One reason I wasn't too concerned about the lack of 192kHz support is that my neighbour is a sound engineer and when I had a chat to him, he said that he thinks that 96kHz/24bit is audibly superior to 48kHz/16bit, but that going to 192kHz gains little. I have some 192kHz content and it would be nice to play it back at full rate but I wouldn't feel bad downsampling it to 96kHz.
Having said that I have a couple of Cirrus Logic/Crystal S/PDIF receiver IC capable of 192kHz on hand and it shouldn't be particularly difficult to design a board to accept them instead of the DIR9001. It won't be utterly trivial but since they both do roughly the same job it wouldn't be too bad. I'll mention this to the SC folks and see whether they think it would be worth publishing a design.
I'll be interested to hear what people think of the performance, and I definitely would like to see any modifications. If someone comes up with a clear improvement to the design they should sent it into the magazine in the "Circuit Notebook".
Well I sure hope the total price for a kit isn't too high, I guess we'll see.
Hmmmm... I agree that active I-V conversion is the way to go - for some reason I though you were suggesting an in-amp as an alternative for this function. If one went this route though, I'm not sure the Iout/-Iout load resistors would need to be referenced to a positive voltage, as the diff-amp doesn't amplify the common mode (negative) DC offset voltage at its inputs.
You'd just have to make sure that resistors used are not too large, as the linear output voltage range of the DAC may be exceeded.
Anyway, I-out DAC's always suffer degraded linearity if the I-V conversion is done passively like this; much better to do it actively and maintain Iout/-Iout at 0V!
You'd just have to make sure that resistors used are not too large, as the linear output voltage range of the DAC may be exceeded.
Anyway, I-out DAC's always suffer degraded linearity if the I-V conversion is done passively like this; much better to do it actively and maintain Iout/-Iout at 0V!
Last edited:
The feedback here has been interesting. I definitely want to ask the SC folks what they think of the stereo imaging issue.
Although it wasn't my original intention, the way the DAC is split up into multiple boards can have some advantages for experimentation. For example, if I'm going to build a 192kHz version I will probably build a new input board prototype by hand and connect it to the other boards as published. That should be easier than trying to modify a more monolithic design. You may similarly find that to be an advantage.
I found some information here:
http://www.sorishop.com/board/gallery/board_view.html?no=1508&page=1&category=300&no_list=
It seems the THD is listed as 0.003% but it doesn't say what sampling rate that's at (or maybe it does, I can't read Japanese). So it should sound pretty good as it is, but adding the SC DAC would improve the distortion somewhat - it would almost certainly halve it, probably do better than that. Whether the difference will be audible I can't say. You'd have to use the internal DAC for 192kHz content.
I'm suggesting to leave the design unchanged except for the final op amp stage and replace that with the instrumentation amplifier. The main intention would be to improve the CMRR of the differential to single ended stage, since as I said when I removed the differential amplifiers from the equation by measuring the performance of the DAC with the differential inputs of the Audio Precision gear, channel separation no longer dropped off with frequency. So I surmised from this that the reason it drops off with frequency in the original design is that as the frequency increases on one channel, the noise being picked up on the differential lines of the other channel increases. With a high enough CMRR this can be rejected when those two lines are subtracted, with a lousy CMRR much of it remains, hence the degraded separation. The reason for the relatively poor CMRR of the op-amp based design is likely the resistor matching, and an instrumentation amplifier uses trimmed resistors for this reason, so it should help.
When I buy some of the chips you recommended and get a chance to wire up a prototype that should indicate whether my theory is correct, but unfortunately I probably won't get a chance to do that for at least a couple of months as I have some other projects to work on (and my day job too).
Nicholas
Silicon Chip DAC Kit
"The feedback here has been interesting. I definitely want to ask the SC folks what they think of the stereo imaging issue."
Nicholas
It's not so much an imaging issue, as the lack of a 3D type soundstage.
