Science Thread: Review of Audioquest Jitterbug and Uptone Regen USB Conditioners

Apr 3, 2010
16,022
0
0
Seattle, WA
#1
Hi everyone. As I promised, here is my review and report on performance of Audioquest Jitterbug (AQ) and Uptone Regen USB conditioners. I had bought the AQ and Michael (Dallasjustice) was kind enough to loan me the Regen. Before I get into the meat of this review, some key notes:

Note 1: This is posted in the science thread. So please don’t post that “but my ears say otherwise.” :) We already know that and subjective reviews have generally been positive on these products. This assessment is designed for the most part to be objective although since I was asked, I will also post some subjective expressions.

Note 2: The information in this post is entirely mine and not the position of WBF Forum.

Note 3: I am not a professional reviewer. I did these tests because I was curious. I make no assurance to accuracy of any of it. It is entirely possible that my test setup is screwed up even though I have repeated them a couple of times to increase confidence in them. If you are one of these manufactures and have data that disagrees with what is posted here, please share them.

Note 4: The test data here is specific to my setup. It is possible and indeed likely that other sources and DACs would act differently. I personally don’t think the overall picture and message will change but I can’t represent that without testing a lot more equipment.

Note 5: As a reported a week ago, my dog “unboxed” the AQ by chewing off its case. Electrically it seems to be intact and it functions by passing through the USB bus. While there is a slight chance that it no longer does what it is supposed to, I think the chances of that are very low.

Executive Summary
There is a lot of data here so I thought I give you the punchline for the less than patient :). In a nutshell, I cannot back any positive reports of these products. At best they do nothing, at worse they add tiny amount of (inaudible) distortions of their own. There are some interesting twists and turns here so I hope you read the rest of the article.

Background
The class of product attempts to clean up/precondition the digital USB bus on the way to your DAC. USB is a simple bus. It has a ground and power line which can supply energy to downstream devices. It also has a pair of lines that are used to transmit and receive data. Being a predominantly digital computer interface, manufacturers are under no obligation to provide laboratory level of precision in either supply power it provides or perfectly clean signals. Compliance tests focus on reliable data communication and that is it. So in theory, cleaning up this type of interface can provide useful gains and hence the creation of this new category of products.

Expanding the last point, computers are noisy devices and some of that noise rides on USB power lines and its signal. If the DAC does nothing to clean up such signals, then it is possible for them to bleed into the DAC. Timing of the USB bus can also change due to noise and if a DAC uses synchronous (lousy) mode of USB operation, it could be impacted negatively.

Here is the strong counterpoint though. We are in domain of “what is best” in this forum. I sure hope anyone buying a DAC is getting one where the designer knows very well that these issues exist and has already put the few cents or dollar or two worth of parts that these products have. As such, I personally don’t expect any high-end DAC to benefit from these products. If any do, I would question the DAC company before resorting to dangling a device like this in front of them.

Anyway, theory is one thing but the proof of the pudding is in backing it with objective data. In this situation, I used my everyday laptop which is an HP Zbook 14. This is a “professional workstation” aimed at high productivity/CAD workers. It cost me $2,200. So not some cheap $300 machine here. I do a ton of listening tests on this laptop and find its audio performance to be good. In this scenario the laptop is feeding the DAC either directly (which I call “nothing”) or through one of these two devices.

Test Setup
I came up with three sets of tests. It is important that you read this section carefully as otherwise you will be lost in a sea of seemingly similar graphs. They are very different and test different things:

1. Filtering of USB power line. I took a USB cable and cut one side of it off. I stripped the wires and then fed that to a “differential probe.” This is a $300 device that acts like the balanced connection on your audio equipment. It rejects common noise that exists all around my test gear from my laptop to different instruments. While it doesn’t guarantee perfect results, it goes way beyond using a simple probe.

The output of the differential probe (which divides the signal by 10) is fed to the BNC input of my Audio Precision Analyzer. I then use a simple FFT to check for the spectrum of the power line. Use of an audio analyzer here is unorthodox as usually a much higher frequency spectrum analyzer is used. Those analyzers unfortunately do not have high precision. And at any rate, we care about seeing the spectrum in audioband and thereabouts. What exists at many megahertz may have some audible impact but I rather know what is happening that our audio equipment can reproduce.

