SMSL SP400 Review (headphone amp)

[Snoopy]

No, 50/60hz is just an example of something that could be cancelled.
Of course the SMSL SP400 itself has SMPS, but 50/60hz could comes externally from source or by EMI from other instruments.

Anyway, the main issue is that it is over processing. If you are balance the do it fully balance and if SE then SE.


I am not arguing with you, I agree that benefits are marginal, especially when my amplifier is SE by construction.

I still don’t see the issue there. You got a example of a dac with 50/60hz hum? Same for EMI actually interfering with anything that is not a tube amp placed next to a WiFi router?

(I’m just curious about it). If you really have 50/60hz hum on a class A/B amp you could try fixing the problem instead or maybe a DC blocker would be already enough

Never had a issue with hum or noise from my SP400. Even when I used it with my Turntable (also a SMPS)

 

[Music1969]

I want to open is SMSL put 470ohm series resistance for each XLR phase at the back panel. I really don't understand why putting so high resistance, even for safety reasons it is too much. I want to decrease this resistance to about 20ohm. I am using the SP400 as preamp

I measured output impedance < 1 ohm...

both headphone out and preamp out

Are you sure you will be decreasing by DIY, not increasing ?

 

[Guy.Cohen]

I do not have a hum problem, I just gave examples where balance connection could help.

As I said, I don't like the over conversion (Bal SE Bal).

I understand that SMSL wanted to offer bal input and bal output.
I don't think that small imbalance (because of balanced volume control and non exactly matched parts) between hot and cold could introduce much distortions later in the chain as it was mentioned.

I opened the SP400. Balance input is summed by buffer to SE, then entered to R2R volume.
SE input is routed directly to R2R volume. The inputs are blocked with electrolytics. The SE by bipolar and the XLR by polar!
After the volume control there is a conversion back to balance. The two conversions above are done with OPA1612.
After the conversion back to BAL, the signal is routed by 470ohm to 4 THX amplifiers to the headphones output. At the same place, right after the 470 ohm the signal is routed to XLR. The THX amplifiers are not used for back XLR.
The 470ohm too much for any reason one can think, safety and speed.

 

[Guy.Cohen]

I measured output impedance < 1 ohm...

both headphone out and preamp out

Are you sure you will be decreasing by DIY, not increasing ?

See what I wrote above.
You can put 1k ohm on hot or cold of back XLR and see reduction of output, you can calculate. It is for sure not 1ohm from the back panel.

 

[Roland68]

Mostly is reduction of common mode noise (including 50/60Hz). It might reduce even order distortions (but might not improve sound quality ).

Firstly, how is this supposed to work within a device like the SP400 without debalancing?
Secondly, due to the construction, one can assume that interference is evenly irradiated into both conductors in a cable. Due to the spatial distance in electronic circuits, this is no longer the case.

Anyway, what is disturbing me is the over processing. Bal > SE >Bal.
Each step might harm transparency, even if you don't see it too much in normal specs (usually, FFT, slew rate, BW).
Maybe it is better to stay all the way SE and not doing these conversions.

No, 50/60hz is just an example of something that could be cancelled.
Of course the SMSL SP400 itself has SMPS, but 50/60hz could comes externally from source or by EMI from other instruments.

Anyway, the main issue is that it is over processing. If you are balance the do it fully balance and if SE then SE.


I am not arguing with you, I agree that benefits are marginal, especially when my amplifier is SE by construction.

In reality, the problem only exists in your head. Do you realize how this often happens when recording in the recording studio? These people are not stupid.
I was once exactly where they are now and thought that a continuous symmetrical chain was the holy grail, also because that was what expensive manufacturers propagated. I was about to purchase some really, really expensive components. We had carried out several tests on the topic in a project in which several developers from the audio sector and 2 application developers from measurement device development were involved. For this purpose, 10 boards were created on which the signal was cleanly debalanced and then balanced again. 10 x cable connections xlr a 1m.
None of us were able to identify the difference in the listening test compared to a 10m XLR cable. The best result, not repeatable, was 4/6. And that despite the additional 20 XLR connectors. The difference to several good 10m RCA connections, however, was clear.
The subsequent measurements led to everyone involved stopping their developments in relation to symmetrical small signal processing.
Draw your own conclusions.

