DXT-MAX - Small active monitors

[silverD]

Hi mga2009,
I agree about the ADC and not wanting to go back to digital after a DAC.
Here is a link to my build of the crossover. I built this before moode came out with the newer kernel that supported multi channel DACs like the Motu M4 and I needed 4 channels for this project. These analog crossovers allowed me to use a 2 channel DAC by offloading the crossover duties as well as the overall gain, tweeter delay, and tweeter gain to the boards. In CamillaDSP I do the BSC and other shaping as needed. The PCBs are very well made, labeled, and utilize through hole components. The crossover frequency is set by the discrete components so changing it later would be difficult. No more miss compared to the ADAU1701!

Edit: A couple of notes on the picture. I have built this into a standard enclosure that will require me to open it to make changes. Once set and closed, the risk of another user making changes is mitigated. The proto board you see is an ESP 48a Electronically Assisted Subwoofer Controller. Since it uses the same power supply as the crossover, I built it in. It sums the signal to mono and provides input for an ICEpower 700as1 which powers a AV123 UFW-12 in a 1.5ft^3 cabinet.

[mga2009]

silverD very nice project! Congatulations!

Looking forward to your comments on the Amplambs's active analog XO board!

I really like active XO, but the main culprit is the ADC. The ADAU1701 ADC is really bad, so it's quite limiting, and there is not much of a choice. Hypex FusionAmps are great, but really expensive.

[silverD]

Update: I've ordered and received the parts to build a 2-way electronic crossover that also allows for time-alignment of the tweeters. Features also include master gain control (can fully utilize DAC volume control), 24dB/oct Linkwitz-Riley high & low pass networks, and volume control on high pass for the tweeter. The XO2W24 PCBs are from Ampslab. This will allow me to have the following audio path: pi running Moode with CamillaDSP for BSC and any other PEQ tweaks -> DS600 DAC -> XO2W24 -> ICEpower amps -> DXT-MAX. I'll see how this compares to the ADAU1701.

[Scarface1]

even if the pre amp overdrive your amp, you can add a limiter in the DSP right?

[ernperkins]

The cheapest miniDSP uses a similar ADAU chip to the DSPB-K which is essentially what I'm currently using but also has opamps on the outputs. The amp powering the tweeter is an Icepower 125asx2 which has a CMoy preamp after the volume pot. I'm looking to utilize a 50asx2SE without preamp or volume in the near future and we'll see if this helps. I have made some changes to the filters, but your suggestion is a good one.

I’ve used several ADAU1701 DSP units (Sure/Wondom/Dayton, miniDSP, 3e Audio and Tinysine). I’ve found a common cause of the hiss problems is not paying attention to matching gain stages. This is especially true for the ADAU1701 chip because it has a few limitations. Looking at the different areas:

1. Input stage. The ADAU1701 can handle either 1 or 2 Vrms on the input, depending on the resistors used on the board. The miniDSP 2 X 4 basic is nice since it is jumper selectable. The basic rule is to use the full input range to reduce the SNR. This is particularly true for ADAU1701 since its ADCs are only rated for a THD+N of -83 dB. Nearly any preamp should be able to deliver 1 or 2 Vrms, but some Bluetooth modules and phones have notoriously low outputs.
2. Output stage. The ADAU1701 DACs can output 0.9 Vrms before clipping. The miniDSP 2 X 4 basic may have op-amps on the output, but is still limited to 0.9 Vrms maximum according to its datasheet. MiniDSP used to have a balanced version with a higher output, but that has been dropped. We want an amp with a gain so that 0.9 Vrms drives it to near clipping, but not over. With too much amp gain you can’t use the full DAC output (higher SNR) and with too little gain you lose available amplifier output. See the ICEpower amp analysis below.
3. Filter stage. The overall filter gain needs to be 0 dB if we’re using the full input range. For example, let’s say we’re using a Linkwitz transform with 7 dB of gain to extend the bass roll off. That means the signal needs to be attenuated 7 dB before the DACs to avoid clipping. The ADAU1701 does have an interesting filter processing approach: after the ADCs 4 bits are added to allow an additional 24 dB of headroom during filter processing. But those 4 bits are removed just before the DAC stage. The takeaway is you have to aware of the filter gain and attenuate accordingly (see Linkwitz’s analysis: https://linkwitzlab.com/LXmini/Design.htm).

Paying close attention to the various gain stages should minimize hiss problems and eliminate digital clipping. No guarantee to completely remove hiss (don’t forget about possible ground loops, poor power supplies, etc.), but it’s a step in the right direction. And I personally don’t put any volume control in the DSP, I always leave it in the preamp driving the DSP.

Here’s a quick output stage analysis I did using the ICEpower amps you mentioned. The good news is the ADAU1701 won’t overdrive those amps. The flip side is you’re losing about 3 dB of available amp output.

