• Disclaimer: This Vendor's Forum is a paid-for commercial area. Unlike the rest of diyAudio, the Vendor has complete control of what may or may not be posted in this forum. If you wish to discuss technical matters outside the bounds of what is permitted by the Vendor, please use the non-commercial areas of diyAudio to do so.

exaU2I - Multi-Channel Asynchronous USB to I2S Interface

The only event that will push price down is more production quantity. Making one custom electronic device is incredibly expensive. You have to make at least 100 to get the first discounts, but usually it's not enough to reach a 'fair' retail price until you make 1000. I doubt that there are even 100 people willing to build their own DAC. I could be wrong about the number of people interested, but I'm not wrong about the costs. Making 100 different products doesn't save any money - you have to have 100 people agree on the same platform and then you get discounted prices.

Quality might be pushed up by competition, especially with a forum like diyAudio to discuss what is learned by each new attempt.

Your talking about " incredibly expensive" why do you think so ?
It is extremely time consuming but if you do not pay for the time (say you do this for a hobby not for a living) then I'm not seeing these extreme costs.
Most electronics is for sale for a low price. :rolleyes:
 
Your talking about " incredibly expensive" why do you think so ?
It is extremely time consuming but if you do not pay for the time (say you do this for a hobby not for a living) then I'm not seeing these extreme costs.
Most electronics is for sale for a low price. :rolleyes:

Hmm I don't know of many designers that give all of their work away for free, sure some do but not all. Time is money to most people. You can break most things down to component value and find the elemental cost is not reflective of the global product.
 
Hmm I don't know of many designers that give all of their work away for free, sure some do but not all. Time is money to most people. You can break most things down to component value and find the elemental cost is not reflective of the global product.

Ofcourse! But then it is not extremely expensive to do/make. The price you pay for it is profit, not the value of the electronics.
 
Update! AK4396 not only plays 352.8 files, it also plays 384 Khz sampling rates!

I've upsampled a 192k file to 384K and I get a hiss free playback.

No the interesting thing is that upsampling a 352.8k DXD file (the ones with hiss) to 384k lowers the hiss by roughly a half.

I've researched a bit regarding the DXD files and their recording is quite different than the standard in that they are having a much more shallower anti-alias filter than the downsampled files. That is needed for DSD conversion.

So I believe I am actually hearing some kind of passband noise, or better said it's effects. Now I don't know why this isn't happening when playing these files with the ESS DAC, but there could be three reasons: a) the ess chip does resampling and this affects this noise, b) the digital filters in the ess are working differently, c) a result of the interaction of the first two features.
 
Your talking about " incredibly expensive" why do you think so ?
Why do I think it's "incredibly expensive?" ... because I actually know how much it costs to make electronics like this! I do this sort of work professionally, and I have seen the numbers on every aspect of manufacturing.

Have you ever designed a piece of electronics? Have you ever manufactured more than one of the same circuit?

I have, and I can say that every single element of manufacturing costs more when you make fewer units. The only cost which is not discounted in quantity is manual labor, such as hand-assembly. But that's not much of a consolation since manual labor is already the most expensive aspect. In this case, I don't think there is much manual labor, so the costs are very dependent upon quantity.

Step One: Order circuit boards. Typically, you cannot order anything professionally without spending $300 to $400 minimum. That means if you need one PCB, it costs $400. You can get that down to $50 with a prototyping outfit, but quality suffers. Sure, DIY folks can order a PCB for as little as $2 per square inch, or even make their own PCB by hand for almost $0, but I can assure you that the quality is not worthy of a product at those prices. exaDevices is surely paying professional rates for PCB manufacturing.

Step Two: Parts ordering. As I mentioned earlier, some chips are not even available in quantities under 1000, and if they are then the cost is higher for small quantities. Even buying 100 of a chip is not enough to get the same discounts that brand name electronics manufacturers have access to. The exact same circuit board can cost several times extra depending upon how many you make.