If these devices completely filter the power line, it should be a flat line at the limit of the analyzer.

2. These next two tests go to the heart of the matter: what the DAC puts out. At some level it doesn’t matter what is going on before the DAC. The proof is what we feed to our amplifier which is the output of the DAC.

In the first instance of this test, I simply look at the output of the DAC when nothing is played. In other words, look at the noise floor of its output. As with Test #1, I show both wideband response to 130 KHz and also limited to 24 Khz.

3. For this test I feed the DAC a J-test signal at 48 Khz sampling. J-test looks like a 12 Khz tone in such a graph. Anything else other than a sharp single spike is noise, jitter or other types of distortion created. An idea DAC would have just a spike at 48 KHz and nothing else. Here I show this spectrum both wideband and audio bands.

The most impact from such devices would be in low cost, USB powered devices. So I tested with the only one I have currently which is the Meridian Explorer DAC. This is a few steps above in performance that Chinese $10 USB sound cards. And the minimum level of performance that I imagine members here would use.

Test 1: USB Power Filtering
So let’s dig in. Here is the Audioquest Jitterbug (AQ) versus straight wire connected to my laptop USB port:

Before getting into the comparison, we see that the laptop USB power does have predicted impurities. Those specific peaks are functions of events inside my laptop running at those frequencies. The tallest one is in the middle at around 52 Khz.

Note that the problem is not as bad as it seems. I have set the vertical scale to -80 dBFS. So this is highly amplified graph. All the other spikes other than the center one are actually below the noise floor of CD/16-bit audio.

Sadly we see that the AQ does nothing for us here. The minor changes are not consequential and the overall picture remains as if we had not used this device.
Let’s now compare the cable to Regen:

Now we are talking. All the tones are gone and the noise floor lowered. This makes sense since Regen provides its own power so what comes out of my PC is inconsequential. In this regard, the Regen is performing something compared to AQ which did nothing.

Test 2: Noise floor of the DAC output
If the computer is electrically coupled to the DAC, one would expect noise to travel to the downstream devices. Let’s look at that and compare it to AQ looking at spectrum up to 130 Khz:

The AQ spikes in green strangely seem worse than piece of wire. It is hard to know the reason for that but clearly nothing good is being done here.

The situation is even worse for the Regen!

Now, these are again at pretty low amplitudes and the vertical scale is highly exaggerated. But one wonders how devices that are supposed to reduce noise, are actually increasing them.

Test 3: Jitter Response
One of the advertised features of these devices is reduced jitter. Usually that is stated in the form of reduced jitter on USB bus but such reduction is not of interest to us (and at any rate is not substantiated by the manufacturer with measurements). What we care is jitter coming out of our DAC. To that end, let’s look that output and compare with and without AQ:

Apologize for the darkness of the graph. What we see there is no difference with insert of AQ. Whatever jitter the Meridian DAC puts out normally is there just the same. As is the noise floor.

Let’s now look at Regen:

Now this is really strange. We now have new distortion spiked in red once we inserted the Regen in the path!!! Again the levels are very low but what the heck? A device implying that it reduces such distortions actually increases them? The Regen uses a USB hub so it is possible that this chip has worse USB performance than one in my laptop.

Let’s zoom into audio band and see if the picture is any different with AQ:

Good news and bad news is that the AQ did nothing in frequencies below 24 Khz. No damage but no improvement either.

Regen on the other hand, did add some distortions:

And this time it seems to be correlated jitter (or modulated reference voltage) at 4 Khz, causing two sidebands at 12-4 = 8 Khz and 12+4 = 16 Khz. Again, the levels are very low but what good is a device that didn’t attenuate any of our problems and added a couple of spikes of its own???

UPDATE: further testing later in the thread with a lab linear power supply and Regen showed this:



We see that the 8 Khz distortion tone has disappeared and the one at 16 Khz reduced too.

So let's compare the Regen with Linear Power supply and nothing in line:



Other than a tiny potential improvement at 16 Khz, there is really no difference to speak of here. In this sense, a Regen with linear power supply becomes like AQ. That is, showing no evidence of improvement in DAC output. All the distortions that were there without these devices, are still there with these devices.

Objective Analysis Summary
Simply put, there is no good news here. Both devices degraded the performance of the DAC a bit and did nothing to improve it. There are reasons for this that I won’t go into but let’s be careful in using our intuition that when something is “filtered,” it is always good. That is not how this interface works.