ICEpower 50ASX2:
SE (stereo) version: 2 X 47W @ 1% THD+N, 20 Hz – 20 KHz, 4 Ohm load
SE version gain = 20.5 dB (= 10.6 voltage gain). Output @ 0.9 V input = 22.75 W (4 Ohm load)
BTL (mono) version: 1 X 170W @ 1% THD+N, 20 Hz – 20 KHz, 4 Ohm load
BTL version gain = 27 dB (= 22.4 voltage gain). Output @ 0.9 V input = 101.5 W (4 Ohm load)

ICEpower 50ASX2:
SE (stereo) version: 2 X 120W @ 1% THD+N, 20 Hz – 20 KHz, 4 Ohm load
SE version gain = 24.8 dB (= 17.4 voltage gain). Output @ 0.9 V input = 61.2 W (4 Ohm load)
BTL (mono) version: 1 X 450W @ 1% THD+N, 20 Hz – 20 KHz, 4 Ohm load
BTL version gain = 30.7 dB (= 34.3 voltage gain). Output @ 0.9 V input = 237.9 W (4 Ohm load)

[silverD]

Hi a4eaudio,
https://sbacoustics.com/product/5x8in-sb15sfcr-00/ shows +/- 11 mm xmech.
To me honest, I haven't had much time play with these. I have a 2 year old and I mentor a high school robotics team and we are in our build season. I'm no expert, but I would recommend them for a small 2-way build. Being able to use some sort DSP to boost the lower end (to reach the 30's) and take advantage of the excursion is a nice plus. Javad built Project Sledgehammer which uses them in a MTM.

[a4eaudio]

Tom, I'd have to pull up the WinISD model on other computer, but I think at 70W/100dB both the woofer and PR (with 30g mass added) are under xmax with a HPF at 20Hz.

I'm so sorry, I modeled the PR with an older Esoteric woofer, not the Epique 5 1/2" sub, my mistake!

Tom, SilverD - did you guys find the xmax for the SB passive radiator somewhere? I can't find it, but I'm thinking that twice the xmax of the matching woofer would make sense.

SilverD - Now that you've had some more time to listen to these, did you make any more tweaks? Regardless of tweeter, any reservations on recommending the E150HE for small two-way builds?

[silverD]

The cheapest miniDSP uses a similar ADAU chip to the DSPB-K which is essentially what I'm currently using but also has opamps on the outputs. The amp powering the tweeter is an Icepower 125asx2 which has a CMoy preamp after the volume pot. I'm looking to utilize a 50asx2SE without preamp or volume in the near future and we'll see if this helps. I have made some changes to the filters, but your suggestion is a good one.

[Scarface1]

Hi Scarface,
Besides the tweaks above, I tinkered with some filters for the tweeter. I'm still learning to take measurements and I don't have a dedicated space to take them. Throw in the fact I have a two year old and it makes it difficult to get space and time to tinker. I think they sound great, but do not have a trained ear. I do notice more hiss than I would like out of the tweeter vs a passive speaker, so I'm looking to change up my DSP a bit. I was running a chromecast audio -> optical -> SMSL SK 10th -> analog -> DSP board (ADC -> DSP -> DAC) -> amps so I thought removing the extra ADC/DAC would help. I connected a pi running Moode to the DSP board via i2s and while it helped some, there is still some hiss. I was looking at the miniDSP nanoDIGI, but they're are no longer being made.

I want to build something with the e150 and a dsp amp myself so i'm really curious how it turn out for you. You can buy pre-amp DSP boards like the miniDSP 4x2 or a dayton DSPB-K ( this one is cheap but if it do the job i wanna use it because it's small AND cheaper )

Have you tried a simple filter or no filter at all for the tweeter, maybe it's not the amp or other equipement but just the filter in dsp. I had some issues with the sound while using dsp and when i re-write the same dsp filters on the chip it got away.

[silverD]

Hi Scarface,
Besides the tweaks above, I tinkered with some filters for the tweeter. I'm still learning to take measurements and I don't have a dedicated space to take them. Throw in the fact I have a two year old and it makes it difficult to get space and time to tinker. I think they sound great, but do not have a trained ear. I do notice more hiss than I would like out of the tweeter vs a passive speaker, so I'm looking to change up my DSP a bit. I was running a chromecast audio -> optical -> SMSL SK 10th -> analog -> DSP board (ADC -> DSP -> DAC) -> amps so I thought removing the extra ADC/DAC would help. I connected a pi running Moode to the DSP board via i2s and while it helped some, there is still some hiss. I was looking at the miniDSP nanoDIGI, but they're are no longer being made.

[Scarface1]

Is there any progress made the last weeks? Really curious how they sound

[tomzarbo]

Thanks for the comments everyone

Tom, I'd have to pull up the WinISD model on other computer, but I think at 70W/100dB both the woofer and PR (with 30g mass added) are under xmax with a HPF at 20Hz.

I'm so sorry, I modeled the PR with an older Esoteric woofer, not the Epique 5 1/2" sub, my mistake!
Just wanted to clear this up to avoid any future confusion.

TomZ

[silverD]

The extra low end boost helped fill in the lowest notes. The healthy xmax allows for this type of EQ with DSP. So far the listening has been in a small room near field and these provide plenty of bass. I need to look at the overall FR and adjust accordingly, but they sound good so far with great imaging.

[Steve Lee]

^ This is what I really like about DSP.

I still think passives are the best solution in the long run though . . .

[silverD]

Changed the x-over slope from LR8-48dB to LR4-24dB to match that of the DXT-MON and added a LF PEQ @ 30Hz: