As the reviews of Topping's AKM based E70 Velvet start to flood the internet, I have seen a few errors here and there. One of the more well known boards, which I will not name, gets something that is important very, very wrong. At the very least, I can set the record straight for my readers.
I thought about naming names, but, I decided intrepid readers can figure it out.
(By the way, if you are one of my readers, I express great gratitude to you. I never thought anyone would really read this new audio blog/review site, but the many thousands of you that have in just a few short months, well, you have overwhelmed me with gratitude for your time and attention.)
The incorrect information comes from a reseller product representative, not so much from the author. The salesman claims by using fixed output as opposed to variable output, the Topping E70V will use bypass mode for Native DSD.
This is incorrect. The E70V never uses the DSD bypass mode, set in the logic as DSDD= '1'. The E70V always, regardless of mode or volume, is set as DSDD='0'. See the diagram below for visual clarification.
There is only ONE DSD filter available in the E70V, and it has a 19khz fc at DSD64. This filter feeds the volume control DSP as a multibit 'intermediary' and is converted to lower bit Delta Sigma via the modulator.
The second DSD filter available with the AKM chipset is a 39khz fc filter at DSD64, and is only available with the 'bypass mode'. None of this is available on the Topping E70V, and this is confirmed by Topping representatives at the source. Not from a sales guy.
Here is the AK4191 diagram that helps spell it out...
Again, there is no alternative filter choice. Whether the Topping E70V is measured in Fixed output or Variable output, there is a single DSD filter with no ability to change it. This option IS available to any DAC that uses the AK4191+AK4499, but Topping does not provide this option in the E70V. Measurements confirm the 19khz fc, with full stopband around 30khz at DSD64. FC (frequency cutoff) will double as the DSD rate doubles. (38khz at DSD128, 76KHZ at DSD256)
It seems some people are confused by my wording and I apologize. To put it as simply as possible, AKM offers two possible DSD paths. A 'direct' path, using DSDD bit "1", that bypasses all Digital Signal Processing and Modulation. The DSD signal is filtered at the output stage with a 'analog' FIR filter, built of shift registers for the delay line and bitswitches for the taps, followed by a analog 'accumulator'.
There is a 'DSP' path that allows for Volume Control, using DSDD bit "0". In this path, the 1-bit DSD signal is first greeted by a digital FIR filter, the output of which strips the majority of the high-frequency noise shaping. The output of any such filter is multi-bit, which allows for the next stage, which is volume control. THEN the multi-bit signal, which has a longer wordlength than can be processed by the output stage, is converted to a lower bit depth at the Delta Sigma Modulator (the same one used for PCM, yes).
Running the Topping E70V in fixed output mode, disabling the volume control, does NOT switch the onboard DSDD logic from bit '0' to bit '1". The Topping E70 never makes use of the direct DSD mode. It is always in logic mode '0', regardless. Fixed output mode simply locks the volume control at 100 percent. The output path of Digital FIR and Delta Sigma Modulator is still followed, emulating what DACs such as ESS do. Why? Because this is more than just a technique to apply volume control to DSD. It can provide for a very precise conversion of DSD, with lower noise levels and taking full advantage of the advanced Delta Sigma output stage. Others, like me, still prefer to keep DSD conversion 'simple'. Keep it at 1 bit all the way until the final FIR filter converts it to analog.
Which is why I prefer DACs made by iFi, which is 1 bit native all the way to the analog FIR output stage, RME, which uses AKM like the Topping E70V, but has the user selectable option to use native DSD Bypass mode. Other DACs off the top of my head that are true native DSD... usually anything made with Burr-Brown DACs, T+A uses either Burr-Brown or their own proprietary output filter for true Native DSD, there seem to be some DIY and Chi-Fi options out there, and I know I am forgetting a bunch.
No ESS DACS don't make that cut. There is no 'bypass' option. All DSD is re-modulated in their Hyperstream converters. Indeed, the sound excellent, but they are NOT Native DSD converters in the strict sense.
In a couple weeks I will be a proud owner of a DSD only DAC built around the Signalyst discrete native DAC-less converter. I cannot wait to give it a spin!!
Also, I am re-evaluating some early scores in reviews that reflect measured performance. The emergence of superior evaluation technology and software skills require some rating changes, to be fair and consistent. Apologies for this, however I feel it is the best way forward for our fledgling publication, as we grow our identity and solidify our standards.
