|
Post by danielquinn on Mar 23, 2015 9:40:16 GMT
I very much doubt if there has been one made with more than £60 worth of parts. I get really uptight when people start mentioning the cost of the parts. Designing a DAC takes a lot of man hours. Making the case, cost of approvals, packaging, etc. quickly adds to the cost. To give an idea, the Caiman MKII took me about 4 years and 7 prototypes to get it right. I am not wrong when I say that if anyone thinks that the cost of the parts should justify an end product costing not much more than the parts is seriously deluded. I accept this on 1 condition , a condition that generally applies to pharmaceuticals and that is once the cost of r/d is recouped the price of the product is significanyly reduced ,look forward to a price reduction any day now p.s - no need to respond with r/d is continuous , the fact is your products are reasonably priced so getting up tight at illegitimate questions is unnecessary .
|
|
|
Post by MikeMusic on Mar 23, 2015 9:57:21 GMT
Thanks How did the sound compare to the Ankaka ? How did your Ankaka fail ? Trying to work out how long I've got !
|
|
|
Post by MikeMusic on Mar 23, 2015 10:00:40 GMT
I very much doubt if there has been one made with more than £60 worth of parts. I get really uptight when people start mentioning the cost of the parts. Designing a DAC takes a lot of man hours. Making the case, cost of approvals, packaging, etc. quickly adds to the cost. To give an idea, the Caiman MKII took me about 4 years and 7 prototypes to get it right. I am not wrong when I say that if anyone thinks that the cost of the parts should justify an end product costing not much more than the parts is seriously deluded. In a similar manner Costs pennies to print. But you need kit and a factory and a skill set. Can always be done cheaper if you work in your garage and don't pay tax ! Our customers also find it useful when we point their mistakes before printing
|
|
|
Post by MartinT on Mar 23, 2015 10:16:33 GMT
How did your Ankaka fail ? Trying to work out how long I've got ! It failed suddenly and finally. I can't hear an SQ quality different, but the XTPower runs the Raspberry Pi 2 as well as the Caiman, so the overall combination sounds better for streaming.
|
|
|
Post by MikeMusic on Mar 23, 2015 10:17:14 GMT
Ta Fingers crossed mine keeps going a long time .............
|
|
|
Post by aurender on May 24, 2015 11:20:10 GMT
As a very new member, I'd like to add my thoughts on the DAC question. The factor that hasn't been mentioned so far is clock accuracy. My digital processing uses DCS Scarlatti DAC, up sampler and master clock. Sources used are either locked to DCS clock or use asynchronous USB so as to minimise clock effects.
the apparent influence of central clocking led me to look at changing clock parameters. This was a purely subjective exercise but proved interesting. The DCS clock uses an off the shelf OCXO clock. People have tried rubidium clocks (actually quite inexpensive as ex telecoms units from China telecoms, bought on the usual auction sites for $100). Rubidium clocks have better long term accuracy (which shouldn't matter at all) but worse short term accuracy (which should). Results were a clear difference in sound quality which I think is for the better. DCS were interested in the idea of using a rubidium clock and borrowed a unit that I had been playing with. Their view was that it changed the sound but their was internal disagreement on whether the changes were entirely for the better.
next step for me was to investigate routes to get much better short term clock accuracy. The only practical alternative was to use a variety of oven controlled crystal in which the suspension is made from a slice of the same crystal. This is known as a BVA clock and whilst simple in concept, is actually very difficult to produce. Short term accuracy is typically 100 times better than the best normal OCXO crystal.
results were astonishing in that sound seems much more analogue with no trace of digital "edge". I know, I know, a very subjective judgement but it does seem consistent from many listeners who are used to the sound of my system which is very revealing. (Aurender,DCS,Ayre, Halcro, YG)
conclusions: I'd say improving clocking to this level has made more difference to "realism" of my system than any other change in recent times. Effects are compaitively easy to check by switching on or off the ability of the DCS master clock to lock to the external 10mHz signal. Settling on switchover is very quick.
|
|
|
Post by MartinT on May 24, 2015 11:46:50 GMT
I've looked at clocks on and off over the years. I tried hard to get a better clock crystal or generator for my Technics turntable project but the frequency is non-standard. I also enquired about the clock in my Ayre but Charlie Hansen said that they had worked hard on the clock circuit and I was only likely to make it worse.
It could be worth revisiting with Stan Beresford for the Caiman-II DAC, although that won't help CD and vinyl playback, my primary sources.
|
|
|
Post by aurender on May 24, 2015 12:10:14 GMT
It seems like it could be worth revisiting. To be effective you would need to be able to lock sources, DAC and any digital processing to the same clock.
