Once this whole thing is set up, it WILL be simple. The analog signal, once it is generated, will have far fewer components in the signal path than most systems - no preamp, and no passive crossover in the speaker.

Yup it might seem awfully complex but as far as analog signal goes it's the simplest bar none!

Sent from my iPhone using Tapatalk

Hi Keith

 

In case I have missed something what is your ultimate objectives with the DSP ?

 

Sometimes to much DSP processing can be a curse according to this white paper from Grimm Audio http://www.grimmaudio.com/site/assets/files/1088/speakers.pdf

 

cheers

 

Yup, read that paper some time ago. My MAIN objective is to linearize the frequency response of the drivers and time align them. The bass will have some room correction, and I will leave the mids and high frequencies alone. As I learn more about DSP, I might have other ideas. Acourate seems to be a vast toolbox of things that I don't understand. 

 

Maybe because with all the knowledge we've built up regarding human biology we can presume there's zero audible improvement. Hifi like any software only requires a certain amount of processing power. That said there are plenty of DSP algorithms we can apply, but as far as needing a modern GPU it seems like overkill gone mad.

 

I am a guy who likes overkill gone mad. Why else would I have a 110W valve amplifier driving only the midrange horn which has a sensitivity of 98dB/W?

 

I'll bet you that those guys who use those GPU's for what they were intended for can't tell the difference between 100fps and 120fps either. But that won't stop them from buying an even more super duper card for more fps. 

 

And besides ... i've heard this argument before. Back in the dim days when Nikon's flagship camera (D2H) only had 4MP and Canon's flagship (1DS Mk.1) had 11MP, Nikon fanboys were all over the forums telling everyone that 4MP is all anyone needs unless they were printing posters. Oh yeah, they were also telling everyone that nobody needs full frame (of course, Nikon did not offer a full frame camera at the time). I wonder if they look back at their old posts and cringe. At least i'm consistent, 10 years ago I thought that 10MP was all I would ever need. It is 2016, and I deliberately bought a 12MP camera even though I could have bought the 42MP version for the same price. 

 

Computing power is cheap. If you used a cheaper DAC like the Exasound E28 (and not the NADAC), the cost of this setup would be lower than the DEQX. And it is more flexible. And much more powerful. And can do things like stream internet radio, act as a transport, and store porn. 

Edited August 16, 20169 yr by Keith_W

A couple of quick points:

A) the computing power is, presumably required largely du to the plan to convert PCM to 256 DSD. (I'm doing 64k FIR filters for a 3-way system with a lot less processing power

B) I guess, Keith, that one way of thinking about the challenge is to 1) get a PC source to perform as close as possible to the Playback Designs and 2) sort out the best possible filters based on measurements. The first theme sounds like the focus given that the RME hasn't arrived (and fact that the reported noises etc from the PC sound very unusual) Also, if you're using HQPlayer then there is the option of using a low power NAA device with the powerful PC removed from the stack but on the network

C) I wouldn't worry about the cost of the RME as I'm guessing that it'll be in the system for quite some time as the measurement process is iterative

Yes Zydeco, (A) is correct. If the studios and audio professionals think they should work in PCM192 or DSD256 when making their mixes - why shouldn't I do the same at home? Studios that use Merging Pyramix software work in DSD256 or higher. Sadly, the files may not be available in DSD256 or PCM192, although there is a limited selection. 

 

I am not sure if (B) can ever be achieved - given that I ultimately think that the NADAC is a downgrade from the Playback Designs. As I have proven to myself with the listening tests, the MPS-5 by itself is massively superior to the PC, even with the PC using the MPS-5 as a DAC. The downside is - no convolution, no DSP, and no ability to make it multichannel. I am buying the NADAC because there is no other cost-effective solution on the market, and the NADAC is probably the best I can get without spending silly money - like buying 8 channels of DCS D-A conversion (now that is absurd money!!!). 