Ahhhh. There is nothing more satisfying than fixing, or rather, bypassing a bug in the system. You may have noted the issue in the white noise filter frequency response graphs. Something about the internal tone generator in the ARTA software creates a 'hiccup' at the Nyquist limit. It does not affect the accuracy of the measurements, at least those 'outside' the hiccup. It is almost as if the white noise started imaging above the original sample rate Nyquist limit as one looks at the oversampled filter plot. The filter quickly brings things back to normal, but that little anomaly shouldn't be there. There should be a smoothly rolling off plot.
Bypassing the ARTA internal tone generator with external tones fixed the issue. Along with fixing that issue, I changed the color patterns quite drastically, which, in my opinion, is much more pleasing to see. The white background with red and blue data plots simply looks better than the default green over black. At least here at Euphonic Review. Opinions vary of course.
See below the results. The GTO filter finally looks like it should. It resembles the MQA filter. No surprise since iFi collaborated with Meridian on this filter. It is personally my filter of choice, and Thorsten Loesch (when he was still actively the iFi engineer) explains it well in the tech notes, which you can access here by clicking. (GTO FILTER PDF)
Below are the iFi NEO DAC fw 3.33 filter profiles. Once again, the only anomaly here is the filter called 'Minimum'. One would likely interpret that as meaning 'Minimum Phase'. It is not a Minimum Phase filter, however. The measurements show it to be a low tap, short linear phase filter.
I hope you like the new filter graph format. Without any further delay, here are what the iFi NEO DAC 3.33 firmware filters look like.
Although the title might evoke images of the Matrix films and its blend of Eastern religious esotericism with gnostic undertones, this is far more simple than that. Thank God. (or should that be Brahma? anyway.....)
After measuring the iFi iDSD NEO v1.1 (original version that came to market), I decided to turn back and re-measure the ZEN DAC V2, which I keep around as a baseline for evaluating all DAC/headamps under $1000. Yes, it sounds that good. Since my original measurements of the ZEN, I have upgraded my lab considerably, and can see deeper and more accurately into what is happening in these hyper sensitive pieces of kit.
What I found was very interesting, although it should have come with no surprises, as already once in this post I have mentioned the veracity of the ZEN V2, and how it punches well above its weight class. The ZEN and the NEO measurements are for all intents and purposes identical.
The only major difference I can see, if it is major, is the falloff of the harmonic distortion is a bit smoother in the NEO. Both, however, produce more even order distortion than odd.
It comes as little surprise then that I think both sound identical. Perhaps one does a certain passage of music better than the other, then the tables are turned when the passage is of a different constitution.
Here for your viewing is the 1khz distortion FFT for each.
As I had mentioned in a previous blog entry, iFi released a version of the NEO iDSD called the 'Performance Edition'. The reality of this release is it is the same hardware as the first NEO iDSD to hit market at end of 2020. The differences are a newer firmware and a couple hardware add-ons.
Eventually iFi released the 3.33 firmware to all NEO iDSD owners. The major feature added by this upgrade is a choice of digital filters, similar to the Gryphon. (they are both xDSD platform based, and much of this also overlaps with the work done a few years earlier with the iDSD Pro.
I caught a small bug in the firmware right off the top. When GTO filter is selected, the info readout shows 'DXD 352' when, for example playing back 44.1khz and its multiples. The other filters simply show the actual sample rate. There is nothing wrong here, just a peculiar anomaly. I suspect that every filter option except for GTO are the onboard Burr-Brown chip filters. (the Bit Perfect setting is not really a filter.. there is no oversampling) The GTO filter, however, is part of the FPGA/XMOS digital processing, and behaves just like it does on the iDSD PRO. When playing anything other than Bit Perfect mode on the iDSD Pro, the display shows the oversampled rate, which is always DXD x2 (705.6khz or 768khz). Coming back around to the NEO iDSD, the small bug is once GTO is selected, the indicator display will always show the oversampled rate rather than the input rate, regardless of which filter is selected. It isn't a big deal when you understand what is going on. Everything is functioning properly, but can understandably be confusing.
Hi-Fi News review of the Gryphon pointed out several of the filters were misidentified, and that the minimum phase filter was not a minimum phase filter at all.
I have confirmed in my testing of the 3.33 firmware that all filters EXCEPT minimum phase are true to their name. As Hi-Fi News uncovered with the Gryphon, the Minimum Phase filter on the NEO iDSD is actually a short linear phase filter, the standard filter only being slightly longer.