One step I'd like to take is to bypass the need for a separate 10mHz clock by having the master clock generate an exact multiple of the 44.1 or 48 clock sets used in digital music. One key difficulty now is that following sale of Swiss company Oscilloquartz, there is currently no available commercial BVA clock producer.
|
|
|
Post by stanleyb on May 24, 2015 14:14:12 GMT
It could be worth revisiting with Stan Beresford for the Caiman-II DAC, although that won't help CD and vinyl playback, my primary sources. Super Firmware has it all built in .
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 24, 2015 14:14:17 GMT
Hi David
I write a detailed post concerning clocking in general but the bloody I pad has just sodding lost it!! Will take me a good hour to do it again and I just do have the time
I stopped uing dedicated aftermarket atomic clocks that connect via BNC. I now install the units directly to the equipment concerned, fetmo clocks seem to be the. Biggest Noise in the forum boards but they need a dedicated set of conditions to generate they ultra low fetmo numbers and very few companies can met these conditions to actually take advantage of this.
I have also found that the power supply to the clock is as important as the clock itself
My server has a dedicated 100mhz atomic clock fitted to for its master clock generator, also one for the North bridge too, the Dac also has a very special GPS global positioning clicking device this produces much more tangible results than fitted an aftermarket clock hooked up by BNC cables
The shorter and more electrically isolated the signal pathway is the more accurate the results will be imho
All good fun though
|
|
|
Post by aurender on May 24, 2015 15:01:09 GMT
Tony That's interesting. Certainly I'd like to avoid the need for two sets of mathematical conversions: first to generate 10mhz, then second set to generate 44.1 or 48 and multiples from the 10mhz number. It would ideally be possible to produce a clocking device which involved whole number division to get from the basic crystal frequency to the digital audio frequency.
If you need best short term accuracy and low phase noise, BVA is very much ahead of the game. So called atomic rubidium clocks do not perform well in that area and GPS clocks perform even worse.
sFZ in Japan produce(d?) a BVA based clock for audio using the now unavailable Oscilloquartz 8607 with a very good psu all mounted in a solid aluminium case at a cost of over $40k. I'm still exploring getting better power supply and replicating similar casework.
In my system, master clock feeds Aurender, Squeesebox Transporter, DCS clock, DCS up sampler, DCS DAC, DCS transporter, so inevitably master clock will need cable connections to all of those (until I can get 44.1/48 direct from my BVA clock.
You are certainly correct on clocking cables where most commercial cables give significant signal reflections at the connectors. WBT BNC connectors are particularly good at minimising reflections.
|
|
|
Post by aurender on May 24, 2015 15:04:00 GMT
Tony, one point I'd like to understand in your post: are you advocating multiple reference clocks in separate pieces of equipment? That would, on the face of it, to totally negate the benefits of reference clocking.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 24, 2015 16:38:11 GMT
Hi David
One of my remits is digital electronics and power supplies I design for a number of manufactures
Now matter how good the clock and supporting psu off the shelf units only go so far you need to look at the whole concept of what is trying to be achieved, the cost, manufacturing, what market it is bring placed in and is it a unique product for a dedicated purpose rather than a generic piece made to fit
Going back to your post regarding dedicated frequencies points relating to either direct multiples or various half multiples of red book standards
Esoteric's Rb clocks have the ability to either give you a industry standard ref of 10mhz or any frequency from 44.1khz to 192khz When using them we found that 176.4khz produced the most pleasing results with esoteric equipment
Have used these with various so called hi end digital products over the years to varying degrees of effect dependant on TX and RX interfaces and the noise rejection.
For example the DCS Vivaldi does sound a lot less contrived and more natural when tying in that particular clock to a full stack than the DCS master clock, the same can be said for an equivalent studio unit as well implementation is the key
Let's look at what you are trying to achieve with having a master clock, the aim is to generate a high quality independent ultra stable square wave with ultra lower variants
This will then supply a CDT, Dac, streamer, adc, digital work bench, pro tools, digital recorder with the idea of tying them together as accurately as possible so they in essence sing from the same hymn sheet
Now the master clock has to supply each piece of equipment in turn, usually by BNC, now this then introduces issues of its own, TX and RX interfaces, cable variations, circuit pathway length, emi/ rfi interference at source point and delivey point digital cables here are indeed very critical and can made a significant difference to how the precived sound is portrayed.
When looking at digital signals particularly clock and PCM signals through a spectrum analyser it is quite easy to observe any reflectional issues with a particular cables weather BNC or USB, some give a decided curved bow shape to the leading edge others are more 'ragged, some look very close to the ideal vertical leading edge of the square wave form.