Subjective Listening Test
I tested the AQ both in “parallel” and “serial” modes. In parallel mode you just plug it into a port next to the one you are using. Doing that to my Berkeley and Audiophilleo USB to AES/S/PDIF to my Mark Levinson DAC did nothing. This was easy to test since it does not interrupt the flow of music.

Testing in “serial” mode was much harder since it would naturally stop the music from playing. Fast switching/comparison was not possible. I tried anyway and couldn’t hear a difference for what it is worth.

I then tested the Regen. Once again, comparison is hard and next to impossible due to multi-second switching. On first insertion, I thought I heard an improvement. I went back and forth and the tables turned and no Regen sounded better :p. In other words, the differences imagined or real, are well within placebo range.

Conclusions
I know we have had our differences before in how audio equipment is evaluated. For those of you sold on these products, I like to implore you that you go into this evaluation with a skeptical mind. Don’t assume that any of the technical explanations given are supposed to improve things. Data here clearly shows that our intuition here is not correct. The circuit theory is much more complex and don’t lend themselves to conclusions we draw. Assume that the device could as well hurt the performance as it can help. Then see if you can reliably hear an improvement.

For my money, I think a rock from the garden put on top of my DAC may do better than either one of these devices. :D For the price of Regen, you can buy 10 high-res downloads and I am confident that would bring more happiness to your music life than this device.

Comments and questions welcome :).
 

Elberoth

Member Sponsor
Dec 16, 2012
1,858
3
38
Poland
#2
+1

BTW Amir - did you use the supplied SMPS with the regen ? I wonder if the added distortions you see may be due to SMPS dumping HF noise, both to the AC line and its DC output.

In my system, I hear a clear improvement when linear PSU is used.

Sound wise, Regen works great with my Trinity DAC, but not so with my Lampi.
 
Last edited:
Apr 3, 2010
16,022
0
0
Seattle, WA
#3
Yes I only used the switchmode power supply it came with. I was going to test with linear supply but per graph I post, what they supplied seemed pretty clean in audio band so I did not do that. When I get some time, I will re-test with my lab supply.
 

Bruce B

WBF Founding Member, Pro Audio Production Member
Apr 26, 2010
6,558
3
38
Seattle, WA
www.pugetsoundstudios.com
#4
Thank-you Amir.. As it so happens, Mike was kind enough to let me borrow an AQ Jitterbug, Regen and a Trinity USB thingy. I tried them both on my Playback Designs and my Lampi Golden Gate.
Subjectively, I could hear no changes using them with my Playback Designs MPS-5. I did a couple of audio captures and could see nothing different as well in my workstation.

With the Golden Gate, I thought I could hear a change with the Regen and Trinity (no change with the AQ). This could also be because the GG is new and tubes/?? are still breaking in and changing changing. Same with the audio capture.

Have you tested USB cables with and without power side from digital?
 

Mike Lavigne

Member Sponsor & WBF Founding Member
Apr 25, 2010
7,085
14
38
#5
Thank-you Amir.. As it so happens, Mike was kind enough to let me borrow an AQ Jitterbug, Regen and a Trinity USB thingy. I tried them both on my Playback Designs and my Lampi Golden Gate.
Subjectively, I could hear no changes using them with my Playback Designs MPS-5. I did a couple of audio captures and could see nothing different as well in my workstation.

With the Golden Gate, I thought I could hear a change with the Regen and Trinity (no change with the AQ). This could also be because the GG is new and tubes/?? are still breaking in and changing changing. Same with the audio capture.

Have you tested USB cables with and without power side from digital?
the Trinity cable is actually a Totaldac USB cable. I was likely not clear enough with Bruce on that issue.
 

Mike Lavigne

Member Sponsor & WBF Founding Member
Apr 25, 2010
7,085
14
38
#6
as far as the Trinity dac; does it matter that it's USB input has true galvanic isolation? as far as how it might work with these devices?
 

Elberoth

Member Sponsor
Dec 16, 2012
1,858
3
38
Poland
#7
Yes I only used the switchmode power supply it came with. I was going to test with linear supply but per graph I post, what they supplied seemed pretty clean in audio band so I did not do that. When I get some time, I will re-test with my lab supply.
John Westlake has done very interesting measureemnts on Regen's PSU.

