Doodlebug - USB Isolator

[glunteer]

Following the thread, and waiting for the final project to buy one

 

[fishski13]

got the ODAC loaner yesterday and powered up via the Doodlebug.  will post impressions soon.

 

[fishski13]

running off a dedicated USB PCI on my desktop without other devices connected, into either a B22 or QRV-08, and AD2000 or modded T50RP headphones, the ODAC is etched, bright, rough and harsh.  makes the BM DAC1 i recently sold after 6 years of enjoyment sound warm, smooth, and analog.  with the Doodlebug inserted, less bright/etched and more fleshed out.  more details and more natural sounding.  still not my cup of coffee, but the Doodlebug is an improvement in my rig.  
 
maybe my desktop USB power really sucks, or the implementation of the secret sauce Sabre is black magic.  either way, isolation and clean power brings improvements.
 
huge props to A_A and tomb.
 
from a build perspective, the pads are generous and require a little extra heat.  high quality board and a perfect first time SMD project. 

 

[JacobLee89]

  running off a dedicated USB PCI on my desktop without other devices connected, into either a B22 or QRV-08, and AD2000 or modded T50RP headphones, the ODAC is etched, bright, rough and harsh.  makes the BM DAC1 i recently sold after 6 years of enjoyment sound warm, smooth, and analog.  with the Doodlebug inserted, less bright/etched and more fleshed out.  more details and more natural sounding.  still not my cup of coffee, but the Doodlebug is an improvement in my rig.  
 
maybe my desktop USB power really sucks, or the implementation of the secret sauce Sabre is black magic.  either way, isolation and clean power brings improvements.
 
huge props to A_A and tomb.
 
from a build perspective, the pads are generous and require a little extra heat.  high quality board and a perfect first time SMD project. 

 
I guess the ODAC's perceived brightness bumps the HD650's subdued highs, which is probably why people enjoy the objective stack + HD650's.
 
I've recently installed the output booster mod on my O2, which has made it more "musical", I can imagine that with doodlebug working on the ODAC, the changes would be noticeable on my setup.
 
Count me in as an interested patron, whenever the next batch is ready!
 
Out of curiosity (question open to anyone knowledgeable), would it be of any benefit to put a ferrite core on the cable from the doodlebug to the ODAC?

 

[tomb]

 

  running off a dedicated USB PCI on my desktop without other devices connected, into either a B22 or QRV-08, and AD2000 or modded T50RP headphones, the ODAC is etched, bright, rough and harsh.  makes the BM DAC1 i recently sold after 6 years of enjoyment sound warm, smooth, and analog.  with the Doodlebug inserted, less bright/etched and more fleshed out.  more details and more natural sounding.  still not my cup of coffee, but the Doodlebug is an improvement in my rig.  
 
maybe my desktop USB power really sucks, or the implementation of the secret sauce Sabre is black magic.  either way, isolation and clean power brings improvements.
 
huge props to A_A and tomb.
 
from a build perspective, the pads are generous and require a little extra heat.  high quality board and a perfect first time SMD project. 

 
I guess the ODAC's perceived brightness bumps the HD650's subdued highs, which is probably why people enjoy the objective stack + HD650's.
 
I've recently installed the output booster mod on my O2, which has made it more "musical", I can imagine that with doodlebug working on the ODAC, the changes would be noticeable on my setup.
 
Count me in as an interested patron, whenever the next batch is ready!
 
Out of curiosity (question open to anyone knowledgeable), would it be of any benefit to put a ferrite core on the cable from the doodlebug to the ODAC?

No.  IMHO, that was a Band-Aid patch on the ODAC to cover up the misfortune that it was designed without any power capacitors.  There is no electrolytic capacitor on the ODAC.

 

[tomb]

 

  This project looks super cool. I'll probably build one for my PupDAC. I recently built a DAC that employs USB isolation and it sounds great. A few questions...
 
1) Is there a possibility of using external battery power with this? If so, what would be required?
 
2) This one might not be able to be answered yet... For the guys who use a Gamma 1/2 is there a benefit to this being that it already has the option for an external 5v supply?

