DIY 24 channel interface? n00b question

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

budney

New member
Joined
Apr 2, 2016
Messages
2
I'm in the process of building my own little music studio, but I'm doing it proper diy, building from kits all my pre amps, eqs, synths, racks, cables etc and then buying broken equipment to repair when I can't build myself, maybe selling off a few to fund some more expensive goodies.

The question is though, is there a way to build myself an interface with 24 analogue channels? Just got myself a lovely 24 channel Allen and Heath system 8 and I would love to be able to multitrack properly. Its just that my ideal interface, Antelope Orion, is far too expensive! Theres also the Motu 24 i/o but its old tech and the pci card i'd need is a bit of a nuisance.
I don't need any thing fancy, no pre amps or digital mixing, just 24 analogue inputs and outputs, ideally d sub! I figured with my handy soldering skills and having plenty of time on my hands I would be able to save a bit of money, and learn something in the process.

I've tried to have a look online and through this forum but its all lost on me...I know nothing about digital except for the basic knowledge I picked up when I was studying, I am a lot more versed with regular audio circuitry. Is this just a pipe dream or is this actually achievable with a bit of guidance? From what I've gathered initially there is no all in one 'kit' that I'd be able to buy, but I don't mind getting my hands dirty in figuring out what I'd need, I just don't want to spend ages researching and planning this to find out its virtually impossible unless I studied programming and pcb design.
 
One project, by user rkn80,  is in this very subforum. It's been slow going though and I'm not really sure of the current status. Nothing is expected imminently  don't think...

A couple of other diy ideas for multichannel ADCs / DACs have floated round online. I can't remember any off the top of my head. Again I think there might be references to some in this sub forum.

However, no one's every got a project working to a state where they've gone on to sell PCBs or kits; as far as I'm aware anyway.

As expensive as the Orion seems, if you start saving now and it takes you two years to buy it, that's still going to be a lot quicker than waiting for a diy project to come along. And if you have no background in digital systems, I'd guess it'd be years before you'd be able to see a project like that through to a successful conclusion. There are people on this board (Andy Peters?) who do this work for a living and are in a better position to talk about feasibility.
 
budney said:
The question is though, is there a way to build myself an interface with 24 analogue channels? Just got myself a lovely 24 channel Allen and Heath system 8 and I would love to be able to multitrack properly. Its just that my ideal interface, Antelope Orion, is far too expensive! Theres also the Motu 24 i/o but its old tech and the pci card i'd need is a bit of a nuisance.

I just don't want to spend ages researching and planning this to find out its virtually impossible unless I studied programming and pcb design.

Well, unfortunately, you'll need to study programming and PCB design. ;)

About the only feasible computer interface for the hobbyist to consider is a USB 2.0 design based on one of the XMOS chips. They offer a development kit that's not too expensive and they have a lot of working evaluation code. But ... the learning curve will be quite shallow. Consider yourself warned.

You will easily spend more money in bare circuit boards and parts than you would for a new MOTU thing. The effort might make the Antelope thing look reasonably priced.

-a
 
Andy Peters said:
Well, unfortunately, you'll need to study programming and PCB design. ;)

About the only feasible computer interface for the hobbyist to consider is a USB 2.0 design based on one of the XMOS chips. They offer a development kit that's not too expensive and they have a lot of working evaluation code. But ... the learning curve will be quite shallow. Consider yourself warned.

You will easily spend more money in bare circuit boards and parts than you would for a new MOTU thing. The effort might make the Antelope thing look reasonably priced.

-a

I figured that this was the case...but just thought I'd check, thanks for the reply! I just thought with all the diy kits popping up in the last few years there may be something similar with interfaces etc, but looks like I'll have to wait a bit longer for all my ideal bits of gear in kit form. Thank you for saving me hours worth of pulling my hair out, now to start saving up!
 
As mentioned, DIY interface is a bit too complicated if you are starting out.  But you can start studying/learning about it now if it interests you.

In the meantime there are good less expensive options than the Orion, get one of those now and get to recording. Then as time goes on you can get more DIY.
 
I am surprised by your writing that the orion is far too expensive. I haven't worked with it, but if it really works reliably and sounds good, then it is a steal for the price. I'm not sure of it's soundcard capabilities, whether it could replace my RME cards with direct monitoring capabilities - but if it does then my soundcards did already cost about as much as the orion.  There seems no way to DIY something like this, the equipment you might need to debug a digital device like a converter with many outputs might cost a lot more already. You certainly won't get away with a multimeter and a cheap oscilloscope. I don't want to discourage you, but I believe if the orion's functionality is the goal for you, then there's no cheaper way to achieve it then buying one. There will be cheaper workarounds if you sacrifice quality, but I believe DIY won't save you money here  :(

Michael
 
I have the first version of the Orion32 (not the 32+).  It really sounds good.  I went from a BLA (and me) modded MOTU 192HD and the difference was really amazing.  Seeing that you can get Orion32's for about $2k used, I wouldn't hesitate at all to jump on one.

