[Modding] Topping DX3 Pro

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Jump to Latest
F

finneybear

Member
Joined 2003
1audio said:
I don't understand what you are trying to say. A log plot is hard to interpret for things like distortion. The multitone plot can be useful except many of the HD and IM products fall on other source tones making it all harder to interpret.

I would only get meaning if it starts with a reference plot showing what you would see without the DUT. Digital in and analog out make that more difficult. You also need to know what the noise floor looks like to be sure you aren't attributing artifacts to the DUT that are in the noise floor of the analyzer.

Also without knowledge of the sample rate, FFT resolution, the windowing etc. its really hard to make comparisons. The FFT of a 1 KHz tone on a modern opamp will usually be limited by even the best analyzer. All the chip vendors use a technique increasing the noise gain to get meaningful distortion plots. The plots you showed were definitely not opamp limitations. Virtually all analyzers have their best performance at 1 KHz and 3V in. Compromises in the design of analyzers will degrade above and below that frequency. Its not necessarily a reflection on the opamp if you get different results at other frequencies. I would expect harmonics of -130 dB or less for a closed loop gain of 10 for any premium opamp designed for this application. The high harmonics in the FFT are from some other source (the DAC?).
Click to expand...

There is nothing complicated here. As far as you have consistent setup across the board, you can still figure out the differences among different multi-tone tests. This is exactly where ASR shows its value. All tests are done by the same person based on the same config. You can always go to FFTs at different frequencies for details later.

As for the choice of 1KHz test, it's a chicken-egg scenario. In the past, for tape decks, 400Hz was the one to use. Later it was moved to 1KHz for CDs to show a bit more on the LPF's impact. Now for some high speed OPs such as OPA1612, they work the best above 1KHz. People start to get the impression that 1KHz is the sweet spot for OPs.
 
F

finneybear

Member
Joined 2003
redjr said:
I agree with you on the AKM sound signature. To me it's simply smooth and silky. My primary listening is done using a RaspPi HAT DAC based on an older AKM4490EQ chip. I love the sound of it - at least in my system. My only 'upgrade' is I'm using an external linear PSU.

I also bought the Topping DX3 Pro for the latest dual-mono chip design, so I'm watching/following this thread with interest. While I like the ESS chips, I keep coming back around to the AKM44xx series DACs.

The Khadas Tone Board has some promise as well, and I'm waiting for their host SBC board for testing and possibly building an all-in-one, network based streamer using roon.
Click to expand...

I have found there's a big jump on SQ from AK4490 to AK4493. Do have a big concern on the coming AK4499 though. The new I/V stage may change everything.

For the past few months I have checked most AK4497 and AK4493 DAC designs available on the market. So far DX3 Pro is one of the very few which are both simple yet can demonstrate how good AK4493/4497 really is after mod.

Dont look down on AK4493 as the new Estoric top machine has four 4493s inside.

Just got a decent linear power supply for DX3 Pro. the new PSU has removed the very last bit of roughness from its sound. Now it's such a sweet, mellow yet dynamic little monster!
 
F

finneybear

Member
Joined 2003
sealand1 said:
I try to consolidate your mod within this spreadsheet, it seems the lpf part is not consistent between two of your post, also I cannot find some parts to change in bom, could you please help us to correct it? Thanks.
Click to expand...

Very sorry for the late reply as I have been super busy with work.

The photo shots were taken at different mod stages. The close-up one is the one to follow.

The BOM looks good to me. Please let me know if you have any questions with it.
 
C

chris719

Member
Joined 2004
finneybear said:
There is nothing complicated here. As far as you have consistent setup across the board, you can still figure out the differences among different multi-tone tests. This is exactly where ASR shows its value. All tests are done by the same person based on the same config. You can always go to FFTs at different frequencies for details later.

As for the choice of 1KHz test, it's a chicken-egg scenario. In the past, for tape decks, 400Hz was the one to use. Later it was moved to 1KHz for CDs to show a bit more on the LPF's impact. Now for some high speed OPs such as OPA1612, they work the best above 1KHz. People start to get the impression that 1KHz is the sweet spot for OPs.
Click to expand...

Where do you come up with this stuff? The datasheet for the part shows you are wrong. Distortion is markedly worse at 10 kHz than 1 kHz, as you would expect from the open loop gain plot. Please, explain with facts not anecdotes.


You’re right that there is nothing complicated here. I’m happy to change my stance when provided explanations that make sense.