This is often a typical comment about SC amplifier designs, and Douglas Self designs in particular. They end up with fabulously low distortion figures, but the amplifiers as published do not normally exhibit a very wide soundstage, or a great deal of depth to the stereo image. The earlier SC ULD was a step in the right direction, with its front end fed from a regulated power supply, but IMO didn't go far enough in that respect. Incidentally, if you checked the last page of the SC ULD2 link that you were given in another forum, you would have noticed that the suggested modification had only a minor effect on ultimate distortion figures, and that the modified version as tested by Jeremy,actually had distortion figures lower than the published figures. If your LTP devices are well matched for both HFE and VBE, you may be quite surprised at the audible improvement in soundstage,with a reduction in harshness, and what appears to be a lower noise floor. These improvements have been verified in posts by several DIYAudio members in several different DIYAudio threads .
Incidentally, in 1999 Halcro patented the addition of an added diode to the normally tied collector -base junction of the current mirror. This modification gives a similar result, as it helps keep the LTP devices balanced with temperature, as both transistors then have virtually identical collector currents. Improvements to the current mirror give similar gains.
SandyK
"The feedback here has been interesting. I definitely want to ask the SC folks what they think of the stereo imaging issue."
Nicholas
It's not so much an imaging issue, as the lack of a 3D type soundstage.
This is often a typical comment about SC amplifier designs, and Douglas Self designs in particular. They end up with fabulously low distortion figures, but the amplifiers as published do not normally exhibit a very wide soundstage, or a great deal of depth to the stereo image. The earlier SC ULD was a step in the right direction, with its front end fed from a regulated power supply, but IMO didn't go far enough in that respect. Incidentally, if you checked the last page of the SC ULD2 link that you were given in another forum, you would have noticed that the suggested modification had only a minor effect on ultimate distortion figures, and that the modified version as tested by Jeremy,actually had distortion figures lower than the published figures. If your LTP devices are well matched for both HFE and VBE, you may be quite surprised at the audible improvement in soundstage,with a reduction in harshness, and what appears to be a lower noise floor. These improvements have been verified in posts by several DIYAudio members in several different DIYAudio threads .
Incidentally, in 1999 Halcro patented the addition of an added diode to the normally tied collector -base junction of the current mirror. This modification gives a similar result, as it helps keep the LTP devices balanced with temperature, as both transistors then have virtually identical collector currents. Improvements to the current mirror give similar gains.
SandyK
Last edited:
http://www.diyaudio.com/forums/showthread.php?t=133018&highlight=current+mirror&page=15
See especially the replies from Ostripper,and on earlier pages the contributions from AKSA.
See especially the replies from Ostripper,and on earlier pages the contributions from AKSA.
Last edited:
Kit availability
Just a heads up, it looks like the Jaycar kit for this project will be available within the next couple of weeks (i.e. late November). The kit should include the three main PCBs and all the components that go on them. They also sell the power supply kit separately (a little over $20 I think) and then you will need to add a transformer or plugpack and a case. The only other components required are miscellaneous things such as the power connector, wiring, bolts, etc.
Just a heads up, it looks like the Jaycar kit for this project will be available within the next couple of weeks (i.e. late November). The kit should include the three main PCBs and all the components that go on them. They also sell the power supply kit separately (a little over $20 I think) and then you will need to add a transformer or plugpack and a case. The only other components required are miscellaneous things such as the power connector, wiring, bolts, etc.
N.Vinen's SC DAC
Nicholas
Last Saturday morning Altronics in Sydney gave a tentative date for the release of their full kit as 20th November. However, since then a friend has been told 1st December by W.A. .
Have you any thoughts on using the Rockby 24.576mHz xtal oscillator modules which are available for $10 (!) each, instead of the simple xtal ? They will be 5V versions, and the output would need to be padded down to suit the DIR9001 which does not handle TTL level . I had in mind something like a small PCB with a 78L05 adjacent to the DIP oscillator, running from the main +15V rail.
SandyK
P.S.
I have ordered 2 of those xtal oscillators, and did you by any chance grab the remaining 2.7nF polyprops from Farnell for use in your prototype ? 🙂
Nicholas
Last Saturday morning Altronics in Sydney gave a tentative date for the release of their full kit as 20th November. However, since then a friend has been told 1st December by W.A. .
