one particular crossover point, would it be feasible to solder a few short leads to those spots on the board to allow for easier changing out of caps?

      Say I wanted a 72 Hz crossover point with two .22 uF caps, I could solder those caps to the leads, and if I wanted to change that in the future, just remove, and resolder two new caps on the leads? Or would the leads change the value somehow or be a potential noise issue?

      At any rate, thanks for doing this, I wanted to pick one of these amps up but wanted to see what could be done with it first. You are the man.

      And thanks Kevin for funding the movie. I love that it was rated 'PE'

      TomZ
      *Veneering curves, seams, using heat-lock iron on method *Trimming veneer & tips *Curved Sides glue-up video
      *Part 2 *Gluing multiple curved laminations of HDF *Cello's Speaker Project Page

      *Building the "Micro-B 2.1 Plate Amplifier -- Part 1 * Part 2 * Part 3 * Part 4 * * Part 5 'Review' * -- Assembly Instructions PDF

      Comment


      • #4
        Tom,

        It may work just fine. In fact my original attempt tried to "piggyback" the leads from a small thru-hole cap onto the existing SMD cap to up the capacitance and reduce the filter's Fc. That didn't work for physical reasons (couldn't get the iron in cleanly). So I went with replacing the caps once I realized I needed to remove some stuff to get at them.

        I think you'll have the same problem with leads unless you remove the heat sink and that one can cap next to the sub volume pot. After that, will the leads pick up noise? Don't know?

        And I bought the $8 USB soldering iron I posted about last week to do it. It has a very narrow body - 4 mm from tip all the way to the handle. And only the very end of the tip is heated.

        Comment


        • #5
          The potentiometer that's used for the tone control, is it a stereo potentiometer? And what is its maximum rating? I wonder if it can be reused for the purpose of increasing the R values in the circuit, thus converting it into a filter control rather than a tone control that's likely never to be used...

          ​I was going to mention piggybacking the existing caps, but I see you're already tried that
          Brian Steele
          www.diysubwoofers.org

          Comment


          • #6
            Originally posted by Brian Steele View Post
            The potentiometer that's used for the tone control, is it a stereo potentiometer? And what is its maximum rating? I wonder if it can be reused for the purpose of increasing the R values in the circuit, thus converting it into a filter control rather than a tone control that's likely never to be used...

            ​I was going to mention piggybacking the existing caps, but I see you're already tried that
            A possibility. You would have to cut the traces on the PCB and hope that there aren't any on internal layers. Then figure what value resitors to put in it's place to get a flat response. It's Kevin's amp so I wasn't going in that direction.

            On the other hand, a couple of trim pots replacing the existing two resistors may be a possibility. I discounted that as the pots (and an associated limiting resistor wouldn't fit in the original space. But it may be possible to find a spot using some leads as Tom mentioned. The unit slides out of the case easy enough by removing 5 screws on the back and the three pot nuts on the faceplate for adjustments.

            Or you could do what I did with a fixed LP filter for my 2.1 boom box. Model a passive LP filter in WinPCD (or Xsim) and then find the Fc ad Q of those passive components (I used the Two-Way calculator in the Excel version of PCD to do that). Then find the values needed for the same Fc and Q in the active filter.

            Attached Files

            Comment


            • #7
              Originally posted by Millstonemike View Post
              A possibility. You would have to cut the traces on the PCB and hope that there aren't any on internal layers. Then figure what value resitors to put in it's place to get a flat response. It's Kevin's amp so I wasn't going in that direction.
              ​I'd be tempted to have a closer look. Alas the DTA2.1 that I ordered was apparently delivered with a bad PSU, and PE says I have to send the entire package back and they'll send a replacement once they receive it. To top it all off, my wonderful country has made it much more difficult to send something overseas for repair or replacement. I think the amp was probably shipped back today by my shipper, so I'm not going to see its replacement for at least a week or two.

              ​I like the trim pot approach though. I've expressed it before - I think any filter changes should be internal to the amp, and not effected via an external knob that any idiot can turn

              Brian Steele
              www.diysubwoofers.org

              Comment


              • #8
                Originally posted by Brian Steele View Post

                ​I'd be tempted to have a closer look. Alas the DTA2.1 that I ordered was apparently delivered with a bad PSU, and PE says I have to send the entire package back and they'll send a replacement once they receive it. To top it all off, my wonderful country has made it much more difficult to send something overseas for repair or replacement. I think the amp was probably shipped back today by my shipper, so I'm not going to see its replacement for at least a week or two.