Here is a graph showing Regen with stock SMPS:


And hre the Regen with linear one:


You can see massive "clumps" of spread spectrum SMPS noise that the provided SMPS is introducing.

Original post:
http://www.pinkfishmedia.net/forum/showpost.php?p=2689811&postcount=1233
 

Elberoth

Member Sponsor
Dec 16, 2012
1,858
3
38
Poland
#8
as far as the Trinity dac; does it matter that it's USB input has true galvanic isolation? as far as how it might work with these devices?
It helps. But Regen is essentially trying address a different problem (packet noise, generated by PHY receiver).
 

Superdad

New Member
Apr 22, 2015
25
0
0
#12
LOL! Quite right. Once again, the emperor has no clothes despite his protestations to the contrary.
I don't have time at the moment, but I can say that the tests are a bit flawed and that you are focusing on the wrong things. The REGEN's main purpose is to improve the signal integrity going into the DAC (such that the DAC's own PHY chip generates less packet-data-noise and ground-plane noise), and this is best measured with a very wide band scope and an eye-pattern test. John Westlake (a competitor in that he is planning, based on what he has seen with the REGEN, to develop a more elaborate, 3-stage version in a larger case that cant be placed at the DAC's input) did just that with a REGEN and had a host of compliments about what he saw. Here are the two eye-patterns he posted:

Without the REGEN--
direct usb.jpg

With the REGEN--
regen usb.jpg

==========
And some quotes from him about the above measurements and his comments:

"Above is a colour graded eye pattern of the Direct PC USB connection to an XMOS based DAC (streaming 1KHz, 0dB FS) High speed USB.
Note the multitude of lines, but more importantly that the background is not black (Black means no data points fell in this area) - the background its dark blue because during the acquisition low frequency "Runt" waveforms where sampled in this area - these "Runt"waveforms occurred as the PC / operating system processed other applications / Operating system house keeping tasks etc. On the Analogue scope it was a horrific mess…
Now with the ReGen USB Data, notice not only the much cleaner waveform but far more importantly that there is very little "Runt" data - the background is Black. The USB Hub IC in the Regen has for the most part cleanly repackaged the data."
"I'd agree that the Regen resulted in the greatest measurable difference to the downstream USB Data."

And a another post from Mr. Westlake with additional appropriate measures:
This evening I tried using a real-time spectrum analyzers to look at the USB data, but basically my unit is design for observing narrower bandwidth RF transmissions then the 500MHz Span which is required to observe the USB spectrum. I’m rather disappoint the unit has proved to be so useless. I’ll replace it with the new unit from the US once it arrives – space is at too much of a premium in the lab.

I made some further measurements with a faster digital analyzer 15GHz Band width – but it lacks the large memory for the deep color grading that the slower scope has, but it still managed to capture the LF events on the Direct USB PC connection.

The Eye pattern of the direct USB connection at 200pS Div – note the LF Runt pulses are just visible despite the short sampling time:-

https://dl.dropboxusercontent.com/u/...SB 200pS.jpg

The same 200pS Div measurement with the ReGen:-

https://dl.dropboxusercontent.com/u/...en 200pS.jpg

No trace of LF Runt pulses – but a slightly higher jitter on the rising edge – but it’s the LF jitter that matters, which the Regen clearly improves upon.

Direct USB connection at 500pS Div – again note the LF Runt data:-

https://dl.dropboxusercontent.com/u/...SB 500pS.jpg

And via the Regen at 500pS Div, showing a cleaner Eye pattern without a trace of LF runt Data :-

https://dl.dropboxusercontent.com/u/...en 500pS.jpg

As above but with the Jitter Bug fitted between the ReGen output and XMOS based DAC PCB.

https://dl.dropboxusercontent.com/u/...JB 500pS.jpg

I’d almost suggest the Jitter bug degraded the crossing point jitter – I suggest its really designed to filter the power supply rail on the USB line which is most beneficial for USB powered DAC’s / Devices, but not so beneficial for self powered DAC’s like the MDAC.

Any RF filtering is a good idea, but IMO the Regen offers better “Performance per pound” then the Jitter Bug for MDAC type designs which are self powered.

