Silicon Chip DAC kit

[kimbo]

The lastest SC mag has the first part of an article describing construction of a DAC....... "High-Quality Stereo Digital-To-Analog Converter, Pt.1".

I've been waiting for the last few days to see if anyone would be posting their comments, but as there have been none I thought I'd start the ball rolling.
Of course the circuit is not available from their website (Understandable, they have to sell their mags), but here is the teaser......
http://www.siliconchip.com.au/cms/A_111597/article.html

I had a very quick look at the newstand and see that it uses a PCM1796 (I think).

And here is a thread in the overclockers forum where the author of the article makes some comments. (note that it is not the author who calls it 'ultra high quality'....he was just trying to design something affordable, straightforward to build and best performance within those boundaries)
http://forums.overclockers.com.au/showthread.php?t=810680

I have no experience with DACs or anything digital having mostly built amps, but occasionally I've thought about dipping my toes into building one, and this may perhaps be a good introduction.

Anyway I'm very interested to hear any comments from members who have actually read the article.

 

[ttan98]

I am currently conducting a survey of the chipset used in DAC and what to look for and what to avoid. Without going into details which will cover a few pages I like to summarise my take/opinion of the chipset to look for. The analog, power supply and receiver chipset is another story.

Oversampling

AD1955
PCM1794

also a lesser know chipset is AD1852 (*) and also wolfson 8740 and AKM 4394

Non-oversampling
TD1541
AD1865(*)

* If you at looking into getting high quality DAC, Dacs based on these 2 chipsets will give you high end quality sound reproduction without the high price tag.

Go to this site for details:
http://www.audiodesignguide.com/DAC_final/DacFinal.html

 

[kimbo]

Thanks TTAN, but I don't think that is particularly helpful.
I am curious to hear from others who may have read this particular article in Silicon Chip. It may be a few days before other Austrailan members get their hands on the issue. Perhaps there will be a rush at the local library this weekend.....its always good to give support to these fine institutions!

 

[ttan98]

There are a LOT of off the shelf DACs from Ebay that are of good value and you can purchased them/it already assembled and plug and play. These boards are based on the chipsets that I mentioned above. I suggest you look into this before building the one from SC.

Log on to US Ebay, ie ebay.com and search for DAC under electronics category. All these boards are of excellent value and sound good too, I purchased a few boards to play with myself. There MANY varieties to pick from. The decision is yours, believe me made in China do not mean inferior quality. Believe me the chipsets you pick determines the type of sound reproduction you want.

Cheers.

 

[kimbo]

thanks for the prompt response TTan.

I supose my problem is that I just like taking over the kitchen table and putting stuff together......I always learn more and I just love that smell of hot resin, but yes, if the SC one gets the thumbs down I'll probably buy something off the shelf.

I'll see if I can sneak a look at the library and post some more details about the chipset used.

Cheers,

 

[col]

I recently built a couple of SC projects, The Multi-Function Active Filter KC4580 and the Bridge Adapter for Stereo Amps KC4569. They were very easy to build and have good documentation. The PCBs are good quality but very large. The kits performed very well. I'm not sure how good quality the design is though, they use a lot of NP caps.

http://minirig.org.au/2009/08/14/minirig-summed-low-pass-bridge-mode-adapter-for-subs/

I wouldn't mind having a go at the DAC kit but have no need for it.

col.

 

[ianrt]

Silicon Chip DAC Project

I've read the article in SC

The DAC project is a good one, but not a very good one.

It uses a DIR9001 re-clocker and a DSD1796 DAC providing up to 24/96k decoding.
It has separate power supplies for analog and digital sections (although these use the one transformer and diode bridge)
Output stage is per 1796 datasheet but the project allows for the use of OPA134 instead of NE5534, and is single-ended only (without the pcb layout it is difficult to know how easy one could mod it for balanced output but it should be do-able)

Output frequency response starts falling above 24kHz (but if modded for balanced output you might be able to shift this up to 70kHz)

There is some strange problem (perhaps from using the one analog ps for both channels) that reduces crosstalk performance by 20dB at 10kHz and 30dB at 20kHz. Again without seeing the pcb layout I don't know whether separate supplies are easily adopted for each channel to try to fix this.

This project would be OK for anyone who doesn't have an outboard DAC to improve sound from mid-level CD players or PC-based material but if you're looking for top performance, try elsewhere.

I hope this helps.

Ian

 

[nvinen]

Output stage is per 1796 datasheet but the project allows for the use of OPA134 instead of NE5534, and is single-ended only (without the pcb layout it is difficult to know how easy one could mod it for balanced output but it should be do-able)

It isn't too hard, this information will be published in a future edition.

