Review and Measurements of Topping DX3Pro DAC and Headphone Amp

[finneybear]

Thank you all for your findings, opinions, modings opinions it is really very interesting reading for me and I think for rest of the forum as well. It would be nice once situation will settle down to publish let's say some tutorial for moding of DX3 Pro let's say maybe for most important mods which will influence listening experience most.

I will say, for most people. the easiest path will be:

Swap the OP, you will hear drastic change right away. Sure, you will have to pick up the right OP.

Limit the listening to PCM768 or DSD512, another big jump on sound quality.

Replace all stock caps with good low ESR caps. Those SMD caps may look intimidating but in reality, it is super easy to remove them. Do this to improve the system stability.

If you are daring enough, replace the 8 DAC output coupling caps with audio grade ones.

Finally, if you use the headphone amp a lot, replace the 8 caps around the amp with good sounding ones such as Elna Silmic II.

 

[777]

Based on what friends learned from AKM4497, OPA1624 and MUSES 8920 are good matching for it.
Teac uses MUSES 8820 in their AKM4497 based streamer.

OPA2134 is very similar to OPA1624 when the latter is a newer design.

Maybe OPA1642, not OPA1624. I was put it instead of OPA1612 and the result is awful, much worse than OPA1612.

 

[y845133y]

The problem is those stock caps are not very good. Their capacitance value drops fast when frequency goes up or temp arises, etc.
At certain point, the voltage would drop due to lack of current reserve then you will run into hang-up or all kinds of problems.
It also depends on various scenarios - run temperature, whether you are in bad luck to have a very bad cap for the MCU, etc.

Increasing the cap value means it will take a while to charge them up during the power on process. The MCU may see the system's voltage is not up yet hence directly goes to the stand by mode. In this case, you need a PS which can charge those bigger caps faster.

I am thinking about just keeping the same cap value. As far as I am using solid polymer caps at 100uf/25V, they can still remove those HF noises very effectively. The 25V rating will have high ripple current, then at the same time, 100uf is easier to drive for the power supply.

Main concern now is 4493 datasheet is calling for 470uf for the analog power. Not sure whether it is OK for the analog section of 4493 to power up late. Otherwise I will have to reduce it to 220uf to see how it goes.

Will switch to linear power supply, LT1085 based, 3A output, just to be safe.

I check the AKM4497's datasheet. The same power (VREF analog power) is recommended to put a 2200uf (Electrolytic Capacitor).

 

[Pavel]

Main concern now is 4493 datasheet is calling for 470uf for the analog power.

Replaced with 680uF https://www.arrow.com/en/products/t520y687m006ate0187280/kemet-corporation
8 output capacitors also replaced with aluminum polymer https://www.arrow.com/en/products/eef-lr0j680r/panasonic
Generally I could not refuse output capacitors. The AK4493 has a non-standardized differential DC offset at the outputs of a few millivolts.

 

[Hyldig]

Not much details, but here is what i got, in the best chinese english

First message:
A happy day, first of all, I am very sorry to bring you trouble, because at present TPPING DX3 PRO we received some buyers to suggest us some problems with the product, we are currently testing all the products in our warehouse, because of the Chinese New Year holiday Just after the end, we received a batch of new products from the factory.
In order to ensure the quality of the products, and solve the customer's problems before the product has been reflected, you can use the forum to understand some of the problems that may arise in order to better serve our buyers and friends, we decided to extend the products you purchased.
Some shipments, of course, we will send you products within the time limit set by AliExpress, please rest assured. We will use these time to conduct a 48-hour broadcast test on the product's possible amp and crash, and use the 110v and 220v voltage tests separately. We will arrange for the test without problems, but we will not be able to ensure that during this test. There is no problem with the product after the test, and our test is only to reduce some of the possible situations.

Second message:
We are very sorry to bring you trouble. For the series of problems that TOPPING DX3 PRO has appeared in the forum, TOPPING officially began to re-complete and improve the product again.
In order to ensure the quality of the products and better serve the customers, TOPPING official requirements, all current DX3 PRO orders will be tested at 110v and 220v for a long time before leaving the factory. It will take some time during this period.

Just got this update from seller
"We just got the good news. Topping has solved this problem. It is expected to start producing new products next month, so please don't worry, I hope I can help you, thank you."

 

[david99]

Just got this update from seller
"We just got the good news. Topping has solved this problem. It is expected to start producing new products next month, so please don't worry, I hope I can help you, thank you."

@TP-Wong

Could you please confirm if this is really correct information?

Thanks.

 

[richpjr]

Just got this update from seller
"We just got the good news. Topping has solved this problem. It is expected to start producing new products next month, so please don't worry, I hope I can help you, thank you."

Great to hear. I have been considering one of these for use at work, but was hesitant after reading this thread.

 

[vincent_emerald]

Just got this update from seller
"We just got the good news. Topping has solved this problem. It is expected to start producing new products next month, so please don't worry, I hope I can help you, thank you."

So what happens to the affected units, would they ship an updated board to all these users ?? Hope they also identify the batches that are affected by this and communicate to them for a replacement. Wish!

 

[gypsygib]

Is anyone else getting noise in their headphones when set on high gain and something is plugged into the RCAs? I get an intermittent faint buzz anytime something is plugged into my RCA ports. Tried different cords and devices. Same intermittent buzz whether I have speakers or an amp plugged in. Noise goes away as soon as the RCA cords are unplugged.

