In the last few years the floodgates have opened to all manner of digital audio tweaks. One of the ones in vogue now is "cleaning" the USB data bus. It makes "intuitive sense" with USB being connected to a noisy computer which conveys that noise to the DAC and hence, damages the sound.
Intuition is fine but it needs to be validated. And validation here is pretty easy since we are talking about noise and noise can be measured with ease. We need to first verify that computers do generate noise that travels to the output of the DAC and that such devices do something to reduce that.
Alas, when I tested one such device, the UpTone Regen, it did anything but this. Not only did it not reduce any noise in the output of the DAC, it actually introduced one that correlated with USB packet timing! Mind you, it was at very low levels but nevertheless, no such tweak better have the opposite effect of what it is supposed to do -- i.e. reduce noise.
Fast forward 18 months or so later and UpTone now has a feature which should have been in the original product. Namely, galvanic (electrical) isolation. At least that way, in theory anyway, it would perform a useful function should noise from the computer be travelling over ground connection from USB. The device is called aptly, "ISO Regen." It has a retail price of $325 plus $9 shipping.
Like the original Uptone Regen, this too is stuffed in a tiny, commodity hobby project aluminum box:
It is half the size of set of playing cards. Other than trying to save money, I can't figure out why a device like this needs to be stuffed into a small enclosure. After all, space is friend of a designer in being able to distance noisy components such as switchmode power supplies.
Anyway, as you see there is not much to the device. It has an external power brick and comes with a set of cables including a stub USB gender converter made out of PC board.
There is a red switch that allows isolation to be turned on and off. It is the chintzy switch that feels like it is going to break any minute when you touch it. But it is good to have it as to test as you see later. Note that its markings are reversed. "1" means isolation is there, and "ON" means not! At least that is what I learned from reading online. The default setting was "1."
As an aside, some of you may know that UpTone cancelled my order of the ISO Regen due not wanting to sell it to me. So I had to resort to a friend of the forum to order it and send it to me. He incurred shipping costs and I will too in sending it back them. Really silly and unprofessional move on their behalf. Note however that none of this has interfered with my judgement of the product.
UpTone ISO Regen Measurements
I plan to do more testing but thought I start by testing the ISO Regen using Mytek Brooklyn DAC which I also have on loan. The Mytek Brooklyn retails for $2,000 which seems like a good fit for a tweak like ISO Regen which sots $325. I will test it later with lower cost DACs that I have in my lab.
For this testing, I used the 24-bit, 48 Khz J-test signal which is essentially a 12 Khz tone for the purposes of this type of testing. In an ideal situation, we would only see a spike at 12 Khz and not much else. The Mytek Brooklyn does very well here but to find something objectionable, I zoomed way into its noise floor. I used 32-pass averaging to lower the random noise floor of my Audio Precision Analyzer and that of the DAC. What is left is any correlated distortion products which would be of interest.
Furthermore, I zoomed way into the waveform to look at the noise floor. The upper part of the graph is at -110 db which is extremely low level. The noise floor is at -130 db. So really nothing here is of audible consequence but for the sake of engineering excellence, we will go ahead and test anyway.
The player was Foobar 2000 on my main audio workstation. It used ASIO interface of the Mytek drivers to play the J-test signal. I started with two runs of Mytek by itself. Both showed a spike at 1.03 Khz. Its level changed from run to run by a 2-3 db. It did so in every condition so do not consider that significant.
I then followed this by these other variations:
1. Mytek alone but with Novabench computer benchmark running concurrently with it. This sharply increased the load on the computer but had zero difference on the output of Mytek. It shows what is to be expected in that well designed DACs isolate themselves from vagaries of USB.
2. Insertion of UpTone ISO Regen into the USB path. At first this caused Foobar to output strange error and would refuse to play. I unplugged and replugged things and it started to work. It seems it requires a specific sequence to work which I think is to plug in the ISO Regen first to the computer, and then the DAC. Perils of USB hubs which this device utilizes.
There was no difference in measured results. Sample to sample variations existed just the same but nothing happened to that spike at 1.03 Khz, nor was the noise floor impacted in any way.
3. Turned off USB isolation in ISO Regen. Again, it made no difference in the measurements.
4. Finally, I replaced the short USB cable between ISO Regen with the stub PCB version. Again, measurements made no difference at all.