Output frequency response starts falling above 24kHz (but if modded for balanced output you might be able to shift this up to 70kHz)

I'm fairly sure there are no (or at least very few) humans who can hear above 24kHz, is there a reason why you care about the frequency response above that?

My own hearing limit is about 18kHz. I don't have particularly bad hearing loss.

There is some strange problem (perhaps from using the one analog ps for both channels) that reduces crosstalk performance by 20dB at 10kHz and 30dB at 20kHz. Again without seeing the pcb layout I d.on't know whether separate supplies are easily adopted for each channel to try to fix this.

It is due to the limited CMRR of the differential amplifier stages and coupling through the power supply. The balanced modification does greatly increase channel separation. However, channel separation performance of the unbalanced outputs is more than adequate. The difference between 70dB and 120dB channel separation will not be audible except under special circumstances - certainly not during regular listening.

This project would be OK for anyone who doesn't have an outboard DAC to improve sound from mid-level CD players or PC-based material but if you're looking for top performance, try elsewhere.

I think you'll find it will be hard to beat the performance in the sense that the output is very clean and low distortion, even compared with a high end CD player. As will be revealed later, typical distortion is actually somewhat lower than originally quoted. If you are looking for something other than low distortion and low noise then all bets are off.

 

[kimbo]

Hi nvinen,

thanks for the post. I see you have dropped a few hints of more to come both here re 'balanced output', and on another forum with "Keep an eye out over the next few months, there will likely be a USB add-on."

I may be wrong but I'd think that any improvement to the crosstalk can only be a good thing as far as imaging goes.....that soundstage illusion is fragile and every bit counts.

For me, I not concerned about USB. I'm just curious to see what improvement I get if I use an outboard DAC on my yamaha DVD/SACD player. To be honest I'd rather splash a whack of cash on a dedicated quality CD player but being a family man, I can only weedle a little spare cash for my hobby. A DAC would be cool as the trouble & strife wouldn't know what it is, but a CD player, it would be ....'what do you need that for?'

As I said, I'm new to DACs and digital in general so I'm pretty keen to get soldering, however cost will probably be the key factor in me following through with a build, so I'm waiting for the usual vendors to come out with a kit before I jump in. If it is too high for my budget I'll probably think about other options....perhaps something from the states when the missus visits her mom in NJ at Xmas.

 

[kimbo]

Au$139.00

however cost will probably be the key factor in me following through with a build, so I'm waiting for the usual vendors to come out with a kit before I jump in. If it is too high for my budget I'll probably think about other options....perhaps something from the states when the missus visits her mom in NJ at Xmas.

Dropped into Jaycar on the way home for something else and picked up the latest flyer. The kit is advertised as a shortform for AU$139. This is sort of OK....but by the time I cased it and provided a transformer......hmmm is the power supply on board? .....can't remember....anyway final cost will be closer to $200 so I may think twice about it.

I'll wait a few weeks for other members to get around to checking it out and perhaps comment on the bangs for bucks.

 

[sandyK]

Sc dac

"I may be wrong but I'd think that any improvement to the crosstalk can only be a good thing as far as imaging goes.....that soundstage illusion is fragile and every bit counts."

I agree with Kimbo. I would also go further and state that just because you can't hear a frequency directly, doesn't mean that it has no effect. Many recent designs ensure that the harmonic structure as high as the 8th harmonic is retained, as it aids in giving instruments their own individual sound, and wider bandwidth also helps preserve the "space" between instruments and voices. The human ear does appear to be sensitive to the rise and falltimes of waveforms. Drums ,as an example , have greater impact in a wideband design.
It would be interesting to insert a good low noise dual polarity shunt in line between the PSU and the DAC. This should also help the 7805 supplying the DAC. (Quite a few DIYAudio members have found that a good shunt such as the Salas and others, even makes a worthwhile improvement to SQ when feeding the 5V and 3.3V DIGITAL regulators of DACs such as the latest Twisted Pear Buffalo DAC.)
It's not so much the improved ripple response that makes the difference, it is the very low, and FLAT impedance of the power supply to >300kHz.
This should also improve channel separation , especially above 10kHZ.

SandyK

 

[nvinen]

Hi nvinen,

thanks for the post. I see you have dropped a few hints of more to come both here re 'balanced output', and on another forum with "Keep an eye out over the next few months, there will likely be a USB add-on."

I may be wrong but I'd think that any improvement to the crosstalk can only be a good thing as far as imaging goes.....that soundstage illusion is fragile and every bit counts.