If someone would be so kind to test I would appreciate it.

 

[Cidious]

Replaced with 680uF https://www.arrow.com/en/products/t520y687m006ate0187280/kemet-corporation
8 output capacitors also replaced with aluminum polymer https://www.arrow.com/en/products/eef-lr0j680r/panasonic
Generally I could not refuse output capacitors. The AK4493 has a non-standardized differential DC offset at the outputs of a few millivolts.

Pavel would you maybe be willing to make a list with a high res photo of all the mods that you did so far ?

Seems the result is pretty good. How does it sound other than the measurements ?

 

[Cidious]

Maybe OPA1642, not OPA1624. I was put it instead of OPA1612 and the result is awful, much worse than OPA1612.

I rolled Muses02 in it. It's a bit less clinical sounding than opa1612.

I also tried 1642 which wasn't a success.

I did replace all the TL072Cs in my active speakers with opa1642 where they perform superb. It's a good opamp. But maybe not for the lpf of this dac

 

[splkn]

Replaced with 680uF https://www.arrow.com/en/products/t520y687m006ate0187280/kemet-corporation
8 output capacitors also replaced with aluminum polymer https://www.arrow.com/en/products/eef-lr0j680r/panasonic
Generally I could not refuse output capacitors. The AK4493 has a non-standardized differential DC offset at the outputs of a few millivolts.

Initially you wanted to install these capacitors https://www.arrow.com/en/products/t520a107m006ate045/kemet-corporation
Why did you change the decision?

 

[tired_guru]

When you change opamps from certain application to something new, always check out if there are no oscillations. It can have devastating impact on many areas once it's there and behaving unstable.

 

[Pavel]

Initially you wanted to install these capacitors https://www.arrow.com/en/products/t520a107m006ate045/kemet-corporation
Why did you change the decision?

It was a compromise solution. There is too little space for installing aluminum capacitors.
But later I found a solution. These aluminum polymer capacitors have a thickness of only 1mm and have an additional 3 output. This design allowed to install them on the PCB.

 

[Pavel]

Pavel would you maybe be willing to make a list with a high res photo of all the mods that you did so far ?
Seems the result is pretty good. How does it sound other than the measurements ?

I have not finished yet. It makes no sense to publish.

 

[david99]

It was a compromise solution. There is too little space for installing aluminum capacitors.
But later I found a solution. These aluminum polymer capacitors have a thickness of only 1mm and have an additional 3 output. This design allowed to install them on the PCB.

View attachment 22319

Hi Pavel,

Thanks for sharing, I see you replaced also capacitors around OPA1612 and smaller ones around AKM chips what are those? Were there any other changes made from capacitors point of view? Thanks again for sharing.

David

 

[bravomail]

want to add to the whole capacitors discussion. From what I see power is coming from DC power supply 15V. My first suggestion to "modders" - get a higher amperage supply (3,4,5A, whichever is higher). This will resolve any power starvation (current drop) issues (again, I'm not saying there are any). 2nd suggestion for more serious, more extreme modders - put a capacitor of big capacity (really big, the size of your fist, 50000mF big) between your + and - lines of DC power line going from PSU to your device. This will both solve power starvation and reduce 50Hz/60Hz humm. Playing with capacitors inside device is not recommended, unless there is a design defect (like in SMSL M3 headphone path). "Rolling" opamps is for noobs. Measurements (and blind AB comparos) showed time and time that there is no discernible difference between opamps. Opamps will amplify.

 

[Pavel]

Thanks for sharing, I see you replaced also capacitors around OPA1612 and smaller ones around AKM chips what are those? Were there any other changes made from capacitors point of view?

I do not think that my experience will be useful to most users. I bought a DX3 not to use it. I needed a prototyping board with AK4493 for exploring the new chip. Over 50$ was spent on additional parts and replacement. And apparently this is not the last expenditure...
I see no reason for most DX3 owners to spend money on this not expensive device.
DX3 fully justifies its factory price. And if Topping corrects errors - this device will be very popular.

 

[finneybear]

I check the AKM4497's datasheet. The same power (VREF analog power) is recommended to put a 2200uf (Electrolytic Capacitor).

It's 470uf. One for each channel.

 

[finneybear]

Replaced with 680uF https://www.arrow.com/en/products/t520y687m006ate0187280/kemet-corporation
8 output capacitors also replaced with aluminum polymer https://www.arrow.com/en/products/eef-lr0j680r/panasonic
Generally I could not refuse output capacitors. The AK4493 has a non-standardized differential DC offset at the outputs of a few millivolts.

For the four analog power caps for 4493, I am thinking of having a 220uf low ESR cap paralleled with a 100uf solid polymer. Low ESR cap takes care of the low frequencies. Solid polymer will remove HF noises. The space is good enough to put through hole parts there.

I cant see the simple output sum-up of 4493 will do much good in DX3. For true mono-mono mode with XLR outputs, it will give you extra 2 or 3db improvement yet this is not the case with DX3.

So I plan just to use one set of outputs for each 4493. This way I can remove the coupling caps. For the second set of outputs, now I can have enough space to use Silmic or Muse caps for coupling. This way, I can add the second set of outputs back as an option.