Have you any thoughts on using the Rockby 24.576mHz xtal oscillator modules which are available for $10 (!) each, instead of the simple xtal ? They will be 5V versions, and the output would need to be padded down to suit the DIR9001 which does not handle TTL level . I had in mind something like a small PCB with a 78L05 adjacent to the DIP oscillator, running from the main +15V rail.
SandyK
P.S.
I have ordered 2 of those xtal oscillators, and did you by any chance grab the remaining 2.7nF polyprops from Farnell for use in your prototype ? 🙂
I don't see it on their web site yet, so the 20th seems premature. Hopefully it will be ready soon.
I suppose it should be possible although I don't really think it's necessary as the 24.576MHz crystal is only used as a frequency reference - the receiver needs a reference so it can calculate the sample rate of the incoming data. However it isn't actually used as a clock source. However if you really wanted to do this, you could run it off the digital +5V supply or a separate +5V supply derived from the +15V. As you say the tricky bit will be converting it to a 3.3V signal as the XTI pin on the DIR9001 is not 5V tolerant. Crystal oscillators don't always output a full swing signal so if you're lucky you could feed it into a 74HC04 running at 3.3V and not exceed its maximum input pin voltage of VCC+1.5, i.e. 4.8V. Then the output would be suitable for connection to the XTI pin. I'd measure the output swing of the oscillator first (ideally using a scope) and that should help reveal the best way to hook it up.
24.576MHz oscillators are actually pretty useful for use in devices where you actually do need a clock source, such as a signal generator or computer sound interface. Maybe it'd be worth keeping them aside for that kind of project. I've certainly had to build such oscillators in the past - in fact I'm working on a new project which uses a 24.576MHz crystal as a clock source, but I kinda cheated there and am using the internal oscillator of a microcontroller to avoid needing a separate oscillator circuit - the microcontroller has a PLL so I'm able to derive a suitable asynchronous clock for the same micro from the same crystal.
No, while I do occasionally order from Farnell (and have recently) I didn't get the 2.7nF capacitors from there. I ordered a bunch of them (25 I think) from Futurlec for the original prototype and had enough left over to build several more.
By the way the Jaycar kit is now available here.
Full kit now available from Altronics
Saw this today...
Altronics - Your One Stop Audio Visual & Electronics Supplier
AU$ 299
I wonder if there are any members considering it?
Saw this today...
Altronics - Your One Stop Audio Visual & Electronics Supplier
AU$ 299
I wonder if there are any members considering it?
I know of 2. I built and working, one still waiting for inspiration. They might chime in with some details. 😀
I hear they got their prototype working today. However it may still be a little while until the kits are ready.
Hopefully it will be worth the wait.
Nicholas
With at least some of the Jaycar DAC boards, there is a missing short track between pin 3 of IC12 and the 180R resistor.
Although most appreciative of the effort you have put into this design, the Input PCB is a PITA ! Due to the VERY thin leads of the Blue LEDs in the pushbutton switches, there is every chance that when trying hard to avoid burning the hell out of the switch itself, even when using a small iron such as the smallest Goot soldering iron, there is a very good chance that solder will not flow on to the leads themselves.😱 Ideally, this PCB only should have been double sided, or at the very least had these critical holes plated through.
I am also very surprised that SC did not requiure Drilling templates to ensure neat looking and properly spaced holes for the switches, LEDs and sockets of the various PCBs when mounted to the front and rear panels.
SC usually do that.
SandyK
P.S.
I understand from a friend who has completed his build, that there are huge gains in soundstage and overall sound quality to be made simply by using polyproplyene capacitors in the DAC PCB time constant areas,(2.7nF,8.2nF,27nF) and especially with a 4.7nF polypropylene on pin 22 of the DIR9001.(There is an error there too, but I will let my friend clarify that.)
I expect that my friend will elaborate shortly.
With at least some of the Jaycar DAC boards, there is a missing short track between pin 3 of IC12 and the 180R resistor.