You may be right, but the way I think of it is this: say you have a channel separation of 60dB. The SC DAC is significantly better than that, especially around 1kHz, but let's just say that's the case for now. Now, clearly you can hear a -60dB signal in a quiet room coming from one speaker. But keep in mind that if you have 60dB channel separation and are playing a full scale sound from a single speaker and the other is silent. This means the amplitude of the signal being sent to the other speaker due to crosstalk is 1/1000th that amplitude. I'm really not sure you'd be able to detect that signal with the other one overwhelming it, especially since if you are not using headphones, some of the sound from each speaker goes to both ears.

I'm away from home at the moment, when I get back I'll hook one of the DAC prototypes up to my best amp and speaker and play some content with good stereo encoding. I'll see if I can spot any issues with it.

Since I'm fairly sure the roll-off in separation with frequency is due to lack of CMRR in the differential stages, if you are still concerned there are a few things that might help. One would be using 0.1% tolerance resistors in the differential network. Another would be increasing the size of the op amp bypass capacitors (larger MKTs) which may reduce any power supply coupling issues. You'd need a good audio millivolt meter to check whether these changes make any improvements.

For me, I not concerned about USB. I'm just curious to see what improvement I get if I use an outboard DAC on my yamaha DVD/SACD player. To be honest I'd rather splash a whack of cash on a dedicated quality CD player but being a family man, I can only weedle a little spare cash for my hobby. A DAC would be cool as the trouble & strife wouldn't know what it is, but a CD player, it would be ....'what do you need that for?'

Do you have any specificaions on your DVD player? Most DVD players the Silicon Chip folks tested had pretty lousy THD, although if yours is also an SACD player then you'd hope it would be a bit better. If you can mention the model number there may be some specs or measurements floating around.

One nice thing about the DAC, though, is it has multiple inputs and you can also use it with computers or other devices which have lousy internal DACs - i.e. you get good quality out of any digital audio source, not just CDs.

As I said, I'm new to DACs and digital in general so I'm pretty keen to get soldering, however cost will probably be the key factor in me following through with a build, so I'm waiting for the usual vendors to come out with a kit before I jump in. If it is too high for my budget I'll probably think about other options....perhaps something from the states when the missus visits her mom in NJ at Xmas.

The Jaycar kit will be cheaper since it won't come with the case, transformer, etc. however once you have added those costs in it may not be worth it. I should point out, though, that there is a "budget option" and that is to install the DAC in a plastic case and use a 15 or 16VAC plug pack to power it. Performance does not suffer. Jaycar sell a suitable plastic case for about $20 (I can probably find the part # if you need it). You will of course have to drill all the holes yourself, whereas if you buy the Altronics kit when it becomes available it should be substantially easier.

I don't know what the kits will cost, hopefully they will be released in the next month or so.

 

[nvinen]

Dropped into Jaycar on the way home for something else and picked up the latest flyer. The kit is advertised as a shortform for AU$139. This is sort of OK....but by the time I cased it and provided a transformer......hmmm is the power supply on board? .....can't remember....anyway final cost will be closer to $200 so I may think twice about it.

I'll wait a few weeks for other members to get around to checking it out and perhaps comment on the bangs for bucks.

Hey, that's a pretty good price, I'm pleased about that. See my last post for information on the budget option. More work, but the result is good - although it doesn't look as nice.

Adding in the cost of the plastic case, plug pack and other bits and pieces you need, it'll still probably be <$200 and I think the performance is great for that sort of price level.

 

[nvinen]

"I may be wrong but I'd think that any improvement to the crosstalk can only be a good thing as far as imaging goes.....that soundstage illusion is fragile and every bit counts."

I agree with Kimbo. I would also go further and state that just because you can't hear a frequency directly, doesn't mean that it has no effect.

OK fair enough. I'd just like to say that listening to it, there doesn't appear to be anything wrong with the stereo imaging as a result of the less-than-ideal channel separation. I also didn't hear anything to suggest that harmonics were missing but there may be some benefit increasing the lowpass cutoff frequency. Doing so shouldn't be too difficult - you could use smaller capacitors in the filter stages but you would have to be careful to avoid introducing passband ripple or additional distortion.

As I wrote earlier I'll do some more critical listening when I get home.

 

[GK]

Not particularly enamoured with that three opamp differential-input I/V convertor circuit suggested in the ‘1796 datasheet.

I think a much better, “single chip” alternative would be with a fully differential opamp, such as this one:

http://focus.ti.com/lit/ds/symlink/opa1632.pdf

The you just have two feedback resistors to match for the best CMRR.

 

[nvinen]

Not particularly enamoured with that three opamp differential-input I/V convertor circuit suggested in the ‘1796 datasheet.

