strange SMSL M500 high 3rd harmonic

[Massimo]

Really bad, my su9 arrived yesterday. I might send it back and get the Gustard X16 if I figure out how to get the infrared codes for Gustards remote control for integration in my universal remote control app.

The Gustard X16 is not without issue and has a problematic high DC offset which could torch Class D amps like Hypex and Purifi.

 

[Massimo]

This kind of measurements should be standard from now on. It is obvious that SMSL has poor QA.

Poor QA is not an exclusively SMSL issue. My Soncoz LAQXD1 cooked itself and nothing but disturbing noise was audible. Many on this forum reported the same fault. Topping has many reported issues on this forum. My Matrix Element could not maintain a reliable WiFi connection. My Khadas Tone 2 Pro firmware upgrade was a 3 hour ordeal and the unit was not operational for a time once upgraded. The Gustard X16 has an issue with high DC offset. The only unit I have owned, and there have been many, that has proven to be bullet reliable in every operational sense is a Lumin player. I cursed at the high price being asked but given my sorry experiences with the above mention brands feel it was money well spent.

 

[Nango]

This has been a topic for a bit, I believe. The brands are very aware of which measurements will be performed by the reviewers, so they can optimize their products for those, even though that may mean neglecting other areas.

It may be necessary to adjust the testing protocols at this point. Wouldn't it be extremely ironic if all the subjectivists who said "test with music" (aka under normal use conditions) would end up being right?

I told ya!!!!

 

[Veri]

The Gustard X16 is not without issue and has a problematic high DC offset which could torch Class D amps like Hypex and Purifi.

Really? Can I read about this anywhere? Seems like a pretty big flaw!

 

[Massimo]

Really? Can I read about this anywhere? Seems like a pretty big flaw!

Discussion on this issue starts from page 123 in Gustard X16 review thread.

 

[mkawa]

i'll take a "maybe possibly audible if we run more tests" over a dc offset that'll cook most things anyday. i would be hesitant to buy an SU-9 over a d30pro (which i did buy over the SU-9, conveniently). of course there are a few AKM options left. the M400 and D90 are still for sale, but expensive. there's also the E30 and of course the tempotec options (best deals in hifi!!). the 9038pro honestly has had SO MANY PROBLEMS in all products that maybe it's just time to retire it. i won't be returning my m500 (3 years later is whaaa) but i'll probably continue on my path to selling in favor of the j2 which just stacks better with my girlfriend's archel 2.5

i really don't care if this affects my sabaj D5 it's followed by an ADAU1701 at the moment in the chain, so it's overkill regardless. i'll focus on getting dirac into that chain instead.

tldr; don't freak out. seriously.

 

[liu]

the 9038pro honestly has had SO MANY PROBLEMS in all products that maybe it's just time to retire it.

Well, actually the 9038 series performs extremely well if designed right. much, much better than AKM series. In addition, it has a lot of features AKM lacks --- ASRC/DPLL de-jitter, IIR/FIR customization, THD compensation, etc, just to name a few. es9038q2m for instance, perform much better than ak4497, at 1/5 of the cost, and offer a lot more features.

Because there are so many register/configuration/features to learn, unexperienced developers are easy to mess things up. Chinese manufactures are poor in R&D capability, so you'll see a lot of failed/problematic products in the market. But it's not 9038Pro's fault.

RME ADI-2 DAC exposed almost all of the feature set of AK4490/4493 to end users.
You can't even imagine how to achieve that with ESS chips.
Right now Chinese manufacturers may only explored 10% of the features of ESS chip, and many of them are implemented wrong.
To do the above thing correctly requires someone who is good at C programming. Those who are capable of programming work for high tech industry, not niche audiophile industry. Chinese manufacturers just don't have those talented people to achieve good things.

 

[JohnYang1997]

Well, actually the 9038 series performs extremely well if designed right. much, much better than AKM series. In addition, it has a lot of features AKM lacks --- ASRC/DPLL de-jitter, IIR/FIR customization, THD compensation, etc, just to name a few. es9038q2m for instance, perform much better than ak4497, at 1/5 of the cost, and offer a lot more features.

Because there are so many register/configuration/features to learn, unexperienced developers are easy to mess things up. Chinese manufactures are poor in R&D capability, so you'll see a lot of failed/problematic products in the market. But it's not 9038Pro's fault.