The impulse response for each filter is below.
As far as overall performance differences? Yes, there were some measurements that showed the fw3.33 version to be slightly more jitter prone, but the results were not repeatable and are of no great a significance to change anything.
Tests are from 3.18 firmware. Next set of testing will be done with the latest firmware installed, which will add several additional filters.
Somewhere here down in the Southern USA where the Euphonic Review lab stands, where my hobby (and primary career) is tickling the ivories, well, these days more like stomping out some Rockabilly Jerry Lee style (Rest in Peace dear Friend), I have been chasing a mystery only audiophile nerds care to chase down.
Turns out I was not the only one.. wow!!! What a consternation was stirred up online regarding the iFi NEO iDSD, a new Performance edition, and various firmware updates.
First, came the admission there is no difference in the original NEO iDSD hardware released in late 2020, and the hardware in the 'new' Performance edition. The differences in hardware consisted only of two add on accessories.. a USB iPurifier3, and a SPDIF iPurifier2. Turns out I already own the USB iPurifier. It is a solid little add on, and I found it to offer subjective improvements with certain USB DACs. The SPDIF version, unreviewed by me personally, I expect it to be and have read it be an excellent product, especially when it comes to eliminating SPDIF jitter, as it is galvanically isolated, and creates a pure, clean SPDIF signal for the receiver PLL, buffers and clocks, making their job with recovering a clean clock much easier.
Of course, this didn't necessarily sit well with many people, that may have bought a Performance version expecting more for their money. And that is understandable. Things were not helped by a chart outlining the new 'upgrades' from the original version. These were later dismissed as, put into my own words, accidental misprints and miscommunication.
I don't mind just a little speculation here at ER, and part of me wonders if part of this goes back to something the keen eye may have observed at the time; that the first release of the NEO to the public was version '1.1'. Perhaps there was one that came before that never made it to production and was quickly upgraded? Possibly. Again, I am just speculating.
I find all the more curious myself though, is the insistence that the NEO iDSD PCM oversampling filter was locked with no other choice, and this filter locked in by firmware was the in-house developed (along with help from MQA) 'GTO' filter. The GTO filter is a short (32 tap) minimum phase FIR filter that you can read all about by clicking here.
But the first major reviews with measurements showed that the filter on offer with the NEO iDSD was not the GTO filter. It isn't even minimum phase. HiFi News caught this but didn't challenge, seeming to assume their measured results were how it was supposed to behave. What they had measured was a moderately long linear phase filter with its expected stopband behavior. Now, there have been many firmware updates, so it seems, in the life cycle of this product, so things could have changed. My particular example of the hardware is version 1.1, with firmware 3.18. And just like HiFi News, my example has a slightly long linear phase FIR filter, with fairly strong stopband rejection. Definitely NOT the advertised GTO filter.
To further muddy the waters, there seems to be multiple past firmware updates, that even allow the end user to choose the single, default filter in their NEO iDSD. PLUS, there is a firmware update NEWER than my version that opens up a host of iFi filter options, ala the Gryphon or iDSD Pro.
I like iFi products. But I also like being a fair reviewer to all sides involved. So to be fair, this was a confusing mess that made a night of scrolling through the wailing and gnashing of teeth on Head-Fi time I wish I had back.
I have already finished most of my benchmarking on the NEO iDSD with the 3.18 firmware. Now, I am off to upgrade to the latest and see what I can find... hopefully what I expect to find.. no trouble down here in the southland.
Y'all come back now, ya hear? Please?
As always THANK YOU for reading and supporting Euphonic Review.com
Well oh well oh well, oh my oh my o my oh my! (Said in my best Matthew Mcconaughey Texas drawl. My Scots-Irish Appalachian drawl makes for a pitiful voice impression).
This SMSL is simply an ENIGMA!
There are things about it not so good, things about it a bit quirky, some a LOT quirky, but, I can already tell you it's the best sounding DAC in the lab under $1000.00. Better sound than either of the impressive Topping E70 twin sisters that have since been 'rehomed' --- separated, but both went into a most loving and caring home, and are surely busy at work providing great musical joy.
Usually I can put anything under $500.00 up against the iFi ZEN DAC V2 and immediately size it up quickly. The ZEN is truly a, well, master guru of the crowded under $500 club. And it does it for 200 bucks, plus it throws in a decent headphone amp practically for free in my opinion.