 

There are a number of things that make me go "hmmmmmm" about the NADAC - (1) it uses an SMPS, (2) the digital and analog circuitry is on the same board, (3) it uses a Sabre ESS9008 DAC (and not the higher end 9018), let alone the discrete FPGA's of the Playback Designs! (4) no isolation between HF generating circuitry and analog. As Tricka said - it's pro audio. They don't think these things matter. Maybe they're right - I am prepared to give them the benefit of the doubt. But all these years of "audiophile" training has more or less told me that all these things are no-no's. 

 

As for © you are probably correct. I will eventually forget how much I paid for it. I hope. 

There are a number of things that make me go "hmmmmmm" about the NADAC - (1) it uses an SMPS ...

 

It's possible to 'tame' a SMPS, Keith.  In the experiments with different PSs for the speed controller on my TT which we did last year, a 48v Meanwell SMPS + isolating transformer + hash filter came out on top.  AIUI, this is bcoz the speed of a SMPS is a good thing for a digital device.  (This is for a speed controller, mind you - which doesn't process the signal!   ).

 

The order of SQ coming from the speakers was (going down in SQ):

1. Meanwell + isotran + hash filter
2. Linear PS + isotran + hash filter
3. SLA battery supply
4. Linear PS
5. Meanwell.

 

Andy

Edited August 16, 20169 yr by andyr

Andy, my new linear PSU will hopefully arrive before Friday. Would you like to come over this weekend and have a listen? Bring your hash filter along? 

Andy, my new linear PSU will hopefully arrive before Friday. Would you like to come over this weekend and have a listen? Bring your hash filter along? 

 

That would be very interesting, Keith - and I am available this weekend.     This is your linear supply for your computer, right?

 

You should have my email address - if not, please PM me your suggested time.

 

I'd also be interested in being told exactly what the NADAC is.  AIUI, it's an 8-channel DAC - which is what i could use as a substitute for the 8 output DACs in my new miniDSP unit (to improve SQ).

 

 

Thanks,

Andy

Edited August 16, 20169 yr by andyr

I'll bet you that those guys who use those GPU's for what they were intended for can't tell the difference between 100fps and 120fps either. But that won't stop them from buying an even more super duper card for more fps.

And besides ... i've heard this argument before. Back in the dim days when Nikon's flagship camera (D2H) only had 4MP and Canon's flagship (1DS Mk.1) had 11MP, Nikon fanboys were all over the forums telling everyone that 4MP is all anyone needs unless they were printing posters. Oh yeah, they were also telling everyone that nobody needs full frame (of course, Nikon did not offer a full frame camera at the time). I wonder if they look back at their old posts and cringe. At least i'm consistent, 10 years ago I thought that 10MP was all I would ever need. It is 2016, and I deliberately bought a 12MP camera even though I could have bought the 42MP version for the same price.

Computing power is cheap. If you used a cheaper DAC like the Exasound E28 (and not the NADAC), the cost of this setup would be lower than the DEQX. And it is more flexible. And much more powerful. And can do things like stream internet radio, act as a transport, and store porn.

Does any audio software available use GPU's? I presume the answer is no.

Comparing video to audio in regard to computing power requirements is kinda interesting, but seemingly irrelevant at the same time.

DEQX is not cheap, but may provide a better result and far less grey hairs because of the support available. I use a HTPC for all audio and video playback which does significant DSP before transmitting to my DEQX.

Edited August 17, 20169 yr by Satanica

Just out of interest how much is a NADAC worth these days ? I presume there is a local distributor for it ?

 

cheers

Does any audio software available use GPU's? I presume the answer is no.

Comparing video to audio in regard to computing power requirements is kinda interesting, but seemingly irrelevant at the same time.

DEQX is not cheap, but may provide a better result and far less grey hairs because of the support available. I use a HTPC for all audio and video playback which does significant DSP before transmitting to my DEQX.

 

Hi Satanica, Signalyst HQPlayer is able to offload processing to NVidia GPU's. This is the software that I am using for playback, convolution, and generation of crossovers. 

 

Since you are on this thread, I have a question for you. If you are using a HTPC for audio and video playback, AND sending the audio to the DEQX (which introduces a latency of about 20-30ms) ... how do you deal with the lip syncing problem? I have been scratching my head at how I am going to solve this when the time comes, and I have to admit I have no idea. 

