The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Custom DAC - Good Design? Monitor Controllers
Old 22nd November 2016
  #1
Gear Addict
 

Custom DAC - Good Design?

I had a custom monitor DAC built for $2000 and I'm having a hard time hearing a difference between it and my Orion 32 DAs, which could very well be a limitation of my system (Drawmer MC2.1 into a pair of BM15A's) or my hearing. DAC is hooked up to an RME HDSPe MADI FX card via AES. Card is clocked by the Orion 32 via MADI.

It's a single PCM1794A DAC chip driven by a CL8416-CZZ digital receiver with 2114D/5534P analog buffers and discrete 2N4037/2N2102 outputs. Separate linear power supplies for the digital and analog boards with LM317 regulators. Pics of the innards below.

Custom Monitor DAC Pictures

What do you think of this design? Is it on-par with other high-end 2-channel monitor DACs? From what I've gathered, it most closely resembles the Forssell MDAC-2, also a single-chip PCM1794A/CL8416 box, except their analog stage is fully discrete JFETs.

Is my MC2.1 or the BM15A's or my own ears the bottleneck here? Or are the differences just not going to be that audible in this tier of conversion? To be honest, I'm a little lost.

Last edited by audiotechnica; 22nd November 2016 at 10:53 PM..
Old 23rd November 2016
  #2
What I see from your photo resembles point-to-point wiring, with long leads on everything and no visible ground planes. It looks like it was built by a technician who can read data sheet schematics and is used to rewiring vintage microphones. There's no shame in that for recreating vintage designs, and I've built breadboards that look similar to what I see here. But I wouldn't attempt any such thing for a converter design.

Frankly, I'm amazed that your DAC works as well as you report it does. I think if you actually pulled the lid off Fred Forssell's DAC, you'd find a multi-layer PCB with serious attention given to routing for signal integrity. Poking around Fred's web site, I wasn't able to find a picture of one of his converter boards. I did see a preamp layout, but I couldn't tell how many layers were used in the PCB. The truth is, preamps don't have the same routing requirements that converters do. With converters, you're not just dealing with audio frequency signals, you're dealing with modulator clocks that are typically running at 24.576 MHz (or higher) and require good signal integrity to at least 5x that frequency if the chip is to work as intended. To make that happen, you need to run the clock signals as controlled-impedance transmission lines, and you can't accomplish that with point-to-point wiring unless you're using coaxial cable or bonded twisted pairs. Neither technique is suitable for connections to a fine-pitch surface mount part, and such parts also need good ground planes and low-inductance bypassing to work correctly.

I've always been impressed by the PCB layouts that John Siau does for his Benchmark Media designs. This article has a picture of his DAC2 layout, using an ESS Sabre chip near the bottom edge. I can't zoom in quite enough to see the clock pins clearly, but it appears to me that he's probably running MCLK as a buried stripline. Also note the multiplicity of high-frequency ceramic bypass caps right next to the chip. This is the kind of careful layout that's required to make a modern converter chip meet its published specs.

David L. Rick
Seventh String Recording
[MSEE, IEEE senior member, past AES board member, ASA member]
Old 23rd November 2016
  #3
Gear Addict
 

Thank you for the detailed reply, David. So you're saying that even with my modest speakers and monitor controller, I should be able to hear a clear improvement over the Orion 32 with a Forssell or a Mytek Brooklyn?

The DAC doesn't sound any worse than my Orion 32, they just sound identical to me. I ordered it on the recommendation of my mastering engineer, who uses custom equipment built by this person and swears by its sonic quality.

BTW, there is no clock in the DAC - it derives clock from the AES signal, so perhaps the PTP connections aren't impacting the clock signal that much? I am not a designer so I have no idea. I just wonder if the $2000 could have been spent better.

