[christianw.]

Zitat:|================================================================
| Staff Comment 2012-10-04 15:56:50 PST
| By: Matthew M
|
| Hello Christian,
|
| I have now had some time to experiment with sequencing this part.
| I used a MAX9714 evaluation kit with a
| AX9704 mounted on it. I matched your configuration (spread
| spectrum, maximum gain, 12V supply) in order to best simulate
| your situation.
|
| To avoid introducing impedance between the device and the
| supply/bypass capacitance I used a current probe to create my
| plots. I am a bit concerned that the 330mΩ resistance could be
| part of the problem. Is it always present between the supply and
| device and where is it located relative to the supply, bypass,
| and device?
|
| The reason I am concerned is that I don't see nearly the inrush
| current you do. When you power up, what is the state of the Input
| pins (INL±/INR±)? A thought occurred to me - maybe your setup
| draws so much power because you have active inputs on power up.
| My concern is that if the inputs are active when the part is
| inactive, and then the device is powered, this could damage the
| device.
|
| The proper sequencing should be:
| 1. supply powers up, device may come out of shutdown during this
| if the /SHDN pin is supply referenced (always on operation).
|
| 2. take the device out of shutdown (may happen in step 1).
|
| 3. Apply an input analog signal. These inputs (INL±/INR±) are
| best left off until the device is fully powered, and the turn on
| time (EC table, page 2) has elapsed.
|
| I will next post my power-on screen shots.
|
| I found that I had exactly the same start-up waveform regardless
| of if I powered the supply with the device held in shutdown, or
| powered with shutdown tied to the supply (so the device activated
| on power-up). In all cases (INL±/INR±) are not driven during
| power on. I drew the same inrush current (1) with a without a
| load (my load was 8Ω + 68µH).
|
| I then tried again with the device allowed to power up active
| (/SHDN goes high with the supply, like your case). The current
| draw was the same (2) as with the device in shutdown. The
| difference was that about 80ms later the output began switching
| (purple line, 3).
|
| In all cases the initial power up draws a peak of about
| 200-210mA.
|
| dark blue = supply (12V)
| light blue = current draw (mA)
| purple = Single ended output, used to show when switching starts.
|
| 1. PU_12V_SHUTDOWN.bmp
| Power up with the device held in shutdown. This was identical
| loaded or unloaded.
|
| 2. PU_12V_ACTIVE.bmp
| Power up with the device coming out of shutdown as the supply
| rises. This also was identical loaded or unloaded.
|
| 3. PU_12V_ACTIVE_SWITCHING_START.bmp
| Same as number 2, but longer time scale allows the timing of when
| output switching starts to be seen.
| I ran out of space in the previous. I tried checking the
| switching as well with and without a load. In both cases the
| single ended outputs switched from ground to 12V. This was
| regardless of load. I didn't see any case where the switching was
| outside of the supply The maximum differential voltage is
| therefore +12V (positive output high, negative low) to -12V
| (positive low to negative high) as expected. This leads to the
| speaker never having a differential voltage of more than 12V
| across it at any time. Again I did not see a difference loaded or
| unloaded. With no input signal, the only bias on the speaker
| output should be when the switching is slightly faster or slower
| on the positive or negative output.
|
| Finally to test survivability, I automated a supply sequence that
| power cycled the device thousands of times. After running over
| night with several thousand cycles the device is still running
| fine with pout any apparent ill effects.
|
| I cannot reproduce your issue unfortunately, and i am at a loss
| to explain it as I simply do not see the large inrush current you
| do (looking at your last figure, if the sense is 0.33Ω, and the
| spike is about 8V across the resistor, thats over 2A of inrush!).
| The possibilities I can come up with are:
|
| 1/ The sense resistor itself is a problem, or the impedance /
| inductance between the device, the bypass, and the supply is too
| high. Remove the sense resistor, and try with a current probe and
| the minimized impedance on the supply bus.
|
| 2/ The device is being powered with a signal already present at
| the input. Do not drive audio through the amplifier until it is
| fully powered up to avoid any chance of latch-up, etc.
|
| 3/ Other external components are not specified correctly or are
| damaged. You did find bad bypass caps, perhaps some other
| components are damaged, or are not rated correctly for the
| operating range of the device. Check the MAX9714 EV-kit build of
| materials against your own for component specification mismatches
| (not just values but voltage and power ratings for components).
|
| 4/ The parts themselves were already damaged or have problems.
| Where did you get these devices? Did you order or sample them
| through Maxim direct or one of our authorized distributors? What
| was your source and do you still have the purchase order
| documentation available? Normally I would not worry about
| counterfeits, but the behavior your observing is vastly
| different.
|
| Outside of this, I am out of ideas for the moment. I may be able
| to schedule some additional lab time early next week to dig a bit
| deeper, and follow-up on your responses.
|
| Again, thank you for your patience and I do apologize for the
| delayed response.
|
| Matthew Mowdy
| Audio Applications Engineer

PU_12V_ACTIVE.BMP
[IMG] https://stromrichter.org/d-amp/content/i...ACTIVE.png[/IMG]

PU_12V_ACTIVE_SWITCHING_START.BMP


PU_12V_SHUTDOWN.BMP