 

Just out of interest how much is a NADAC worth these days ? I presume there is a local distributor for it ?

 

cheers

 

I think the NADAC list price for the 8 channel version is about $17-18k. Might be a little cheaper since the AUD$ strengthened against the USD$. And yes, there is a local distributor - ATT Audio Controls. They are pro audio and very pleasant to deal with. I happened to meet the owner of ATT at the SNA hifi show. Hugh Dean (AKSA) introduced us. 

Since you are on this thread, I have a question for you. If you are using a HTPC for audio and video playback, AND sending the audio to the DEQX (which introduces a latency of about 20-30ms) ... how do you deal with the lip syncing problem? I have been scratching my head at how I am going to solve this when the time comes, and I have to admit I have no idea.

Hi again, most video playback software seem to have some sort of A/V Sync Correction. I use JRiver MC and it allows both positive and negative numbers, others might too. Displays and projectors seem to introduce at least about 30ms of delay on their own due to their video processing. My JVC projector introduces about 130ms and my madVR video settings from JRiver MC introduce even more, so I need about 300ms of audio delay set in JRiver MC.

Thanks, Satanica. After I read your post, I bought a JRiver license. But I have been a little too busy to play with it due to the arrival of a couple of new toys: my RME Fireface UC and Uptone Audio JS-2 Linear PSU. 

 

I bought the JS-2 LPS from SNA member @JDWest who runs Sublime Hi-Fi. Uptone owner Alex Crespi gave me his contact details and suggested I speak to him. After taking into account the difference in currency, shipping, and GST - the price comes out to be the same as what it would cost to get the unit shipped from Uptone themselves. 

 

I have read that some people do not believe in linear PSU's. Well, all I can say is - WOW what a difference. AndyR came over today with his hash filter hoping to fix the clicks and pops that I was complaining of. He listened for a while and said, "you mentioned you had a problem with computer noise. Where is it?". I told him it's fixed. By the PSU. 

 

Since he brought his hash filter, we replaced the Uptone with the SMPS that came with the PC. The clicks and pops came back. The hash filter did a good job of suppressing these, but they were still audible. 

 

We then reinstalled the Uptone, but this time with the hash filter in place. At first, I could hear no difference. But when the music went quiet, I could hear that the noise floor was much lower. I commented on this to Andy - he agreed. 

 

Since Andy was here, I repeated all the experiments I conducted above. Andy heard exactly the same things that I did.

 

I came into digital audio with no prejudice. If there was bias, I am biased towards finding that I have not wasted my money. Unfortunately, honest listening tells me that in some ways I took a step back, and I have to do some work to claw the quality back. The Uptone JS-2 goes a long way in improving the sound of the PC. On its own, it massively reduced the noise floor of the PC. It is now nearly as quiet as the Playback MPS-5. With the Uptone, instrument separation was much improved, clarity was heaps better, and the midrange is no longer smeared. 

 

There are still a number of upgrades in the pipeline: 

 

- waiting for the NADAC to arrive. Once it arrives, the analog active crossover will be removed from the system. Using the NADAC as my active crossover, I will then be able to implement a few advanced functions of Acourate. This will hopefully yield more gains.

- the PC is still not optimized. I have not yet installed Audiophile Optimizer (i.e. turn services off), nor have I done things like turn off the wifi on the PC. 

- the Uptone is the only "audiophile" hardware upgrade I have done so far. There are more to come. 

 

... once all these are in place, I will hopefully begin to exceed what I had with my old system. The fact that the Uptone made such a difference was VERY encouraging. If I had listened to the naysayers, I wouldn't have made this $1500 leap of faith. @JDWest offered a money back guarantee if I wasn't happy. Well, the money is yours!! I might even buy another one for the NADAC. 

 

As for the RME Fireface UC, this is essentially a fancy measurement device, similar to what I was using before, a Focusrite Scarlett 2i2. The difference is - the RME has 8 DAC channels and thus capable of 8 channels of analog output. Instead of the NADAC, I could use the RME and implement a digital active crossover right now! I have all the equipment I need to do it. 