Although most appreciative of the effort you have put into this design, the Input PCB is a PITA ! Due to the VERY thin leads of the Blue LEDs in the pushbutton switches, there is every chance that when trying hard to avoid burning the hell out of the switch itself, even when using a small iron such as the smallest Goot soldering iron, there is a very good chance that solder will not flow on to the leads themselves.😱 Ideally, this PCB only should have been double sided, or at the very least had these critical holes plated through.
I am also very surprised that SC did not requiure Drilling templates to ensure neat looking and properly spaced holes for the switches, LEDs and sockets of the various PCBs when mounted to the front and rear panels.
SC usually do that.
SandyK
P.S.
I understand from a friend who has completed his build, that there are huge gains in soundstage and overall sound quality to be made simply by using polyproplyene capacitors in the DAC PCB time constant areas,(2.7nF,8.2nF,27nF) and especially with a 4.7nF polypropylene on pin 22 of the DIR9001.(There is an error there too, but I will let my friend clarify that.)
I expect that my friend will elaborate shortly.
Silicon Chip DAC kit: a personal experience Dec 09
The DAC Kit comes with three circuit boards with printed overlays and component values, so mistakes are minimized. There was an overlay of the 3.3v Reg., which was reversed this, is noted in the instructions. The surface mounted components are mounted and no shorts were detected. This eliminates any possible problems arising from those who may not have the skills or eyesight to do this with success. There was however a piece of track missing on pin 3 of IC12 which resulted in no amplified output in the right channel, I informed Jaycar but have heard no comments about this.
Building the kit with the components supplied is a little disappointing, I have an Elecktor DAC 1992 vintage using two PCM 63’s, which I have tweaked with various mods, this was my control.
The OPA 134 op-amps were generic and not true Burr Brown chips, the polyester caps in the output stage were green-caps and grey rectangular things. As all tweakers know these are not the best performers for sound quality.
Having started this quest with a fellow DIY tweaker SandyK who generously supplied polypropylene caps, polystyrene and some LME 49710 Op-amps. When these were replaced the difference was chalk and cheese, the resulting improvement in sound quality was astounding, however I was still not satisfied with the result and set about further tweaks.
The next step was a Rock Grotto special using a John Linksley Hood “Ripple Eater” to supply the plus and minus 15 volts, this cleared up the details and moved the soundstage forward a few feet.
The next obvious one to me was the output from the DAC and the op-amps, sharing a blobby ground, I created an analogue ground across the output ground tabs of the RCA output sockets and cut all the tracks to ground for the op-amps and capacitors and soldered wires back to analogue ground and then grounded this point to a star ground in the box. This already had the grounds from both input, DAC and ripple eater. This improved the details and sound even more.
Still not satisfied I put another ripple eater set at 5v volts onto the DAC board and things improved even more, especially detail. Not satisfied with the result I was surfing DIY audio and found someone had tweaked a PCM63 with Oscons and had especially good results, so I replaced the 47uf electros with 100uf Oscons and had more revelation.
Still not satisfied I sourced some polypropylenes from RS for the filter circuit on the DIR 9001 and had even more improvement, so I replaced the two 10uf electros on the Dir 9001 with the last two 100uf Oscon and had more improvement. Please note that in the SC DWG the filter has a 6.8k resistor which is 680ohms in the BB DATA sheet, so far Silicon Chip have not replied as to why the difference and Texas Instrument have not replied as to what ramifications may arise. So nothing to report from either of the originators of the system.
My next evolution of this construction was to build a PFM Flea and replace the clock on the input of the DIR9001, this necessitated setting the Flea to 3.3vdc and using a CMOS clock module which I obtained from HyQ in Melb. This turned out to be a revelation the soundstage moved forward another few feet and the detail improved beyond belief. Spurred on with the success of this mod I made another Flea and replaced the clock in the CD transport, this however was the first heartbreak. The output was glassy and fatiguing to listen to. My remedy was to remove all the Oscons as they seem to be inductive and returning to conventional electros smoothed out the sound. The eventual result is a far more pleasant sound with detail, instrument separation, more clarity and detail on the voice with a great deal more image and presence than I started with.
