ADAU1701 crash audio out plugged in. weard

Hello,

Re the 10uF ceramic capacitor: where I work, we use ceramic caps for decoupling without any problems. The ams1117 data sheet recommends a 22 uF tantalum output capacitor for best stability. Sometimes the ESR characteristic of tantalum and electrolytic capacitors damps the control loop of linear regulators, thus helping with stability. However, the data sheet doesn't go into sufficient detail to tell.

Just as you said, the op-amp buffering would prevent issues with plugging and unplugging the RCA output. In fact, I haven't seen any issues even with passive output filtering. As a test, try just touching your RCA plug's ring (shield) to the outside of your board's RCA jack repeatedly to see if this causes the same problem. In other words, I suspect that connecting and disconnecting the ground of your output cable is all you need to make it happen. Likely this cable leads to an amplifier of some sort, such as amplified speakers. If these in turn are powered by a switching power adapter, it may have quite a bit of voltage on this "ground" input. Many switching adapters with a two-prong plug have EMI caps from both sides of the line to their DC output "ground" side. This makes a voltage divider resulting in 55 volts AC (!) common-mode riding on their DC output cable. You can easily check this with a multimeter connected between the shell of the DC output cable and the 3rd prong ground of a nearby outlet. Come to think of it, the 9V adapter powering your project may have this "feature" as well. The EMI caps are typically only 470 pF so the current behind this voltage isn't dangerous, yet a sudden connection may cause a transient which may upset your system.

If this turns out to be the problem (and you can always try different adapters to test), the fix is the punch line in the old joke, "Doctor, it hurts when I do this." If you're making a product for production and wish to make it more "bullet-proof" from this possibility, adding more bypassing may help. When the RCA ground is connected, current from that jack to where your 9V adaptor connects develops transient voltages along this path. Thus, keep this path direct and away from the low-level circuitry around the chips.

Best regards,

Bob

wow I never expected to see this voltage. I had noticed sometimes I got shocked from the housing of my project.

so interestingly I am using tannoy revel active speaker. its quite a high quality speaker. the 9V plugin adaptor is quite old, I found it randomly in a box of stuff I have horded for maybe 16 years.

I think I tried on a 9V battery though and it seemed to still do it as I suspected the PSU already. I will try that again.

I noticed in the speakers mains cable it had been butchered and a after market plug fitted. opening this I found there is no earth wire to the speaker.

I have a hantek scope here. I put my probe in x10 mode and checked it seemed calibrated with its 2KHz 2VPP output although with alot of overshoot.

referenced to my houses earth pin in the extension lead

The shell of the 9VDC has a ugly 100V PP wave at 50Hz which is almost square in character but not a clean square.

My speakers RCA gnd has around 10V PP of 50Hz dirty wave that looks more sine. Probably due to the missing GND

The RCA on my PC which feeds the music is very close to the earth of the house and I think its properly done.

maybe I need to consider my projects metal housing and how its all gnd. the RCA is isolated from the case, anything on the RCA will go near the ADAU1701. I think though if I try GND via the case the noise might increase and it just dont seem like a reliable thing to do.

Hello pedro,

I agree with what KJBob said and I think he is spot on. If you look up the ADAU1452 datasheet near the end there is the PCB Layout section and also earlier there is the DVDD voltage regulator discussion. There I detailed our findings where the regulator was designed years ago when all larger caps were electrolytic and had a lot of ESR. So that was factored into the design. Changing to a lower ESR ceramic 10uf cap reduced the phase margin in the design to a very low level. Although I did try in the lab to get the regulator to become unstable, I was never able to make it oscillate but the simulations showed the effect of the reduced ESR. Therefore we recommend to add around 0.5 ohms in series with the cap to add back in some ESR.

Regarding the grounding or lack of grounding. With the advent of switching supplies and USB powered devices the presence of a good low impedance ground is a dying breed. USB interfaces often use ferrite beads in series with the ground lead to go to the chassis ground so at high frequencies you have no ground. This causes issues like bussing noise from digital communications.

Then as you found out it can get worse if there is no actual reference to a chassis ground. Then you can develop the dangerous potentials you have encountered. I do not have a good answer for preventing this since the design of other devices is out of our control. The best you can do is try to make your device as stable as possible and immune to noise.

Look at the PCB Layout section and see how we recommend how bypass caps are connected to the pins of the part and how it connects to the vias to the power and ground planes. Do not let the ground pin of the part tie to a top layer ground fill directly. Force the PCB program to suppress the fill so you only have one trace from the pin going past the bypass cap therefore making the bypass cap much more effective. Do this with all the bypass caps in your design.

Then look at your ground paths from the RCA jacks to the power input of your board. This is where I agree with KJBob that you need to think that current will be flowing and do not let it share a path with the current for the DSP circuit. Route it around and let it have its own path. That should help a little. With the extreme voltages you are seeing it may very well not be enough but it is worth trying to make the DSP circuit a little less sensitive to noise spikes.

Dave T