Last edited by audiotechnica; 23rd November 2016 at 06:12 PM..
Old 23rd November 2016
  #4
Lives for gear
 

Hi
On the basis it works and you are at least quite happy with it it is a worthy attempt. It is not however a 'production' design as David points out a lot of serious thinking and design needs to go in to make it truly function to the limit of the chipset used. Simply buying an 'evaluation board' from a chip maker and bolting that in a box with a supply is likely to have been a little better and certainly repeatable on the basis they want to sell chips and expect them to be tested and analysed almost to death.
You have to wonder how much digital 'noise' is present on the analogue outputs that is getting filtered by your monitor amp/speakers. In itself not an issue (as you like it) but you might get significantly different results if you use other gear. Inconsistency is a problem.
Matt S
Old 23rd November 2016
  #5
Gear Addict
 

Thanks for your insight, Matt. This was supposed to be a long-term investment that I wouldn't have to upgrade for years, a pristine monitor DAC that would outperform my Orion 32 in terms of fidelity - at least those were the specifications I gave to the person who built it.

Given that I can't hear any improvement over the Orion, I am rather skeptical. Especially for $2000, I can't imagine one would allow the layout issues you're describing, permitting digital noise to seep into the analog stage. Looks like I might be having a chat with the maker about a refund and put that money towards a Forssell MDAC-2 (from my research, seems to be exactly what I need).

Last edited by audiotechnica; 23rd November 2016 at 08:17 PM..
Old 23rd November 2016
  #6
Lives for gear
 
deuc647's Avatar
 

In all seriousness, Give Val at Qes Labs a call. That man makes some stellar stuff. I have the first ADC converter hes made in a while and it is pretty damn transparent.
Old 24th November 2016
  #7
Lives for gear
 

I would borrow/rent a high end commercial design and see if you hear a difference.

But as others have mentioned, hand wiring a DAC is really not the way to high performance.
Old 24th November 2016
  #8
Gear Addict
 

Quote:
Originally Posted by johnnyc View Post
I would borrow/rent a high end commercial design and see if you hear a difference.

But as others have mentioned, hand wiring a DAC is really not the way to high performance.
This is actually my next step. I'm in touch with Scott Forssell already (he's on vacation in Australia till December 6th), and I'm looking to test drive the Burl B2 and the Mytek Brooklyn in the next couple of weeks.

Last edited by audiotechnica; 24th November 2016 at 03:46 PM..
Old 24th November 2016
  #9
Quote:
Originally Posted by audiotechnica View Post
Thank you for the detailed reply, David. So you're saying that even with my modest speakers and monitor controller, I should be able to hear a clear improvement over the Orion 32 with a Forssell or a Mytek Brooklyn?
First off, I wouldn't expect your Orion 32 to sound bad. If there's one thing Antelope does understand, it's low jitter clocking. But what the designer of a higher-end monitoring DAC has going for him is a higher per-channel parts budget, which may enable use of a high-end stereo ADC (and sometimes two in parallel) instead of a multi-channel part and still leave budget for a nicer output stage and more careful power supply design. These things do lead to incremental improvements, but if the difference is "night and day", then one of the products was of very poor design to begin with. Compare your Orion to a first generation MOTU portable unit and you'll certainly hear a difference.

I don't think your monitoring chain is at fault, either. That Drawmer controller is much like the SPL one that I use. I don't have personal experience with Dynaudio near fields, but they are well respected. How are your room acoustics?

Quote:
BTW, there is no clock in the DAC - it derives clock from the AES signal, so perhaps the PTP connections aren't impacting the clock signal that much? I am not a designer so I have no idea. I just wonder if the $2000 could have been spent better.
Well, I certainly think so. That price point is where a lot of excellent DAC+headphone amp/monitor controller boxes from Benchmark, Grace, Lavry, and Mytek live. The competition between these designers is truly fierce! A colleague has purchased each new release from Benchmark and Grace for several years now, and ranks them (from good to better) like this:

Benchmark DAC1
Grace m903
Benchmark DAC1HDR
Grace m920
Benchmark DAC2

What you'll notice about this list is that his ranking goes strictly by release date. If we added Mytek and Lavry to the list, I think that would probably still hold true. These designers (John, Michael, Dan, and Michal) are clearly learning from one another and trying to push the envelope a little further with each new product release.