The following is the output of the Audio Precision with all the graphs on top of each other. As you see, there is absolutely no meaningful difference between any of these runs. Nothing is fixed. Nothing is broken. The ISO Regen in these measurements is the same as not having it there.
("32-bit" should say "24-bit')
Again, note that we are going way out to find something of note there. Even if there were differences in these measurements, simple psychoacoustics would show them to be completely immaterial. But we did not even get there.
If someone has seen any measurements that show the output of the DAC to vary with ISO Regen, I love to see it. For now, I have no choice but to declare it ineffective.
Subjective Listening Test
There is a great line in Bill Cosby's fantastic HBO Special where he was wishing for a boy but his wife gave birth to a girl. While changing the diaper on the girl, his father gives him advice that "he did not put a stem on the apple!" Watch it here at 39:00
And Cosby says, "there comes [....] the time where you give the ridiculous a chance. " So here I am, doing a listening test to see if the ISO Regen makes an audible difference.
Much like folks who do this kind of thing for a living, I am going to brag about my listening system
which in this case included the Stax SRM-007t differential tube amplifier driving the Stax "Pro" headphones. These I think retailed for $6,000 when I bought them. They were connected to the the unbalanced output of the Mytek Brooklyn.
For music, I selected an AIX well recorded track that was put out by Mark for test of high-resolution audio against CD. And a track from Reference Recording.
I started to play these tracks and moved the red switch up and down. I detected absolutely no audible difference. These tracks sounded fantastic with the switch in either position.
This was not an exhaustive test of course but given the combined force of measurements, I say this device does not perform any usable function in my system.
Summary
Despite so many defenders of this class of products, measurements and objective analysis of the devices lends to no useable benefits. High performance DACs are designed to sound and measure excellently without such tweaks (after all, if there was something to these devices, they would include them in the DACs themselves at the prices they are charging us).
My opinion of hearing improvements from users is simple: when you test things you tend to focus more and with that, you hear details that were always there but you did not note them. Such difference can be quite convincing but unfortunately is not real. The soundwaves out of the box do not need to change to hear such differences.
My suggestion would be to save your money and put it toward purchasing music.
As always, I very much welcome corrections, comments, etc. from manufacturers or anyone else with feedback. Such testing is prone to errors and I am happy to correct any that may exist here.
Intuition is fine but it needs to be validated. And validation here is pretty easy since we are talking about noise and noise can be measured with ease. We need to first verify that computers do generate noise that travels to the output of the DAC and that such devices do something to reduce that.
Alas, when I tested one such device, the UpTone Regen, it did anything but this. Not only did it not reduce any noise in the output of the DAC, it actually introduced one that correlated with USB packet timing! Mind you, it was at very low levels but nevertheless, no such tweak better have the opposite effect of what it is supposed to do -- i.e. reduce noise.
Fast forward 18 months or so later and UpTone now has a feature which should have been in the original product. Namely, galvanic (electrical) isolation. At least that way, in theory anyway, it would perform a useful function should noise from the computer be travelling over ground connection from USB. The device is called aptly, "ISO Regen." It has a retail price of $325 plus $9 shipping.
Like the original Uptone Regen, this too is stuffed in a tiny, commodity hobby project aluminum box:
It is half the size of set of playing cards. Other than trying to save money, I can't figure out why a device like this needs to be stuffed into a small enclosure. After all, space is friend of a designer in being able to distance noisy components such as switchmode power supplies.
Anyway, as you see there is not much to the device. It has an external power brick and comes with a set of cables including a stub USB gender converter made out of PC board.
There is a red switch that allows isolation to be turned on and off. It is the chintzy switch that feels like it is going to break any minute when you touch it. But it is good to have it as to test as you see later. Note that its markings are reversed. "1" means isolation is there, and "ON" means not! At least that is what I learned from reading online. The default setting was "1."
As an aside, some of you may know that UpTone cancelled my order of the ISO Regen due not wanting to sell it to me. So I had to resort to a friend of the forum to order it and send it to me. He incurred shipping costs and I will too in sending it back them. Really silly and unprofessional move on their behalf. Note however that none of this has interfered with my judgement of the product.
UpTone ISO Regen Measurements
I plan to do more testing but thought I start by testing the ISO Regen using Mytek Brooklyn DAC which I also have on loan. The Mytek Brooklyn retails for $2,000 which seems like a good fit for a tweak like ISO Regen which sots $325. I will test it later with lower cost DACs that I have in my lab.