1. Sure, you'd need a battery capable of maintaining over 9V at whatever current your USB load is going to draw, for however long to want it to work - so likely something fairly hefty.
The power supply on the doodlebug in it's current form won't be overly efficient for battery use. It will draw 10-20mA or so through the LEDs depending on the exact devices and where you set the output voltage. But with some component changes you could get this down to 1-2mA at the expense of possibly raising the noise floor a tiny amount.
 
2. It depends on whether or not an isolated ground will give any rewards. Some computers (especially laptops) have noisy ground rails and this will get into the analog output of your DAC if it has the digital and analog grounds connected. If there isn't a problem with a noisy ground, or ground loop(s), then a full USB isolator shouldn't offer any benefit compared to power only, everything else being equal.

2. I'm treading on very thin ground here, because I respect AMB and all of his designs.  However, to suggest that the ground is noisy, but the voltage side is not, seems a bit of conjecture.  If the noise exists, the voltage side is always referenced against the ground, so there's no way that one would be "contaminated" and the other not. (They both come through rectification equally to form the DC plus and minus references.)
 
There is some support in the conjecture when it comes to PCB layouts that convert AC to DC.  In those cases, I agree that as little of a connection to ground from the AC side to the DC side is best.  However, the connected ground is always less noisy than the non-connected ground.  I speak from experience in the months and months of experimentation and measurements that Colin Toole and I did on the Millett Hybrid MAX and MiniMAX designs.  The original completely deleted the ground plane.  Later versions connected the ground plane.  Finally, ground planes that were minimally connected had the least noise.  We went from most noise (no AC ground plane) to less noise (ground planes contiguous) to least noise (ground planes minimally connected).
 
In the case of the DoodleBug, we're talking about cleaning up DC that already exists, period.  Trash is on both sides of the voltage differential.

 
 
EDIT:  As you may see from my future posts, I think the difference lies in the sophistication of the DAC.  For instance, does the design depend on numerous regulators throughout the design, or is everything sort of taken as is, and the basic performance of the DAC chip implemented as simply as possible?  I think the small testing I did shows that sophisticated DACs can make very good use of the DoodleBug for a greatly increased performance.  On the other hand, those DACs that are simplest in nature (simplified DAC chips, very limited onboard regulation) don't have the resources in the circuit design to take advantage of the DoodleBug.

 

[tomb]

http://www.diyforums.org/Doodlebug/tests/SkeletonDAC-DoodlebugComparison.htm
http://www.diyforums.org/Doodlebug/tests/PupDAC-DoodlebugComparison.htm
http://www.diyforums.org/Doodlebug/tests/Comparison-Doodlebug-all.htm
 
The links above are self-explanatory.  They're also available on the main prototype web page, here:
http://www.diyforums.org/Doodlebug/tests/Comparison-Doodlebug-all.htm
 
What you will see is a not-so-much improvement on a very simple DAC, the SkeletonDAC.  Nevertheless, the improvement is there and was measured.  However, for a sophisticated DAC like the PupDAC, you'll see that the improvement is both large and significant.  Noise and dynamic range approach the virgin specs of the PCM1794, which is 129dB.  Distortion, both THD and IM, were improved by entire magnitudes:
 
NOISE LEVEL:

 
DYNAMIC RANGE:

 
THD + NOISE:

 
IM DISTORTION:

 
Finally, CROSSTALK:

 
Basically, performance across the entire range of measurements were improved with the DoodleBug.  More than that, my entire testing regime was improved.  The reason for that is that the M-Audio Transit that I've used all of these years was also USB-powered.  The typical testing source that I used was the same thing I would take to meets: a couple of Acer NetBooks.  It is more than obvious now that I have been working under a sub-standard environment to both test and demonstrate Beezar DACs.  Sorry to mention AMB again, but all this time I thought it was his suggestion that my cables were not shielded enough, or that my house contained too much Wi-Fi and wireless telephones, and that none of this was done in a steel enclosure.
 
With the DoodleBug, I've proved to myself that it was the substandard USB and power connection to the M-Audio Transit.*

 
 
 
* For all of these tests, I had a DoodleBug on the M-Audio Transit at all times.  The tests were then conducted with the DACs fed directly from the Netbook USB or from another DoodleBug interjected between the DAC and the Netbook.  It looks like I will have to re-run tests on all Beezar's products, because the noise level is so much lower with the tests on the M-Audio Transit equipped with the DoodleBug.
 