Step Three: Assembly is handled by robots called a pick-and-place machine, but setup time is expensive. Tape reels with thousands of components must be manually loaded into each individual slot, and the computer files must be manually adjusted to create a functioning board - this never works right the first time, and the failures cost money, too, so the assembly shop charges a setup fee. It costs a great deal of money to buy 1000 capacitors when you only need 1 per board, and only have 100 units to make. Assembly shops might share a reel of 1000 capacitors among several customers if the values are common, but interesting electronics always have at least one chip that is unique to your product and not on any other customer's boards. So, the cost of buying 1000 of these chips is immense, and must be absorbed into the unit cost. Also, the assembly shop charges for the setup time and the process they go through to tweak the pick-and-place until it can pop out a functioning circuit. All of the above is way more expensive when you start the whole process only to stop short of 100 or 1000 units. Nobody sells 10000 DIY kits.

Step Four: Finishing touches. Here is where you might expect a discount, but don't count on it. Custom enclosures have the same overhead as everything above. The exaU2I might not have an enclosure, but that doesn't mean it can be much cheaper. Besides, there's still the process of boxing the product and shipping it out to customers. Admittedly, that's not a major component of the price, but don't expect to save hundreds.

The only hope to get prices down is to manufacture more units at one time, because doing 100 here and 100 there doesn't save any money. Mr. exa would have to take advance order payments from 1000 people to get the prices down significantly.

It is extremely time consuming but if you do not pay for the time (say you do this for a hobby not for a living) then I'm not seeing these extreme costs.
Mr. exa is not getting any money for his time. My comments have nothing to do with getting paid for design time. I am talking purely about manufacturing costs of professional PCB and parts assembly.

Most electronics is for sale for a low price. :rolleyes:
I'll ignore the bad grammar and assume this is a joke. Do you understand the difference between mass-produced consumer electronics versus very small production runs? Most electronics that you see for cheap prices are manufactured in amounts of hundreds of thousands of units, if not millions. The price overhead for each unit is tiny when divided by one million. As I said, exaDevices is probably lucky if 100 people buy this thing, so unit costs are much higher.
 
Update! AK4396 not only plays 352.8 files, it also plays 384 Khz sampling rates!

I've upsampled a 192k file to 384K and I get a hiss free playback.

No the interesting thing is that upsampling a 352.8k DXD file (the ones with hiss) to 384k lowers the hiss by roughly a half.

I've researched a bit regarding the DXD files and their recording is quite different than the standard in that they are having a much more shallower anti-alias filter than the downsampled files. That is needed for DSD conversion.

So I believe I am actually hearing some kind of passband noise, or better said it's effects. Now I don't know why this isn't happening when playing these files with the ESS DAC, but there could be three reasons: a) the ess chip does resampling and this affects this noise, b) the digital filters in the ess are working differently, c) a result of the interaction of the first two features.

I start to believe that your DAC are performing some kind of downsampling like clocking in only every second sample or something like that :confused:

With a ESS DAC there are first a upsampling circuit with a digital filter before the jitter removal resampler etc...
The ESS DAC can disable the upsampling circuit and digital filter for use with other external upsampling filters..

I can play the DXD files with and without the upsampling filter..
The ONLY difference are the "negative" sonic effect of the upsampling filter.

I have upsampled DXD 352.8k/24bit files to 384k/32bit files without dither (truncate) for test purposes.
These files plays equally good - and without any noises or hiss...

As you tries to use your DAC at unsupported samplerates I would have asked the developers of the chip for an explanation to why/how the DAC chip "plays" 352.8k and 384k files, and what issues this may cause - like hiss etc..
 
With a ESS DAC there are first a upsampling circuit with a digital filter before the jitter removal resampler etc...
The ESS DAC can disable the upsampling circuit and digital filter for use with other external upsampling filters..

I can play the DXD files with and without the upsampling filter..
The ONLY difference are the "negative" sonic effect of the upsampling filter.

This apparently refutes my little theory on why you can play the DXD files.

I start to believe that your DAC are performing some kind of downsampling like clocking in only every second sample or something like that

Well this might be true, although I don't see how's that possible. Furthermore none of our explanations do something regarding the fact that I can play 384 and 352.8 files without hiss if these files are upsampled and why the DXD files (about which we now know that have a different noise content than other files) are affecting the hardware like this.

From my point of view, this whole thing is definitely linked to the specific content being played and not only the hardware.