I think a much better, “single chip” alternative would be with a fully differential opamp, such as this one:

http://focus.ti.com/lit/ds/symlink/opa1632.pdf

The you just have two feedback resistors to match for the best CMRR.

The output is still differential though, so it doesn't entirely solve the problem.

What would probably be best would be an instrumentation amplifier with good audio performance. However, I didn't want to use especially exotic/expensive/hard-to-get parts in the DAC project where avoidable.

 

[GK]

The best "audio" instrumentation amp for the application that I can think of would be this one:

http://focus.ti.com/docs/prod/folders/print/ina217.htm ($8 at Farnell)

The I-to-V would just have to be done with a pair of resistors from the DAC outputs to (signal) ground.

That fully-differential opamp would still perform better here though, IMO.

Anyway, I'm just talking from a perspective of achieving the best measured performance; I don't think there is anything to be gained (audio-wise) over the circuit you've got.

 

[sandyK]

Nicholas
I hope that at a future date, you will consider a replacement PCB for the input that is 24/192 capable, even if the figures aren't quite as good.
Quite a few people I know (even in the U.K.) would have constructed your design if it had this capability. One of the most common requests in forums these days , is a request for a DAC that will natively handle the 24/192 downloads, from Linn Records, and the 24/176.4 downloads from Reference Recordings etc.
Nevertheless, your design fills a big gap in the area of affordable DIY DACs, and is certainly a fine building block, as are many of the Silicon Chip designs.
In my case, I use a heavily modified Musical Fidelity X-DAC V3, which I use mainly with the P.C. , but also move to the speaker system for serious music listening. Your design, after perhaps a minor amount of tweaking (yes, an inline JLH PSU addon module too) will save me a lot of hassle moving everything around, and will hopefully approach the SQ of the X-DAC V3, which incidentally, only does up to 24/96 , although it does upsample to 24/192.
Thank you for presenting your design, and the effort involved in getting it available via Jaycar and Altronics.
BTW, the Jaycar ad is not very clear, and only shows the DAC board in the photo.
The quoted price kit seems too good to be true for the full complement of boards.
Being so new, it does not show anything on their website as yet.

SandyK

 

[kimbo]

Nicholas

Thank you for presenting your design, and the effort involved in getting it available via Jaycar and Altronics.

Yes, thanks Nicholas. I hope I did not sound too negative earlier. I was able to borrow the Sept issue yesterday and have a good look at it. I can see what a lot of work went into it.

I've spent a few hours looking at other threads and designs and I've got to say that I've become bit overwhelmed by the variety and complexity of DAC design. The availability of a kit and clear directions about it's implementation into a finished complete solution is really appealing. I toyed with purchacing one of the Chinese units from eBay, but it's hard to get a handle on their value, ease of implementation and importantly, reliability especially as the issue of counterfit parts is also part of the mix. I find myself more drawn to your design.

My needs are relatively modest like my system and yours almost fits the bill. You asked about my player. I'm at work so I can't check (sure hope the boss isn't reading this post!) but I think it is a yamaha S657 from around 2005. The published specs don't include THD and are :

DVD-Audio/Super Audio CD Playback
CD Upsampling
192kHz/24-Bit Audio D/A Converter
Audio Direct ( the video circuitry can be turned off.....so they say)
AUDIO DAC 108 MHz / 12 bit

Cheers, Kim

 

[nvinen]

The best "audio" instrumentation amp for the application that I can think of would be this one:

http://focus.ti.com/docs/prod/folders/print/ina217.htm ($8 at Farnell)

The I-to-V would just have to be done with a pair of resistors from the DAC outputs to (signal) ground.

The INA217 looks pretty good, I should pick some up and try them out.

The DAC outputs only sink current so you would have to connect them to a positive voltage and it would have to be well filtered or else supply noise would get into the output. I'd stick with active I/V converters. I suspect replacing the final op amps with instrumentation amps (and changing the resistors appropriately) would be all you'd need to do in order to improve the channel separation and an interesting project.

That fully-differential opamp would still perform better here though, IMO.

It might be possible to combine them, although I don't think it would be necessary. I don't know what instrumentation amplifier the Audio Precision gear uses internally but when I rely on that instead of the OPA134s to handle the differential signal coming from the filter the performance does get a bit better. That's why I thought using an instrumentation amplifier might be a good idea.

Anyway, I'm just talking from a perspective of achieving the best measured performance; I don't think there is anything to be gained (audio-wise) over the circuit you've got.

That's basically why we stopped tweaking it at the stage we did - while we know it's possible to make further improvements it was felt that they would have little audible effect. I wanted to avoid adding extra complexity or using parts which are difficult to get for improvements which were measurable but unlikely to be audible.