Audio HP Amp TI LM4811 - am I getting crazy?

[Saimoun]

Hi all!
So glad to join the community here - I have been following Dave's videos for a while now, really really great stuff!! 

My first post - an issue of course!

I am trying this IC from TI called LM4811 - datasheet here: https://www.ti.com/lit/ds/symlink/lm4811.pdf

On the paper it's all good, it seems very simple, here is the "typical application" from the datasheet.




So I made a PCB and ordered it, soldered it, no shorts and everything seems perfectly ok! But when it comes to testing, it seems that the Bias voltage (internally created by the IC) is *heavily* influenced by the CLOCK pin! yes - the CLOCK pin!

Here are my schematics:




To test I tried measuring the voltage on the Bias pin directly or on the Vin pin - since this should be at the same average voltage (they are identical since for now I am sending a 1.65V DC from the DAC). Here are measurements when the CLOCK pin is at 0 (GND) and when the CLOCK pin was at 1 (5V) while starting up. The yellow line is the Vin pin and the blue line is the CLOCK pin.


CLOCK at GND:



CLOCK at 5V:



Sometimes the bias voltage changes by itself, oscillating between 0 and 1V when the CLOCK is at GND, and between 5V and 3V when the CLOCK is at 5V.

I double-triple checked the PCB layout and the soldering. Nothing complicated, just the IC and a few through hole components around. Is there something I missed in the datasheet maybe?

Thank you in advance! 

Cheers
Simon

[Saimoun]

Ooh and I forgot to say that the CLOCK pin seems like it pulled down somehow - if I put a 1K resistor between 5V and the CLOCK pin it reads at around 0.5V. If I put 100Ohm resistor it reads at around 1V. To put it at 5V I need to connect it directly to 5V.

[magic]

Are you sure the chip is genuine?
Are you sure you didn't connect shutdown wrong?
What the clock pin is for and why are you connecting it to some fixed voltage? I'm lazy and didn't download the datasheet

[Saimoun]

It's all good - we're all lazy

Chip is genuine, bought from Digikey. I already tried two of them and both have the same behavior.

The CLOCK and UP/DN pin work to change the volume. As I understood they should be at GND when "resting". Here is all the datasheet is saying about it:



The Shutdown pin is connected to GND and just to check again I am looking at the datasheet that the device is active when Shutdown is low... And actually this is very confusing, I am not sure anymore :/

How do you guys understand it, should it be LOW or HIGH to put the device active?

In order to reduce power consumption while not is use, the LM4811 features amplifier bias circuitry shutdown.
This shutdown function is activated by applying a logic high to the SHUTDOWN pin. The trigger point is 1.4V
minimum for a logic high level, and 0.4V maximum for a logic low level. It is best to switch between ground and
VDD to ensure optimal shutdown operation. By switching the SHUTDOWN pin to VDD, the LM4811 supply current
draw will be minimized in idle mode. Whereas the device will be disabled with shutdown voltages less than VDD,
the idle current may be greater than the typical value of 0.3µA. In either case, the SHUTDOWN pin should be
tied to a fixed voltage to avoid unwanted state changes.

[magic]

Ooh and I forgot to say that the CLOCK pin seems like it pulled down somehow - if I put a 1K resistor between 5V and the CLOCK pin it reads at around 0.5V. If I put 100Ohm resistor it reads at around 1V. To put it at 5V I need to connect it directly to 5V.

It looks like there is a diode from CLOCK to GND inside the device. With 100Ω to 5V it's consuming 40mA of current if the pin is at 1V
And the datasheet says that it should be pulled to 1.4V to register as high? Makes no sense.

If it isn't fake, are you sure you didn't mount it backwards? Does it work at all, by the way? Do the VIN pins really track the voltage at BYPASS or was that just your speculation? Can you get any sound from it?

Shutdown seems correct. "Shutdown active high" means that high will make it shut down.

[Saimoun]

Ok about the shutdown - thanks.

What makes you think there is a diode from CLOCK to GND?

These values were just what I remembered, I checked again and it's more like 0V (or almost) with a 10k resistor, and 0.3V with a 100Ohm resistor. But there is clearly current flowing because the 100 Ohm resistor got really hot really quickly!!

