Hello,
Once in a while this question pops up.
Some people will write it is a real update.
Doede will probably react soon and say its effect will be limited and that you want the shunts to be protected from a to high input voltage.
Once i used the Tent shunt in another choke input application but all other things not comparable to the DDDAC circuit. Guido told me to put a big 5 watt zenerdiode across one of the caps so the input voltage could never rise above a certain limit. When nothing is wrong i think a big 11 volt zener will not add its sonic signature when it will be mounted on the last cap in the supply where the voltages has to be 10 volt.
In my situation it was put on the first cap after the input choke in a LCRC network. Dont know if it will work too in a LCLC network because in the DDDAC set up there will only be the right voltage if the correct current is being drawn by the load connected to the last capacitor.
Skipping the 7810 is a risk. If the zener will work i would try it.
greetings, Eduard
Hello,
Doede started making his own DDDAC because he wanted to make something that surpasses the products available in the market.
Some of the choices he made maybe were there to make it possible for most of us to build it. If only one out of ten people will fry the shunts within a few weeks after removing the 7810 he will feel sad i presume.
THAT is why i waited for the prints with the shunts incorporated to be available.
If it would be possible to use a well chosen zener as protection i will give it a try.
Greetings, eduard
Doede started making his own DDDAC because he wanted to make something that surpasses the products available in the market.
Some of the choices he made maybe were there to make it possible for most of us to build it. If only one out of ten people will fry the shunts within a few weeks after removing the 7810 he will feel sad i presume.
THAT is why i waited for the prints with the shunts incorporated to be available.
If it would be possible to use a well chosen zener as protection i will give it a try.
Greetings, eduard
Thanks for the point you make on safety. It is important.
But, if you use a regulated PSU, I suppose the risk would be minimal?
I was just wandering about the audible benefits.
ok, here we are 
indeed we had this discussion before and it is good to recap so now and then...
You wouldn't believe what can go wrong and hence will go wrong.
So as a DIY DAC which can be hobbied and DIY around with, I needed to make some compromizes, which I did. Nevertheless, even in the standard form it is a great piece of joy
It is a general truth that if you are carefull to make sure nothing goes wrong you will limit damage. But believe me, I also fried stuff more than once
Now, everyone feeling confident that they will not fry stuff - or don't mind buying new boards (?) - should do this; it is a hobby and it is a lot of fun to try out. And rewarding to get that extra quant of SQ out of it.
switching gears:
now one thing never popped up (I AM surprised it did not...) No one ever suggested so far, to replace the standard 7810 (which is suggested by Guido Tent as PRE-Regulation) with one of the HIGH END regulators in a 10 Volt version (in stead of completely removing it)
BTW: I will come back this week with final report on the SQ impact (at my ears) for the 5 volt mainboard regulator
It took the French post 4 weeks to get me a small parcel with the last regulator I wanted to try out in my A_B test gear...
..
indeed we had this discussion before and it is good to recap so now and then...
You wouldn't believe what can go wrong and hence will go wrong.
So as a DIY DAC which can be hobbied and DIY around with, I needed to make some compromizes, which I did. Nevertheless, even in the standard form it is a great piece of joy
It is a general truth that if you are carefull to make sure nothing goes wrong you will limit damage. But believe me, I also fried stuff more than once
Now, everyone feeling confident that they will not fry stuff - or don't mind buying new boards (?) - should do this; it is a hobby and it is a lot of fun to try out. And rewarding to get that extra quant of SQ out of it.
switching gears:
now one thing never popped up (I AM surprised it did not...) No one ever suggested so far, to replace the standard 7810 (which is suggested by Guido Tent as PRE-Regulation) with one of the HIGH END regulators in a 10 Volt version
(in stead of completely removing it)BTW: I will come back this week with final report on the SQ impact (at my ears) for the 5 volt mainboard regulator
It took the French post 4 weeks to get me a small parcel with the last regulator I wanted to try out in my A_B test gear...
..
Yes, very curious about that outcome
and a new chapter beginsI'll wait and see (read) what comes out of this very interesting idea, currently very happy with my setup, even without FiFoPi/Accusillicon.
Who'll be the first one to tackle this
The LF50 replacement tweak - A-B test report
So, finally I concluded this test after many weeks of listening on several music tracks and using a nice selection of different regulators at the LF50 position of the mainboard.
Just to get this straight, what is the function of the LF50 and why could it make a difference?
