I just bought the Topping EX5 and got it today. here's my take.
A few month ago I emailed Topping's boss and JohnYang, asking for a device that is
1. combo unit with DAC XLR output and headphone out. For HP out, just use the L30 circuit.
2. low power. use switch power supply so it won't produce a lot of heat. sleep mode should not consume > 0.5W.
3. use single ES9038Q2M and achieve good jitter result
4. small. no desktop clutter, no cable management issue.
5. no MQA shit
6. no traditional ugly topping display.
7. has negative gain for IEM users.
8. <$400
and I can see EX5 implements part of them. Thanks Topping.
Analysis:
1. Topping device does not use a switch power supply that produces both +15V and -15V rails like their competitor SMSL does. So it makes it very hard to deliver -15V rail in a cheap way. On Premium devices they could use premium components such as larger inductors/discrete DC-DC converters, etc. However on this cost saving device they can't. So they just use discrete charge pump followed by a tps5430 to step down into -12.3V.
No, this is not a measurement error. On the PCB I clearly see the feedback resistors are 9.1k and 1k so 1.221*10.1 = -12.33V. So you have a +15V / -12.33V rail for all op amps and headphone amplifier.
I'm not saying they are doing it wrong. but you see who the product targets. In the end you get what you paid for.
2. There's no premium grade components. You won't find things like OPA1612. All op amps in the signal path are LM4562.
LM4562 is not a rail to rail op amp. so with a -12.33V negative rail it would deliver even lower output voltage.
Also, based on my measurement of XLR output and HP out, the power supply is a bit noisier than I expected.
3. The DAC chip is powered by, you guess it, an LM4562 op amp. No ES9311 (which is usually used as power supply chip in ESS designs) is used. The op amp also serve for bias input for LPF op amp. In a classical op amp based power supply, to get a stable voltage without ripple usually zener diode is used for voltage reference. Not here. Usually to do a linear power supply a BJT is used with the op amp to supply more current. Not here. so I'm quite surprised one piece of LM4562 has enough current to support two DAC chips + bias input in a stable way.
Later I saw a PCB photo of Topping D90SE. the 9038Pro is powered by two pieces of OPA1612. OK...
4. The signal path is interesting. works like the following:
View attachment 139157
1) DAC differential outputs to LPFs using LM4562.
2) differential outputs from the LPFs go into a LM4562 for summing
3) The LM4562 in stage 2 is used with a TPA6120A2 and form a nested feedback composite amplifier. The architecture is exactly the same as the Topping L30. It's just one chip for both L and R, while in L30 you have two buffers.
4) There's another LM4562 which gets the HP out and inverts it.
Why this is interesting? Well. For rear XLR output. it gets the XLR + phase signal from the class AB headphone amplifier discussed in 3). and the - phase signal is from the inverting signal of a regular op amp 4).
Also more interesting is there's no separate op amp for the RCA output. they share the same headphone amp output.
So for people use XLR output, the + and - signal is getting from chips for different purposes. Also for people who connect XLR and RCA simultaneously, remember they are powered by the same buffer.
In reality maybe it's fine --- After all TPA6120A2 has a lot of power, and can power RCA + XLR simultaneously without any issues. The capacitive load issues (if your speaker/preamp/amp has AC coupled input) may also be fine thanks to the zobel network used in the L30 architecture. But just something you need to be aware of.
5. The last disappointment is from the Jitter performance. In my measurement jitter is quite poor. I wonder why topping was not applying their D10B into this product.
Overall Topping delivers a product in a good price tag. They cut a few corners and able to sell it for $350 but can still get plenty of profit.
I would recommend them further improve this product by
1. using better power supply that can output +/- rails directly. this won't cost more.
2. reduce DAC chip to one. Save more cost.
3. kill MQA. Why would one need it?
4. solve the jitter issue.
I haven't do stability test though. Will report back after a month of using it.