Which makes me think: what the hell happens when I set the pin to high without any resistor?!

The chip seems to work, the VIN pins get to the bypass voltage (yes I measured), and if I put a sine way on VIN I have the same wave (ish - it's cut because of the wrong bias voltage) inverted on the output.

[magic]

That's some paranormal stuff

Check if there is continuity from power supply to VDD.
Do you have a ceramic capacitor between VDD and GND close to the chip?

[tooki]

Did you build the “typical application” circuit or the demo board circuit? The big difference are the pull down resistors.

Also, in your schematic, Vin1 and Vout1 are both floating. Is this actually the case? It’s possible that’s an unstable configuration.

And the datasheet says that it should be pulled to 1.4V to register as high? Makes no sense.

Why does that not make sense? It’s a device that supports a 2V supply voltage, and 1.4V is 0.7x2V.

[magic]

Well, it makes no sense that it's impossible to pull this pin above 1V

[Saimoun]

Well, it makes no sense that it's impossible to pull this pin above 1V

I agree with that!

And yes tooki I agree the 1.4V makes sense, it's just the minimum value to be considered "high" (shutdown, up/dn and clock pins are digital inputs).

Did you build the “typical application” circuit or the demo board circuit? The big difference are the pull down resistors.

Also, in your schematic, Vin1 and Vout1 are both floating. Is this actually the case? It’s possible that’s an unstable configuration.

To be honest I tried everything at this point! Pull-down, pull-up, and directly tied to GND and to +5V.

The schematics I showed was the simplified version, but my original version - the one where I started experiencing the issue - is the one I put above. Let me get the schematics and the PCB for the original version, I'll put it here.

[magic]

Again, are you sure there is no problem with VDD and the chip isn't taking power through ESD diodes from other pins?

[Saimoun]

Here is the original schematics and PCB.

Schematics (ignore the fact that all nets have a "2-" in front of the name):



PCB top:




PCB bottom:

[OM222O]

the pcb is very messy and non standard (just look at the 10 pin SOIC, looks like a 6 pin jobby  ) and there was a problem with your schematic it seems.

Please upload the schematics.
Also this seems like easyeda. If so please share the link rather than photos.

[magic]

No idea
Any chance they sent you a wrong chip?

BTW, simply rotating C12 180° would decrease the length of connection from C12 to the GND pin for free. Probably not important here.

[Saimoun]

the pcb is very messy and non standard (just look at the 10 pin SOIC, looks like a 6 pin jobby  ) and there was a problem with your schematic it seems.

Please upload the schematics.
Also this seems like easyeda. If so please share the link rather than photos.

Sorry the picture upload went wrong - the post above has the schematics now.

And yes the PCB is a bit messy - it's just a PCB to test the IC. And actually the strange solder pads for the IC is on purpose, it is my way to be able to solder a tiny package as I have never soldered SMD before. But as I said the soldering went really well, no shorts and I tested each individual pin to make sure it was connected.

Here is a picture of the soldered PCB - the "dot" of the IC is at the top right.

[Saimoun]

Again, are you sure there is no problem with VDD and the chip isn't taking power through ESD diodes from other pins?

That's a good point - I did check on the VDD pin directly and it's getting +5V. But yes very interesting when I remove the +5V and let the VDD pin float - I get the exact same result = the voltage on the voltage pin follows the CLOCK pin.
But it's only the clock pin (whether I have VDD connected or not), if I move the up/dn or the shutdown pin nothing happens.

[magic]

 
Can you desolder the chip and check if VDD has any internal connection? You should be able to detect something using diode test between different pins.

Maybe you damaged it somehow

[Saimoun]

I still have two more chips so I could test with them. I tested with two already tho - and both have the same issue.

But I am starting to think maybe Mouser (yes it was Mouser not digikey actually) sent the wrong chips? I would never think that would be possible but the marking says 8565 G11 for a chip named LM4811?!

I have been googling for a SSOP-10 package with 8565 marking but without any luck. Quite terrible that the LM4811 datasheet does not show the marking! 