In reality not all those that project an ideal pattern sound they way you feel they should.
The BNC connectors are only half the story, you will need to have the same attention to detail all the way through the TX and supply circuitry if you wish to eliminate as much of the errors as possible. Do you have matching board mounted BNC Recpticals etc? With regard to asynchronous USB data transport I find this flawed, I realise that 99% of pc etc have this form of connections think about the latency the very best this method can offer is 4msec.
Currently I am listening to Marillion @ 41usec of latency not a USB in sight Sorry I digress
Presavation of data integrity as well as accurate clocking is paramount in quality digital replay
You also raised the point of multiple ref clocks which is a valid one.
Could it be with the new clock designs being far more stable, accurate, smaller and consuming less power that local master clocks for specifics tasks could be the lesser of two evils?
I have a master ref clock for the CPU cores, a more accurate clock here is obvious in a PC and computational engine if you are sending your audio data via dedicated LAN bus then clocking issues are global within the network stream.
The Dac clock is a rock solid design which produces a very natural, open, totally and utter grain free natural flowing sound that makes engaging unforced music
I have the ability to use a Rb clock in this system on the digital delvery platform oddly enough it is still sat in the next room, again implementation and design specifics are instrumental here
You would feel the using two or more master clicks would be counter intuitive and make things worse, I would say the results speak to a different tune
As with all things in the 21st century there is more than one way to skin a cat
Did you get round to building your main listening room David as I see you dropped the Rockports for the YG's? Size wise more accommodating
|
|
|
Post by aurender on May 24, 2015 17:24:11 GMT
Tony Interesting debate.
Esoteric has a decent rubidium clock. 176.4 is fine but how is that going to help you with a source generating a signal at 48ks?. Similarly if you are using 176.4 with a 44.1 signal you will have some speed problems.
No contest on clock accuracy. Locking everything to a BVA is, to my ears, a huge improvement. Spec of the Oscilloquartz 8607 way ahead of rubidium or best "fe to" clock. Until you've heard the effect of a BVA master clock, you will not see the difference. Sadly commercial availability is now a serious problem.
For my understanding, it would be good to what the time signal from the source is actually contributing when you lock onto the time signal in the final component?
Agree with you about asynchronous sub. It's a less than perfect solution but for some components, it's a reasonable solution.
Different topic: music room project hit major delay just over a year ago when house suffered major flood damage following the heavy storms we had. Repair work still not complete and the amount of disruption to our lives has led us to a different solution which doesn't involve living any longer in a building site. Hopefully the new concept should enable a very nice listening area.
|
|
|
Post by MartinT on May 24, 2015 20:10:30 GMT
Super Firmware has it all built in . Care to expand on that a little, Stan?
|
|
|
Post by MartinT on May 24, 2015 20:12:53 GMT
Certainly I'd like to avoid the need for two sets of mathematical conversions: first to generate 10mhz, then second set to generate 44.1 or 48 and multiples from the 10mhz number You'd be better off starting with an 11,289,600Hz clock.
|
|
|
Post by aurender on May 24, 2015 20:42:42 GMT
Certainly I'd like to avoid the need for two sets of mathematical conversions: first to generate 10mhz, then second set to generate 44.1 or 48 and multiples from the 10mhz number You'd be better off starting with an 11,289,600Hz clock. That's right. I discussed making a rubidium with Quartzlock, especially for DCS. If Oscilloquartz had continued with their 8607 BVA, I would have commissioned a BVA at that frequency which could have been the basis for a 44.1/48 master clock.
|
|
|
Post by stanleyb on May 24, 2015 21:50:24 GMT
Super Firmware has it all built in . Care to expand on that a little, Stan? Not really. I prefer if others worked it out for themselves. It did take me several years to figure it out and get it right after all. I do use three clocks in the CMII by the way.
|
|
|
Post by aurender on May 24, 2015 22:29:10 GMT
Care to expand on that a little, Stan? Not really. I prefer if others worked it out for themselves. It did take me several years to figure it out and get it right after all. I do use three clocks in the CMII by the way. thats obviously in the spirit of sharing information which is the basis of a good forum!
|
|
|
Post by stanleyb on May 25, 2015 5:57:13 GMT
Not really. I prefer if others worked it out for themselves. It did take me several years to figure it out and get it right after all. I do use three clocks in the CMII by the way. thats obviously in the spirit of sharing information which is the basis of a good forum! Where does sharing end and subtle shilling start? It's best to play safe in cases like this and avoid any conflict of interest. I have been accused over the last ten years of talking up my products in order to sell more. So now I just stay away from things like that.
|
|