As you tries to use your DAC at unsupported samplerates I would have asked the developers of the chip for an explanation to why/how the DAC chip "plays" 352.8k and 384k files, and what issues this may cause - like hiss etc..

Yes, I should contact their technical support. Maybe they would help with some valuable input.

By the way, anyone here tried to use the exa device with a non-upsampling DAC (Ad1955, PCM1794, wolfson parts, Cirrus-Logic)?
 
hi,

can we implement digital xo on 8ch for mutliamp speakers? just like we can do in APOGEE ENSEMBLE.

if using exa21 + ess9018 chip

cheers
henry

Hi Henry,

There are two ways to do digital crossovers with exaU2I. In the early days of this thread they were discussed, however we haven't reached a consensus on best case scenario. At the present time exaU2I offers a USB / Windows / ASIO solution. This set of technologies comes with certain advantages and limitations and if we assume that they are acceptable we can focus the discussion on implementing 8 channel digital crossover.

exaU2I accepts 8 ASIO output channels, so the crossover functionality can only be implemented in hardware or software before the data is fed into the ASIO driver.

Scenario one is to use a software crossover.
One immediately available software solution is to use Foobar crossover plug-in: foo_dsp_xover.
Another possibility is to use Foobar VST host and Allocator:
George Yohng's VST wrapper for Foobar2000 - Hydrogenaudio Forums
Transient Perfect loudspeaker DSP tools . Windows DSP speaker crossovers.
I am experimenting with this configuration and I will post my results. I've also emailed the author of Allocator and I hope to be able to provide simplified direct integration between Allocator and the exaU2I software.

Scenario two is to use third party hardware for DSP processing.
Integration with exaU2I can be achieved using ASIO drivers. There are third party software solutions to route ASIO outputs from one device to the input of another device. For example: ReaRoute - CockosWiki.

I hope this information can give us a good starting point to discuss digital crossover solution with the exaU2I multichannel USB to I2S interface.

Cheers,

exa065
 
EXA feeding BuffaloII

hello folks
happy easter greetings from Norway
got my exa pcb s today after 4 days in transit...not bad!
mounted it in my BuffaloII dac and Foobar with asio,dll and exa driver installed found the device at once
very nice sounding right out of the box
lush, rich with very convincing harmonics and 3D
no problem hearing the improvement from the battery modified HIFACE, which by no means is bad sounding at all
highly recommended
god knows what´s in it if replacing the usb power?
I´ll certainly try
best
Leif
Norway
 
god knows what´s in it if replacing the usb power?

Maybe more than you can imagine..
I demonstrated LIFEPO4 battery vs USB powered for a friend and he did not believe the improvement was possible.
When you have heard the exaU2I battery powered the USB powered version sounds destroyed in comparison :D
I expect my linear SIC diode rectified and JFET regulated power supplies will sound equally good when I have time to connect them.
I will also replace the 5 Volt USB power to the USB receiver parts and disconnect the USB power from the PC..
 
can you let us know in more detail what needs to be done to the exaU2I device to power the USB part externally ?
It should be fairly simple: USB consists of four wires. The outer two are Power and Ground. The inner two are bidirectional data, and because they are differential they are not referenced to Ground. All you would need to do is maintain the data connection with the host computer but disconnect the power and replace it with another source such as a battery. Just make sure you stay within the 5 V nominal range, or 4 V minimum to about 5.25 V maximum. Check the USB Specifications for pin numbers.

However, I do not recommend making permanent changes to the exaU2I under any assumption that there will be a difference in the sound, since there are no analog circuits in the exaU2I anyway. What you really want to do is make sure that your DAC board has isolated analog power, and the data isolation chips on the I2S outputs of the exaU2I will ensure that the USB host computer supply noise does not make it to the DAC analog outputs. In other words, there's no point in focusing on the exaU2I hardware - the DAC board is where power and noise improvements will be successful.
 