I understand the SP400 can drive headphones in a balance way (BTL) but we can ask also here for what reason.

Better channel separation, twice the power, halved rise time because each power stage only has to go half the way.
Isn't that enough?

 

[Music1969]

See what I wrote above.
You can put 1k ohm on hot or cold of back XLR and see reduction of output, you can calculate. It is for sure not 1ohm from the back panel.

Yes i know how to measure

what measurement did you get?

Also i know you said you dont use headphones but what measurement you get from headphones ?

 

[Guy.Cohen]

Yes i know how to measure

what measurement did you get?

Also i know you said you dont use headphones but what measurement you get from headphones ?

I didn't measure headphones out output impedance, that I believe close to zero. I measured the output impedance from the back XLR and it is about 470ohm per phase as appears on the PCB.

 

[Guy.Cohen]

Firstly, how is this supposed to work within a device like the SP400 without debalancing?
Secondly, due to the construction, one can assume that interference is evenly irradiated into both conductors in a cable. Due to the spatial distance in electronic circuits, this is no longer the case.



In reality, the problem only exists in your head. Do you realize how this often happens when recording in the recording studio? These people are not stupid.
I was once exactly where they are now and thought that a continuous symmetrical chain was the holy grail, also because that was what expensive manufacturers propagated. I was about to purchase some really, really expensive components. We had carried out several tests on the topic in a project in which several developers from the audio sector and 2 application developers from measurement device development were involved. For this purpose, 10 boards were created on which the signal was cleanly debalanced and then balanced again. 10 x cable connections xlr a 1m.
None of us were able to identify the difference in the listening test compared to a 10m XLR cable. The best result, not repeatable, was 4/6. And that despite the additional 20 XLR connectors. The difference to several good 10m RCA connections, however, was clear.
The subsequent measurements led to everyone involved stopping their developments in relation to symmetrical small signal processing.
Draw your own conclusions.


Better channel separation, twice the power, halved rise time because each power stage only has to go half the way.
Isn't that enough

I agree that the common mode cancellation is achieved at the point of the differential pair summing.
At the SMSL it is really at the input.
Then it could pick up something along the path and again could be cancelled at the second differential pair (like power amplifier input).

Listening tests of course are subjective, but you were a group. So it has a meaning.

For the BTL connection at headphones out. Power and rise time could be achieved the same anyway, even with SE connection.
Another amplifier is more complicated then putting 4 times powerful.
For crosstalk this is something that you could improve by balanced and can't get it with SE the same.

 

[Guy.Cohen]

Exactly, loosen 6 screws and then lift the top plate with a suction cup. For the bottom plate you need to remove another 6 screws from the board.
Remember that the XLR preamp outputs are fed directly from the THX circuit with the TI OPA564.

By the way, after your help on how to open the box, as I mentioned already, the preamp output (back panel is not taken from the 4 THX amplifiers.
It is taken from the balancing circuit after the R2R volume, but with series 470ohm resistor (per phase). This 470ohm resistor is also the series resistor between the balancing circuit (OPA1612) and the 4 THX amplifiers.
When you have time you can check, maybe I am wrong.

 

[Roland68]

I agree that the common mode cancellation is achieved at the point of the differential pair summing.
At the SMSL it is really at the input.
Then it could pick up something along the path and again could be cancelled at the second differential pair (like power amplifier input).

Read my last post again about why this is a problem, and no, it doesn't work.

Listening tests of course are subjective, but you were a group. So it has a meaning.

You missed something here too. Our hearing tests always run in parallel with complex measurements and must lead to the same results. In this case, the measurements were done by engineers, to whom developers, industry and universities go when problems with measurements arise. In this case it was necessary and expensive, but fruitful.

