Lepai 2020TI behaving strange with audio analyzer

[Thinker]

Hi Everyone,

I recently picked up a cheap Lepai 2020TI to mess around with. I connected to my bench equipment since I was curious to see if it would meet spec. I connected it to my Variac set to 120 volts, my HP8903B audio analyzer, and my 8 ohm 1% non inductive dummy load. I've tested a bunch of equipment before never with an issue.

When I tried sending a 1.2 volt 1 KHz signal to the Lepai with the tone controls out of signal path I got nothing out of it and I noticed the the Lepai's LED's were periodically dimming. I played around with different input voltages and I would start each time with the volume all the way down and then bring it up slowly and I would get no usable signal according to the analyzer. So I connected my scope to see if I anything was coming out. All I saw was a flat line that was periodically going up and down but staying flat.

I discovered that there was -5.68 voltage present on the chassis when in reference to ground. When I disconnected the dummy load it disappeared. When I measured the connections to the dummy load in reference to ground there's no negative voltage, when I connect the dummy load but disconnect the BNC between the dummy load and the analyzer the -5.68 volts disappears. I then measured the BNC connection on the front of the analyzer in reference to ground and there was no negative voltage.

I discovered if I send a 1 KHz tone into the Lepai using my ipad instead of the audio analyzer, I get a normal test wave at the output that meets spec, I'm guessing that there's some kind of grounding problem in my testing setup but, I don't understand why it hasn't been a problem in the past. That or something isn't right in the Lepai.

Side note, I've always noticed when testing gear at it's rated spec you hear some of the test tones coming from the gear. I played music through the Lepai into the dummy load, without clipping and was surprised how much of the music I could hear coming out of the Lepai, it sounded like tinny headphones, I'm guessing that the sound was coming from the small inductors on the ouput? Just curious what causes that.

 

[OMGCat!]

I know a lot of those chip amps are internally bridged so if you're using a common ground between channels you'll cause problems.
Hearing test tones from inside gear isn't all that unusual. I don't think I've ever been able to hear actual music but I don't think I've tried to.

 

[Thinker]

I know a lot of those chip amps are internally bridged so if you're using a common ground between channels you'll cause problems.

I tried it with only one channel and there was still the -5.68 volts, I'm pretty confused as to what's causing it, doesn't matter if it's the left or right channel. Is it a digital amp thing?

I'm thinking about replacing the 4 capacitors with decent ones, not sure if that will make any difference. I did hook it up to a pair of speakers and it sounds pretty good, only problem is it has turn off thump.

 

[OMGCat!]

That really is strange. Possibly some sort of current leakage with your analyzer? Seems like everything started to work normally when that was taken out of the picture.
I don't have a Lepei so I'm not sure if turn off thump is normal. I do have a few different chip amps I bought from eBay and some thump a bit and others don't.

 

[restorer-john]

Test tones are often heard in the heatsinks, the inductors and the output transistors when testing at high powers.

You clearly have an effectively BTL style output stage in the Lepai (class D) and you are essentially shorting out one half of it as the sig gen and the analyzer are both ground referenced.

We've all done it in the past and scratched our heads...

The ipad is not ground referenced, whereas the sig gen and the analyzer are tied together at 0V.

 

[elnaldo]

I was thinking the same, some problem with the test equipment ground. I run my oscilloscope "ground lifted" and only use 2 probes at the same time when I'm 100% sure both ground references can be shorted.

 

[Thinker]

You clearly have an effectively BTL style output stage in the Lepai (class D) and you are essentially shorting out one half of it as the sig gen and the analyzer are both ground referenced.

We've all done it in the past and scratched our heads...

The ipad is not ground referenced, whereas the sig gen and the analyzer are tied together at 0V.

I was wondering about that, so I switched the input and output to float for the ground but still got -5.68 volts on the chassis, I even thought perhaps the float switch wasn't working properly so I cobbled together a patch cable to only connect the center conductor and still got -5.68 volts on the chassis. I thought by lifting the ground that the negative voltage should disappear?

 

[Bill Ferris]

Well Thinker, FWIW, this method has served me well on both benches for decades.
I power ground float all three prong power grounded test equipment(Dist. analyzer, scope, sig gen. but have their ground post`s connected leads/w leads to alligator clips available for connecting to the DUT, and a heavy gage(12) rope wire connected to the ground of my grounded Variac for reference to power ground, if needed, and depending on the audio gear being tested have found that various combination of these test equipment ground lead connections, via this method, provides me with the lowest noise, and THD measurements with no signal reading/scope display weirdness or issues, unless the amp has inverted outputs, like my ARC D76A.
And that just required me to invert the amps output connections to the floating dummy load, because of the now ground referenced test equipment connected to it.

This ground connection scheme can vary, depending whether the DUT is very high gain, line level, power amp, power grounded, or non power grounded.
I don`t have to break, or custom make any signal, or test equipment returns, shields, just a ultimate single ground point reference for test equipment and the DUT to make all happily hand shaking on my benches for accurate results ..
This has worked very well for me since the early eighties.

Good luck Sir. whatever your method.

Kind regards, OKB