If you check my circuits, which are published on my website, you can see it in more detail, but basically there are two functions. First, to optionally power the WaveIO isolator (I am not using this anymore btw) and secondly to power the logic and SPDIF section on the Mainboard. I know there are thoughts on “all the logic on the board introducing extra Jitter”. Well it is less influential as you might think. Reason for this is, that the majority of the logic is SPDIF and to reshuffle the data and the R/L clock and not for the bit clock (BCK). This signal sees a very nice path to the dac chip by only moving it through the 74VHCT244 which is a level shifter from 3,3V to 5V logic and a buffer function (which I found out that for parallel use of the PCM1794 this is advantageous) The 244 chip is powered by the LF50 on the mainboard as well. So this position might be influenced by the solidity of the voltage supply. The rest (SPDIF, LR and Data) are not so critical, as with LR and Data, it is just filling the shift registers of the PCM1794A and the final D/A conversion actually takes place on the change of the BCK clock edge. This is the most jitter dependent point in time… I checked this with an engineer from TI by the way, as the datasheet is not very clear on this and some people thought or claimed it is the SCK or L/R signal doing this.
Disclaimer:
Do not expect BIG differences (play against the wall BS and alike) or bad and great sound. It all sounds great, but there are nice tweaks possible and something clearly occurred to me, it can very much depend all on the rest of your equipment. So this test is to my ears and with my equipment ( mostly Tube: Cleo6, 300BSE for midrange/high and TUMOS for low range, but high definition loudspeakers (Avalon sentinel clone)). I can imagine, if you are playing with class-D amplifiers, you prefer a somewhat more mellow sounding regulator, or, when you play with old Quad/Rotel or vintage warm mellow tube gear and old KEF speakers, a more detailed sounding one will be the choice)
I used the following Regulators in my test:
LF50 (of course)
UA7805
LM317
LDOVR LT3045-M 78xx replacement
Per Sjostrom super regulator SSR01
Sparkos discrete voltage regulator
NewClass D 78xx replacement
Tentlabs Shunt (a somewhat lower current version set by Guido)
The test setup:
I wanted to rule out the effect of time lag comparing audio stuff. Be honest, we all know how tricky it can be to have a new build or modification and we listen hours, days, or even weeks later to the new situation. Even when you compare stuff by standing up, walk to your gear and flip a switch or reverse a cable or whatever, there is already a time lag. Although I agree that this is doable and our memory seems to work still pretty ok on these short time lags. But still…. I wanted to be sure….
So what I did is using one of those Bluetooth relay boards driven by a smartphone app, where you can switch between A&B from your phone. I soldered wires from the LF50 position to the board and some support structure and the two regulators to test A-B. There is picture attached. Looks like a mess, but it works just fine. It might be the 10cm of wire could influence the performance? May be, but the differences of SQ are still the same I believe for comparison purpose. After soldering my final choice back on the LF50 position, at least I could not hear anything different. So I guess it will be alright.
I must say, such an “I sit in my chair and can switch back and forward without even moving my head-solution” is really a great tool to compare detail, special moments in the music etc. With one hand I have the smartphone switching, in the other the remote control for the music streamer (in this case the DDDAC1794 with Roon/Ropieee and FiFoPi)
The only “downside” is that I could “only” compare two at a time. May be an A-B-C-D-E setup would have been better, but hey, I combined many regulators in A-B and in the end you get a very good idea on their sonic capabilities and how they compare.
The test itself:
I am not selling Audio magazines, so I will refrain from long descriptive sound checks and reviews of the albums played and just give my conclusions. I listened to normal pop music, typical high end live tracks, classical, jazz, male voice, female voice etc.
LF50
Reference, you all know this… and for the price unbeatable (I could very well live on with this)
UA7805
Compared to the LF50 a bit duller, but kind of close and sonically the same in terms of voices and instruments – not recommended
LM317
Well, this was kind of similar to the LF50. I could not make out too much and quickly skipped as I saw no pathway forward anyway. I was surprised to be honest it was better in than the 7805
LDOVR LT3045-M 78xx replacement
Based on the extremely low noise LT3045M chip. Very detailed, sometimes you think it paints to sharp lines around everything. But it puts a very open sound stage and especially male voices get a clarity and texture you cannot hear with the LF50. The sound stage moves forward a bit – this one heavily depends on the rest of your gear I would say
Per Sjostrom super regulator SSR01
A serial regulator designed by Per Sjostrom. Hand built PCBAs. Serial transistor and low noise Opamp doing the error correction. Basically this is a similar setup as the dddac power supply and all other serial regulators. Pretty costly by the way and not a 78xx drop in format of course. This sounds very good. Excellent soundstage, nice soft voices and lots of small lower level details. In my set up I would prefer this over the LT3045
Sparkos discrete voltage regulator
Basically the same as the SSR01 principle, but the error amplifier is discretely built up. Also this one sounds excellent – I would say identical with the Per one. I could not hear a lot of difference. Very close. But cheaper and drop in replacement which is a plus
NewClass D 78xx replacement
Not sure on the technology behind this one but it is some kind of serial regulation. Compared to the others it is pretty noisy, but somehow that seems to disturb the sound stage. This regulator is for me the mellowest one. Sounds beautiful, but can be too much laid back (in my setup and ears, remember?) The sound stage is great like the others before and especially female voices are showing good and very analog texture.