I also checked my order and it was the only SSOP-10 package so I could not have mixed it up with other chips. And the antistatic bag does say LM4811MM/NOPB.

EDIT: my bad the datasheet *does* show the marking, it says G11 and maybe with another marking, so it kinda seems it fits?

[Saimoun]

But actually on LCSC the picture shows a 9235 G11 marking... Not sure how much we can trust this anyways :/

https://lcsc.com/product-detail/Audio-Power-OpAmps_Texas-Instruments-Texas-Instruments-LM4811MM-NOPB_C529566.html

[magic]

It is possible and it rarely happens but the table in the datasheet says that marking is G11.

[Saimoun]

Yeps I saw as well - thanks

I made a few more measurements - now with a completely new chip, which I was super careful with not to touch in case I could have fried the previous ones. There are no shorts and all pins are soldered as they should. There is nothing else on the PCB, just the chip. Measuring resistance between the digital pins and ground, and between the digital pins and VDD, with my multimeter.

UP/DN and SHUTDOWN pins always give something above 1M (to both GND and VDD pins) while the CLOCK gives around 10k to GND (and around 100k to VDD)?!

Then I tested the following: 10k pull-down resistors on Shutdown, CLOCK and UP/DN, and give 5V on VDD (and 0V on GND). All other pins floating.
I get the following, yellow is the Bypass pin and blue is VDD.



(at around 7s I removed the pulldown resistor on CLOCK, making it float, and again this affected the Bypass pin).

I am getting out of ideas of what to test, and out of chips... 

[OM222O]

those pins are meant to be driven actively. I don't understand why you're "testing" the IC with EVERYTHING FLOATING. connect it to an MCU, drive the volume control and clock pins, add your audio source and speakers, see if it works as expected or not. many ICs may not function well under random circumstances. I remember having a seven segment driver which kept burning out under "testing" with floating pins, and I thought it was a bad chip, but when actually connected it to everything, worked as expected.

[Saimoun]

Fair point OM2020 - but at the end the only floating pins were the inputs since the outputs are internally tied to GND and I pulled down the digital ones. And the inputs are meant to be analog anyways so it should not make much difference?

And tbh I would rather not drive the pins from a MCU since the amount of current the clock pin seems to sink.

But ok I will try to follow their Demo board schematics then - it has buttons and jumpers to control the digital pins.

[tooki]

Let’s review some assumptions and facts:
1. It’s extraordinarily unlikely that the chips are fake since they were bought from a reputable vendor
2. It’s extraordinarily unlikely that multiple chips in a row would leave the factory faulty
3. The datasheet lists a maximum current draw of 3mA for the chip
4. Given that the IC is specified as being a low-power part, it wouldn’t make sense for its support components to draw large amounts of current.

So, what’s left?
5. Does the schematic actually make sense and work in theory?
6. Does the PCB match the schematic?
7. Are all the support components good, and of the correct values?
8. Was the PCB assembled properly?
9. Was care taken to ensure the chip isn’t damaged during soldering, due to excessive heat or ESD damage?
10. Is the device being tested under sensible conditions?


You probably have more than one PCB. Try building up another one.

Try leaving out the low pass filters on the inputs (omit C1/9 and bridge R17/21).
Make sure you have a load on the outputs.

What are you powering it from?

Remember that if it’s not the IC that’s bad, then it’s something else, i.e. the rest of the circuit.

Take the original populated PCB and remove just the IC. What happens to the voltage at the pads for the clock, up/down, shutdown, bypass, and VDD pins as you feed in high and low signals on the clock, up/down, and shutdown inputs? Somehow, I suspect that there is a short elsewhere on the board acting as a near short to ground.

[OM222O]

An MCU won't "burn" or "source" amperes of current! They're internally limited by the small fets that activate them. They have a "max current rating" that is usually less than 40mA and even then, the voltage will drop to close to 0. You can literally short out an MCU pin to GND forever and nothing happens. These pins are not meant to be analog. CLOCK definetly means digital, there are timing diagrams for them in the datasheet! Floating pins cause unreliable and random operation modes. Even if they were analog (op amp inputs for example) "best practice" is to tie them to GND or VCC if they're not used. Floating pins are always bad.