This has literally been a revelation.

 

[DingoSmuggler]

  2. I'm treading on very thin ground here, because I respect AMB and all of his designs.  However, to suggest that the ground is noisy, but the voltage side is not, seems a bit of conjecture.  If the noise exists, the voltage side is always referenced against the ground, so there's no way that one would be "contaminated" and the other not. (They both come through rectification equally to form the DC plus and minus references.)
 
There is some support in the conjecture when it comes to PCB layouts that convert AC to DC.  In those cases, I agree that as little of a connection to ground from the AC side to the DC side is best.  However, the connected ground is always less noisy than the non-connected ground.  I speak from experience in the months and months of experimentation and measurements that Colin Toole and I did on the Millett Hybrid MAX and MiniMAX designs.  The original completely deleted the ground plane.  Later versions connected the ground plane.  Finally, ground planes that were minimally connected had the least noise.  We went from most noise (no AC ground plane) to less noise (ground planes contiguous) to least noise (ground planes minimally connected).
 
In the case of the DoodleBug, we're talking about cleaning up DC that already exists, period.  Trash is on both sides of the voltage differential.

 
 
EDIT:  As you may see from my future posts, I think the difference lies in the sophistication of the DAC.  For instance, does the design depend on numerous regulators throughout the design, or is everything sort of taken as is, and the basic performance of the DAC chip implemented as simply as possible?  I think the small testing I did shows that sophisticated DACs can make very good use of the DoodleBug for a greatly increased performance.  On the other hand, those DACs that are simplest in nature (simplified DAC chips, very limited onboard regulation) don't have the resources in the circuit design to take advantage of the DoodleBug.

I'm think you may have read way to much into my post, I'm a bit confused by your reply.
 
If were talking about non-bus-powered DACs then only the ground is connected, not the 5V. The doodlebug will simply provide an isolated ground, which will only help when there's an issue. (not talking about internal grounding schemes at all)

 

[JacobLee89]

 
No.  IMHO, that was a Band-Aid patch on the ODAC to cover up the misfortune that it was designed without any power capacitors.  There is no electrolytic capacitor on the ODAC.

Thanks  

 
I'm also interested on the "spikes" found on the 1k mark on the charts "dynamic range" and "THD and noise" for the doodlebug tests. I'm quite new to reading these types of charts and can't quite make sense of them.

 

[tomb]

 

  2. I'm treading on very thin ground here, because I respect AMB and all of his designs.  However, to suggest that the ground is noisy, but the voltage side is not, seems a bit of conjecture.  If the noise exists, the voltage side is always referenced against the ground, so there's no way that one would be "contaminated" and the other not. (They both come through rectification equally to form the DC plus and minus references.)
 
There is some support in the conjecture when it comes to PCB layouts that convert AC to DC.  In those cases, I agree that as little of a connection to ground from the AC side to the DC side is best.  However, the connected ground is always less noisy than the non-connected ground.  I speak from experience in the months and months of experimentation and measurements that Colin Toole and I did on the Millett Hybrid MAX and MiniMAX designs.  The original completely deleted the ground plane.  Later versions connected the ground plane.  Finally, ground planes that were minimally connected had the least noise.  We went from most noise (no AC ground plane) to less noise (ground planes contiguous) to least noise (ground planes minimally connected).
 
In the case of the DoodleBug, we're talking about cleaning up DC that already exists, period.  Trash is on both sides of the voltage differential.

 
 
EDIT:  As you may see from my future posts, I think the difference lies in the sophistication of the DAC.  For instance, does the design depend on numerous regulators throughout the design, or is everything sort of taken as is, and the basic performance of the DAC chip implemented as simply as possible?  I think the small testing I did shows that sophisticated DACs can make very good use of the DoodleBug for a greatly increased performance.  On the other hand, those DACs that are simplest in nature (simplified DAC chips, very limited onboard regulation) don't have the resources in the circuit design to take advantage of the DoodleBug.

I'm think you may have read way to much into my post, I'm a bit confused by your reply.
 
If were talking about non-bus-powered DACs then only the ground is connected, not the 5V. The doodlebug will simply provide an isolated ground, which will only help when there's an issue. (not talking about internal grounding schemes at all)

Yes - I didn't realize that you were talking about externally-powered DACs.  I think the original question was y1/y2, so I took that to also mean the power options available,  one of which is USB-powered (or even a walwart, only).  If someone used a Sigma25 or similar, then yes, I agree completely that the difference from the power side of the DoodleBug would probably be negligible.
 
