Electronics > Projects, Designs, and Technical Stuff
LM1875T Amplifier Problems
AdamP2209:
Hi, new to the forum!
So, I'm working on a bluetooth speaker and have been doing the proof-of-concept stuff recently and I'd like some help diagnosing an issue.
From the 240V wall I have a PSU that provides 30V (max 1A) to:
A voltage divider dual PSU (Image of rough scheme attached) using 4700uF Caps and 1k 2W resistors which feeds +/- 15V to one of those chinese LM1875T boards.
I bought the board, and replaced all the components with genuine parts (TI LM1875T NOPB & ELNA caps).
Hooked it up to an Up2Stream Pro bluetooth/wifi board and a small speaker in a quick enclosure to test.
It worked initially - good level of output with a clean sound. However I stupidly scoped the output wrong shorting the output to ground.
After this, on power-up, the speaker will play for 1-2 seconds and cut out - measuring the voltage across the V+/ G / V-, I find the PSU has shifted to one side:
(28V+/ G / 2V-) instead of (15V+ / G / 15V-), and this naturally stops the amplifier doing it's job.
Thinking I'd damaged the LM1875T board, I soldered up a copy, exact same components but new. Connecting this new board yields the same result.
Diving a bit deeper I found the voltage shift doesn't need an input to occur - scoping the input from the Up2Stream shows it working normally. Simply powering up the LM1875T board causes this shift.
I also found disconnecting the speaker prevents this issue from happening - with no load connected to the output, the voltage is normal (15V+ / G / 15V-) on the PSU.
Checking the speaker with another amp shows it working normally.
Lastly, I measured all components in the dual PSU and amp boards and they all seem normal.
At this point i'm really confused and was wondering if anyone has the intuition/experience to see whats going on here.
Many thanks in advance!
Adam
magic:
Perhaps the amplifier outputs negative DC for some reason. This flows through the speaker to the virtual ground and discharges it down faster than the 1kΩ resistors are able to keep up.
The top resistor sources 28mA into ground and the bottom sinks 2mA when equilibrium is reached. That's 26mA through the speaker and over 200mV offset if it's 8Ω. You should be able to detect it with a DMM.
AdamP2209:
After measuring, this is exactly what is happening.
With no input connected, and no load, the potential across the output terminals in -107mV, voltage on PSU normal at (15V+ / G / 15V-).
With the speaker load connected, I get (28.9V+ / G / 1.1V-) and 107mV potential across the terminals, 12mA DC through the speaker (yes it's 8 ohm).
So with this apparent, what could be wrong with the Amp? Is it possible to get the IC too hot when soldering, destroying it and explaining why both the original board and a copy have the same error?
Really appreciate this btw, I'll solder a third board up with a heat sink to keep temps low and see if this resolves the issue.
Someone:
--- Quote from: AdamP2209 on November 30, 2019, 06:20:48 pm ---With no input connected, and no load, the potential across the output terminals in -107mV, voltage on PSU normal at (15V+ / G / 15V-).
With the speaker load connected, I get (28.9V+ / G / 1.1V-) and 107mV potential across the terminals, 12mA DC through the speaker (yes it's 8 ohm).
So with this apparent, what could be wrong with the Amp?
--- End quote ---
1k Ohm * 12mA is roughly the 14V shift you see in the virtual ground, everything makes sense there.
Why are you trying to make a virtual ground with such a high impedance? Why not use the single supply circuit from the amplifiers data sheet?
AdamP2209:
That was simply because I couldn't find a pre-made PCB for it at the time, a single supply would be much nicer though!
Wrt to the high impedance, 1k keeps the idle power very low for a 30V input, and I wasn't expecting a DC draw.
Just hooked the amp up to a benchtop dual PSU, and all the boards I've soldered up do actually work well.
I see the issue with using that voltage divider PSU when there's any DC on the output, I'm only confused by why the circuit initially worked now (enough to play loudish music for 10mins before the scoping accident).
Navigation
[0] Message Index
[#] Next page
Go to full version