THS4531: Glitching observed at -40 C ambient

Hello Bhav,

I can help investigate this behavior with you.

My first request would be to see if a known good unit exhibits the glitch on a 'known issue' board; similarly if a 'known issue' IC exhibits no issue on a 'known good' board.  This A-B swap would help understand if rework/handling and IC/board variance are to be considered.

15% is a fairly high rate for a non-ideal behavior to occur; could you share any layout or PCB files?  I can message you privately on e2e to exchange emails if you prefer.

My first thought would be if there has been any issues with storage/moisture for the units?  Older units or units which have had exposure to moisture can exhibit odd behavior at the temperature extremes (for operating temperature).  

Whatever the cause may be, I think we need more information to help out.  If you can check the A-B-A swap and also work with me on more design details (like PCB), we can progress on resolving this behavior.  If needed, we can consider additional options for official return, good-faith shipment to TI (my office) for debug, and debug/analysis via email/exchange of information.

I will also check across my team to see if any cold temp glitches are known about for either this device family or the process technology.

Best,

Alec

Hi Alec,

Thank you for your prompt response.

Have you had any feedback from your team regarding cold temperature glitching? Have any of your customers experienced this before?

Regarding sharing the layout, we are discussing it internally and we will confirm as soon as we get authorisation to share.

Since my initial post, we have tried the following (all tested as per original schematic, unless otherwise stated):

1. Removed output connection to 'another TI IC' and substituted it with 40 k differential resistor - No improvement

2. Removed feedback capacitors - Glitching disappeared

3. Changed the supply to +/- 2.5 V - No improvement

4. Probing single-endedly with a standard 10 x passive probe directly on input or output pins - Glitching disappears

5. Adding 22 pF in parallel with R4 - Glitching disappeared

Look forward to hearing from you.

Many Thanks,

Bhav

Hello Bhav,

Thank you for the update.

Do you see the glitching if you proportionally reduce RF and RG resistors (14k and 10k respectively) to keep same gain configuration?  Sometimes in FDA circuit the high resistor values can increase noise contributions and effects.

You do not need to do this if you need these specific values for any filtering, but for now I just observe a typical FDA configuration with feedback capacitors.

It is interesting adding 22pF in parallel with R4 removed the glitch; this is effectively decoupling higher frequency noise.  What are the target frequencies of operation for this design? 

Best,

Alec

Hi Alec,

Thanks for your speedy response.

We have not tried reducing RF and RG resistor yet, but will do.

This in an audio application and this stage requires a ~16 kHz cut-off frequency.

One thing we noticed when cooling the part is the drop in output common-mode from 1.25 V to ~0.9 V.  Is this what you would expect to see?

Many Thanks,

Bhav

Hello Bhav,

The data point of output common-mode voltage changing over temp is good to know.  I will check with the team on this, but it would be expected for VOCM to stay at the level set by the VOCM pin, due to the internal output common mode buffer.

Can you confirm the IQ of the FDA at cold temp?  I also wonder if the output load remains consistent over temperature.

Regarding your debug step #4 above: when you probe each output there is no glitch, how are you measuring the FDA outputs when you do observe the glitch?

Best,

Alec

Hello Alec, I work with Bhav who is on vacation today. As this problem is quite serious and we are on ship hold, I wanted to answer your questions now and not wait till next week.

1. As we use power planes, we are not able to measure IQ without lifting the power pin. We were concerned that this would them change the chip behaviour. We can measure IQ it it will help you get an answer.
2. The probing is done after the 16k resistors. Apologies we should have explained that. We have disconnected the other TI chip and used an equivalent load resistor, but see no difference in the behaviour, so do not suspect the other TI chip to be cause of the problem.
3. We definitely see the "output" common mode change with temperature. It is ~1.25 V at room temperature and dips down to 0.9 V at cold and then recovers, ie back to 1.25 V at the temperature is increased back to ambient. The VOCM pin remains at 1.25 V. Looks like the common mode voltage control inside the chip is temperature dependent.

Please let us know if we can provide any other information.

Thanking you for your continued support

Dipak

Hi Alec,

One correction to the information above;

The VOCM input and Output Common-mode does not change with cold temperature and remains at 1.25 V.

The observation of the ~0.3 V shift was isolated to a single-board and could not be seen on the few others we've since tested.

Any initial thoughts from you or your team?

Many Thanks,

Bhav

Hello Bhav,

Thank you for the update.  It is good to confirm VOCM input and output common mode voltage do not deviate from 1.25V.

Initial thought: the THS4531 outputs may be oscillating/unstable due to driving the output capacitors.  The stability issue may only present itself at cold temperatures.

A way to check this would be to add small series resistance at the outputs of the FDA, providing isolation resistance between the output impedance of the FDA and the capacitors.  This resistance could be in a range as described in Section 9.2.8 Driving Capacitive Loads in the THS4531 datasheet.

You could also check this by removing the capacitive loading if you are unable to add small series output (isolation) resistors at OUT+ and OUT-.  When you probe the device pins directly and notice the glitch disappear, the inductance of the probe may be offsetting the load capacitance and changing the amplifier's stability.

Best,

Alec