Then I finally received a reply from TI.
Hi Alfred,
I apologize for the delay in responding to your inquiry, it is due to some technical reasons. I appreciate your patience.
You can use a buffer to provide level translation for the +5V clock module. The SN74LVC1G125 can be operated from +3.3V and has a +5V tolerant input. The output of the buffer can then be used to drive the XTI input as shown in Figure 9 of the DIR9001 data sheet.
DIR9001 data sheet Figure 8 / Table 5 and Figure17 indicate 680 ohms for the loop filter R value. The DIR9001 EVM user's guide schematic also shows a value of 680 ohms for the loop filter R in Figure 5. The 6.8K resistor may have been installed in error. Please replace it with 680 ohms. The R value impacts the clock recovery PLL loop filter response and stability. I would recommend the value shown in the data sheet and EVM users guide.
Best Regards,
Prasanna Dayala,
TEXAS INSTRUMENTS.
Original message to TI
Problem:
There is little or no information in the DATA sheet to implement the use of a 5v clock module, what steps should be taken to interface with the dir9001 as it is not 5v tolerant, could you suggest a circuit? Recently a unit designed locally with the PLL pin 22 Fig 17 of data sheet using a 6.8k resistor instead of the 680R you advise, it works but what are the implications of a resistor 10x the value?]
Armed with this information I change the resistor and all the base came back and the glassiness disappeared, however the muddiness of electros was present again and I started with a prudent application of Oscons, replacing one at time with listening tests in between, this returned detail and sharpened up the response of the unit. This is amazing that one resistor has such a severe effect on the response when a good clock signal is used. So I am back to where I want to be in terms of increased sound quality and I now have other parts of my system exposed as needy of improvement. All in all a significant improvement on the old Elecktor DAC, which has not been modified in the digital department as much as this unit has.
Whilst the listening tests did reveal a better response, I found that a low ESR Suntan 47uf cap for the DAC filter seem to be better than the Oscons, the other capacitor that was recommended to be removed was the 2.2nf on the audio out, this revealed more detail with a larger sound stage and much more image. Another recommended improvement was the using of a polypropylene on the digital input, as RS no longer stock a 100nf, I used a 68nf from the ones I bought for the digital filter. This enhanced all the previous highlights, with emphasis on the sound stage and especially image.
Parts used
Op Amp LM 49710
4.7 nf Polypropylene Capacitor RS 401-4766
68 nf Polypropylene Capacitor RS 401-4918
2.7nf Polypropylene Capacitor RS 401-4867
8.2nf Polypropylene Capacitor RS 401-4794
2.7nf Polypropylene Capacitor RS 401-4722
2.2nf Polypropylene Capacitor RS 401-4716 - later removed
100uf Oscon Farnell 121-7611
47uf Lo ESR Jaycar RE6332
Bead Ferrite WES FB12
Oscillator HyQ QXO8CBA 24m576
“Flea” Pink Fish Media
“Ripple Eater” Rock Grotto Forum
The original Kit from Jaycar can be improved 100 fold from that supplied.
The DAC Kit comes with three circuit boards with printed overlays and component values, so mistakes are minimized. There was an overlay of the 3.3v Reg., which was reversed this, is noted in the instructions. The surface mounted components are mounted and no shorts were detected. This eliminates any possible problems arising from those who may not have the skills or eyesight to do this with success. There was however a piece of track missing on pin 3 of IC12 which resulted in no amplified output in the right channel, I informed Jaycar but have heard no comments about this.
Building the kit with the components supplied is a little disappointing, I have an Elecktor DAC 1992 vintage using two PCM 63’s, which I have tweaked with various mods, this was my control.
The OPA 134 op-amps were generic and not true Burr Brown chips, the polyester caps in the output stage were green-caps and grey rectangular things. As all tweakers know these are not the best performers for sound quality.
Having started this quest with a fellow DIY tweaker SandyK who generously supplied polypropylene caps, polystyrene and some LME 49710 Op-amps. When these were replaced the difference was chalk and cheese, the resulting improvement in sound quality was astounding, however I was still not satisfied with the result and set about further tweaks.
