Neutrik NA2M-D2B-Tx balancing transformer adapter review

[solderdude]

Yep Lundahls are about 5x more expensive. They can be used for 4V XLR though and can solve the same issues (galvanic separation).

So those that believe they need one for the hifi systen need to fork out $ 300.- (for stereo) these would be a suited option and would certainly get thumbs up from Amir.

 

[pma]

For those who think they can hear the NA2M adapter in the chain, here is the test with two files, one of them has the adapter inserted into the chain.

https://pmacura.cz/trafotest.zip

The level is 1.5Vrms during the drum booms and bass guitar. If you hear the difference, please make a foobar abx test and post the protocol here.

 

[voodooless]

For those who think they can hear the NA2M adapter in the chain, here is the test with two files, one of them has the adapter inserted into the chain.

https://pmacura.cz/trafotest.zip

The level is 1.5Vrms during the drum booms and bass guitar. If you hear the difference, please make a foobar abx test and post the protocol here.

You should have made a comparison with a ground loop, that would have been much easier to spot

 

[solderdude]

It could be interesting to do the same thing but with 4V input (the way Amir tested it).
Of course 4V in RCA is kind of high by itself ...

 

[pma]

It could be interesting to do the same thing but with 4V input (the way Amir tested it).
Of course 4V in RCA is kind of high by itself ...

Understood, but 4V input makes no sense for an adapter with RCA-CINCH single ended input and XLR male balanced output. There is no question and no option in which direction the adapter is to be connected. You would not press male-to-male XLR no matter how hard you would try . And I am sorry, I would not be obedient to some eventual noobish suggestions. Not me.

 

[pma]

Just to mention - this, this and this are examples of clever use of this NA2M adapter .

 

[pma]

I have just measured inter-winding capacitance between primary and secondary of the NA2M and it is very low 43pF. This explains excellent CMRR of this adapter and why it yields very good isolation between input and output.

 

[Noisea]

The statement that the load resistor (secondary side) does not have impact on performance is false. Read the paper from Bill Whitlock (posted by somebody else) to learn about audio transformer basics.

The 47k load built into the Neutrik box is in parallel with the load that the input connected to the transformer secondary has. 47k is rather on the high side and merely ensures compatibility with high impedance inputs. Since the connected input impedance is unknown and variable from one equipment to another, the frequency response (both LF and HF) will vary as well. This may explain different measurement results by different people.

To make things worse, the transformer inductance also interacts with the cable capacitance, which also influences frequency response. A short cable with low capacitance is preferable on the secondary (output) side.

Often a snubber is placed at the transformer secondary for frequency compensation. The snubber together with the load resistance determines the frequency response. This may be difficult to optimize.

The typical low pass filter of any amplifier input may look like a snubber to the transformer. The amplifier input is then tapped from in between the R and C of the filter, which further complicates things. This filter does interact with the transformer and has a lot of influence on the frequency response. The exact component values of such an input low pass filter may vary a lot from amplifier to amplifier. As a result, actual frequency response is almost unpredictable.

Getting a flat frequency response from transformers may be tricky in general. Correct loading is important for both LF roll-off and HF roll-off. There may be limited frequency response both LF and HF or a gain peak in the HF region in case termination is not optimal.

 

[solderdude]

The statement that the load resistor (secondary side) does not have impact on performance is false. Read the paper from Bill Whitlock (posted by somebody else) to learn about audio transformer basics.

Not quite so.
You see the resistor (47k) does not provide a correct load (should be in the 2k-4k range) and as that resistor is on the XLR side only it really is only a load with the RCA->XLR version and not in the XLR->RCA version.
Not much thought seems to have gone in the design. It is not intended for hifi applications so it is moot anyway.

Often a snubber is placed at the transformer secondary for frequency compensation. The snubber together with the load resistance determines the frequency response. This may be difficult to optimize.

yep, and this is load dependent as well as source dependent. As no one knows where this device will end up in there is no snubber present.
Besides, when used as expected (with proper levels) a snubber does not seem needed anyway.

Getting a flat frequency response from transformers may be tricky in general. Correct loading is important for both LF roll-off and HF roll-off. There may be limited frequency response both LF and HF or a gain peak in the HF region in case termination is not optimal.

Yep, but this device is a 'universal' problem fixer (galvanic separation for low signal levels + connection conversion) so it is possible it well be used in less than ideal circumstances.

 

[Noisea]

Not quite so.

Hmm, you may be right about this.
Following common sense, distortion should rise with increased loading.
But:

The total harmonic distortion is mainly a function of the operating flux density in the core at the lowest operating frequency. Reducing the flux density will reduce the distortion. (Wolpert)

I need to measure this myself to believe it.

You see the resistor (47k) does not provide a correct load (should be in the 2k-4k range) and as that resistor is on the XLR side only it really is only a load with the RCA->XLR version and not in the XLR->RCA version.
Not much thought seems to have gone in the design. It is not intended for hifi applications so it is moot anyway.

They should have put load resistors on both sides then.
I would not call the design mindless. It is just impossible to properly design for a completely unknown environment.

Besides, when used as expected (with proper levels) a snubber does not seem needed anyway.

Why is the need for a snubber dependent on level?

Agree, there is no simple universal fix without downsides. This is what I wanted to point out. Proper application of such transformers is complicated.

 

[solderdude]

They should have put load resistors on both sides then.

I think they should have loaded the transformer on the output side only so on the XLR in one converter and on the RCA on the second one.
That, however, would mean the input would always be loaded with that same resistance. As it is now a 10k load would be 'seen' as a 10k load on the input as well.
The 47k will do nothing really and could have been left out. It is in the XLR-> RCA case anyway.

I would not call the design mindless. It is just impossible to properly design for a completely unknown environment.

Yes, impossible to account for all purposes it can be used in.

Why is the need for a snubber dependent on level?

It isn't. With proper levels I meant the levels it is supposed to be used with.

Agree, there is no simple universal fix without downsides. This is what I wanted to point out. Proper application of such transformers is complicated.

Yes, but as simple as it is (and limited in usage) it can do the job it is supposed to do. It will add some distortion, will be limited in lowest frequency and highest frequency and might be peaking above 20kHz when not properly loaded but at least you have galvanic separation, balanced signal and good enough performance for most professional cases when it is needed as a problem solver.
Hifi people should look elsewhere ... but will have to pay a lot more.