RME ADI-2 DAC exposed almost all of the feature set of AK4490/4493 to end users.
You can't even imagine how to achieve that with ESS chips.
Right now Chinese manufacturers may only explored 10% of the features of ESS chip, and many of them are implemented wrong.
To do the above thing correctly requires someone who is good at C programming. Those who are capable of programming work for high tech industry, not niche audiophile industry. Chinese manufacturers just don't have those talented people to achieve good things.

Well. So right yet so wrong.

 

[liu]

Well. So right yet so wrong.

Well, why not just make one product that is feature rich and works well?
You don't need to reach the level of those pro products that do fancy DSP in FPGA.
Just make it reliable (that means, play music without any hiccups), and expose useful features a chip has.
For instance, I have quite a few early 1980 classical cds.
Can I get the de-emphasis feature that almost all chips (akm/ess/cirrus) have?
Look, I just need a product designed for human, the music listener, not for amir's leaderboard.

Oh, most Chinese manufacturers are really poor at C programming. You know what I am talking about.

 

[JSmith]

i'll take a "maybe possibly audible if we run more tests" over a dc offset that'll cook most things anyday.

Well not for me, I'd rather not have either...

requires someone who is good at C programming

It's certainly hard to find good C programmers now...

So right yet so wrong.

Are you suggesting SMSL do have the required skill to rectify this defect?



JSmith

 

[liu]

It's certainly hard to find good C programmers now...

There's nothing fancy in firmware dev.
XMOS provides all templates already.
One just need to do simple modifications.
Most of the above modifications are related to searching/changing strings (such as product name) and numbers (such as channel count), and a little bit bit-manipulation in setting registers.
Topping didn't know how to do the previous for years until I gave step-by-step guide to @JohnYang1997 ,
And it seems that SMSL doesn't know how to do the latter.
That is expected though --- experienced programmer who can do that prefer working in higher paid industry, on more challenging projects.

 

[Katji]

I think he meant that generalising can be problematic.

 

[JohnYang1997]

So many things to argue, I don't know where to start. Your perception is bent.

 

[Veri]

I think the "10% of ESS chip potential" statement is also pretty bizarre.

 

[JohnYang1997]

Well, actually the 9038 series performs extremely well if designed right. much, much better than AKM series. In addition, it has a lot of features AKM lacks --- ASRC/DPLL de-jitter, IIR/FIR customization, THD compensation, etc, just to name a few. es9038q2m for instance, perform much better than ak4497, at 1/5 of the cost, and offer a lot more features.

Because there are so many register/configuration/features to learn, unexperienced developers are easy to mess things up. Chinese manufactures are poor in R&D capability, so you'll see a lot of failed/problematic products in the market. But it's not 9038Pro's fault.

RME ADI-2 DAC exposed almost all of the feature set of AK4490/4493 to end users.
You can't even imagine how to achieve that with ESS chips.
Right now Chinese manufacturers may only explored 10% of the features of ESS chip, and many of them are implemented wrong.
To do the above thing correctly requires someone who is good at C programming. Those who are capable of programming work for high tech industry, not niche audiophile industry. Chinese manufacturers just don't have those talented people to achieve good things.

Ak4497 is roughly the same performance as es9038q. 4497 can have better SNR than 9038q at full level.

Everyone in China can do C programming. You don't have any idea what's being done in these products. Do you really think that CPLD, MCU firmware upgrade just came out like magic?

You just focusing on those little things. It's like CGI, you only see it when it's done badly. You never notice when the things work.

MQA requires special filter in ESS fir. So anyone who has MQA+ESS chip have done it. The issue is probably actually related to MQA even.

X16 use asynchronous mode for 384khz and lower and synchronous mode for higher than 384khz. Clock division also affects distortion harmonics and noise. 9068 also has crosstalk cancellation.

On the RME thing, if you just give an Adi2dac to most of our customers, they don't even know how to use it regardless of having good documentation. Heck many don't even read manual.

 

[JohnYang1997]

Well, why not just make one product that is feature rich and works well?
You don't need to reach the level of those pro products that do fancy DSP in FPGA.
Just make it reliable (that means, play music without any hiccups), and expose useful features a chip has.
For instance, I have quite a few early 1980 classical cds.
Can I get the de-emphasis feature that almost all chips (akm/ess/cirrus) have?
Look, I just need a product design for human, the music listener, not for amir's leaderboard.