The next step was a Rock Grotto special using a John Linksley Hood “Ripple Eater” to supply the plus and minus 15 volts, this cleared up the details and moved the soundstage forward a few feet.
The next obvious one to me was the output from the DAC and the op-amps, sharing a blobby ground, I created an analogue ground across the output ground tabs of the RCA output sockets and cut all the tracks to ground for the op-amps and capacitors and soldered wires back to analogue ground and then grounded this point to a star ground in the box. This already had the grounds from both input, DAC and ripple eater. This improved the details and sound even more.
Still not satisfied I put another ripple eater set at 5v volts onto the DAC board and things improved even more, especially detail. Not satisfied with the result I was surfing DIY audio and found someone had tweaked a PCM63 with Oscons and had especially good results, so I replaced the 47uf electros with 100uf Oscons and had more revelation.
Still not satisfied I sourced some polypropylenes from RS for the filter circuit on the DIR 9001 and had even more improvement, so I replaced the two 10uf electros on the Dir 9001 with the last two 100uf Oscon and had more improvement. Please note that in the SC DWG the filter has a 6.8k resistor which is 680ohms in the BB DATA sheet, so far Silicon Chip have not replied as to why the difference and Texas Instrument have not replied as to what ramifications may arise. So nothing to report from either of the originators of the system.
My next evolution of this construction was to build a PFM Flea and replace the clock on the input of the DIR9001, this necessitated setting the Flea to 3.3vdc and using a CMOS clock module which I obtained from HyQ in Melb. This turned out to be a revelation the soundstage moved forward another few feet and the detail improved beyond belief. Spurred on with the success of this mod I made another Flea and replaced the clock in the CD transport, this however was the first heartbreak. The output was glassy and fatiguing to listen to. My remedy was to remove all the Oscons as they seem to be inductive and returning to conventional electros smoothed out the sound. The eventual result is a far more pleasant sound with detail, instrument separation, more clarity and detail on the voice with a great deal more image and presence than I started with.
Then I finally received a reply from TI.
Hi Alfred,
I apologize for the delay in responding to your inquiry, it is due to some technical reasons. I appreciate your patience.
You can use a buffer to provide level translation for the +5V clock module. The SN74LVC1G125 can be operated from +3.3V and has a +5V tolerant input. The output of the buffer can then be used to drive the XTI input as shown in Figure 9 of the DIR9001 data sheet.
DIR9001 data sheet Figure 8 / Table 5 and Figure17 indicate 680 ohms for the loop filter R value. The DIR9001 EVM user's guide schematic also shows a value of 680 ohms for the loop filter R in Figure 5. The 6.8K resistor may have been installed in error. Please replace it with 680 ohms. The R value impacts the clock recovery PLL loop filter response and stability. I would recommend the value shown in the data sheet and EVM users guide.
Best Regards,
Prasanna Dayala,
TEXAS INSTRUMENTS.
Original message to TI
Problem:
There is little or no information in the DATA sheet to implement the use of a 5v clock module, what steps should be taken to interface with the dir9001 as it is not 5v tolerant, could you suggest a circuit? Recently a unit designed locally with the PLL pin 22 Fig 17 of data sheet using a 6.8k resistor instead of the 680R you advise, it works but what are the implications of a resistor 10x the value?]
Armed with this information I change the resistor and all the base came back and the glassiness disappeared, however the muddiness of electros was present again and I started with a prudent application of Oscons, replacing one at time with listening tests in between, this returned detail and sharpened up the response of the unit. This is amazing that one resistor has such a severe effect on the response when a good clock signal is used. So I am back to where I want to be in terms of increased sound quality and I now have other parts of my system exposed as needy of improvement. All in all a significant improvement on the old Elecktor DAC, which has not been modified in the digital department as much as this unit has.