Tentlabs Shunt (a somewhat lower current version set by Guido)
Tentlabs is the only real SHUNT regulator here, which is totally different technology and likely thanks to the very low output impedance and shunt function it seems to produce the least jitter on the HCT244 chip. Reason I say is that with this regulator the bass definition was the best and this is considered as a result of low jitter (for reasons I don comprehend by the way, but it is experienced in many listening test in the past decades) In terms of the other aspects the shunt is very similar. Open sound stage, bit forward, great texture on the voices and instruments. After many A-B switches and with the knife on the throat (= very small difference, LOL) I believe I hear a somewhat better contour of all instruments and voices being placed in the room. But if you secretly would change it at night against a Per or Sparkos without me knowing it, I most likely would not notice…
Conclusion:
This was very due to test out and I had great fun doing it. On top learned a lot. Foremost I learned that the differences are sometimes so small, you need to be careful to not kid yourself into anything… Still the differences are there and it can be a personal choice between a few at the very end (like with me)
So this is the final ordeal from DDDAC:
All regulators sound great – basta, everyone should be able to get old with this
7805
I see no need to use this
LF50
Standard and thanks to low cost, keeps the kit price reasonable and sets a great sound for live
LM317
I see no need to use this
LT3045 LDO
This can be too much detail… Use when it fits to the rest of gear – in mine it was a tad too much and others were better overall to my ears
SSR01 and Sparkos
Could not hear any real difference this would be the most general compromise when you are not sure if your total gear is more warm or detailed sounding. This will never disappoint
NewClass D
Great sounding and most warm/mellow/analog (?) Especially in low frequencies. You really need to try this yourself I believe.
Tentlabs
As described, it does all the SSR01 and Sparkos does but in the bass it is a little bit better defined and has more punch (that probably goes hand in hand) So the Tent Shunt stays. In my set up this was photo finish with the ClassD, SSR01/Sparkos in that order. I would use any one of these as replacement for the LF50 and know that my SQ would have increased again a tad. To compare, the FiFoPi was clearly a larger jump forward and more worth it!!
I hope you enjoyed reading this (at the end LONG text) and it helps you making a choice if you want to do this tweak…
On a last note, now that this book is closed for me I am getting curious what the 3,3 Volt regulation on the FiFoPi does. Based on Ian’s recommendation to use lowest noise I blindly put the LT3045 on it. But I also know Dick has the NewClass D in this position and is very content with it. Both are wide apart in noise figure, so how can that be ? Meaning versus Ian recommendation… so I will need to do the same A-B test (I have it now) on that position. Some time, for now I have other things to do. Not even mentioning the 7810 position, LOL
So, finally I concluded this test after many weeks of listening on several music tracks and using a nice selection of different regulators at the LF50 position of the mainboard.
Just to get this straight, what is the function of the LF50 and why could it make a difference?
If you check my circuits, which are published on my website, you can see it in more detail, but basically there are two functions. First, to optionally power the WaveIO isolator (I am not using this anymore btw) and secondly to power the logic and SPDIF section on the Mainboard. I know there are thoughts on “all the logic on the board introducing extra Jitter”. Well it is less influential as you might think. Reason for this is, that the majority of the logic is SPDIF and to reshuffle the data and the R/L clock and not for the bit clock (BCK). This signal sees a very nice path to the dac chip by only moving it through the 74VHCT244 which is a level shifter from 3,3V to 5V logic and a buffer function (which I found out that for parallel use of the PCM1794 this is advantageous) The 244 chip is powered by the LF50 on the mainboard as well. So this position might be influenced by the solidity of the voltage supply. The rest (SPDIF, LR and Data) are not so critical, as with LR and Data, it is just filling the shift registers of the PCM1794A and the final D/A conversion actually takes place on the change of the BCK clock edge. This is the most jitter dependent point in time… I checked this with an engineer from TI by the way, as the datasheet is not very clear on this and some people thought or claimed it is the SCK or L/R signal doing this.
