SMSL M500 Teardown & ESS ES9038Pro DAC Thermal Analysis

[mjwin]

This DAC needs a peltier on the DAC chip. A possible tweak that really works.

As I was reading Amir's report, I was thinking *exactly* the same thing. (Great minds think alike

What a great aftermarket product this would be: The peltier element would be bonded to the chip, and would form part of a water cooled assembly. The transparent coolant pipes (illuminated by blue LED) would pass through a user-created "thermal wormhole" in the rear panel and thence to the master control box. This would be an exquisitly styled unit (milled from a solid ingot of Al/Mg alloy, of course). A nice OLED display and temperature control knob would complete the assembly. Users would be encouraged to dial in the temperture to suit their mood or according to the type of music to which they were listening, or... The possibilities are endless.

It could be the first ASR brand product & then Amir will be able to afford that fancy speaker test gear he has his eye on.

But, given that marketing is everything, what shall we call it?

 

[Tks]

Typical sign of some cancellation going on. (At least) two mechanisms with different tempcos. The point where they neutralize best can be very process-dependant with large chip-to-chip variations and probably ESS didn't implement a individual trimming to get the minimum at a typical expected operation temperature.

Interesting how no one using the chips was able to catch on to this possibility. Leave it up to a bunch of hobbyists to do the work corporations seemingly ever won't. (The same is true especially in the computing sphere with every single damn month it seems a new vulnerability, a broken firmware, a inadequate upgrade, all within the hardware realm is discovered). Look at this shit for example.. Do these OEM's simply take whats given to them and slap on whatever on their products hoping everything works? Why is it users are always the ones finding these sorts of horrendous issues. Is this simply a case of neglegence, and also not wanting to rock the boat much in the first place as a client?

 

[beefkabob]

Jeff "Piece of Crap" Bezos spent the early days of Amazon purchasing books from publishers in large enough quantities to get a discount then returning most of them after he sold a few books. Feel free to fuck over Amazon as much as you want.

 

[Tks]

I could not help but imagine @MC_RME reading this and smiling over all the nonsense they do not have to deal with, thanks to using AKM DACs.

His company is beyond this sort of stuff for the most part. The PCB they use has pretty good thermal transfer with the rest of the case anywho. So thermal considerations were taken in the design of the RME ADI 2 product line to some extent. (though it would be interesting to see nonetheless how AKM chips now stack up in light of this ESS ordeal).

Granted this whole situation probably has no audible impacts when we're dealing with these levels of distortions and such. So for all intents and purposes, this is just us here being anal about engineering design, though such massive swings in performance are not insignificant enough to not have ESS made aware of it, last thing we need is obliviousness and further potential regressions in their unreleased chips still in development.

This has quite serious impacts on sales for instance when products are competing neck and neck, and someone looks up any published specs/measurements only to find a potentially lower performing device on paper, but in reality has a much higher potential (but because temps were taken into account, no one is wiser to the fact).

For instance, you have people asking on the forum sometimes "Hey is this better than X" lots of replies are pretty sensible in explaining to people that some of these devices for the sake of audibility are the same, and you should focus on features. But then there are folks that will nudge someone to a better performer slightly simply due to better measured metrics.

I just sympathize with any company that thinks their product measures better than the results here. Though the only cases of that are very slight differences, or differences raised by reps of companies who don't even have measurements themselves for us to even compare and validate their claims against.

 

[mkawa]

what happens if you use component cooler on the retail sample? does it hit the same 117db sinad as the review?

with the temperature envelope unknown between samples, it's difficult to even size a heatsink properly.. this is ridiculous on ESS's part.

 

[NTomokawa]

peltier

The issue with Peltier cooling/below-ambient cooling is the potential for condensation.

This is also why there are very few Peltier CPU coolers out there.

 

[amirm]

Folks, this is a few dbs change at the end of the rainbow. While it is important with respect to bragging rights (i.e. who is on top of the SINAD graph), it is not something material you want to mess with to get better fidelity. There is a reason I put DACs in just four quadrant, not 200!

 

[NTomokawa]

this is a few dbs change at the end of the rainbow

That may be, but the interesting part here is how two identical products display opposite temperure-dependent behavior.

 

[Killingbeans]

I wonder if the AKM chips are easier to work with, or that they have their own faults/troubles.

I'm thinking the same thing. Right now my go-to choice for designing a total overkill DAC would be the AK4499. I'd love to see/do some thermal analysis on that chip. One thing that might be an advantage is that it has 2½x the surface area (10x10 vs. 16x16mm), but who knows whether the design is horribly inefficient.

 

[BeepPeep_61]

If heatsink is needed cab anyone recommend a good one to stick on top of 9038Pro?

