Toscanini is an evolution of Kubelik DAC (no longer available) but with a couple of tweaks. The main improvement is that although its still using six TDA1387 DAC chips, each is being 'overclocked' by about 50% to gain a higher SNR, noise being the primary artifact I work at reducing in my designs. In effect, six overclocked chips are doing the work of nine normal ones. Also Kubelik's power supply was a bit non-standard 20V, Toscanini lowers that to 15V so it'll be easier to find an off-the-shelf supply. The PCB offers a choice of the manufacturer of the filter inductors - Coilcraft SMD, Bourns SMD or through-hole - using the SMD ones makes for a lower profile PCB - better suited to vertical stacking for creating a multichannel DAC. Opamps can be rolled to taste, for the stock design I'm going for ADA4075-2 of which Mouser had plenty in stock when I last checked.
Schematic now uploaded, BOM & Gerbers too.
There are 3 inductor choices on the Mouser BOM, only two of which are available from Mouser. Coilcrafts like in the photo below don't seem to be available from Mouser but can be gotten direct from Coilcraft I believe. Note that I built the proto with 8.2mH as I couldn't obtain any 6.8mH - Coilcraft now only send out samples to registered companies here in China and Chinese copies have poor (20%) tolerance. The higher value needs to be paired with a higher I/V resistor and thus leads to a higher max output voltage.
The total cost of all the Mouser parts is about 32USD to which you'll need to add the cost of the TDA1387s (from Aliexpress or eBay) and the PCB (presumably from JLCPCB).
Schematic now uploaded, BOM & Gerbers too.
There are 3 inductor choices on the Mouser BOM, only two of which are available from Mouser. Coilcrafts like in the photo below don't seem to be available from Mouser but can be gotten direct from Coilcraft I believe. Note that I built the proto with 8.2mH as I couldn't obtain any 6.8mH - Coilcraft now only send out samples to registered companies here in China and Chinese copies have poor (20%) tolerance. The higher value needs to be paired with a higher I/V resistor and thus leads to a higher max output voltage.
The total cost of all the Mouser parts is about 32USD to which you'll need to add the cost of the TDA1387s (from Aliexpress or eBay) and the PCB (presumably from JLCPCB).
Attachments
Last edited:
Subscribed. Do you use the existing Kubelik boards for the modifications? I still have an unbuilt Kubelik kit waiting.
I think on a Kubelik board you could do some of the tweaks but it would be far better to start with a new Toscanini PCB and re-use some of the components for Kubelik on that. I've not done a side by side SQ comparison between the two btw so I'm not making claims that this is going to be across-the-board superior to Kubelik.
Hi abrax
I see that the main change is in connecting pin 7 of DAC chips to Vcc through 100k resistors.
George
I see that the main change is in connecting pin 7 of DAC chips to Vcc through 100k resistors.
George
George - yes that's the 'overclocking' that I spoke of in the first post. It increases the DAC's output current. The value of 100k is subject to amendment when I've tried the design with a wider spectrum of chips, so far I've only built DACs with my underperforming chips - those that are giving less than 1mA without any pin7 connection.
Hi Richard,
I just registered an account to write this message. I have been reading DIYAudio for many years but have been learning and generally don't have anything valuable to add. I have built a number of Nelson's Amp designs, learned a few things and this year have been following miro1360's AD1862 thread which I have built many versions of... I never expected I could build a DAC as digital seemed like a whole new world and I was still very much learning analog.
I just recently found your threads when using the search fucntion on the forum for your user name as I was impressed with some of the comments from you made on miro's thread.
Anyway, all this to let you know that I sent the Toscanini gerbers off to JLCPCB and should get them in a couple of days. Ordered the parts from LCSC as most of them were availible except using almost all Yageo ThinF resistors as Susumu had limited stock. Assuming I can get the TDA1387's this weekend as well, I will likely build it up this weekend. I am certinly not someone qualified to give opinions on sound but I know what I like with the music I like to listen to, so am interested in hearing your creation.
Thanks for supporting the DIYAudio community, even those of us that are still on a long learning curve.
