Misbehaving Purify ET400-based Class D Amp (Request for help)

[mcebrian]

Hi all, I bought the exact same class D amp based on Purify ET400 modules reviewed by @WolfX-700 here (https://www.l7audiolab.com/f/min400a/).

Problem: With certain specific tracks like the one below, the amp goes into protection mode and fails to output any sound for 1-2 seconds irrespectively of the source (different dacs, etc) or volume.

With 99.9% of the tracks, it works incredibly well.

Amp configuration: No buffer and no op-amp used in the input card before the purify modules. (Same as the one reviewed by Wolf)

Potential solution: Subsonic filter cutting off anything < 10Hz. Then, the problem is solved but I cannot always apply EQ, for example, streaming from Airplay certain tracks.

Request: Any help with intermediate board checks of modifications to stop it from oscillating? or entering DC protection? when <10Hz signals are processed. I am lost now.

Simple measurements of the amp using COSMOS ADC-1

 

[Sokel]

Checked the datasheet of the modules and to be honest I didn't expect it to be DC coupled,but it is.

If I was you I would certainly search for a buffer that would eliminate DC before the input.
It will have a small hit to your nice chart but peace of mind is priceless.

 

[mcebrian]

Thanks @Sokel, I have tried with the included buffer and it behaves exactly the same. The second problem is that I do not really know what is sending in the output during those two seconds or what could be the consequences (if any) for the speakers.

 

[Sokel]

Thanks @Sokel, I have tried with the included buffer and it behaves exactly the same. The second problem is that I do not really know what is sending in the output during those two seconds or what could be the consequences (if any) for the speakers.

You can record it with Cosmos and see.

 

[mcebrian]

I recorded it now with an accurate multimeter and I just see a DC offset of 300mV at the speaker output for a few ms when the issue is happening and then everything goes to 0.100mv till it starts playing music again properly. This does not reproduce when the track being played doesn't have significant content < 8 Hz.

 

[Sokel]

300 mV at the amp's output does not look like much but it is.
The thing is,what will happen (even in a short duration) if something up the chain decides to misbehave and send some serious voltage down the road.
Speakers won't like that.

 

[Kim Nordtorp Madsen]

mcebrian

When you have issues with any of the Purifi Audio products you - and all our customers - are allways welcome to write to sales"at"purifi-audio.com and we will help.

The 1ET400A module is DC-coupled so there is linear phase down to 20Hz or lower.

To protect the loadspeaker there is a DC-sense circuit that disables the output stage if there is excess DC on the output.

The 1ET400A module is a single ended class-D amplifier, so it can generate supply pumping, that is move energy from the positive supply to the negative and vice versa depending on the output signal.
The 1ET400A measures the supply voltage and in case of too high voltage shuts down. A DC-signal or low frequency signal can generate supply pumping that leads to the supply voltages beeing too high and the module will then shut down.

Some power supplies have protections agains supply pumping, like the Hypex SMPS1200A400, and others not.

You can test if the cause of your shutdown is the supply pumping or the DC-protecttion - by disconnecting the loudspeaker then there is no energy beeing pumped so if the module shuts down its the dc-protection.

If you have supply pumping issues - it can be lowered on stereo amps by reversing the phase on one channel e.i. flipping the +/- on both input and output of one amplifier channel

rgds
Kim

 

[capslock]

The LMAudio (and similar Chinese amps with a different brand name) seem to use a home grown SMPS the topology and specs of which are unknown. I would not be surprised if supply pumping was the problem here.

The other thing that surprises me is that in your test and in Wolf's, HD2 at 5 W is about -114 dB and HD3 about -121 resp. -124 dB, excellent per se but much worse than other implementations of the 1ET400 modules tested here on ASR.

What might be the reason for this? Knock-off modules, poor wiring or ferroelectric binding posts or case? Wolf has measurements for 4 and 8 R and all kinds of output levels, and these distortions stay pretty constant or decrease even with higher output, so I am inclinded to rule out a nonlinear mechanism such as ferreelectric materials.