                ​I like the trim pot approach though. I've expressed it before - I think any filter changes should be internal to the amp, and not effected via an external knob that any idiot can turn
                Look at this pic. Follow the red lines from the front, past the pots to th back by the 3.5mm AUX IN port. That is the signal route for the amp and avoids the power switching (as much as possible). Mount two trim pots on the AUX casing. Drill two holes in the back face plate for screwdriver adjustment. Back there there's plenty of room to daisy chain the limiting resistor in series with the trim pots. Maybe multi-turn trim pots may make it easier to keep the resistances equal.

                Or remove the AUX In port (use the RCA IN) and put a regular dual gang pot there with a short shaft and no knob. Hey, at least it's in the back at least

                New Board.png

                Back to the tone control. Remove it, turn it 180o and epoxy it to the PCB. Wire the exposed, upright pins to the filter circuit just behind, and to the left of it. Deftly solder two 25K resistors on the flip-side of the PCB just onto the original thru-holes to balance the tone.

                Comment


                • #9
                  Originally posted by Millstonemike View Post
                  PE's new DTA-2.1BT amplifier seems ideal for small to mid-sized 2.1 systems. The one glaring negative is the sub's channel's LP filter. The first issue [...]
                  This is simply awesome work.

                  Mike, would there be any practical way to mod a high-pass filter onto the L&R speaker outputs? Or would that be best accomplished with an outboard high-pass network between the amp and the speaker?

                  Can you speculate as to why FXAudio/SMSL/Dayton and other manufacturers seem so clueless when it comes to implementing a "proper" (ie, LP'd sub outputs and HP'd speaker outputs) crossover in these mini Class D subs? The passive subwoofer outputs on SMSL's amps (Q5 Pro, AD18, etc) have similar issues (fixed low pass filters at impractical frequencies).

                  So many of these amps are really well designed little units in all other respects, which just makes it bizarre when they flub bass management so badly.

                  Comment


                  • #10
                    Originally posted by JohnBooty View Post

                    This is simply awesome work.

                    Mike, would there be any practical way to mod a high-pass filter onto the L&R speaker outputs? Or would that be best accomplished with an outboard high-pass network between the amp and the speaker?

                    Can you speculate as to why FXAudio/SMSL/Dayton and other manufacturers seem so clueless when it comes to implementing a "proper" (ie, LP'd sub outputs and HP'd speaker outputs) crossover in these mini Class D subs? The passive subwoofer outputs on SMSL's amps (Q5 Pro, AD18, etc) have similar issues (fixed low pass filters at impractical frequencies).

                    So many of these amps are really well designed little units in all other respects, which just makes it bizarre when they flub bass management so badly.
                    Maybe 10 lbs of customer requirements in a 5 lb PCB, no money for real product development marketing and/or they don't expect the mains to need HP. Or just plain poor-clueless engineering. That's my call on the 150 Hz LP Fc.

                    I've identified another bare board that I will evaluate in the next month or two. It has variable sub and a fixed HP option based on a preliminary view. No Bluetooth but that can be added for ~$20 using the Sure BT4.0 w/ APTX module.

                    Comment


                    • #11
                      What's really frustrating is that a lot of these chip amps (and/or their companion chips) actually have this functionality built right in!

                      But, the OEMs building these amps simply don't use/expose this functionality.

                      Check out page 28+29 of the manufacturer's specs for the chip used in the SMSL AD18. Note, this is isn't the chip amp itself. It's a processor chip. $12 at Mouser.

                      http://www.ti.com/lit/ds/symlink/tas5508c.pdf
                      "For 32-kHz to 96-kHz data, the TAS5508C provides 56 biquads across the eight channels (seven per channel)"

                      "From 32-kHz to 96-kHz data, the TAS5508C has four bass and treble tone controls. Each control has a 18-dB control range with selectable corner frequencies and second-order slopes. These controls operate four channel groups"

                      These are all programmable via the I2C protocol in real time. For sure, that is plenty of firepower. With that, you could implement crossover functionality and still have plenty of filters left over for tone controls, loudness compensation, etc.

                      For me it's not a huge problem. I own several MiniDSPs that handle the job just fine. But, dropping $100 a pop on MiniDSPs is not very ideal.

                      Do I need to dive into the world of I2C programming? hmmmm

                      Comment


                      • #12
                        Originally posted by JohnBooty View Post
                        What's really frustrating is that a lot of these chip amps (and/or their companion chips) actually have this functionality built right in!

                        But, the OEMs building these amps simply don't use/expose this functionality.