 

After Andy left, I spent a few hours trying to get this thing to work, and it was an exercise in frustration and plenty of time consulting the manual. Compared to the Focusrite, the RME was exceedingly difficult to set up. With the Focusrite, you install the drivers, plug in the machine, and all the inputs/outputs show up nicely in the ASIO device list. 

 

Not so the RME. First, you have to set the firmware to "PC" (out of the factory, it defaults to Apple mode). Install the drivers, and nothing shows up. Quick check, and the RME has decided to set itself back to Apple mode. I hate Apple with a passion, and I reckon the RME did it deliberately to piss me off. Set it back to PC, and reboot the computer. 

 

At the back of the RME are six analog outputs labelled ANALOG 1 2 3 4 5 6. Well, guess what devices show up on ASIO? Analog (3,4), Analog (5,6) and Analog (7,8). What happened to Analog (1,2)??? After tearing my hair out reading the manual, it turns out that Analog (1,2) is actually "RME Fireface Speakers". 

 

Then I could not get the microphone interface to work. It just refused to supply phantom power to the mic. After consulting the manual, I found an obscure entry telling me to open the Mixer software, which looks like this: 

 

 

Find the microphone panel, then click on the tiny settings icon. In there you will find a button to enable Phantom power. 

 

Microphone still doesn't work!!! ARGHHHHHHHHHH!!!!!!!

 

Then I found another entry that said that by default, the microphone input routes nowhere. I have to manually route the mic input to one of the analog inputs using their mixer software. So once again, I open up that intimidating looking panel, find the setting, and voila. Microphone now works. 

 

Welcome to the world of pro audio. Not exactly friendly to us amateurs. 

 

After a few hours of futzing around, I was able to take my first REW measurement and generate new filters with Acourate. It sounds good! 

 

Tomorrow i'll have a go at creating a multichannel setup. But for now it's time for bed. 

 

Keith

 

Yes, the TotalMix interface is daunting but powerful once the concepts make sense. There are some good instruction videos on youtube https://www.youtube.com/watch?v=4M23E1Njrc4. I'd suggest removing the analogue x/o and using the RME as both a measurement and playback tool instead of waiting for the NADAC. My experience is that it took a bit of time to be able to get the process of measurements / analysis / deployment sorted so the earlier that you start the better. A

 

Also, whilst I remember, under the outputs there are little tool-bars (similar to that you used to set the mic phantom power) that are used to set the voltage level of the analogue outputs and, to avoid clipping, you'll want to get this aligned with the NADAC or power amplifiers.). And the "matrix" view (as opposed to "mixer" view in your screen shot) is also useful in seeing the channel routing.

 

Zydeco

 

P.S. Does the NADAC have a "mute" function? The reason for the question is that it probably makes sense, ultimately, to have some form of back-stop should the computer do something odd while the amplifiers are powered on so as to avoid damage to the drivers

Thanks for the tips, Zydeco. This morning I am staring at my coffee, wondering if I have the determination to go through this exercise again. I did try listening to music through the RME last night. Of course it was not as good as the Playback Designs, but the real deal killer is the intermittent tics I hear through the RME (no such problem with the Playback MPS-5 since the Uptone LPSU was installed). I suspect that I need to increase the buffer size to get rid of the tics. 

 

I don't know if the NADAC has a mute function. But there is always the power switch! 

I admire your patience Keith. An I thought adding a new amp to my active system has taken a lot of effort...

Sent from my iPad using Tapatalk

Strange - I've never had these ticks. Are you using USB between the PC and RME (as opposed to the PCIe sound card)? If it helps I could take a picture of my settings latter this morning.

It's OK zydeco, don't worry about it. I decided that I am not in the mood for more audiophile hair tearing this morning, so i'm off to the markets to buy something nice for dinner. 

Just checked and I'm using 1024 sample buffer which is working fine with multi-channel USB

Re: digital ticks / clicks: have you installed a PCIe USB card into the PC? The reason is that I found that a SoTM PCIe card was incompatible either RME Fireface. And, just to be clear, have you installed the RME USB driver? (You should see a little fire like icon in the system tray)

I don't have a PCIe USB card. At the moment, I am feeding the Fireface and all USB devices with motherboard USB. I do not intend to buy a PCIe USB card, because the PC will interface with the NADAC via Ethernet. 