Realize also that there's been debate in the past about whether to connect the USB shield (and connector) through the cable to the DAC's PCB ground.  I thought perhaps some of that was implied in your post, too.
 
Sorry that I misinterpreted.

 

[tomb]

 

 
No.  IMHO, that was a Band-Aid patch on the ODAC to cover up the misfortune that it was designed without any power capacitors.  There is no electrolytic capacitor on the ODAC.

Thanks  

 
I'm also interested on the "spikes" found on the 1k mark on the charts "dynamic range" and "THD and noise" for the doodlebug tests. I'm quite new to reading these types of charts and can't quite make sense of them.

RMAA uses a 1K carrier signal.  Unless the testing software specifically filters it out (RMAA doesn't), you'll see it on every USB DAC/device:  The same thing occurs on the IM plot at 60Hz.
 
Here's AMB's y1 THD plot (Config. B):

 
y1's IM distortion plot (Config. B):

 

[mcandmar]

Hi Tomb,
 
Nice to see somebody quantifying the difference an isolator/clean power source can make, nice work. For me it verifys what i found with my own two ears, always good to know it wasn't my imagination after all.
 
Quote:

Realize also that there's been debate in the past about whether to connect the USB shield (and connector) through the cable to the DAC's PCB ground.  I thought perhaps some of that was implied in your post, too.

 
In relation to the above, if you study the ODAC you will find the USB shield is not connected to the circuit, however there is an unpopulated solder pad to do so which implys to me the designer was in two minds about it, but ultimately decided to leave it disconnected in the end.  I dont see the need for it personally unless you are using a battery powered device and want its power supply ground referenced.

 

[KimLaroux]

Oh, that's a very exciting project. Great to see the objective measurements backing up the theory. I'm blown by the improvements made by this simple circuit. (forgetting the complexity of whatever is inside the chip.)
 
But it seems a bit counter-intuitive that simpler DACs will see less improvements than more complex ones. One would think that DACs with better PSUs would be less affected by the quality of their supply. Unless it's not a matter of clean supply, but supply overload or outright data noise.
 
Have you tested it with the grubDAC? I'm curious if this would improve the performance of a grubDAC+Carrie.

 

[tomb]

  Oh, that's a very exciting project. Great to see the objective measurements backing up the theory. I'm blown by the improvements made by this simple circuit. (forgetting the complexity of whatever is inside the chip.)
 
But it seems a bit counter-intuitive that simpler DACs will see less improvements than more complex ones. One would think that DACs with better PSUs would be less affected by the quality of their supply. Unless it's not a matter of clean supply, but supply overload or outright data noise.
 
Have you tested it with the grubDAC? I'm curious if this would improve the performance of a grubDAC+Carrie.

I think simpler DACs probably have nowhere to go in performance.  Not to insult everyone's favorite cheap tube amp, but it's sort of like the Starving Student.  Once the inherent noise in the tube circuit exceeds the switcher power supply, there's little performance gain achieved by substituting a better power supply.
 
Beezar is moving this month.  Unfortunately, all of my GrubDACs are packed and I can't get to them right now.

  However, I suspect the results would be somewhere in-between that of the SkeletonDAC and the pupDAC, because the GrubDAC has an onboard regulator.

 

[KimLaroux]

  I think simpler DACs probably have nowhere to go in performance.  Not to insult everyone's favorite cheap tube amp, but it's sort of like the Starving Student.  Once the inherent noise in the tube circuit exceeds the switcher power supply, there's little performance gain achieved by substituting a better power supply.
 
Beezar is moving this month.  Unfortunately, all of my GrubDACs are packed and I can't get to them right now.

  However, I suspect the results would be somewhere in-between that of the SkeletonDAC and the pupDAC, because the GrubDAC has an onboard regulator.

 
I really have no idea what's you're talking about.
 
LOL 

 
 
But the next question is, would the Carrie gain performance?
 
But then it would be ironic to add a wall-wart and a middle-man to a two-in-one whose meaning of existence is to be small and solely powered by USB.
 
Have fun moving.