Here's an example of one that sold recently:

https://www.gearslutz.com/board/gearslutz-secondhand-gear-classifieds/1098218-antelope-audio-orion32.html?highlight=orion

For that money....You can't touch it.

The only downer with the Orion 32 is the inability to create multiple cue mixes.  It only has one mixer that you can use, but you can save multiple patching "presets" and recall them from the front panel.  Since you have a real analog mixer, this shouldn't be an issue as you would just create cue mixes on the mixer itself.

Even at the real retail price of an Orion 32, you'd be hard pressed to DIY something comparable.  At the used prices you can get an Orion32 for...There's nothing else out there in the ball park.
 
Orion is a steal for a price. It's roughly 100$ for each in AND out. I doubt it's possible to DIY it cheaper. And there's no guarantee that the DIY converter will sound and perform as good.

I foresee a lot of problems with something as simple as an 8 I/O box.

The learning experience is another matter completely. But for business this won't work IMO
 
Can it be done?  sure... but.... (given caveats above)

some things to consider:

Since this is ADC, DAC, USB, AES-EBU or whatnot, this is getting into Signal Integrity  high-speed digital design (like RF phenomena) on PCBs....  6MHz to 12MHz or even 25 MHz / 33 MHz  might be required depending on chipset selected and ADC oversampling rates.....

high-speed digital design on PCB is not intuitive at first, but with experience can be enlightening.

I'd skip any notions of 2-layer PCB for (un)obvious reasons; four or even six layers are likely.  Yes, some ADC/DAC chip manufacturers may have reference designs for two-layers (for economy, for evaluation in a _laboratory_ environment) that may be copied, but be mindful of electromagnetic interference (EMI) notions, especially with high gain or high impedance analog electronics in the same room...  a 4-layer board with proper planes and high-speed design motifs (matched impedances) can knock down EMI by 20dB or so (rule-of-thumb)....

High-speed digital PCB designs work best with a solid (physically uninterrupted) ground plane underneath high-speed digital traces, for any semblance of EMI reduction.  This is because the "return current" of these high-speed signals will desire to flow (or concentrate) directly underneath the actual signal trace on the other layer... this is the path of least impedance, usually, the path of least inductance....two-layer PCBs physically constrain the design such that at some point, power, a connector pin-field, or some other signal will need to cross under the high speed trace (and physically interrupt the contiguous ground plane) thus increasing the path of inductance for the "return current" that would have to go around the obstacle (pin-fields on connectors also can break up a ground-plane).. This is also a contingency recipe to create electromagnetic interference... 

keeping analog and digital ground planes separate is another interesting topic...

Also PCB dielectric material and THICKNESS dictate some other RF/high-speed parameters (see below) AND trace width (keeping in mind limitations of PCB fab house for allowable minimum trace width)....  keeping in mind that usual overall thickness for standard PCBs might be 0.062 inches (1.575mm) thick, so the PCB layer stackup should be considered ahead of time....

With high speed digital, the frequency is important, but also the rise time (how sharp the digital signal square waves are) is extremely important. The sharper the rise time, the more harmonic frequency content that must arrive from "transmitting output of logic gate on chip" to the "receiver input of logic gate on another chip"..  For example, some FPGAs have slew-rate limiting on their digital output pins for this reason (to slow down rise time, and moderate some design requirements of high-speed digital traces / power requirements)...

PCB traces at high-speed/frequency are Transmission lines that have a characteristic impedance and like to see that "RF" energy be terminated into matching impedance; max power transfer motifs as well...  Otherwise reflections (or ringing) occur, which can cause timing or threshold level errors in the digital system...

Splitting of PCB traces (e.g., multiple tee's or branches on a clock line for distribution of the clock to multiple chips) is tricky...  each branch or split needs to be mindful of total impedance, reflection, etc.... or use a clock buffer/distribution chip, at the expense of perhaps some noise/jitter, power draw, layout etc....