For the BTL connection at headphones out. Power and rise time could be achieved the same anyway, even with SE connection.
Another amplifier is more complicated then putting 4 times powerful.
For crosstalk this is something that you could improve by balanced and can't get it with SE the same.

With equivalent amplifiers, the rise time for Balanced is always half that of SE. Each amplifier branch only has to perform half the power and half the voltage swing, and since this happens in both amplifiers at the same time, the rise time is only half as long as with SE. If you develop a more complex amplifier, this would again have the advantages if it were used for balanced amplification. And no matter how complex the amplifier is, the components and the additional board space for the second branch are not expensive.

No matter how, you have (almost) all of my information on the subject, and you can also imagine why I have equipped all of my normal devices with symmetrical inputs and outputs and deliberately forego symmetrical small signal processing. Except for the directly necessary one in front of real symmetrical power amplifiers.

I had long planned to convert the completely symmetrical Singxer SA-1 to a better quad pot or symmetrical relay-based quad volume control, as all 4 channels are completely different.
But here too, the better way will be a desymmation stage at the input, high-quality 2-channel volume control, a symmetry stage in front of the power amplifiers. So the food for thought was helpful for me too, sometimes you forget what you've learned and fall into old patterns.
By the way, the reason for buying the SA-1 was the completely symmetrical design, total irony...

 

[Guy.Cohen]

Indeed ironic and the Singxer is not cheap. I looked at SA1, but it was too expensive vs SP400 and the published specs are worse in compare to SP400 (also what Amir published).
I also looked at the Gustard H16 that I think FDA all the way but measurements are not as good as SP400 (but at least H16 is cheaper).

When you are saying that the 4 symmetrical channels are completely different in SA1, in what sense? Phase? Amplitude? THD? Noise? All together?

Last, in your post one thing I didn't understood. So, you can't hear the difference between 10m XLR and 10 times 1m XLR with debalance and balance.
10m of RCA cable was much inferior? I understand correctly?

Read my last post again about why this is a problem, and no, it doesn't work.


You missed something here too. Our hearing tests always run in parallel with complex measurements and must lead to the same results. In this case, the measurements were done by engineers, to whom developers, industry and universities go when problems with measurements arise. In this case it was necessary and expensive, but fruitful.


With equivalent amplifiers, the rise time for Balanced is always half that of SE. Each amplifier branch only has to perform half the power and half the voltage swing, and since this happens in both amplifiers at the same time, the rise time is only half as long as with SE. If you develop a more complex amplifier, this would again have the advantages if it were used for balanced amplification. And no matter how complex the amplifier is, the components and the additional board space for the second branch are not expensive.

No matter how, you have (almost) all of my information on the subject, and you can also imagine why I have equipped all of my normal devices with symmetrical inputs and outputs and deliberately forego symmetrical small signal processing. Except for the directly necessary one in front of real symmetrical power amplifiers.

I had long planned to convert the completely symmetrical Singxer SA-1 to a better quad pot or symmetrical relay-based quad volume control, as all 4 channels are completely different.
But here too, the better way will be a desymmation stage at the input, high-quality 2-channel volume control, a symmetry stage in front of the power amplifiers. So the food for thought was helpful for me too, sometimes you forget what you've learned and fall into old patterns.
By the way, the reason for buying the SA-1 was the completely symmetrical design, total irony...

Indeed ironic and the Singxer is not cheap. I looked at SA1, but it was too expensive vs SP400 and the published specs are worse in compare to SP400 (also what Amir published).
I also looked at the Gustard H16 that I think FDA all the way but measurements are not as good as SP400 (but at least H16 is cheaper).

When you are saying that the 4 symmetrical channels are completely different in SA1, in what sense? Phase? Amplitude? THD? Noise? All together?

Last, in your post one thing I didn't understood. So, you can't hear the difference between 10m XLR and 10 times 1m XLR with debalance and balance.
10m of RCA cable was much inferior? I understand correctly?

