INA1650: INA1650 - one channel damaged after 1-2months when connected with Eversolo DMP-A8

Hello Mateusz,

Thank you for the detailed description of your setup and the troubleshooting steps you have already performed. Based on the information provided, the behavior you are seeing is unusual but there are a few things that may help identify the root cause.

One important observation is that the issue appears to occur only on the right output channel of the Eversolo DMP-A8, while the left channel continues to operate normally. Since the INA1650 devices function correctly again after replacement, this suggests that the input stage of the affected channel may be experiencing a stress condition coming from the connected source.

Although the INA1650 includes internal protection structures, repeated exposure to large input transients or excessive input current can potentially degrade the input stage over time. The symptom you described (very low output level but not completely silent) is consistent with partial degradation of the input stage.

To help determine the cause, it would be useful to check the following:

1. Swap the left and right XLR connections from the DMP-A8 to the amplifier input board. If the problem moves to the other channel of the INA1650, this would indicate that the source output is likely responsible.
2. Check for power-up or power-down transients on the right channel output if possible. Large spikes during startup or shutdown could stress the input.
3. Measure DC voltage on the DMP-A8 XLR outputs (with nothing connected):
   • Pin 2 to Pin 1
   • Pin 3 to Pin 1
   • Pin 2 to Pin 3
4. Ideally these values should be very close to 0 V (typically less than a few tens of millivolts).

As an additional test, you can try adding small series resistors (for example ~100 Ω) in line with each INA1650 input. These resistors can help limit surge current into the device during transient conditions. This test would be telling if the part survives longer than what you have observed. 

Example:

• XLR Pin 2 → 100 Ω → INA1650 IN+
• XLR Pin 3 → 100 Ω → INA1650 IN−

Your power supply implementations (±12 V or ±15 V using linear regulators or TPS7A39) should not cause this behavior, so the issue is more likely related to the signal interface conditions.

If you are able to provide the DC measurements from the DMP-A8 outputs or any scope captures of the signal during power-up, we would be happy to review them.

Best regards,

Regards,

Chris Featherstone

Hello Chris,

thank you so much for your reply.

1 - I have already checked and it same situation, low level signal. One channel of INA1650 is already damaged. So swaping cables changes nothing
2, 3, 4 - I will check next week.

Additional test looks ok, but I have also 10uF caps on input of INA1650. So it should like this:
a) XLR pin 2 or 3 -> 100R -> 10u -> input of INA1650

or

b) XLR pin 2 or 3 -> 10u -> 100R -> input of INA1650

minimelf 0,4w will be ok? 


I will provide scope captures, Im waiting for package with DMP A8


Have a nice weekend



Regards
Mateusz

Hi Mateusz, 

Ideally the resistor would go closest to the input.  XLR pin 2 or 3 -> 10u -> 100R -> input of INA1650. This way any transient current even from the cap would be limited. 

Have a nice weekend. 

Best Regards, 

Chris Featherstone

Hi Chris,

Tomorrow I will receive one of the A8 units mentioned earlier. Could you please let me know what measurements I should take?

I will go through your list and perform all the necessary measurements so that we can gather all the details needed to identify the problem.

For example:

1. A8 startup:

• A8 powered off

• probes connected to the RIGHT channel + and -

• power on the unit and record the signal without playing any music
  
  

2. Shutdown while the A8 is on:

• probes connected to the right channel + and -

• no music playing

• power off the unit and record the signal

If needed, I can also take measurements with the INA1650 in the circuit. Tomorrow I can build the same module as shown in schematic “b” (same PCB, same PSU).





One of the main measurements will be the one described in your third point:

Measure DC voltage on the DMP-A8 XLR outputs (with nothing connected):

• Pin 2 to Pin 1
• Pin 3 to Pin 1
• Pin 2 to Pin 3
  
  
  
  

Regards,
Mateusz

Hi Mateusz, 

Thank you for performing the measurements. Your proposed approach looks good. Below is a slightly refined set of measurements that would help us capture the most relevant information.

1. DC Offset Measurement (no connection to the amplifier) With the A8 powered on and no audio playing, please measure with a DMM:

• Pin 2 → Pin 1
• Pin 3 → Pin 1
• Pin 2 → Pin 3

If possible, please record these values for both the left and right channels.

2. Startup Transient Measurement (oscilloscope)

Setup: 

• CH1 probe → Pin 2
• CH2 probe → Pin 3
• Ground → Pin 1

Procedure: 

• A8 powered OFF
• Connect probes
• Power ON the unit and record the signal until it stabilizes.