Are you aware that Samuel Groner did very detailed measurements of OPA1611 (rebranded OPA211). His measurements, which go beyond the datasheet, disprove every single statement you have made about it in this thread.
 
Last edited:
C

chris719

Member
Joined 2004
finneybear said:
I have found there's a big jump on SQ from AK4490 to AK4493. Do have a big concern on the coming AK4499 though. The new I/V stage may change everything.
Click to expand...

AK4493 is on a new process it shares with AK4497 (and presumably 4499) according to what I saw on the website somewhere. The AK4499 is going to have the same stupidly high output current like ES9038. Other than the digital section, this DAC is going to be more like the Sabre than the 4493 it seems.
 
C

chinook9

Member
Joined 2013
Paulkouhan said:
I get the BOM, the hex 2mm key, a solder station for smd.
I will practice first on an old graphic card !
Click to expand...

An old graphics card could work but you might find it easier to purchase an SMD practice kit off ebay. They are very cheap and include various sizes of smd and the PCB to solder and desolder on.

You may not need it, but if money is no issue it might be nice to have a cheap hot air SMD rework station. They are cheap and I have found many uses for mine. 858D works well.
 
D

david99

Member
Joined 2019
Hi All, Thank you for sharing all you knowledge here.

@finneybear, is your mod - especially LPF part safe also if I leave original OPA1612 in place? Also I noticed that you are using only two lines instead of four as in original setup between each AKM4493 and OP what is the reason?

Thanks.

David
 
Terry Demol

Terry Demol

Member
Joined 2002
finneybear said:
CS43198 is the first high end all in one design DAC chip. It is built mainly for portable audio devices/cell phones, etc. It has everything inside, even headphone detection. It has PLL based jitter removal, configurable digital filter, and only needs 1.8V power supply, etc.

The charge pump allows it to output 5V signal, and since its output is H-calss based, the LPF is given hence no need for external LPF. A CS43198 with output buffer, that's all what you need.

Is the H class output bad? Not really. Remember that DSD itself is PWM as well. If DSD could sound good, H class can sound good, too, as far as the switching frequency is high enough. The beauty of this architecture is that, the H class output will remove the need of external LPF hence there will be no active I/V, no active LPF involved. No active LPF means no NFB, the whole signal chain now has no NFB. The only NFB happens in the digital filter inside CS43198, in the digital domain. In the analog domain, it's a simple straight path, no NFB. Definitely this will give a very unique sound to CS43198, and I am not surprised it can sound good now. What a brilliant design idea!
Click to expand...

At least you are an optimist, if only life were so simple :)

I think you might want to re study what is actually inside the 43198, read, a whole bunch of Cmos opamps, and reconsider a lot of that reply.
 
Last edited:
C

chris719

Member
Joined 2004
Terry Demol said:
HaHa - that's a pretty funny post by any measure - at least you are an optimist, (sadly needed around here), if only life were so simple :)

I think you might want to re study what is actually inside the 43198, read, a whole bunch of Cmos opamps, and reconsider a lot of that reply.
Click to expand...

I do think the CS43198 is a nice part, but you're right. There is approximately zero chance it doesn't have NFB in the internal output. Drives 600 ohms with -115 dB THD but doesn't use NFB... riiight.

I'm not sure he knows what Class H is, either. It's also completely optional and can run in normal Class A/B with a register setting.
 
Terry Demol

Terry Demol

Member
Joined 2002
chris719 said:
I do think the CS43198 is a nice part, but you're right. There is approximately zero chance it doesn't have NFB in the internal output. Drives 600 ohms with -115 dB THD but doesn't use NFB... riiight.

I'm not sure he knows what Class H is, either. It's also completely optional and can run in normal Class A/B with a register setting.
Click to expand...

You have to be fast around here, I edited that (Haha bit) out 10 sec after posted- so you don't need to reiterate it.

Yes, agreed on all counts. Maybe the real story here is how good they have made such an integrated device with obviously some constraints, perform.

In mono, fully balanced mode, this part would be pretty darn good.

T
 
C

chris719

Member
Joined 2004
Terry Demol said:
You have to be fast around here, I edited that (Haha bit) out 10 sec after posted- so you don't need to reiterate it.

Yes, agreed on all counts. Maybe the real story here is how good they have made such an integrated device with obviously some constraints, perform.

In mono, fully balanced mode, this part would be pretty darn good.

T
Click to expand...

Sorry, didn't realize it was edited after. I can't edit mine now :(.

Guerilla said:
Someone ought to find out;)
Click to expand...

I might if I can get some free time over the next few months - looking unlikely though.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Top Bottom