Disclaimer:
Do not expect BIG differences (play against the wall BS and alike) or bad and great sound. It all sounds great, but there are nice tweaks possible and something clearly occurred to me, it can very much depend all on the rest of your equipment. So this test is to my ears and with my equipment ( mostly Tube: Cleo6, 300BSE for midrange/high and TUMOS for low range, but high definition loudspeakers (Avalon sentinel clone)). I can imagine, if you are playing with class-D amplifiers, you prefer a somewhat more mellow sounding regulator, or, when you play with old Quad/Rotel or vintage warm mellow tube gear and old KEF speakers, a more detailed sounding one will be the choice)
I used the following Regulators in my test:
LF50 (of course)
UA7805
LM317
LDOVR LT3045-M 78xx replacement
Per Sjostrom super regulator SSR01
Sparkos discrete voltage regulator
NewClass D 78xx replacement
Tentlabs Shunt (a somewhat lower current version set by Guido)
The test setup:
I wanted to rule out the effect of time lag comparing audio stuff. Be honest, we all know how tricky it can be to have a new build or modification and we listen hours, days, or even weeks later to the new situation. Even when you compare stuff by standing up, walk to your gear and flip a switch or reverse a cable or whatever, there is already a time lag. Although I agree that this is doable and our memory seems to work still pretty ok on these short time lags. But still…. I wanted to be sure….
So what I did is using one of those Bluetooth relay boards driven by a smartphone app, where you can switch between A&B from your phone. I soldered wires from the LF50 position to the board and some support structure and the two regulators to test A-B. There is picture attached. Looks like a mess, but it works just fine. It might be the 10cm of wire could influence the performance? May be, but the differences of SQ are still the same I believe for comparison purpose. After soldering my final choice back on the LF50 position, at least I could not hear anything different. So I guess it will be alright.
I must say, such an “I sit in my chair and can switch back and forward without even moving my head-solution” is really a great tool to compare detail, special moments in the music etc. With one hand I have the smartphone switching, in the other the remote control for the music streamer (in this case the DDDAC1794 with Roon/Ropieee and FiFoPi)
The only “downside” is that I could “only” compare two at a time. May be an A-B-C-D-E setup would have been better, but hey, I combined many regulators in A-B and in the end you get a very good idea on their sonic capabilities and how they compare.
The test itself:
I am not selling Audio magazines, so I will refrain from long descriptive sound checks and reviews of the albums played and just give my conclusions. I listened to normal pop music, typical high end live tracks, classical, jazz, male voice, female voice etc.
LF50
Reference, you all know this… and for the price unbeatable (I could very well live on with this)
UA7805
Compared to the LF50 a bit duller, but kind of close and sonically the same in terms of voices and instruments – not recommended
LM317
Well, this was kind of similar to the LF50. I could not make out too much and quickly skipped as I saw no pathway forward anyway. I was surprised to be honest it was better in than the 7805
LDOVR LT3045-M 78xx replacement
Based on the extremely low noise LT3045M chip. Very detailed, sometimes you think it paints to sharp lines around everything. But it puts a very open sound stage and especially male voices get a clarity and texture you cannot hear with the LF50. The sound stage moves forward a bit – this one heavily depends on the rest of your gear I would say
Per Sjostrom super regulator SSR01
A serial regulator designed by Per Sjostrom. Hand built PCBAs. Serial transistor and low noise Opamp doing the error correction. Basically this is a similar setup as the dddac power supply and all other serial regulators. Pretty costly by the way and not a 78xx drop in format of course. This sounds very good. Excellent soundstage, nice soft voices and lots of small lower level details. In my set up I would prefer this over the LT3045
Sparkos discrete voltage regulator
Basically the same as the SSR01 principle, but the error amplifier is discretely built up. Also this one sounds excellent – I would say identical with the Per one. I could not hear a lot of difference. Very close. But cheaper and drop in replacement which is a plus
NewClass D 78xx replacement
Not sure on the technology behind this one but it is some kind of serial regulation. Compared to the others it is pretty noisy, but somehow that seems to disturb the sound stage. This regulator is for me the mellowest one. Sounds beautiful, but can be too much laid back (in my setup and ears, remember?) The sound stage is great like the others before and especially female voices are showing good and very analog texture.