BTW, I don't suggest you take the top-end rankings as gospel. Depending on what your priorities are, you might rank them differently. Remember this oft-quoted maxim: "The best DAC's sound more alike than they sound different." But each brand has a different philosophy about how the front panel user interface should work. Each designer has a different ideas about how to design an output stage or a headphone drive amp. Mytek is an "early adopter" when it comes to new audio formats, while Grace takes a "wait and see" stance. Michael Grace favors current-mode amplifiers, while John Siau tends to pick the very best voltage-mode parts he can find. Some designers use analog gain control, some use digital, and some use both. They make different choices about the rear panel input options. Their clocking approaches vary as well.

Let's talk clocking, since you asked about it. Any competently-designed DAC will sound better when running on its own internal clock than it will when locked to anything else. (If it doesn't, sell it and buy something better!) That's why USB DAC's got so much better sounding when they started using asynchronous data transfer. But the reality of studio work is that we often do need to use external clocking. Cleaning up a jittery AES or word clock feed is a tough problem. The typical approach is a phase-locked loop, but every PLL has a frequency point below which it stops filtering the incoming phase noise and starts tracking it instead. The designer can push the crossover frequency lower, but then his PLL may not lock to some signals. So a few years back people started building dual-stage PLL's. The first stage can lock to anything have a good enough clock output that a less noisy but more finicky second stage PLL can (usually) lock up. A variation of this idea is adaptive control of the PLL loop gain and bandwidth. (I think Prism Sound does this.) John Siau's DAC1 design was novel in that he replaced the PLL with an Asynchronous Sample Rate Converter chip, so the internal DAC chip gets to use its own clock. You can feed a DAC1 the worst clock on earth, and it will still sound ok. But not everyone wanted asynchronous sample rate conversion done on an audio signal undergoing final Q/C. So the analog camp pushed back, improving their PLL designs. I think Michael Grace's current m905 implementation now uses three PLL stages in series. (I'm a fan of this approach, and also of analog gain control, because I'm trying to listen to my DSP, not somebody else's, however good it is.)

Meanwhile, the ASRC camp has pushed back, because ASRC chips have gotten better. Crane Song's Dave Hill is the latest convert to this approach. His latest-generation Avocet controller uses ASRC to a fixed 207kHz sample rate. (This happens to be relatively prime to any common audio clock frequency.) So the external clocking approach is one significant differentiator between an Avocet IIA and a m905. But, for a mastering engineer, it might not be the most significant difference. Some people love the Avocet remote control and some people hate it. It's perfectly valid to let something like that influence your decision, because ease of operation is a big deal when you have your hands on something five days a week.

Most of these makers offer some kind of trial program. My advice is to max out your credit card and get two or three demo units in house at the same time. Listen to them with different input sources and a wide variety of program material. But, equally important, put your hands on them and see how the knobs and switches feel, and how carefully (or not) you have to think about using them. Then make your choice and go back to making music!

David L. Rick
Seventh String Recording
Old 25th November 2016
  #10
Gear Addict
 

Thank you for this super-thorough reply, David, I really appreciate you explaining things in detail. This "custom" box is going back to its maker.

My room is decent but a bit bright - I might add some diffusors and extra absorption.

Funny, the Orion 32 was an upgrade from a MOTU 2408mk3+24IO system and boy, I didn't realize how bad those units sounded until I heard the Orion. That really was night-and-day difference.

Now, with a stereo monitor DAC, I am simply hoping to get that last bit of quality out since I mix 100% in the box and don't have money or space for analog outboard or a console, hence I figured that investing in my monitoring chain would be the wisest.

On the other hand, I don't have the cash to buy a pair of ATCs, so I'm wondering if investing in a high-end monitor DAC would even make sense in my situation... but I guess no way to tell other than to demo a couple units, which I'm planning to.