Last edited:
this technique does not always work, depending on the receiver it will sometimes use the power lines to be sure there is something at the other end before sending data. will do this by pulling high or low in a designated fashion. given RayCtech has been using LiFeP04 already its likely the exadevice is tolerant of the power wires not being there, but its not an across the board solution
 
As lack of Mac support is one of my major reasons for not getting the device I would like to know if drivers are under development and a possible realistic ETA.
Cheers,
Nic

Nic,
The OS X drivers for exaU2I are the top priority for us. Unfortunately we are unable to commit to a delivery date at this time. I wish I could be more specific but there are risks when something is done for the very first time.

Regards,

exa065
 
Exa, I think many people would love that since for now with Allocator I have to use S/PDIF loop in order to use crossover. cPlay supports VST plugins and there is no need to go through the loop. That is also great way to test what S/PDIF conversion does to a sound, although its a double conversion. It is very noticeable, so with your device that is already I2S plus VST... man, that will be just plain awesome. I have been reading a lot on the subject lately, and I came to sensory overload. I tried Jack and virtual cable, to avoid S/PDIF and it is very difficult to align everything... support for W7, support for sampling rates, to recognize Allocator... than add to it convolver, or console and... I mean really... I just try to play sound not to launch rocket.

So yes, that would be such a logical step for Exa to became host for VST plug ins, so that within the same driver we have crossover app and room Eq app. Allocator is good up to 192 KHz and is really good and well executed app.

Allocator's developer Jan, better known on this forum as Thuneau, is great guy and very approachable, so maybe you two guys have a chat. It seems like we are up to something big! Exa you have open such a burst of innovation with your device that I have not seen in a while!

Hello AR2,

Following your request I've been playing with Allocator to familiarize myself with it. I attempted to use it with the prepackaged VST shell, Jack etc. This can become quite complicated and in my opinion detrimental to sound quality. The exaU2I ASIO driver uses an unusually small buffer (512k) and this confuses the VST shell packaged with Allocator. I haven't heard back from Jan but I have pretty good idea what is required on my side. The good news is that I've managed to integrate Allocator with Foobar using a third party Foobar / VST host component. It was tricky because Allocator accepts 2 inputs and produces 8 outputs. I don't have the right speakers and amplifiers to test properly a digital crossover, so at the end of the day I had the base channels driving my front speakers and the midrange playing from the rear side of the room. My setup is described here. Hopefully someone can try it with the right speakers and report back to us.

Allocator appears to be working at 352.8 kHz. Unfortunately exaU2I is limited to 4 channels at this sampling rate.
 
Hello AR2,

Following your request I've been playing with Allocator to familiarize myself with it. I attempted to use it with the prepackaged VST shell, Jack etc. This can become quite complicated and in my opinion detrimental to sound quality. The exaU2I ASIO driver uses an unusually small buffer (512k) and this confuses the VST shell packaged with Allocator. I haven't heard back from Jan but I have pretty good idea what is required on my side. The good news is that I've managed to integrate Allocator with Foobar using a third party Foobar / VST host component. It was tricky because Allocator accepts 2 inputs and produces 8 outputs. I don't have the right speakers and amplifiers to test properly a digital crossover, so at the end of the day I had the base channels driving my front speakers and the midrange playing from the rear side of the room. My setup is described here. Hopefully someone can try it with the right speakers and report back to us.

Allocator appears to be working at 352.8 kHz. Unfortunately exaU2I is limited to 4 channels at this sampling rate.

Thank you Exa for the fast work you put into this. This is awesome "DIY customer support". I will try to duplicate your set up in Foobar. I quite well understand and agree with you on the fact that it becomes too complicated and the fact that Allocator needs a lots of buffer. The only reason I would prefer Allocator implementation within the exadevice USB2I2S is because that way you are independedant of certain player. That means if I would like to use some kind of measurement, an example for room correction, than in that case I could have an app running through the exadevice and crossover with no problems. Regardless, that is a small issue. Quite well understand the difficulties various apps could present. I also appreciate your detailed instructions you posted on the exa forum http://forum.exadevices.com/Forum/tabid/41/forumid/3/threadid/3/scope/posts/Default.aspx
for anyone to try.

Allocator have been battling its buffer hungry nature and that is why Jan made Allocator Lite. For anyone watching movie, big buffer and sound delay was an issue, so he published Allocator Lite to solve that problem.

On the another note, you can also see my system that is slowly coming together and literarily shaping up.

Best
AR2