Tentlabs Shunt (a somewhat lower current version set by Guido)
Tentlabs is the only real SHUNT regulator here, which is totally different technology and likely thanks to the very low output impedance and shunt function it seems to produce the least jitter on the HCT244 chip. Reason I say is that with this regulator the bass definition was the best and this is considered as a result of low jitter (for reasons I don comprehend by the way, but it is experienced in many listening test in the past decades) In terms of the other aspects the shunt is very similar. Open sound stage, bit forward, great texture on the voices and instruments. After many A-B switches and with the knife on the throat (= very small difference, LOL) I believe I hear a somewhat better contour of all instruments and voices being placed in the room. But if you secretly would change it at night against a Per or Sparkos without me knowing it, I most likely would not notice…
Conclusion:
This was very due to test out and I had great fun doing it. On top learned a lot. Foremost I learned that the differences are sometimes so small, you need to be careful to not kid yourself into anything… Still the differences are there and it can be a personal choice between a few at the very end (like with me)
So this is the final ordeal from DDDAC:
All regulators sound great – basta, everyone should be able to get old with this
7805
I see no need to use this
LF50
Standard and thanks to low cost, keeps the kit price reasonable and sets a great sound for live
LM317
I see no need to use this
LT3045 LDO
This can be too much detail… Use when it fits to the rest of gear – in mine it was a tad too much and others were better overall to my ears
SSR01 and Sparkos
Could not hear any real difference this would be the most general compromise when you are not sure if your total gear is more warm or detailed sounding. This will never disappoint
NewClass D
Great sounding and most warm/mellow/analog (?) Especially in low frequencies. You really need to try this yourself I believe.
Tentlabs
As described, it does all the SSR01 and Sparkos does but in the bass it is a little bit better defined and has more punch (that probably goes hand in hand) So the Tent Shunt stays. In my set up this was photo finish with the ClassD, SSR01/Sparkos in that order. I would use any one of these as replacement for the LF50 and know that my SQ would have increased again a tad. To compare, the FiFoPi was clearly a larger jump forward and more worth it!!
I hope you enjoyed reading this (at the end LONG text) and it helps you making a choice if you want to do this tweak…
On a last note, now that this book is closed for me I am getting curious what the 3,3 Volt regulation on the FiFoPi does. Based on Ian’s recommendation to use lowest noise I blindly put the LT3045 on it. But I also know Dick has the NewClass D in this position and is very content with it. Both are wide apart in noise figure, so how can that be ? Meaning versus Ian recommendation… so I will need to do the same A-B test (I have it now) on that position. Some time, for now I have other things to do. Not even mentioning the 7810 position, LOL
Attachments
Excellent write up Doede and a clever way to compare them
So many possible tweaks
I assume the shunt is this....
TentlabsShop
So many possible tweaks
I assume the shunt is this....
TentlabsShop
Thx Doede. As always, thorough and factual from you. Small differences, and what’s best depends on the rest of the setup. I will stick with Ldover, based on your experience with it. The last thing I want in my setup is a more mellow/warmer sound. Ldover matches the warm sound from the Mullards in my pre. It’s now warm and detailed, pure pleasure.
Excellent write up Doede and a clever way to compare them
So many possible tweaks
I assume the shunt is this....
TentlabsShop
yes, this is the one - there is no special "doede" version for 5 Volt I see, but just ask Guido....
Thx SD
yes, your conclusion is exactly what I meant, try to combine what fits best for you 
Hi Supersurfer,
A year ago I made the decision to buy a DCS Dac or build the dddac and decided to buy the DCS delius DAC with the Purcell upsampler.
I am very curious what type dcs dac you compare to the dddac?
I have also the old dddac, maybe I will build also the new dddac.
Streamer I use is the allo usbridge signature and the digione signature with the shanti power supply.
Roon rock on the nuc8i7beh
Regards,
Rudy
A year ago I made the decision to buy a DCS Dac or build the dddac and decided to buy the DCS delius DAC with the Purcell upsampler.
I am very curious what type dcs dac you compare to the dddac?
I have also the old dddac, maybe I will build also the new dddac.
Streamer I use is the allo usbridge signature and the digione signature with the shanti power supply.
Roon rock on the nuc8i7beh
Regards,
Rudy
Last edited:
dCS Elgar en Purcell
Why are you using the digione on the dddac? You can get direct i2s out of the usbridge.
A reclocker is highly recommended!
I also use the Shanti For the usbridge sig.
Hi supersurfer,
I don't use the old dddac enymore.
In the past I use the i2s output from the kali to the dddac but the old dddac is retired now
Thank for your answer I try to find somebody if they want to bring the new dddac so I can compare it also with my dcs dac with upsampler.
Regards,
Rudy
I don't use the old dddac enymore.
In the past I use the i2s output from the kali to the dddac but the old dddac is retired now
Thank for your answer I try to find somebody if they want to bring the new dddac so I can compare it also with my dcs dac with upsampler.
Regards,
Rudy