So for best quality, you'd recommend a DAC with its own built-in clock? Some people are swearing by the Forssell MDAC-2 as the most transparent and clean they've ever heard, and it does not have a clock. This really looks like a game in which the only way to win is to try all the available boxes one by one.

I have an RME HDSPe MADI FX card with the Orion 32 currently clocking the card over MADI. The card has AES I/O that I want to use to add a monitor DAC. If I get a DAC with its own clock, would it be best to have it be the master clock and slave the card and the Orion 32 to it?

Last edited by audiotechnica; 25th November 2016 at 05:28 AM..
Old 25th November 2016
  #11
Motown legend
 
Bob Olhsson's Avatar
 

As I understand it, the only problem with ASRC is added latency. It really comes down to the specific application.
Old 25th November 2016
  #12
Lives for gear
 

Quote:
Originally Posted by audiotechnica View Post
Some people are swearing by the Forssell MDAC-2 as the most transparent and clean they've ever heard, and it does not have a clock.
All converters have a clock. Whether they have an external clock output or not is different but all converters have an internal clock otherwise they couldn't be run standalone in any form.
Quote:
Originally Posted by audiotechnica View Post
If I get a DAC with its own clock, would it be best to have it be the master clock and slave the card and the Orion 32 to it?
Yes. Slave all other devices to your master output clock -- unless you have a system clock that is better with equal legged distribution to everything that needs to be clocked. Then clock everything to that (taking into consideration what ASRC does as described by David.) Preferably all of this should be over WC if possible -- not input audio clock (like AES, MADI, Spdif, or ADAT)
Old 25th November 2016
  #13
Gear Nut
 

Why, for those loving numbers, don't you perform a loop test?
Something like:
a) using orion DAC and orion ADC
b) using your custom DAC and orion ADC

then you compare a) vs b)
You can check for the Audio Diffmaker thread in case you need more details.
Old 25th November 2016
  #14
Gear Addict
 

Quote:
Originally Posted by pentagon View Post
All converters have a clock. Whether they have an external clock output or not is different but all converters have an internal clock otherwise they couldn't be run standalone in any form.
I see. So if locking a DAC to AES is bad, and a DAC sound better when running off its own internal clock, then why does the Forssell, which does this as its only way of syncing to the digital source, get such high praise? I am confused whether I should get a DAC with a clock output and slave the RME card and the Orion to it, or get the Forssell and slave it to the Orion's clock?

I'm just trying to understand because I am very confused about these things. Thank you for taking the time to answer my questions and help me in my quest for better audio.
Old 25th November 2016
  #15
Gear Addict
 

Quote:
Originally Posted by mudseason View Post
Why, for those loving numbers, don't you perform a loop test?
Something like:
a) using orion DAC and orion ADC
b) using your custom DAC and orion ADC

then you compare a) vs b)
You can check for the Audio Diffmaker thread in case you need more details.
The custom DAC is already boxed up for shipment back to its creator. In my testing I also discovered some transformer hum leaking into the audio signal when applying high gain to the output, which made me rather unhappy. My Orion is silent as a tomb on the outs.
Old 25th November 2016
  #16
Lives for gear
 

