Odd issue. DX3Pro radiating noise into powered speakers, at multiples of 48k only

[mtnbrit]

Hi, I am enjoying the DX3Pro very much with these highly rated IKmultimedia iLoud Micro speakers on my desk, but I have found a very strange issue. The DX3Pro is fed by USB from MacBook Pro. I mostly listen to Radio Paradise via Roon Internet Radio with no oversampling, and the DAC runs at 44.1k perfectly. The other day I switched to listen to JB Radio that runs at 192k, and when the DAC switched sample rates, I suddenly heard a nasty high pitched noise from the left speaker (the one with the plate amp). The noise is a high frequency tone around 10k or 12k, not at a high level, but definitely audible over music and well above the noise floor. After some fiddling around with changing sample rate going to the DX3Pro, I deduced the noise is generated by the DAC only when its at 48k, 96k, 192k, and 384k, but not at all at 44.1, 88.2, 176.4, 352.8 and 705.6k. Defying the pattern somewhat, the noise was not audible with Fs at 768k, but it may still be there just at some inaudible higher frequency. I have not yet broken out the scope or spectrum analyzer to look at the shape of the waveform and its harmonics.

The odd thing about this is that the noise is not passing over the RCA jacks to the powered speakers input as one might immediately suspect. If I disco the RCAs at the DAC or the speaker input, the noise remains. The iLoud Micro speakers are bi-amped with a single 4-channel plate amp in the left speaker and a thick 4 conductor cable with Kycon connectors going from the left plate amp to the right speaker. I have the dac in the middle of my desk in between the speakers, and this speaker cable runs right behind it. If I disconnect the speaker cable from the right speaker so its effectively unterminated, the noise continues on the left. Only if I pull the open-circuit unterminated speaker cable out of the plate amp does the noise stop. I tried relocating the speaker cable to be further away from the DAC but the noise persists, its obviously got quite a bit of energy.

So what we have here admittedly is some dubious issue with the iLoud micro speaker design, where its somehow susceptible to radiated noise getting picked up on the cross-connecting speaker cable going back into the right channel output stage on the plate amp, being amplified and audible on the left channel. I plan on doing a tear-down and high level reverse engineer of the plate amp to see why this is happening. Unclear yet if the speaker cable is screened or if the shell of the Kycon connector is grounded in the plate amp.

Until I get around to that, does this issue ring any bells in that this DAC, or DACs in general, might radiate noise when its sample clock is running at multiples of 48k but not at 44.1? Sorry its bit of a long post about a possibly obscure issue, but it may be of academic interest to see why the DAC is acting like this.

 

[restorer-john]

Get some of these before you do a teardown of the iLoud amplifier. Put them on the interconnecting cable.

 

[daftcombo]

In case someone doesn't know, it is called "ferrite".
I'll try it myself on my USB cable from PC to DAC, since I have a nasty treble sound in my headphones. Didn't think about it, thanks.

 

[Pluto]

the noise is generated by the DAC only when its at 48k, 96k, 192k, and 384k, but not at all at 44.1, 88.2, 176.4, 352.8 and 705.6k

DACs often have two clock units, one for 44.1kHz and its multiples, the other for 48kHz and its multiples. This kind of issue might be explained by a fault with the latter.

 

[restorer-john]

DACs often have two clock units, one for 44.1kHz and its multiples, the other for 48kHz and its multiples. This kind of issue might be explained by a fault with the latter.

It's also possible there is an interaction/intermodulation with the switching frequency of the internal class D amplifiers in the iLouds. There may be a direct feedback issue in those amplifiers as they pick up the DX3's noise.

 

[mtnbrit]

Ill gladly try the ferrite chokes on the speaker cable and report back, thanks.

I feel like I may have opened up a can of worms though. If you think of these iLouds that appear to be susceptible to radiated noise as a kind of EMF detector, we may have exposed a potential design flaw in the DX3Pro. Why is it blasting out this 48k and multiples clock noise at all? Knowing how much audio folks are concerned with eliminating all forms of extraneous signals and noises, both real and imagined, maybe @amirm should consider including a "radiated noise" test in his reviews? Im thinking of something like a de-tuned am radio placed next to the device under test. Not very scientific. Could be improved with a RF spectrum analyzer to look at signals from dc to several GHz.

 

[restorer-john]

Ill gladly try the ferrite chokes on the speaker cable and report back, thanks.

Put them on the DX3 power and USB leads too.

 

[mtnbrit]

I put several ferrites on the speaker cable, usb cable and DC power cable. No effect at all at reducing the induced noises.

I set up a mic in front of the left speaker and set the DAC to 192k. I took a quick recording of the sound, amplified it and made these spectrum plots in Audacity. Freq axis is linear to expand the higher freqs. The spikes are at the following frequencies.

2400
6000
7200
8400
9600
10800
12000
13200

For comparison I include the plot right after switching sample rate of the DX3Pro to 176.4k.

My next step is to fabricate a new speaker interconnect cable to where I can isolate each of the 4 conductors in turn and see which one may be induction the noise. I may try a ferrite on the individual cores, but I'm not optimistic as this is audio band stuff, not RF stuff that ferrite would be able to kill.

 

[restorer-john]

Try an offtuned AM pocket/small radio as an RF sniffer on the DX3Pro when you change sampling rates.

I reckon it's likely to be intermodulation products (from the Class D amp and the D/A) ending up down in the audible band. Each one is quiet without the other right?

 

[BtK]

I am using DX3pro and ILOUD Micro monitor and I have the same symptoms.

Can you tell if you've found a solution?

 

[pos]

A simple solution would be to resample everything to 176.4kHz or 352.8kHz in your player.

 

[mtnbrit]

Yep, that's what I ended up doing. Less than satisfactory. I even fabricated a new screened speaker cable and modded the plate amp to ground the screen of the mini DIN plug. This may have attenuated the interference marginally, but certainly not enough to claim any kind of victory. I also picked up a 24V 5A power brick with grounded AC connector and modded it inside to tie the DC ground to mains ground (the JS ground mod). No effect on the 48k interference.

 

[DDF]

Yep, that's what I ended up doing. Less than satisfactory. I even fabricated a new screened speaker cable and modded the plate amp to ground the screen of the mini DIN plug. This may have attenuated the interference marginally, but certainly not enough to claim any kind of victory. I also picked up a 24V 5A power brick with grounded AC connector and modded it inside to tie the DC ground to mains ground (the JS ground mod). No effect on the 48k interference.

If both devices have earth ground on the AC, see if you can borrow to try a class 2 power supply without earth ground (fully safe, its class 2) for one of the devices

 

[mtnbrit]

For anyone that's interested, here are some shots of the plate amp from the iLoud MM. The class D amp chips are TPA3118. The 2 channels going to the remote speaker pass through the two common mode chokes on the back of the amp board. Interesting they found the need for these only on the remote channels, im guessing to counter effects of the long cable run on the LC filter. The 48k radiated noise is apparently being picked up on the speaker inter-connect cable, and must be getting back into the amp. Its audible on both the plate-amp channel and the remote channel, until you disconnect the cable, then its 100% gone from the plate-amp side.

Other ICs identified:
CSR8635 electrodragon bluetooth module
ADAU1707 50MIPS DSP which has 2 ADCs, 4 DACs and a 4 input mux
N79E815 8-bit MCU
4580 op-amps

 

[bravomail]

topic should be renamed to iLoud issue

 

[BtK]

topic should be renamed to iLoud issue

To be honest, I don't know which side is the problem.

ILOUD has no problems with other DAC.