Greg
I just registered an account to write this message. I have been reading DIYAudio for many years but have been learning and generally don't have anything valuable to add. I have built a number of Nelson's Amp designs, learned a few things and this year have been following miro1360's AD1862 thread which I have built many versions of... I never expected I could build a DAC as digital seemed like a whole new world and I was still very much learning analog.
I just recently found your threads when using the search fucntion on the forum for your user name as I was impressed with some of the comments from you made on miro's thread.
Anyway, all this to let you know that I sent the Toscanini gerbers off to JLCPCB and should get them in a couple of days. Ordered the parts from LCSC as most of them were availible except using almost all Yageo ThinF resistors as Susumu had limited stock. Assuming I can get the TDA1387's this weekend as well, I will likely build it up this weekend. I am certinly not someone qualified to give opinions on sound but I know what I like with the music I like to listen to, so am interested in hearing your creation.
Thanks for supporting the DIYAudio community, even those of us that are still on a long learning curve.
Greg
Last edited:
Hi Greg - thanks for dropping by and letting me know you're actively pursuing this. I'm happy you're building Toscanini, so far I haven't heard of anyone else taking it up so you might be the first. I wish you luck with your build!
I wouldn't expect Toscanini to play in the same league as miro's AD1862 as that chip has much lower noise. But given Toscanini isn't using an opamp directly on the output of the DAC it might give you an interesting alternative perspective. Do please share the results.
I wouldn't expect Toscanini to play in the same league as miro's AD1862 as that chip has much lower noise. But given Toscanini isn't using an opamp directly on the output of the DAC it might give you an interesting alternative perspective. Do please share the results.
Hi Richard,
Exactly the fact that this particular thread was not so active encoraged me let you know I was building it. Every creator needs some feedback for encoragment.
I'm very interested if I can hear the difference especially in the output filtering.
For the AD1862 I'm using an AD797 following nearly eaxctly the DS from AD, so rather than being exactly miro's design I would say it was inspired by miro. I recently bought a scope as my experince and learing curve have progressed to the point of needing one and when I looked at the output with all those steps I was shocked and thought I had done something wrong. Only to read more that it is exactly what it should be doing. This made me more interested if I could filter some of these HF harmonics out with a sallen key filter and have been playing with that, which is when my reading lead me to your Kubelik design.
No pretenting I can fully understand it, but I have some idea and the filter before the IV looks like a very interesting concept.
Cheers.
Exactly the fact that this particular thread was not so active encoraged me let you know I was building it. Every creator needs some feedback for encoragment.
I'm very interested if I can hear the difference especially in the output filtering.
For the AD1862 I'm using an AD797 following nearly eaxctly the DS from AD, so rather than being exactly miro's design I would say it was inspired by miro. I recently bought a scope as my experince and learing curve have progressed to the point of needing one and when I looked at the output with all those steps I was shocked and thought I had done something wrong. Only to read more that it is exactly what it should be doing. This made me more interested if I could filter some of these HF harmonics out with a sallen key filter and have been playing with that, which is when my reading lead me to your Kubelik design.
No pretenting I can fully understand it, but I have some idea and the filter before the IV looks like a very interesting concept.
Cheers.
AD797 is certainly a good choice from the point of view of its ultra-low noise. But I have misgivings about using it directly on the output of the DAC based on it seeming to be rather susceptible to RF. Walt Jung (along with Jan Didden) used it in one of their very low noise regulators but then swapped to another ADI opamp because of (how I interpreted their writings) RF pick-up.
Yes, stair steps are what the opamp output is supposed to look like. Scott Wurcer introduced a cap on the output of the DAC in the AD797 DS to try to slow down the DAC edges. My reservation about that approach is it increases the noise gain of the opamp at higher frequencies. On my latest TDA1387 design (Stack DAC) I've gone back to passive I/V and the (passive) filter will be a separate module so the user can decide how much filtering is appropriate. So far I've found more filtering sounds better.
If you like playing around with I/Vs then have a look at my closely related 'Dorati' design which uses passive I/V and no opamps. It definitely sounds different to get rid of the opamps but not all aspects of the sound seem to improve according to a couple of keen listeners.
Yes, stair steps are what the opamp output is supposed to look like. Scott Wurcer introduced a cap on the output of the DAC in the AD797 DS to try to slow down the DAC edges. My reservation about that approach is it increases the noise gain of the opamp at higher frequencies. On my latest TDA1387 design (Stack DAC) I've gone back to passive I/V and the (passive) filter will be a separate module so the user can decide how much filtering is appropriate. So far I've found more filtering sounds better.