Still no headamp in this SMSL D300, but man, the SOUND. We are EUPHONIC Review and this thing truly nails the euphonic part like nothing I have heard close to this price. BEAUTIFUL sound. So much so I had to break out the heavy gear to truly place its sound and make an honest assessment.
My reference head-fi system begins, as many a system, with a Roon front end. Roon pushes audio data into the Roon Ready iFi ZEN STREAM, running in its Roon exclusive mode. Many people at this point would simply connect via ethernet to the Stream, guaranteeing no audio drops that one might very reasonably expect over Wifi when streaming say, high bandwidth DSD256 files. I prefer the opposite strategy and go with Wifi, because that is essentially an isolated system that shares no physical connections or grounding points that can bring who knows what unwanted 'heebie jeebies'.
Lucky for me, I have never had a single drop out of ANY file of any kind, regardless of bandwidth including DXD and DSD256.
For even MORE overkill, there is an Intona USB High Speed 2.0 isolator between the Stream and my reference DAC, the iFi iDSD PRO. Yes, the iDSD PRO has internal galvanic isolation, but I can show on the spectrum analyzer how some spurious anomalies completely disappear when I apply my overkill solution.
Oh, and my headphones of choice over here in Referlandia are FOCAL Clear MG Pro. (But for long listening comparisons while reviewing, I will still default back to the venerable and ever competitive wonder of a product Sennheiser HD650 cans.)
I say all this, just to get to the point that this SMSL is pushing Referlandia just a bit. No, it isn't close enough that instant regret sinks in my soul over how much money could have been saved. But, it was necessary to establish a new baseline of sorts for where we are going when it comes time to rate the sound quality of this little kit with the alternative DAC chip from ROHM.
Tonight? Direct battle between my second reference system, the RME ADI-2 PRO FS R Black Edition, with the SMSL. Some of you may recall in the Topping E70V review, The RME's excellent DAC section was not worlds better than the Topping. It sounded better than the Topping, yes, but not enough so.... that one might pause and think hard about it, if all you want is a DAC and have no need for the ADC functions.
Cannot wait to dive into more of this little SMSL DAC that can! Assuming it keeps working. One of those little quirks is obvious cost cutting in chassis design. We will see as it is torture tested how it continues to fare!
What's next for the test bench?? A slight deviation from plan, and I will tell you why. I happened to be perusing Amazon looking at various source components that I find interesting, and I came across the SMSL D300, which uses a Japanese chipset I have not yet heard of.. the ROHM BD34301EKV.
It caught my eye, because it appears to convert DSD via native analog FIR, and ONLY that way. No 'DSD processor' or 'DSD direct' to choose from. The block diagram seems to say it is ONLY DSD Direct. I also like what the data sheet has to say about the DSD FIR filters. It is very straightforward.
It is also a segment DAC. It doesn't say exactly what those segments are, but one of them must be a Delta Sigma segment. Which means the other is some kind of true PCM conversion, most likely. The diagram gives no indication of any but a Delta Sigma modulator for PCM, but since it does say its a segment DAC/filter then I would say its very likely all a thermometer code DAC with PCM'S msb's converted with a number of equally weighted switches that do indeed bypass the DS-modulator. Just a guess, though. If I am correct, then its very much like the Burr-Brown DACs developed in Japan before TI took ownership.
It arrives tomorrow, and testing will begin after after a 48 hour break-in over the weekend.
Can't wait to see how it compares to AKM and ESS based products at this $400 price point!
One of the backbones of my audio lab is the RME ADI-2 PRO FS R Black Edition. Quite a title, eh? I suppose lots of numbers and letters appended to the end is appropriate for a top of the line ADC/DAC.
The RME can't quite dig deep into the noise floor of the latest Chi-fi DACs that are being custom built, by their own admission in their 'mission' statement, to perform on the bleeding edge of the very latest Audio Precision test analyzers.
For my own bleeding edge dive into the world of the noise floor, (kind of like sinking into Challenger Deep), I decided to fight fire with fire. I purchased the E1DA ADC and APU (Audio Processing Unit with pre-analyzer notch to assist in measurements of THD and SINAD) from the Silicon Valley of the East, the Shenzhen 'Special Economic Zone', from whenceforth Topping, SMSL, Gustard, etc all come.