Whilst the listening tests did reveal a better response, I found that a low ESR Suntan 47uf cap for the DAC filter seem to be better than the Oscons, the other capacitor that was recommended to be removed was the 2.2nf on the audio out, this revealed more detail with a larger sound stage and much more image. Another recommended improvement was the using of a polypropylene on the digital input, as RS no longer stock a 100nf, I used a 68nf from the ones I bought for the digital filter. This enhanced all the previous highlights, with emphasis on the sound stage and especially image.
Parts used
Op Amp LM 49710
4.7 nf Polypropylene Capacitor RS 401-4766
68 nf Polypropylene Capacitor RS 401-4918
2.7nf Polypropylene Capacitor RS 401-4867
8.2nf Polypropylene Capacitor RS 401-4794
2.7nf Polypropylene Capacitor RS 401-4722
2.2nf Polypropylene Capacitor RS 401-4716 - later removed
100uf Oscon Farnell 121-7611
47uf Lo ESR Jaycar RE6332
Bead Ferrite WES FB12
Oscillator HyQ QXO8CBA 24m576
“Flea” Pink Fish Media
“Ripple Eater” Rock Grotto Forum
The original Kit from Jaycar can be improved 100 fold from that supplied.
I heard about that, it's unfortunate, I think it was a problem with the PCB manufacturers since I know that the first prototype Electus built worked correctly without any modifications (other than having to insert the 7805 the right away around, as opposed to what it shows on the silkscreen).
Sorry that you had trouble with that. Personally I prefer using double-sided boards universally, and as you say it would have made it substantially easier to build. However I think the folks at Silicon Chip prefer for me to use single sided boards because then people can etch them at home. How many actually bother to do this I don't know; personally I find it too much effort so I have them made at a factory.
I didn't find building the switch board too difficult - I made four - although I must admit I have a lot of practice with soldering. Maybe your iron was not hot enough. I turn mine up above what is generally recommended (to 350 degrees C) since it makes solder flow much more easily. At that temperature and with good quality solder, once there's enough solder on the pad it will just adhere to the component leg without having to coax it.
Anyway I agree it's more difficult than it should be. Hopefully you were able to built it without injuring yourself.
I agree. We should have mentioned it in the text, but the photo of the switch board that was published was at life size, so you should be able to photocopy it at 1:1 and use that as a template. What I actually did was print out the PCB layout, which I believe you can download from the web site, although I'm not 100% sure about whether you need an online subscription or not.
While technically there are differences in dissipation factor between polyester, metallised polyster and polypropyelene capacitors, and polypropylene are superior, I'm not convinced that you can hear the difference. However, I don't want to get into an argument over it. It certainly won't hurt to use polypropylenes, and I used some in the prototype simply because that's what I could easily get in the appropriate values. So if you prefer them, go ahead and use them.
Yes, I made a typo and specified a 680 ohm resistor as 6800 ohm (i.e. 6.8k). I recommend using the correct value, but I haven't been able to measure any difference in performance after fixing it. It might have an effect if your source has a lot of clock jitter.
Thanks for posting your experience building this kit.. it's very detailed and I can't reply to everything you've said. Do be careful about the regulator orientation if you buy the first run of the Jaycar kit. The original overlay published was correct in that respect so the Altronics kit should not suffer from the same problem.
I have to point this out though:
I did extensive component testing when designing this project, including trying many different op amp types, including NE5534, OPA134, OPA2227 and LME49710. In each case I swapped the op amps on the DAC board and re-measured the THD+N and SNR figures. I assume you are referring to the LME49710. It had the highest distortion of all those I tested, despite what the data sheet claims. Distortion with LME49710 was about three times higher than NE5534 (0.07% if memory serves), and NE5534 had significantly higher distortion than OPA134.
I think the reason for this is that the LME49710 is designed to work as a voltage follower or simple voltage gain stage. It's probably very good in that configuration (it was used in the SC preamp after all, in its dual form), but it has other problems, possibly weak output drive strength, that make it far from ideal in this project. So if you're after low distortion I don't recommend using LME49710 in the DAC.