Phase noise test of the ReGen's clock - measured from the Clock output pin on the USB Hub IC - I've seen worst

Some odd components at 3.3KHz and beyond... but really not so bad.

https://dl.dropboxusercontent.com/u/...MHz Xout.jpg

-114dBC at 1KHz offset -133dBC @ 100KHz.

----------------------

And with regards to the graphs of SMPS spread spectrum noise that Elberoth posted on this thread's first page, note how far down they are and how high in frequency. I won't defend the $12 SMPS we include with the REGEN (other than that this 7.5V/2.93A/22W model is better than the pile of others we tried in the range), and many people are using the REGEN with LPS units from modest to fancy (our own choke-filtered, dual-output, 5-7A JS-2 being one of them). But let's keep perspective here.

I defies logic for there to be measured increases in jitter from insertion of the REGEN, so I suspect something is amiss with Amirm's testing. I will ask my partner, engineer John Swenson to review and comment upon that. By the way, you did not indicate what DAC or USB input you used in your testing.

I would welcome further meaningful dialog about our popular product. With just under 2,000 units sold around the world and being used with DACs from modest to mega, do dismiss our device without really understanding how it works does not seem appropriate.

Thank you and regards,

--Alex Crespi
UpTone Audio LLC
 

Elberoth

Member Sponsor
Dec 16, 2012
1,858
3
38
Poland
#14
LOL! Quite right. Once again, the emperor has no clothes despite his protestations to the contrary.
I think it is way too early for such a conclusions. The plots posted by Amir clearly show that it improves the quality of USB power A LOT. Now we need to find out why on the two DACs tested (or just one, Amir ?) Regen introduced some other, unexpected problems.

Wether the benefits outweight the problems - that is a completely different story and may be very much system dependent.
 
Apr 3, 2010
16,022
0
0
Seattle, WA
#15
Welcome to the forum Alex. Appreciate your participation and data added to the conversation.
I defies logic for there to be measured increases in jitter from insertion of the REGEN, so I suspect something is amiss with Amirm's testing. I will ask my partner, engineer John Swenson to review and comment upon that.
It was surprising to me too but is what I got. Note that their levels is very low.

By the way, you did not indicate what DAC or USB input you used in your testing.
I did actually. I used the Meridian Explorer:



I am unclear what you mean by "USB input." I used the short cable that came with the box to connect it to my USB port on my laptop.

I would welcome further meaningful dialog about our popular product. With just under 2,000 units sold around the world and being used with DACs from modest to mega, do dismiss our device without really understanding how it works does not seem appropriate.
This is a puzzling remark. What is "appropriate" for any audiophile is what sound is produced by the DAC. What the waveform looks like coming into it is of no concern. What is next? Checking to see what the CPU is doing???

Eye pattern analysis is very useful for data communication where long lines are involved or very high speeds. Neither is the case here. No one is having reliability issues here with data loss. It is extreme misdirection to focus the issue on eye pattern instead of what is produced by DACs.

I am looking to substantiate statements like this on your web site:

"If you are familiar with the variations in sound quality that come from different computer configurations, USB cables, and power supplies (no, "bits are bits" really does not apply when pursuing the audio summit), then you will immediately recognize the often dramatic effect that the REGEN can have on the the connection you feel with the music."

How would you connect the eye pattern on USB bus to "connection you feel with music?" None of us feel those pulses. We feel the analog signals coming out of the DAC through the rest of our system. You go on to say:

"Although the REGEN does not eliminate all sound quality differences between USB cables in a high-end system, it significantly reduces those differences. "

Is there some evidence of this? Any measurements you like to share?

I like to understand how we can demonstrate this to be true:

"From all the time we have spent listening to the REGEN in our systems we can say that it is exceeding our expectations. Often it sounds like you are listening to a different DAC! "

None of my measurements indicate I am listening to a different DAC. Do you have some that do that?

If the only advertising on your site was about the eye pattern, then you would have a case showing it such even though in our applications, data integrity is not a concern. But currently you talk about audible benefits with a technical explanation. We need to be able to connect these two somehow. I hope you appreciate that.

Thank you and regards,

--Alex Crespi
UpTone Audio LLC
Thanks again for joining us. I welcome any and all feedback you have on this topic.
 