On an aside, Shenzhen (once called the less than sexy "Sham Chun"), looks like something straight out of Star Trek. The 'home base' on Terra in the 24th century would look like this in my dreams.
But now, it isn't Topping under the knife. I finished all those measurements and am planning on having those reviews (E70 and E70V) up by midnight Eastern Time this day. A day late, but better late and somewhat informative than on time and a bunch of crap. Under the knife is the RME ADI-2. (We will just call it that from now on, and we all will know the reference is to RME's state of the art top model)
By any reasonable standard, the RME ADI-2 STILL is an impeccable piece of engineering, with measurements that completely blow past the human ear's resolution. That is why it is also marketed as a home lab measuring device, since the vast majority of hi-fi kit even to this day will not reach the specifications of the RME.
Here is what I found in my non-literal look under the hood. Although we could do that too, someday.
A special ADC/DAC or ADDA. One that will always be here. One of two products I will never sell. This RME, and my iFi iDSD PRO. (my iDSD has a iffy OLED display, so maybe someday I will just buy another one, but it will be a DAC I have until it dies.)
Pictured above is every reviewer's dream piece. A very select few are lucky, lucky to have them. These are typically reserved for large audio corporations who need the best of the best for prototyping the latest product on the drawing board. The most wealthy magazines have one in the office, and maybe even an older spare in someone's home. Maybe someone even invented their own proprietary measurement interface and software and sells it as well as uses it to let their audience in on what is going on underneath the hood of the kit of the day under review.
Surely, Surely such results can't be replicated in the amateur enthusiasts lab? Well, that would be correct. They cannot. There are certain things that just require this level of perfection. Alas, all seems lost. But, wait a moment! NO! It isn't!
For all the things we can't do as well, there are quite a few things that we can do as professional reviewers stuck in amateur labs! Some recent products out of China that take direct advantage of the latest ESS chipset's harmonic distortion cancellation, along with creative minds and somewhat eclectic collections of software can provide results of which any reviewer should be proud!
Now, That day IS coming when I own something like the Audio Precision, although I think I will stick with the new Dscope M1 from Spectral (formerly Prism).
I am hoping within the next year or so to have such a fine state-of-the-art Audio Analyzer in the lab.
In the meantime, I MORE than make do with E1DA products, that are a BOON for budget audio analysis!!
What kind of boon you say???? Well, lets just say MOST average hi-fi equipment, no problem. With the combination of the E1DA COSMOS ADC and APU (Audio Processing Unit), most average equipment is no problem. Which brings me to the problem. The latest two DACs in the lab made by Topping. The E70 and E70V sister DACs, the only difference being one's heart is the ES9028, while the other is the very latest AKM AK4191+AK4499EX split chip DAC. I say 'only difference', but internally, that difference is pretty major.
The big problem for EuphonicReview is how to get accurate measurements on these two DACs. It wasn't easy. Every day was a learning experience, but, finally, right before my publication deadline of February 10, everything clicked into place.
Again, there are lots of measurements that I cannot match-up with an Audio Precision, but, you might be surprised how close in the end some of the most important measurements can come.
I give a big shoutout to Amir at audiosciencereview.com. We are going to stick to the Topping E70 DAC with the ES9028 chipset, because I know he recently reviewed it with his Audio Precision analyzer. Below is a comparison of his results with the AP, and mine with the E1DA Cosmos.
So, less than 500 dollars vs I am guessing 20,000 dollars. Not bad! Did you expect the results from the Cosmos to be so very off the mark? I did. Because the Topping E70 DAC has measurables that push the limits of the Audio Precision. Pushes them hard. And it pushes my lab to the very, very limit. But they are NOT that far off the mark. Impressively in the neighborhood IMO. Anything more, and I don't believe I can provide useful commentary. The highest SINAD DAC I have yet measured happens to be the sister DAC of the E70, the Topping E70V. 'V' as in 'Velvet'. I measured its SINAD with same setup at 121.3db, or 0.000086% THD+N.
At the moment, yes, I am satisfied with the marks I can reach. But I know the day is coming when my marks will fall more and more behind as technology keeps advancing. So eventually an upgrade is in order. But, for NOW, I can say with great confidence I can provide useful measurements in my reviews for anyone considering the product I am evaluating. Most especially when compared INTERNALLY to other products reviewed here at www.euphonicreview.com.