Measuring with oscilloscope probes also involves different techniques.  The "long" flying ground clip-lead on the probe is usually bad for measuring high frequency stuff due to loop area, but can be good for sniffing... Usually taking the cap off of the scope probe to expose the ground ring (and design the PCB to be probed with oscope probe tip cap removed, with ground nearby for the ground ring)

Also, typical 'scope probes cannot probe a crystal directly because the probe impedance is too low. (and stops the crystal from oscillating)... 

indeed, 12MHz square wave will have harmonics that hopefully the scope can capture or measure within its bandwidth

Arrival time (and skew) topics are also important for high-speed digital signals... it does take a finite time for a signal to travel on PCB traces (and is different propagation times for outer layer PCB traces versus inner layer PCB traces)....

Vias have (parasitic) inductance.... ground planes in sensitive areas may increase parasitic capacitances, that are not represented in the schematic (hence the term parasitic)...  these may not be critical at lower "high-speed" frequencies, but....

Signal integrity analysis tools typically use IBIS models (which are different than SPICE models)... These tools are used before PCB layout (schematic design time) to figure out termination schemes on high-speed nets... They also have post-layout capabilities to tune routing and terminations.....

<shameless plug>  I use Altium, but am not employed or involved with any of the below links....

http://www.altium.com/video-altium-presents-signal-integrity

https://www.youtube.com/watch?v=FXjVrYwSAGc

even 10MHz with a fast enough rise time can be considered high-speed...

</shameless plug>

Of course power (and decoupling capacitor with proper PCB layout, and decoupling capacitor TYPE - e.g., film, electrolytic with ESR and ESL, tantalum, ceramic etc.) is important, and higher frequency digital signals need to have their power sourced from the chips using the chips' power (and ground) pins... phenomena like ground bounce happens if the ground is feeble (conductance-wise), given also that the "ground" (reference voltage, especially for logic threshold concepts) bond-wire inside the chip from the pin to the silicon is a (parasitic) inductance, especially higher up in frequency...

decoupling capacitors have RF properties and resonances and help counteract trace inductance on power distribution traces.... 

Just some abbreviated random ramblings for high-speed digital design... concepts to look up on the Internet....

https://en.wikipedia.org/wiki/Signal_integrity

http://electronics.stackexchange.com/questions/75368/how-can-pcb-trace-have-50-ohm-impedance-regardless-of-length-and-signal-frequenc


 
Andy Peters said:
About the only feasible computer interface for the hobbyist to consider is a USB 2.0 design based on one of the XMOS chips. They offer a development kit that's not too expensive and they have a lot of working evaluation code. But ... the learning curve will be quite shallow. Consider yourself warned.

-a

Indeed, the learning curve is very shallow. They're too busy marketing the XMOS  micro-controller awesome architecture, but they forget to publish XMOS C and Assembly Programming Language: The Complete Reference. A handful of board design, and working source code is not good enough.  If Microsoft didn't publish Windows API: The Complete Reference to anyone, there wouldn't be any Windows programs except those made by Microsoft.
 
metalb00b00 said:
Andy Peters said:
About the only feasible computer interface for the hobbyist to consider is a USB 2.0 design based on one of the XMOS chips. They offer a development kit that's not too expensive and they have a lot of working evaluation code. But ... the learning curve will be quite shallow. Consider yourself warned.

-a

Indeed, the learning curve is very shallow. They're too busy marketing the XMOS  micro-controller awesome architecture, but they forget to publish XMOS C and Assembly Programming Language: The Complete Reference.

It took exactly two clicks from the XMOS home page to find the XMOS Programming Guide..
 
I read that Programming Guide PDF, and I still think it's rather confusing. It's a lot more like a work in progress, than a complete reference. Perhaps if I print it out, it'll help???

One confusing example:

The <: operator outputs a value to a port.  Other than this, you'll also see a bunch of operators like :> -> @
after reading some more it became clear that the :> operator assigns a value to a defined variable, but until the end of the guide, -> @ operators are still a mystery.

Another confusing example:

"case c :> int i:"

what does that even mean? assigning an integer value to i while comparing c to certain conditions?

"case t when timerafter ( timeout ) :> void :"

ugh...............



 
metalb00b00 said:
I read that Programming Guide PDF, and I still think it's rather confusing. It's a lot more like a work in progress, than a complete reference. Perhaps if I print it out, it'll help???

OK, three clicks..

You want confusing? Have you opened up a manual for an ARM Cortex-M3? Or any of the newer Xilinx or Altera FPGAs? Thousands of pages ...
 
When I was an engineering intern, I went through the first task they had ready for me and didn't have much else lined up to do yet, so they handed me an ARM9 programmers guide and said "Here, familiarize yourself with this".    :eek:

That kind of thing I can spend about 15 minutes perusing before it's either time to move on or dig in deeply, which would have been a bit above my level anyway (and probably still is).
 
Back
Top