                        Check out page 28+29 of the manufacturer's specs for the chip used in the SMSL AD18. Note, this is isn't the chip amp itself. It's a processor chip. $12 at Mouser.

                        http://www.ti.com/lit/ds/symlink/tas5508c.pdf
                        "For 32-kHz to 96-kHz data, the TAS5508C provides 56 biquads across the eight channels (seven per channel)"

                        "From 32-kHz to 96-kHz data, the TAS5508C has four bass and treble tone controls. Each control has a 18-dB control range with selectable corner frequencies and second-order slopes. These controls operate four channel groups"

                        These are all programmable via the I2C protocol in real time. For sure, that is plenty of firepower. With that, you could implement crossover functionality and still have plenty of filters left over for tone controls, loudness compensation, etc.

                        For me it's not a huge problem. I own several MiniDSPs that handle the job just fine. But, dropping $100 a pop on MiniDSPs is not very ideal.

                        Do I need to dive into the world of I2C programming? hmmmm
                        It's not the chip that is really expensive. It's everything you need surrounding it to allow end user control.

                        Comment


                        • #13
                          The AD18 already has all the user interface hardware needed. It's got a LED (OLED?) lit screen. By default the big knob controls volume, but you can cycle thru settings like "bass" and "treble" and so forth.

                          SMSL's older Q5 Pro, as well as some of FX Audio's other amps, have a similar screen + settings adjustment interface, though unlike the AD18 don't know which chips those models use.

                          That's why this is so frustrating. Allllll the necessary hardware (both chip level, and user interface level) there on some of these amps. =)

                          Comment


                          • #14
                            Mike,
                            would you mind double checking this for me if you could before I order these.... I think this is right for a roughly 72 Hz crossover point. I'd need two caps to do the swap... I'm ordering a few extra because I'm at least modding two of these units, possibly three. I think I got the size right too, 1.25mm?

                            Click image for larger version

Name:	22 uF Cap Mouser.JPG
Views:	1
Size:	80.8 KB
ID:	1372059


                            Man, finding these parts is a little complicated... it actually took me 45 minutes to locate this believe it or not.

                            TomZ

                            *Veneering curves, seams, using heat-lock iron on method *Trimming veneer & tips *Curved Sides glue-up video
                            *Part 2 *Gluing multiple curved laminations of HDF *Cello's Speaker Project Page

                            *Building the "Micro-B 2.1 Plate Amplifier -- Part 1 * Part 2 * Part 3 * Part 4 * * Part 5 'Review' * -- Assembly Instructions PDF

                            Comment


                            • #15
                              Originally posted by tomzarbo View Post
                              Mike,
                              would you mind double checking this for me if you could before I order these.... I think this is right for a roughly 72 Hz crossover point. I'd need two caps to do the swap... I'm ordering a few extra because I'm at least modding two of these units, possibly three. I think I got the size right too, 1.25mm?

                              [ATTACH=CONFIG]n1372059[/ATTACH]


                              Man, finding these parts is a little complicated... it actually took me 45 minutes to locate this believe it or not.

                              TomZ
                              Tom,

                              You want 2.00 mm (L) x 1.25 mm (W). The way I read your order the cap is 1.25 mm in height (which doesn't matter much).

                              With two 0.22 uf caps, you are going to have a LR2 LP filter at 72 Hz (e.g., Q = 0.5). You may want a steeper roll off. When I did the mod for Kevin, he was thinking small, sealed, desktop speakers with a sub, so the higher LP Fc and shallower roll off seemed like the right way to go. But I did take note of your comment "I can still hear vocals through the sub". While I don't think that will be an issue at 72 Hz, you may want to compare the filter slopes.

                              If you use WinPCD (or xsim?) you can model the LR2 roll off as well as a steeper "near" Butterworth roll off (Q=0.74). Below is the basic difference for the those two filters at 72 Hz as shown in WinISD. Once you settle that, I can pick the caps from Digikey (couple dimes each plus $3.75 USPS shipping). The steeper filter will require a 0.15 uf and a 0.33 uf cap.

                              Lastly, I'm about to mod my second SMD board. But I don't have a hot air station or "Tweezers" soldering iron. It's easy enough to remove the existing SMD caps by heating up one side and then the other while to the first side is still molten with a relgular iron. But that leaves solder on the pads. So it's difficult to get both sides of the new cap flat against the PCB. So I also ordered some copper braid to remove the original solder from the PCB pads. I hope it will be like soldering the new caps to a PCB that's only been fluxed. That cost me $1.50 at DigiKey with my latest cap order.

                              Filter Qs.png

                              Comment

                              Working...
                              X