Which schottkys? I have seen cases where leaky schottky diodes caused more damage during power on/off than they helped prevent.
The data sheet shows clamping schottkys on the inputs so if these don't work, it would be a reason to avoid ESS altogether.
BAT54SW. But I doubt the damage was caused by inputs as at least one chip failed without any input voltage.
Silly way to comment. Why not just tell us what you use.
T/H part is SR108-TP which was mostly used to protect Arduino I2C pins. Never had an Arduino fail with those clamp diodes. Some other schottky diodes caused immediate Arduino failure at turn on. Problem was reverse leakage current, which allowed the pin to receive Arduino Vcc faster than the Arduino could power up internally. Checking diode reverse leakage with a Fluke 87 showed measurable resistance. OTOH SR108-TP reverse leakage measured as out of range, unmeasurable.
Anyway I have BAT54S in my schematic, to clamp the single-ended inputs within the op-amp supply voltage (+/-7V). And the gain of the SE-to-diff converter is about 0.21 so it will be around +/-1.5V max on the input pins. And as I mentioned above, ES9820 datasheet shows internal clamping diodes on inputs anyway.
It shouldn't need to be said, but if anyone is not using proper ESD precautions then that would be the cause of failures.
It shouldn't need to be said, but if anyone is not using proper ESD precautions then that would be the cause of failures.
So as I understand it you have protection diodes on the op-amp inputs.
Do you also have protection diodes directly on the ADC inputs (after the op-amps)? Preferably with a voltage clamp to avoid passing high currents to the supply, which may increase the supply voltage, perhaps beyond the allowed maximum voltage. Op-amps can have overshoot on the outputs during power on/off and it can have voltages close to the op-amp supply voltage in case of DC on the input.
Do you also have protection diodes directly on the ADC inputs (after the op-amps)? Preferably with a voltage clamp to avoid passing high currents to the supply, which may increase the supply voltage, perhaps beyond the allowed maximum voltage. Op-amps can have overshoot on the outputs during power on/off and it can have voltages close to the op-amp supply voltage in case of DC on the input.
These are the internal ESD diodes. They may not be able to handle the 60-80 mA you can get from e.g., the OPA1612. OPA1612 has a typical short circuit current of +55/-62 mA. So external Schottky diodes may be needed. And if you connect one of the diodes to VDD you may increase the supply voltage, because the short circuit current is higher than the current consumption of the ADC. This may destroy the ADC.
An opamp output briefly hitting 7v may not current limit until, say, maybe > 50mA is flowing. At 50mA, BAT54 may have a forward drop at room temperature of around 450mV, which could be enough to overcome some of its its protective effect. Sometimes it is recommended to limit IC over-voltage to 300mV maximum.
So we need to think about a combination of things possibly including: clamp diode forward voltage under fault current conditions, reverse leakage effects, and maybe in some cases, diode junction capacitance. Unfortunately low schottky diode forward voltage ratings tend to go with high leakage current ratings. Using a larger low leakage junction may help lower forward voltage, but then capacitance tends to go up.
Note: Was in the process of typing the above when JensH replied first. Oh, well.
So we need to think about a combination of things possibly including: clamp diode forward voltage under fault current conditions, reverse leakage effects, and maybe in some cases, diode junction capacitance. Unfortunately low schottky diode forward voltage ratings tend to go with high leakage current ratings. Using a larger low leakage junction may help lower forward voltage, but then capacitance tends to go up.
Note: Was in the process of typing the above when JensH replied first. Oh, well.
Last edited:
IME protection circuitry does audibly affect the sound of 24-bit data converters. However, a little loss of performance in return for better reliability seems sensible.
That's more or less what AKM seemed to be saying about audibility of I/V opamp output protection for AK4499. Rather than use protection diodes, they recommended to power on the opamps only after the DAC chip was powered up, and to power down the opamps first at shutdown. They also said that if power could not be sequenced then use protection diodes.
That said, ES9820 doesn't appear to be as opamp output voltage tolerant as AK4499 was.
That's more or less what AKM seemed to be saying about audibility of I/V opamp output protection for AK4499. Rather than use protection diodes, they recommended to power on the opamps only after the DAC chip was powered up, and to power down the opamps first at shutdown. They also said that if power could not be sequenced then use protection diodes.
That said, ES9820 doesn't appear to be as opamp output voltage tolerant as AK4499 was.
You cannot really compare an ADC to a DAC. The ADC will have the inputs driven directly from the op-amp or perhaps with small series resistors. The DAC (e.g., the AK4499) will typically look into the input of the op-amp + the feedback resistor. But yes, AKM had some notes about the AK4499 and protection.
Thanks all for these excellent tips. I am learning a lot... those documents from bohrok2610 are particularly useful since they explain it all in detail. The reference circuit in the ESS datasheet (which I mostly copied) has a 36R series resistor to limit current from the op-amp, but this is not sufficient if the op-amp output is the full 7V. I need to add BAT54s on the ADC inputs as suggested. Let me study these documents a bit more 
Another question... is it worth putting an LC filter on the SMPS outputs which feed the op-amps to reduce noise? I was thinking a 3x3mm 47uH coil with 100uF electro to make cutoff frequency ~2.3kHz. I have done a new board layout where I implemented the suggestions from before ... the separation between the analog and SMPS is much improved, the SMPS & related components are on the bottom, and the ADC regulator is on the top layer. There is enough room for these LC filters with this new layout.
Another question... is it worth putting an LC filter on the SMPS outputs which feed the op-amps to reduce noise? I was thinking a 3x3mm 47uH coil with 100uF electro to make cutoff frequency ~2.3kHz. I have done a new board layout where I implemented the suggestions from before ... the separation between the analog and SMPS is much improved, the SMPS & related components are on the bottom, and the ADC regulator is on the top layer. There is enough room for these LC filters with this new layout.
Good news... the datasheet specifies that the ADC inputs will tolerate +4.5V. So it should be reasonable to use BAT54s for clamping, as they should start to activate around 3.8V.
The maximum rating for VDD is +3.6V though, so really I don't want to clamp to that rail (this is quite small tolerance, and might be why these chips die). Ideas are welcome!
The maximum rating for VDD is +3.6V though, so really I don't want to clamp to that rail (this is quite small tolerance, and might be why these chips die). Ideas are welcome!
There are various ways to protect against overvoltage conditions. An old idea is the precision crowbar: https://axotron.se/index_en.php?page=26
There are voltage monitor chips such as: https://www.ti.com/lit/ds/symlink/tps3702.pdf?ts=1664554327788&ref_url=https%3A%2F%2Fwww.google.com%2F ...although you have to decide what you want it to do if an overvoltage condition is detected.
Such things add complexity and cost however. More time could be spent looking into options if you are interested.
Regarding an output filter for SMPS, are you going to use an LDO to clean up the SMPS output? Thought about leaving room for shielding in case it turns out to be needed? https://www.mouser.com/new/harwin/harwin-ez-emi-rfi-shield-cans/
There are voltage monitor chips such as: https://www.ti.com/lit/ds/symlink/tps3702.pdf?ts=1664554327788&ref_url=https%3A%2F%2Fwww.google.com%2F ...although you have to decide what you want it to do if an overvoltage condition is detected.
Such things add complexity and cost however. More time could be spent looking into options if you are interested.
Regarding an output filter for SMPS, are you going to use an LDO to clean up the SMPS output? Thought about leaving room for shielding in case it turns out to be needed? https://www.mouser.com/new/harwin/harwin-ez-emi-rfi-shield-cans/