3. Shutdown Transient Measurement

Using the same probe setup:

• A8 powered ON with no music playing
• Begin recording on the oscilloscope
• Power OFF the unit and capture the waveform during shutdown.

4. Optional signal verification

If convenient, you may also play a 1 kHz sine wave and observe the signals on Pin 2 and Pin 3. These should appear as equal amplitude signals with opposite polarity centered around 0 V.

If you are able to repeat the startup measurement with the INA1650 input board connected as well, that information could also be helpful.

The goal of these measurements is to check for any DC offset or large startup/shutdown transients that could stress the input stage over time.

Best Regards, 
Chris Featherstone

Hi Chris,

Unfortunately, it’s difficult to capture on the oscilloscope. Once, during power-off, something was visible. It was tested at 1ms, 5ms, and 10ms, with 50mV, 100 mV, 200 V, and 1V settings. Maybe you could suggest other settings that would help us detect it more quickly, if anything is actually present? Additionally, the trigger was set to detect the falling or rising edge.

What we did manage to measure was, of course, the DC offset, which looks quite interesting in the right channel:
a) LEFT
2 - 1 = -0,42mV DC
3 - 1 = 0,37mV DC
2 - 3 = 0,78mV DC

b) RIGHT
2 - 1 = -0,1mV DC
3 - 1 = 0,34mV DC
2 - 3 = -0,44mV DC

Results after powering on the Eversolo and waiting approximately 5 minutes. WIthout music.


(On the Eversolo, it was tested with the output port settings set to “BAL-XLR” and “XLR/RCA” - the measured values were the same in both cases.)

Additionally, one of the probes stopped working, so I’ll upload the full capture of both the positive and negative signals the day after tomorrow - a replacement probe has already been ordered. Today, I performed the measurements and tests using a single probe, switching it between the positive and negative signal of XLR while observing the behavior during power-on and power-off.

Photo shows a 1kHz signal from USB source:
image uploader stopped working so please check via this link -> https://ibb.co/6JvBvMY4

Regards
Mateusz

Hi Mateusz, 

You might try longer time scales such as 20us/div, 50us/div, and 100us/div (micro seconds) to see if you are able to capture any power up or power down transients. For the volts per division you may try 1,2, 5 V/division. I would try using the single shot mode on the scope and do the edge trigger on Ch1 pin2. If the scope supports math you can do Ch1-Ch2 in order to see the differential signal. 

Try power sequencing such as

• A8 off
• INA1650 Board off
• Turn on A8

Second test

• INA1650 board on 
• A8 off
• Turn A8 on

Is the XLR cable ever connected and disconnected while the devices are powered on? This may be another potential cause for transient spikes. 

Best Regards, 
Chris Featherstone

Hi Chris, 

thanks to a friend’s help and his 4-channel oscilloscope, we managed to take some more measurements.

The first measurement is of the left and right channels, with no oscilloscope ground connected - meaning only the “+” and “-” of both channels were connected to the probes, without the oscilloscope ground. The measurement was taken while plugging the Eversolo into mains power - the main 230 V supply. As you can see, there is a clear glitch on both channels at around 12V. (5V/div)

The second measurement was taken with the oscilloscope ground connected to pin 1 of the XLR. The same test was performed - switching on the 230 V power using the switch on the back of the Eversolo unit. Scale: 2 V/div. There is less of it.

So far, there are no additional effects when turning the unit on or off using the front button (the encoder). It is possible that the problem in the right channel occurs, for example, only once every 100 power cycles, and then there is one strong spike.

Pin 1 of the XLR in the Eversolo is connected to... the RCA ground - in other words, to the ground of the entire power supply (not PE - chassis, just shorted withour additional resistor to RCA GND) and this GND also without additional resistor or RC to PE - chassis.

To keep all the information in one place: pin 1 at the input of my modules is always connected either directly to the chassis PE, or to PE through an RC network (10Ω + 100nF).


Regards

Mateusz

Hi Mateusz, 

Thanks for providing the information. I am not sure if a revision can be done on your board but if there is the possibility of large transient spikes that can cause electrical overstress we generally recommend using TVS diodes to clamp the inputs. We have a video training series that provides guidance on selecting these diodes that can be found under the electrical overstress section in the link below:

https://www.ti.com/video/series/precision-labs/ti-precision-labs-op-amps.html

Best Regards, 
Chris Featherstone