Oh, most Chinese manufacturers are really poor at C programming. You know what I am talking about.

Using FPGA for DSP is not very smart.

Many "professional" grade equipments have issues. Recently M audio air 192|4, audient id14 mkii. One has output cliping at 2V and it's clipping before the volume control. One has two input channel with polarity inverted to each other and time delay. And time delay changes with samping rate. RME adi2pro over heats, one channel got noisy in the low frequency after powering up 30mins, and hiccups many times when reboot.

 

[JohnYang1997]

There's nothing fancy in firmware dev.
XMOS provides all templates already.
One just need to do simple modifications.
Most of the above modifications are related to searching/changing strings (such as product name) and numbers (such as channel count), and a little bit bit-manipulation in setting registers.
Topping didn't know how to do the previous for years until I gave step-by-step guide to @JohnYang1997 ,
And it seems that SMSL doesn't know how to do the latter.
That is expected though --- experienced programmer who can do that prefer working in higher paid industry, on more challenging projects.

That's just poor documentation on XMOS part. What about upgrading firmware? There are many issues that needed to avoid in order for the thing to work properly. Remember the M500 lock into version 1.05 after two updates? We only know it's related to the size of the flash but no documentation explains it clearly. Same with encryption. Poor documentation. We are just so busy working on other stuff that no one really care about a space after the name.

 

[JohnYang1997]

Well not for me, I'd rather not have either...

It's certainly hard to find good C programmers now...

Are you suggesting SMSL do have the required skill to rectify this defect?



JSmith

I'm not sure what rectify means here. They can certainly fix it if they want to. There are a lot of things they can achieve if they put their mind to it.

 

[liu]

So many things to argue, I don't know where to start. Your perception is bent.

start with providing firmware updates to your products then.
Simple things:
- remove all dx3pro relay clicking. You don't need a relay for ak4493. the hiccups are caused by your software bug.
- I already taught you how to fix the ending extra space. release the firmware to all of your past products released in the past 5 years.
- adding the deemphasis feature is trivial. just do that.
Otherwise your company is just releasing one after another useless gadget that is supposed to be thrown away in a few months.

There are many issues that needed to avoid in order for the thing to work properly. Remember the M500 lock into version 1.05 after two updates? We only know it's related to the size of the flash but no documentation explains it clearly. Same with encryption. Poor documentation.

This is how mediocre incapable programmers complain about documentation and example code.

The string issue: I didn't have your code but I knew exactly where you went wrong. this means something.
Of course, I am a professional. You are not. That's understandable.
What I said in previous post means exactly that --- software professionals work on more important projects. Why would someone capable of creating the next uber or facebook works in the niche audiophile industry?

Nothing is bug free. Apple and Google are many, many magnitude bigger than xmos so they have tons of resources to pour employees to work on libraries, docs and sample code. Yet Apple's and Google's frameworks still have tons of bugs. Their documentations are poor. Take a look at Apple's CoreAudio's documentation, (the most important piece of audio technology in their operating system), for instance, you almost get no useful info on what a function does/ why things fail / how things work. Good developers know how to overcome those. If all mobile phone app developers work in the way you do, the internet industry cannot get anything meaningful done.

We are just so busy working on other stuff that no one really care about a space after the name.

Yes, that's why I won't choose your products, because I can't get long term support on any of your products.
You are busy with new products. I understand.
But you use that as an excuse not to support existing customers. That's something unacceptable to me.

You just focusing on those little things. It's like CGI, you only see it when it's done badly. You never notice when the things work.

I don't need anything fancy. Just need a device to play music flawlessly. that's it.
Imagine a pro device failing in the middle of an Abbado's concert or Biden's inauguration speech.

RME adi2pro over heats

I'm with you on this. This is a product only usable in places where climate is similar to Germany's.

 

[MarsianC#]

@JohnYang1997 you know about multiquote? This forums software is great! Mark what you want to reply to and make your life (and mine) a little better
Thanks @liu for keeping the forum clean

Nice to see a argument between our friends in China. Keep going, I'm listening