Review and Measurements of Topping DX3Pro DAC and Headphone Amp

[ChuckT]

LOL The transformer can easily handle 10-15mA.

DC isolation/line transformers are widely used in telecommunication industry. As mentioned above, tube amp uses transformers and the DC bias can be in the range of a few hundred volts.

So how about I declare you as the winner? Will this make you happy now? LOL

Each of the ak4493 out L/R + and L/R- sums after the 4 coupling cap. I had tried transformer output for the ak4495 and it improves the sound quite a lot. But in mono mode with 4 output per channel, I am not sure how to connect to transformer.

 

[simonchretien]

I assume your issue is phantom shutdowns? Personally I had 3 since I bought my unit on 11.11, last one 3 days ago. It's quite bizarre, not a big deal overall, but I would like this thing not to die on me, and those shutdowns as you say don't seem normal.
Just got my mean well 1.66A psu today (GS25A15-P1J) but that apparently does nothing anyway, oh well.

Yeah I wouldnt mind the shutdowns but too many dead units so far, since its too expensive to return I'm stuck with it. Like you said, I can only wish it wont die. I was expecting Topping was aware of the problems and working on something, but they dont.

 

[777]

Ha ha, I did exactly the same. I replace all 8 tan caps with the green Nichicon Muse ES. And I removed the 2 tan cap and 2 ceramic caps around opa1612 and replace with a small value Nichicon KZ. It brought a more organic sound to the dac where before it has a relatively clinical sound.

Yes, it's all about the low frequency distortion. Here the tantalum caps are very bad. Bipolar electrolytic caps are very good for that job. Cyril Bateman is the Bible of the capacitors measurements.

 

[noobie]

So many experiments being run to improve the sound quality and potentially reliability. Might be an interesting side business if someone feels up for it to mod these units... I have the DX3 Pro, DX7S, and the D50, so I would be game to send in my DX3 Pro...

Of course I think that it would be a good idea if whoever wants to get into the side business first send a unit to Amir for testing to see how it measures against his original unit (or random sample if someone else want to send a virgin unit).

 

[splkn]

Each of the ak4493 out L/R + and L/R- sums after the 4 coupling cap. I had tried transformer output for the ak4495 and it improves the sound quite a lot. But in mono mode with 4 output per channel, I am not sure how to connect to transformer.

Not certainly in that way. They are summarized in pairs. Look at the diagram and understand where to connect. The right part of the scheme Topping have done differently.

 

[MacDac]

No no, I want to keep it decently hot. Hot machines sound better. This is what Nelson Pass said.

Sure, especially the black hotties do

 

[MacDac]

After one hour of music on RCA-pre output, with pre at -25dB, both AK4493 had the temperature around 35C-40C. Ambient temperature is 22C. May be wrong but AK4493 does not seem to have temperature problems.

I doubt that overheating is an issue of these DA chips per se, as there are many other devices equipped with (similar) AKM chips on the market, that work OK. But to find out if we're dealing with overheating here we need measurements of defective DAC(s), since healthy ones obviously won't show any abnormalities (caused either by application of low quality parts, manufacturing errors or poor circuit design or whatever other reason).

 

[777]

The PCB tracks between voltage regulators and OPA1612 had 0.15 ohm.

 

[splkn]

The PCB tracks between voltage regulators and OPA1612 had 0.15 ohm.

Can you show in the photo where these regulators?

 

[777]

Do you have any idea what is the output impedance of TPS54331 ?

 

[777]

Can you show in the photo where these regulators?

There are near headphones output.

 

[noel_fs]

Mostly using my HD6XX with it, sound is pretty good.
So even considering my negative rambling about the unit I like the unit

Got a reply from Topping, their solution is to send it back to the seller, oh yeah I'll surely pay 45$ for air shipping plus 40$ of restocking fee lol.

They say they dont have any problematic unit so they cannot test or do anything about it yet...

where did you buy it? you can dispute in your bank if you think they are kinda abusing you.

 

[splkn]

Do you have any idea what is the output impedance of TPS54331 ?

«Integrated 80-mΩ High-Side MOSFET Supports up to 3-A Continuous Output Current» From datashit. Maybe this will help you knowing the output voltage.

 

[finneybear]

OK, some exciting news to share with fellow DX3 owners!

When I looked at the 1KHz FFT test done by Amir, those little spikes around 1KHz were a big concern to me. Then people started to report edgy highs, shrills, all of those are pointing to LPF problem. Indeed, when I just got the DX3, I ran into the same problem with highs.

So to verify this, I tried different sampling rates – 44.1K, bad; 96K, still bad; 192K, highs improved; 352K, the edginess almost gone; then 705K, wow, the top is smooth and mellow now. Punchy sound, superb image!

Next is DSD. DSD64 (SACD is), no edginess yet the sound is a bit muddy and flat. DSD128, 256, 512, things are improving along the way. It’s at DSD512 this has totally changed! Wow! The sound stage is now wide opened! Beautiful top! Everything comes to alive! Each instrument is so three dimensional! You can even tell the size of the instrument. There’s a clear shape around it. Just so 3D! Stable, pin point image (looks like the clocks are not bad). When compared to ES9038 Pro running at DSD512, 9038 has a tad more etchiness on the top, more analytical. DX3 has more coherence, is more organic, and roundish. It can do better on vocals. The difference is subtle yet it’s there. Now I can see why AKM called this the velvet sound!

Yeah, still no mid-bass but I am not too worried about it now as this can be fixed. The sound still needs refinement here and there yet I am confident to say, after simple mods, this thing at DSD512 can easily beat most DACs under $2K!


A few extra points to mention:

. Obviously the cut off frequency of the LPF is set too high. Will have to open the machine to check the parameters to confirm this.