By the way I also tested many configurations that used different combinations of the mentioned op amp types for the I/V converter and differential amplifier stages. So, the choice of OPA134 all-round was not made at random. The next best configuration I found, if I remember correctly, was NE5534 for the I/V stage and OPA134 for the differential amplifiers. I think that had slightly higher distortion than all OPA134.
I have to point this out though:
I did extensive component testing when designing this project, including trying many different op amp types, including NE5534, OPA134, OPA2227 and LME49710. In each case I swapped the op amps on the DAC board and re-measured the THD+N and SNR figures. I assume you are referring to the LME49710. It had the highest distortion of all those I tested, despite what the data sheet claims. Distortion with LME49710 was about three times higher than NE5534 (0.07% if memory serves), and NE5534 had significantly higher distortion than OPA134.
I think the reason for this is that the LME49710 is designed to work as a voltage follower or simple voltage gain stage. It's probably very good in that configuration (it was used in the SC preamp after all, in its dual form), but it has other problems, possibly weak output drive strength, that make it far from ideal in this project. So if you're after low distortion I don't recommend using LME49710 in the DAC.
By the way I also tested many configurations that used different combinations of the mentioned op amp types for the I/V converter and differential amplifier stages. So, the choice of OPA134 all-round was not made at random. The next best configuration I found, if I remember correctly, was NE5534 for the I/V stage and OPA134 for the differential amplifiers. I think that had slightly higher distortion than all OPA134.
Silicon Chip DAC Kit
In my career in PMG/Telstra , I would have soldered many tens of thousands of joints in Telephone Exchange maintenance.I also used a high quality low power Goot soldering iron for this area .
The problem was due to the very thin leads of the Blue LEDs sitting in a much larger drilled hole.Most likely the gap between lead and hole hindered solder flow. I doubt that there would have been a problem if the LED leads were the usual thickness.
I have since had a report from a friend in Brisbane who has now also got the kit successfully working, that he had exactly the same problem with all 3 Blue LEDs.
SandyK
P.S.
I can confirm the findings regarding the use of polypropylene capacitors. Especially with the 4.7nF at pin 22 of the DIR9001.
I also got a marked improvement in soundstage by reducing the 2n2 capacitors at the output to 100pF,as did the other 2 constructors..
100pF is normally adequate to avoid most of the effects of different cables. I also chose to retain 4 of the OPA134, but use LME49710 at the output. In my experience this helps avoid the "house" sound from using all of the same type.It is also normally an advantage to have a faster opamp at the output.
It is interesting to note that my Musical Fidelity X-DAC V3 uses similar 100R and 100pF at the output.`
Hi Nicholas
In my career in PMG/Telstra , I would have soldered many tens of thousands of joints in Telephone Exchange maintenance.I also used a high quality low power Goot soldering iron for this area .
The problem was due to the very thin leads of the Blue LEDs sitting in a much larger drilled hole.Most likely the gap between lead and hole hindered solder flow. I doubt that there would have been a problem if the LED leads were the usual thickness.
I have since had a report from a friend in Brisbane who has now also got the kit successfully working, that he had exactly the same problem with all 3 Blue LEDs.
SandyK
P.S.
I can confirm the findings regarding the use of polypropylene capacitors. Especially with the 4.7nF at pin 22 of the DIR9001.
I also got a marked improvement in soundstage by reducing the 2n2 capacitors at the output to 100pF,as did the other 2 constructors..
100pF is normally adequate to avoid most of the effects of different cables. I also chose to retain 4 of the OPA134, but use LME49710 at the output. In my experience this helps avoid the "house" sound from using all of the same type.It is also normally an advantage to have a faster opamp at the output.
It is interesting to note that my Musical Fidelity X-DAC V3 uses similar 100R and 100pF at the output.`
Last edited:
Any UK based DIY'ers might be interested to know that this is DAC kit is available Jaycar UK for £40.50, which is phenomenal value. Ignore the blurb on the website about having to solder the SMD's, as they come pre-soldered. Also, if it follows Jaycars other kits, It will contain a copy of the article to build it.
Link to UK Jaycar SC Dac Kit
Link to UK Jaycar SC Dac Kit
- Status
- Not open for further replies.