In my OPPO Sonica, the ES9038 chip has a ceramic heat sink as standard, but to be more foresight, I added a thermal cut-to-size piece of heatsink, recovered from a used mother board, with a double-sided adhesive pad.
I enclose a photograph

 

[Blumlein 88]

As I was reading Amir's report, I was thinking *exactly* the same thing. (Great minds think alike

What a great aftermarket product this would be: The peltier element would be bonded to the chip, and would form part of a water cooled assembly. The transparent coolant pipes (illuminated by blue LED) would pass through a user-created "thermal wormhole" in the rear panel and thence to the master control box. This would be an exquisitly styled unit (milled from a solid ingot of Al/Mg alloy, of course). A nice OLED display and temperature control knob would complete the assembly. Users would be encouraged to dial in the temperture to suit their mood or according to the type of music to which they were listening, or... The possibilities are endless.

It could be the first ASR brand product & then Amir will be able to afford that fancy speaker test gear he has his eye on.

But, given that marketing is everything, what shall we call it?

Cooler Sound

or maybe

Sounds Cooler

 

[NTomokawa]

Cooler Sound

or maybe

Sounds Cooler

Now watch as "audiophiles" blast the product as making their DACs sound "cold, dry and analytical"!

 

[VeerK]

Just got my M500 in the mail today after a month, it’s a cold day so I guess I’m going to have to bake it in the sun for about 15 minutes before I listen to it lol. Fun fun fun, looking forward to seeing more 4499 DACs come out next year

 

[1audio]

I have worked with AKM for a number of years and found them and their chips quite straightforward. No hidden special stuff (Except their DSP chips which you will need some help with). My ESS experiece resembles this with the myriad of ways it can misbehave.
The AKM demo boards are a great starting point but can be improved upon with straightforward engineering. And you can apply your own DSP if you so desire. The AK4499 is their first current output DAC and will get you an improvement over the voltage output DAC's. Getting that improvement (claimed -114 dB THD+N to -126 dB THD+N) down your audio chain will be a herculean task let alone getting it out of any known transducer.

 

[Tks]

Folks, this is a few dbs change at the end of the rainbow. While it is important with respect to bragging rights (i.e. who is on top of the SINAD graph), it is not something material you want to mess with to get better fidelity. There is a reason I put DACs in just four quadrant, not 200!

One problem bossman, lets say you measure a device that makes it to the lower end of the blue graph, while another makes it to green. But due to behaviors like this that we've seen, I can imagine a scenario where one device ought have been in the green, while the one in the green ought have been in blue.

While most folks here understand none of this really matters too much at the end of the day in terms of listening to some music. You have to understand some folks either really place value on engineering (in the same way another might put value on aesthetics), and other people simply have NO idea what any of your measurements mean, and simply like to narrow their choices down to "just buy blue graph stuff", budgets permitting of course.

 

[Doodski]

In my OPPO Sonica, the ES9038 chip has a ceramic heat sink as standard, but to be more foresight, I added a thermal cut-to-size piece of heatsink, recovered from a used mother board, with a double-sided adhesive pad.
I enclose a photograph View attachment 42357

I always use thermal epoxy. It works great.

 

[mansr]

This is what you get when you have no technical specification sheets for your product. Maybe they exist in some form for clients, but idk.. Someone ought have leaked the details by now.

I got some ESS "data sheets" from Charley Hansen once. Absolutely terrible. I assume it was the best he had access to.

In my experience, the data sheet quality ranking goes something like this:

1. Linear Technology (now ADI)
2. Analog Devices (ADI)
3. Texas Instruments
4. Maxim, NXP, ON Semi, etc
5. Single-market outfits like AKM
426. Secretive companies like ESS

The top 3 are all very good. At level 4 the quality varies greatly between parts, but there's always enough information to get the job done. At level 5 enough information is available to make the device work, but important details may be missing or hard to find. Below that, you're essentially guessing.

 

[NTomokawa]

426. Secretive companies like ESS

Perhaps ESS provides more detailed datasheets to their biggest clients only? The plebeian companies have to make do with far worse datasheets?

I wouldn't be surprised if this were the case.

Meanwhile I did enjoy Analog Devices' datasheets, especially when I was trying to mess around with their AD8662/AD8672 chips.

 

[Tks]

I got some ESS "data sheets" from Charley Hansen once. Absolutely terrible. I assume it was the best he had access to.

In my experience, the data sheet quality ranking goes something like this:

1. Linear Technology (now ADI)
2. Analog Devices (ADI)
3. Texas Instruments
4. Maxim, NXP, ON Semi, etc
5. Single-market outfits like AKM
426. Secretive companies like ESS

The top 3 are all very good. At level 4 the quality varies greatly between parts, but there's always enough information to get the job done. At level 5 enough information is available to make the device work, but important details may be missing or hard to find. Below that, you're essentially guessing.

One thing I don't get about the choice with ESS, aren't they more expensive to use as well compared to most others? Aside from appeals to audiphiles wanting "US" chips or something of the sort, why would a client even consider them in the first place?

 

[NTomokawa]

why would a client even consider them in the first place?

Because SABREDAC SABREDAC SABREDAC. I think an iBasso MP3 player was the first mobile music player to incorporate an ESS chip and that helped immensely with the Sabre's popularity.

Seriously though, ESS is half built on top of hype, but at least their chips deliver great performance, aside from the (as seen here) fickle implementation.