Elberoth

Member Sponsor
Dec 16, 2012
1,858
3
38
Poland
#16
I just thought it would be interesting to quote John Westlake again on this topic:

The Analogue scope measurements of the USB direct and ReGen USB revealed the truly "horrid" nature of the USB traffic from a PC with heavy "CPU Process" jitter modulation - and I've been trying to think how to graphically capture it so I can post it here on PFM.

So I dug up a Digital scope - rearranged the lab to make space for it... I think its the first time I've really used it - I hate them, but it proved useful to capture the data.



Above is a colour graded eye pattern of the Direct PC USB connection to an XMOS based DAC (streaming 1KHz, 0dB FS) High speed USB.

Note the multitude of lines, but more importantly that the background is not black (Black means no data points fell in this area) - the background its dark blue because during the acquisition low frequency "Runt" waveforms where sampled in this area - these "Runt"waveforms occurred as the PC / operating system processed other applications / Operating system house keeping tasks etc. On the Analogue scope it was a horrific mess...



Now with the ReGen USB Data, notice not only the much cleaner waveform but far more importantly that there is very little "Runt" data - the background is Black. The USB Hub IC in the Regen has for the most part cleanly repackaged the data.

The colour grade data sample hit size is the same for both plots (20K Hits on the Red).


Those plots clearly show that the Regen gets its primary job done (reducing USB jitter which should lead to reduced PHY noise). What is more John basicly validated Regen's concept by saying:

Yes, I strongly believe that the USB Packet jitter cause "second order" effects in the MDAC (PSU, Ground plane and RF modulation).

I realise that the MDAC's USB device (the TAS1020B) not only decodes the USB, but its also the house keeping processor on the MDAC mainboard - so its directly connected to the ESS DAC (via the ESS's I2C MCU configuration port) allowing multiple RF coupling paths between the external USB Host device (PC / Computer) and the ESS DAC.


So what we see here is:

1. greatly reduced noise on power line
2. greatly reduced USB data jitter

Now the 1 million dollar question everyone is asking now is, how it is possible that greatly reduced jitter and noise, induced more distortion in the DAC. Either we didn't measure the right thing, or there is other variable, that Amir's tests didn't account for (like the SMPS).

I would very much like to see how this topic develops.
 

Elberoth

Member Sponsor
Dec 16, 2012
1,858
3
38
Poland
#17
Superdad's post just mysteriously appeared over my post - it was probably waiting for moderator's approval (1st post by Superdad) - chence double posting of the same graphs.
 
Apr 3, 2010
16,022
0
0
Seattle, WA
#18
I just thought it would be interesting to quote John Westlake again on this topic:

The Analogue scope measurements of the USB direct and ReGen USB revealed the truly "horrid" nature of the USB traffic from a PC with heavy "CPU Process" jitter modulation - and I've been trying to think how to graphically capture it so I can post it here on PFM.

So I dug up a Digital scope - rearranged the lab to make space for it... I think its the first time I've really used it - I hate them, but it proved useful to capture the data.



Above is a colour graded eye pattern of the Direct PC USB connection to an XMOS based DAC (streaming 1KHz, 0dB FS) High speed USB.

I came to post what he had written here and you saved me the typing :).

I must say, I am very surprised that a PC USB port would have such horrid eye output. Then I read the initial line that he is using some kind of "heavy "CPU Process" jitter modulation?" I don't know what the heck that is but this is not what I expect the eye pattern to be on a USB port with a short cable. I can't imagine the above eye pattern being from a USB compliant port.

There is unfortunately no USB police to speak of so it is entirely possible for him to have an implementation that is not well implemented.

Note that my laptop was simultaneously driving the DAC and my Audio Precision analyzer over USB. In typical server playback, assuming your music is not coming from USB, you would have a much quieter environment than mine. The CPU load in playing music is nothing, certainly lower than what I was doing in this measurement.

Anyway, even if we are going to talk about eye patterns, we better show a broader spectrum of USB ports than one unknown one here.
 

RBFC

WBF Founding Member & Super Moderator
Apr 20, 2010
5,121
0
36
Albuquerque, NM
www.fightingconcepts.com
#20
I approved his post a bit ago. There were posts directed at Alex personally, and the post did not violate forum rules.

Interestingly, Mark Waldrep of AIX records has already sent out his daily email complimenting Amir's post on this topic.

Lee