If you like playing around with I/Vs then have a look at my closely related 'Dorati' design which uses passive I/V and no opamps. It definitely sounds different to get rid of the opamps but not all aspects of the sound seem to improve according to a couple of keen listeners.
Ah, I see.
I saw the thread dates and thought Toscanini must be the latest, but certainly would be foolish of me not to take your advice on trying Dorati. Plus it is an excellent opportunity for me to support your work a bit.
Since I beleive I am too new to have PM access if you could PM me so we could work out the details, that would be excellent.
I live near you so quick SF shipping would get it to me quickly.
G
I saw the thread dates and thought Toscanini must be the latest, but certainly would be foolish of me not to take your advice on trying Dorati. Plus it is an excellent opportunity for me to support your work a bit.
Since I beleive I am too new to have PM access if you could PM me so we could work out the details, that would be excellent.
I live near you so quick SF shipping would get it to me quickly.
G
Looking forward to your thoughts on the sounds G. Richard certainly knows how to design great sounding kit. Be sure to try his amps.
I think Richard is under selling himself....or the abilities of his DACs/1387. I have both a Miro and a Kubelik and a Dorati...(I think!) and they sound on par with an 1862 to my (average) ears.
I think Richard is under selling himself....or the abilities of his DACs/1387. I have both a Miro and a Kubelik and a Dorati...(I think!) and they sound on par with an 1862 to my (average) ears.
Thanks for the feedback Jim, I was basing my comments on feedback I'd gotten from another customer who had a miro DAC which he upgraded with Sparkos discrete opamps. So it wasn't stock, I defer to your ears as my remarks were just conjecture, I've not listened to a miro DAC.
Hi, Richard, I’m interested to read about your selection of those inexpensive 220uF Wurth electrolytic output coupling capacitor.
From my pov there doesn't seem to be much to the selection. I need a cap, film caps are way too bulky for the credit-card like form factor. X7Rs are small but don't sound that great, that only leaves 'lytics.
Did I understand your question in the manner in which you meant it Ken? Quite possibly I have missed the point.
Did I understand your question in the manner in which you meant it Ken? Quite possibly I have missed the point.
I think so, yes. I was curious about whether you felt that Wurth cap. helped enable a sound you maybe were looking to obtain. So my question was mostly about your subjective assessment of it's performance in selecting it. As you well know, audiophile coupling caps. are typically extremely expensive, so I'm always interested in reading about reasonably priced discoveries for signal coupling application, and which are competitive in subjective transparency/musicality.
Last edited:
Maybe this is a bit more audiophile than Wurth? Still affordable and a few left before they are unobtainium.
https://www.digikey.com/en/products/detail/elna-america/RFS-35V221MI6-5/970227
https://www.digikey.com/en/products/detail/elna-america/RFS-35V221MI6-5/970227
I must confess I haven't actually built the DAC with those Wurth caps. Here I used another manufacturer, I'm not sure which. I selected them solely on grounds of availability and cost when browsing Mouser.
In terms of cap audibility, I can't say I have noticed much difference between standard 'lytics from various mainstream manufacturers (like Nichicon, Rubycon). They do need to form up though over time or they let too much leakage current through and that doesn't play well into the trafos I have. I haven't experimented with boutique types.
Interesting subject. I must admit when Richard and Nelson Pass ( and I'm sure others) design such good sounding kit with simple run of the mill electrolytics in the signal path, it has altered my opinion of what can and should/shouldn't sound good. Loads of people are in the no cap is best camp...and certainly not an electrolytic. ACA Mini has a 15000uf on the output, ACA 4700uf. No detriment to my ears.
My own suspicions about electrolytic capacitor sound are more in the area of noise than distortion. In harmony with this notion, I recently came across the following from Jim Williams. He's talking about choosing capacitors suitable for coupling to the input of an ultra-low noise preamp :
https://www.analog.com/en/technical...rement-for-a-low-noise-voltage-reference.html
https://www.analog.com/en/technical...rement-for-a-low-noise-voltage-reference.html