 

[Snoopy]

Hm I got both the Singxer SA1 and the SP400. Pretty sure the SP400 was a bit more expensive when I purchased it

 

[Guy.Cohen]

See here the price

Singxer SA-1 Review (Balanced Headphone Amp)

This is a review and detailed measurements of the Singxer SA-1 balanced headphone amplifier. It was sent to me by shenzhenaudio.com and costs US $599. While many desktop products look the same these days, I liked the more stout and positive feeling toggle switches on the SA-1: Unusual to...

www.audiosciencereview.com

I bought my SP400 at about $400 from hifi express, recently. Maybe the last units....

 

[Roland68]

Indeed ironic and the Singxer is not cheap. I looked at SA1, but it was too expensive vs SP400 and the published specs are worse in compare to SP400 (also what Amir published).
I also looked at the Gustard H16 that I think FDA all the way but measurements are not as good as SP400 (but at least H16 is cheaper).

Singxer's prices are much more stable. SP400 and SA-1 are on the same level.
If the H16 is actually completely and consistently symmetrical, then the poorer values could be due to this.
Our tests back then showed that there is nothing to be gained with symmetrical small signal processing, but the costs increase by 1 to 2 zeros before the decimal point per device if you don't want to lose too much. But you definitely lose to SE.

When you are saying that the 4 symmetrical channels are completely different in SA1, in what sense? Phase? Amplitude? THD? Noise? All together?

It's all about different volume and it comes exclusively from symmetrical small signal processing and the quad pot. The individual power amplifiers are very well tuned, otherwise I wouldn't have kept the SA-1.

Last, in your post one thing I didn't understood. So, you can't hear the difference between 10m XLR and 10 times 1m XLR with debalance and balance.

Yes, and anything else would be fatal for all recording studios in the world and there would be no great music recordings, because that is exactly the situation from the recording microphone onwards. Because it goes through many devices in which the processing is done SE, but the cable routes go via balanced XLR connections. And there are enough critical musicians, sound engineers and producers who would not accept losses that they suspect.

Here you will find 2 threads from an experienced and valued member on a very similar topic.
8th generation ADC/DAC loop vs the original.....can you hear it?
Listen and choose the 8th generation digital copy
I was amazed, but was able to verify it in my own test.
And with this topic the effects should be even greater, but that also only exists in our heads...

10m of RCA cable was much inferior? I understand correctly?

And yes.

 

[Hal Rockwell]

Can someone please tell me what OPA564 and OPA1612 do in SP400?

 

[Roland68]

Can someone please tell me what OPA564 and OPA1612 do in SP400?

They are both OPAmps from TI and are part of the THX AAA-888 amplifier circuit in the SP400.

 

[Hal Rockwell]

They are both OPAmps from TI and are part of the THX AAA-888 amplifier circuit in the SP400.

That I know. I can read spec sheets. I want to know why there are two of them and what is the purpose of each one.

 

[Doodski]

That I know. I can read spec sheets. I want to know why there are two of them and what is the purpose of each one.

Hi @Hal Rockwell. You will be required to @ the least provide details and images clearly indicating what exactly that you want to know.

 

[Hal Rockwell]

Hi @Hal Rockwell. You will be required to @ the least provide details and images clearly indicating what exactly that you want to know.

The big one is OPA564. The small one is OPA1612. What is their purpose in the THX888 module? Is the output of one feeds the input of the other? Is one of the controls the feedback of the other?

 

[Guy.Cohen]

The inputs enter the R2R volume. The XLR inputs are debalance using OPA1612. The SE input goes directly (all with DC blocking). After the R2R volume there is a balance circuit using again OPA1612. From that place the signal is going through 470ohm resistors to two places: to back panel XLR and to the 4 THX amplifiers. The outputs of these THX amplifiers feeding the headphones outputs.
The THX amplifiers are composite amplifiers using OPA1612 and the power opamp OPA564.
The THX module suppose to have some sort of feed through nulling error correction ala Hawksford.
For that last sentence I am not sure it is implemented in SP400,
but I think the use of error correction (not only simple feedback) is part of the THX patent