 

And yes, I do have the RME USB driver installed. 

When I changed from one USB to SPDIF device to another (Musical Fidelity V-Link to a Musical Fidelity V-Link 192) all of my pops and clicks on the DEQX went away, wonderful!

Last night I removed the preamp and analog crossover from the signal chain. Right now it is the RME providing 6 channels of DAC, and potentially 8 channels if I feel like it (only a matter of going out to buy the cable). I took new measurements with Acourate, generated the digital crossovers, and had an absolute beast of a time trying to install the crossovers into HQPlayer. After emailing the author of HQPlayer, the digital active crossover is now up and running! 

 

For those who want to know how to implement digital active crossovers in HQPlayer, this is how you do it. 

 

1. Generate mono .WAV filters in Acourate as per this guide. You need one per channel. 

 

2. Set HQPlayer to 5.1 mode. 

 

3. Plug your cables in using this convention: 

    - Left: Channel 1 to left subwoofer* 

    - Right: Channel 2 to right subwoofer* 

    - Center: Channel 3 to left mid

    - LFE: Channel 4 to right mid

    - Left back: Channel 5 to left high

    - Right back: Channel 6 to right high

    * In my case, because I only have one subwoofer (for now), I plugged both cables into the JL Audio sub and let it do the summation. 

 

4. Install the convolution filters using the Matrix Pipeline setup (columns Channel In, Mix Channel, and Filter). Acourate users will know the file naming convention of Acourate - "Cor1L192.wav" means, "Corrected Filter, Left, 1st channel, 192,000Hz": 

    - 1 - 1 - Cor1L192.wav

    - 2 - 2 - Cor1R192.wav

    - 1 - 3 - Cor2L192.wav

    - 2 - 4 - Cor2R192.wav

    - 1 - 5 - Cor3L192.wav

    - 2 - 6 - Cor3R192.wav

 

5. Using the Speaker Setup feature of HQPlayer, verify (AT LOW VOLUME!) that the channel assignments are correct. Once done, switch off Speaker Setup. 

 

6. Sit back and listen to music! 

 

So, what does it sound like? Note that I am using the RME instead of the Playback Designs (i.e. I am comparing a $200 per channel DAC vs. a $8000 per channel DAC). There are many aspects of the sound which are not as good. The noise floor is higher. The sound is more edgy. It sounds more rough than the Playback. Yet, there are many aspects of the sound which are massively improved. Transients hit with real impact, and sound is more coherent. 

 

Since the arrival of my Uptone JS-2, the sound from the computer is quite acceptable. I might be able to wring out more performance from it. I should start hanging around that "Building the ideal-ish music server" thread. That thread seems to be full of nutjobs. Can I join the club? Suggestions for tweaks for the PC are welcome! 

 

The RME is not capable of decoding DSD, so I am sending it 192kHz PCM. Out of curiosity, I opened up task manager to see how much CPU was being consumed by upsampling six channels to 192kHz. Only 2.5%!

 

DSD is supposed to be more CPU intensive than PCM, so when it comes time to use the NADAC, it should probably be able to handle it. And since I have CPU power to spare, I might turn on a few of the advanced algorithms of HQPlayer as well. 

 

I also compared HQPlayer to JRiver (I have bought licenses for both). No contest, HQPlayer wins hands down. It just sounds more refined and relaxed. 

 

Thus far I have climbed only the first couple of rungs up the steep Acourate learning curve. Next step: driver linearization. Then time alignment. Then i'll try the quasi-anechoic measurement technique via beamforming. 

 

And when my NADAC and new subwoofers arrive, I will do it all over again. 

So is the RME like $1600?

Sent from my iPad using Tapatalk

Yes. The RME is $1520 - $1599 depending on where you buy. 

@@Keith_W you must be a very patient man who is on a mission to listening nirvana. Enjoying your posts but have to say I would have thrown in the towel ages ago.