. Ranking of sound quality - DSD512 > PCM705 > DSD256 >> PCM352 > DSD128

. Even not all DSD512 sounds the same. The ranking – XiSRC > HQPlayer >> foobar2000

. I do have problem to play DSD512 from XiSRC under foobar2000. Still need to figure out what has gone wrong.

. When running foobar2000 for upsampling, make sure it’s set to high CPU priority mode otherwise you will hear clicks/hiccups.

. So far I am using a plain USB cable. The PC has 8 USB devices connected to it.

. If your computer can not handle DSD512, at least try PCM705K. Otherwise DSD256 is the one to recommend. Yes, DX3 is really meant for DSD.

. This thing is a beast!!! An unbelievable bargain at $200!!! 

 

[finneybear]

It doesn't have heat problems. It runs cool enough. Might be other issues though. I sometimes get into trouble when going to another track or skipping within a track.. hear the click inside and need to wait a few seconds and restart the song.. hardware.. or software. hard to say.

For the rest I have to admit that my LKS 9018 is better in some areas of refinement. sounding more smooth overall and more holographic but the potential of the AKM having more detail and imaging is definitely there.. I think with some mods the AK4493 chips will be able to sound very decently.

And there is a huge gap between LKS 9018 and LKS 9038. There is another big jump when the 9038 is doing DSD512. Now the DX3 is getting close!!!

 

[finneybear]

I don't understand why you want ultra-low ESR in the signal path.

For coupling, you want low leakage current. Those KO caps are not bad either.
Next thing to look at is the noise and non-linearity.

 

[finneybear]

Each of the ak4493 out L/R + and L/R- sums after the 4 coupling cap. I had tried transformer output for the ak4495 and it improves the sound quite a lot. But in mono mode with 4 output per channel, I am not sure how to connect to transformer.

Just use one set of the outputs. Need to verify how the sum up is done in DX3 yet most likely they are just tied together. Pretty much useless. For voltage output DACs, the sum up needs active device to double the amplitude.
It will be better off to have each 4493 drive one XLR output.

 

[finneybear]

Not certainly in that way. They are summarized in pairs. Look at the diagram and understand where to connect. The right part of the scheme Topping have done differently.

From the photo I assume DX3 is doing pretty much the same thing. Instead of using two OPs, the two summed up outputs are fed to one OP.

 

[pejakovic1]

OK, some exciting news to share with fellow DX3 owners!

When I looked at the 1KHz FFT test done by Amir, those little spikes around 1KHz were a big concern to me. Then people started to report edgy highs, shrills, all of those are pointing to LPF problem. Indeed, when I just got the DX3, I ran into the same problem with highs.

So to verify this, I tried different sampling rates – 44.1K, bad; 96K, still bad; 192K, highs improved; 352K, the edginess almost gone; then 705K, wow, the top is smooth and mellow now. Punchy sound, superb image!

Next is DSD. DSD64 (SACD is), no edginess yet the sound is a bit muddy and flat. DSD128, 256, 512, things are improving along the way. It’s at DSD512 this has totally changed! Wow! The sound stage is now wide opened! Beautiful top! Everything comes to alive! Each instrument is so three dimensional! You can even tell the size of the instrument. There’s a clear shape around it. Just so 3D! Stable, pin point image (looks like the clocks are not bad). When compared to ES9038 Pro running at DSD512, 9038 has a tad more etchiness on the top, more analytical. DX3 has more coherence, is more organic, and roundish. It can do better on vocals. The difference is subtle yet it’s there. Now I can see why AKM called this the velvet sound!

Yeah, still no mid-bass but I am not too worried about it now as this can be fixed. The sound still needs refinement here and there yet I am confident to say, after simple mods, this thing at DSD512 can easily beat most DACs under $2K!


A few extra points to mention:

. Obviously the cut off frequency of the LPF is set too high. Will have to open the machine to check the parameters to confirm this.

. Ranking of sound quality - DSD512 > PCM705 > DSD256 >> PCM352 > DSD128

. Even not all DSD512 sounds the same. The ranking – XiSRC > HQPlayer >> foobar2000

. I do have problem to play DSD512 from XiSRC under foobar2000. Still need to figure out what has gone wrong.

. When running foobar2000 for upsampling, make sure it’s set to high CPU priority mode otherwise you will hear clicks/hiccups.

. So far I am using a plain USB cable. The PC has 8 USB devices connected to it.

. If your computer can not handle DSD512, at least try PCM705K. Otherwise DSD256 is the one to recommend. Yes, DX3 is really meant for DSD.

. This thing is a beast!!! An unbelievable bargain at $200!!! 

How would I upsample my FLAC's to DSD512? Let's say I use MusicBee program.

 

[gypsygib]

For those who have problem with the device, can you post the product number (only the first 4 digit is fine, mine is 1812 which means December 2018 batch)? It's on the side of the device with a barcode and DX3 PRO - BLACK or SILVER label. Maybe we can see which batch is problematic.
I got mine from local dealers so return is trivial, fortunately mine doesn't have any problem yet aside from weird interaction with linux (which exist with the D50 anyway).

Mine is 1812 as well

Received it on January 11th, no issues so far (hopefully never), it's been on playing sound from either headphone or RCA about 8 hours a day, first three days it was on about 12 hrs a day with cheap headphones to test for any weird voltage/issues. Measured zero on both left and right channels and the stock PS measured 15.1 volts, which I'm using. Although my meter is pretty cheap so I don't know how accurate it is.

Only issue is that my remote is dead and it came cracked and looked like it had dropped hard somewhere as there are deep scratches on the back (maybe customs damaged it while removing the battery). Finally got a reply from Topping after over a week and two emails, they said contact the seller. Wasn't impressed with that CS.