Quote:
Originally Posted by audiotechnica View Post
I see. So if locking a DAC to AES is bad, and a DAC sound better when running off its own internal clock, then why does the Forssell, which does this as its only way of syncing to the digital source, get such high praise?
Re-read what David posted. You are making the assumption that there is only one way to do things and that doing each thing has some kind of inherent value of making things better like an additive point system. In engineering and design that is often not the case -- it is often the case the whole is greater than the sum of its parts (or in this case: partial design choices.)
As I wrote, clocking to dedicated WC is preferable to clocking over digital audio input signal. Not bad -- preferable. It's just because WC is specialized for output clocking so both ends (sending and receiving) have higher defined standards and is dedicated for this purpose. (This is an oversimplification but you can look up the standards for WC BNC output.) Since we are talking about a DAC which is an endpoint (vs ADC which is passing signal with a clock and you can see Forssell does have WC in/out on the ADC), clocking of the signal past this point is meaningless; it only matters the designer has the DAC do what is necessary. The Forssell may by asynchronous, it may be designed that way so you can send (upstream) a variety of signal to it and it doesn't need to rectify incoming signal to a house clock on a separate WC input, and there could be more reasons for this choice. It looks like on the MDAC4, you can have each stereo pair on different clocks so that may be the reason for the design choice (modularity.) You should just examine the DAC based on the audio results.
Quote:
Originally Posted by audiotechnica View Post
I am confused whether I should get a DAC with a clock output and slave the RME card and the Orion to it, or get the Forssell and slave it to the Orion's clock?
Get what sounds best and don't worry about the connectors. However, if you have a choice between WC and AES/Spdif/ADAT/etc for sending clock, choose the WC. That's it.
Old 25th November 2016
  #17
Gear Addict
 

Thank you everyone for your replies. I'll be doing some listening tests in the coming weeks (provided I can get a hold of the DACs I'm interested in for a test drive).
Old 25th November 2016
  #18
Motown legend
 
Bob Olhsson's Avatar
 

Quote:
Originally Posted by pentagon View Post
...As I wrote, clocking to dedicated WC is preferable to clocking over digital audio input signal. Not bad -- preferable...
According to Daniel Weiss, this was only true of early generation SPDIF-AES/EBU chips and the modern ones are actually better than word clock.
Old 25th November 2016
  #19
Gear Addict
 

Quote:
Originally Posted by Bob Olhsson View Post
According to Daniel Weiss, this was only true of early generation SPDIF-AES/EBU chips and the modern ones are actually better than word clock.
Good to see you here, Bob! This would explain why the Forssell is getting such high praise.
Old 25th November 2016
  #20
5534 opamps won't reveal what the PCM1794 can do, you need better. I use Analog Devices ADA 4898-1 in mine. .9 nv noise, 60V slew rate, .00003% THD.
Old 26th November 2016
  #21
Lives for gear
 
Casey's Avatar
 

The Bricasti M1 uses a fifo clocking method. It uses a precision variable speed clock that changes frequency up or down in imperceptible steps over minutes as samples in the fifo grow or recede due to the jitter in the source signal.

Because the precision clock drives the DAC directly it is completly decoupled from the input clock. It translates the high frequency jitter that we do hear to very very low frequency jitter that we don't here. It makes more difference in the sound than one might imagine.

We specifically avoid using the filter in the DAC that we use. To our ears it just does not sound as good as the filters that run in the SHARC DSP that we use.

Using an SRC as invented by Benchmark imposes a similar filter to the one we avoid in the DAC. So we use the clocking system described above to avoid using an SRC and avoid using a PLL.



-Casey


.
Attached Thumbnails
Custom DAC - Good Design?-20161125_205652.jpg  
Old 26th November 2016
  #22
Gear Addict
 

Wish I had $9k, Casey. My budget is $2k-3k.
Old 26th November 2016
  #23
Lives for gear
 

Since you are running separate A to D and D to A converter units . . . While the current rule of thumb is that your A to D converter should be set for internal clock (running as master) as this is the most critical stage, once you are done recording and in the mixing stage you can simply change your D to A converter's clocking from slave to master to get it's best performace for that stage. It's a bit of a hassle but workable. Obviously a mastering guy would already have their D to A converter set for internal clock (as master) to best hear the changes they make. It would be an interesting test for the mastering guys using analog outboard however to have their A to D converter if separate set as master to see if that additional recording step also benefits from the rule of thumb.
Old 1st December 2016
  #24
Gear Addict
 

Just ordered a Dangerous Convert-2 for a test drive. From what I've read about it, it seems to have the qualities I'm looking for - depth, width, space, and detail. Plus, built-in clock, and it's $1000 cheaper than the Forssell. I shall report once it arrives.
Old 6th December 2016
  #25
Gear Addict
 

Spent the evening with the Convert-2 and have to say it's a noticeable improvement in clarity over my Orion 32, even on headphones (much more noticeable on speakers though). Not night-and-day, because the Orion 32 is already very good (and was a night-and-day improvement from my old MOTU boxes) but cloudy day vs. sunny day or cloudy night vs. clear, starry night in the countryside.