For this testing, I used the 24-bit, 48 Khz J-test signal which is essentially a 12 Khz tone for the purposes of this type of testing. In an ideal situation, we would only see a spike at 12 Khz and not much else. The Mytek Brooklyn does very well here but to find something objectionable, I zoomed way into its noise floor. I used 32-pass averaging to lower the random noise floor of my Audio Precision Analyzer and that of the DAC. What is left is any correlated distortion products which would be of interest.
Furthermore, I zoomed way into the waveform to look at the noise floor. The upper part of the graph is at -110 db which is extremely low level. The noise floor is at -130 db. So really nothing here is of audible consequence but for the sake of engineering excellence, we will go ahead and test anyway.
The player was Foobar 2000 on my main audio workstation. It used ASIO interface of the Mytek drivers to play the J-test signal. I started with two runs of Mytek by itself. Both showed a spike at 1.03 Khz. Its level changed from run to run by a 2-3 db. It did so in every condition so do not consider that significant.
I then followed this by these other variations:
1. Mytek alone but with Novabench computer benchmark running concurrently with it. This sharply increased the load on the computer but had zero difference on the output of Mytek. It shows what is to be expected in that well designed DACs isolate themselves from vagaries of USB.
2. Insertion of UpTone ISO Regen into the USB path. At first this caused Foobar to output strange error and would refuse to play. I unplugged and replugged things and it started to work. It seems it requires a specific sequence to work which I think is to plug in the ISO Regen first to the computer, and then the DAC. Perils of USB hubs which this device utilizes.
There was no difference in measured results. Sample to sample variations existed just the same but nothing happened to that spike at 1.03 Khz, nor was the noise floor impacted in any way.
3. Turned off USB isolation in ISO Regen. Again, it made no difference in the measurements.
4. Finally, I replaced the short USB cable between ISO Regen with the stub PCB version. Again, measurements made no difference at all.
The following is the output of the Audio Precision with all the graphs on top of each other. As you see, there is absolutely no meaningful difference between any of these runs. Nothing is fixed. Nothing is broken. The ISO Regen in these measurements is the same as not having it there.
("32-bit" should say "24-bit')
Again, note that we are going way out to find something of note there. Even if there were differences in these measurements, simple psychoacoustics would show them to be completely immaterial. But we did not even get there.
If someone has seen any measurements that show the output of the DAC to vary with ISO Regen, I love to see it. For now, I have no choice but to declare it ineffective.
Subjective Listening Test
There is a great line in Bill Cosby's fantastic HBO Special where he was wishing for a boy but his wife gave birth to a girl. While changing the diaper on the girl, his father gives him advice that "he did not put a stem on the apple!" Watch it here at 39:00
And Cosby says, "there comes [....] the time where you give the ridiculous a chance. " So here I am, doing a listening test to see if the ISO Regen makes an audible difference.
Much like folks who do this kind of thing for a living, I am going to brag about my listening system
which in this case included the Stax SRM-007t differential tube amplifier driving the Stax "Pro" headphones. These I think retailed for $6,000 when I bought them. They were connected to the the unbalanced output of the Mytek Brooklyn.For music, I selected an AIX well recorded track that was put out by Mark for test of high-resolution audio against CD. And a track from Reference Recording.
I started to play these tracks and moved the red switch up and down. I detected absolutely no audible difference. These tracks sounded fantastic with the switch in either position.
This was not an exhaustive test of course but given the combined force of measurements, I say this device does not perform any usable function in my system.
Summary
Despite so many defenders of this class of products, measurements and objective analysis of the devices lends to no useable benefits. High performance DACs are designed to sound and measure excellently without such tweaks (after all, if there was something to these devices, they would include them in the DACs themselves at the prices they are charging us).
My opinion of hearing improvements from users is simple: when you test things you tend to focus more and with that, you hear details that were always there but you did not note them. Such difference can be quite convincing but unfortunately is not real. The soundwaves out of the box do not need to change to hear such differences.
My suggestion would be to save your money and put it toward purchasing music.
As always, I very much welcome corrections, comments, etc. from manufacturers or anyone else with feedback. Such testing is prone to errors and I am happy to correct any that may exist here.