The Convert-2 adds that extra clarity, separation of instruments, space between them, and overall focus throughout the spectrum. Hearing the Orion 32 takes the smile off my face (even when clocked by the Convert-2), not because it sounds bad, but because it removes that extra bit of "life" from the music.

Another place where the Convert-2 wins is playing low-level material - it's very clear even when playing stuff that barely hits -24dB on the digital scale, whereas the Orion 32 has some subtle but clearly audible cloudiness there. The Convert-2 also reveals the ends of reverb tails clearly whereas they get kind of lost in the haze on the Orion 32.

I got the Convert-2 used for $1950 shipped so I am super-happy both on the money front and the audio fidelity front.

Last edited by audiotechnica; 7th December 2016 at 02:45 PM..
Old 6th December 2016
  #26
Gear Addict
 
mmarra's Avatar
Quote:
Originally Posted by audiotechnica View Post
I got the Convert-2 used for $1950 shipped so I am super-happy both on the money front and the audio fidelity front.
Where you able to get a used convert
-2?
Old 6th December 2016
  #27
Gear Addict
 

Quote:
Originally Posted by mmarra View Post
Where you able to get a used convert
-2?
Perfect Circuit Audio, they had it listed on eBay. Sweetwater also has demo units for usually 10% off.

Last edited by audiotechnica; 6th December 2016 at 02:31 PM..
Old 7th December 2016
  #28
Gear Addict
 

Another day of listening. I have to say that I do not agree with the guy who called the Convert-2 harsh (and preferred his Crane Song Solaris instead). After listening to almost a hundred tracks, I find that the Convert-2 is extremely revealing of the flaws and strengths of each particular mix. Harsh mixes do indeed sound harsh on it, fatiguing even, because it reveals their harshness so well - a lot more so than the Orion 32. Mixes that are smooth and silky sound simply fantastic - smooth, natural, not fatiguing at all. It's really showing you the true nature of your material.

I'm relistening to so much stuff and hearing how bad some of the commercial mixes are and how great some others are! Most commercial electronic music from the past decade sounds very gritty, but stuff like Vangelis or Mike Oldfield sound absolutely silky/airy/smooth/magical. Sarah Jarosz "Build Me Up from Bones" sounds just as beautiful as it sounded at The Mastering Lab when Doug played it for me when he was cutting the lacquer for it. The detail on the vocals and the transients on all the plucked instruments is amazing.
Old 7th December 2016
  #29
Lives for gear
 

Quote:
Originally Posted by Casey View Post
The Bricasti M1 uses a fifo clocking method. It uses a precision variable speed clock that changes frequency up or down in imperceptible steps over minutes as samples in the fifo grow or recede due to the jitter in the source signal.

Because the precision clock drives the DAC directly it is completly decoupled from the input clock. It translates the high frequency jitter that we do hear to very very low frequency jitter that we don't here. It makes more difference in the sound than one might imagine.

We specifically avoid using the filter in the DAC that we use. To our ears it just does not sound as good as the filters that run in the SHARC DSP that we use.

Using an SRC as invented by Benchmark imposes a similar filter to the one we avoid in the DAC. So we use the clocking system described above to avoid using an SRC and avoid using a PLL.



-Casey


.
do you feel that the SRC is detrimental at 192k? i'm curious to know your thoughts.
Old 8th December 2016
  #30
Lives for gear
 
Casey's Avatar
 

Do you mean 192 in and 192 out?



-Casey
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump