## Diode input protection when power rails have series diodes

From circuitbending to homebrew stompboxes & synths, keep the DIY spirit alive!

Moderators: Kent, luketeaford, lisa, Joe.

DozenCrows
Common Wiggler
Posts: 95
Joined: Sun Jun 04, 2017 5:28 am

### Diode input protection when power rails have series diodes

On a module circuit that has protection against power rail reversal (e.g. series diodes), if the inputs are protected with clamping diodes connected to the incoming power rails, that would provide a path for significant current to flow if the rails were reversed - right through the input protection diodes, and likely enough current to burn them.

So it seems to me that connecting input protection diodes to the power rails after their in-series reversal protection would be the thing to do - rail reversal is fully protected against, with the caveat that the input protection now clamps to one diode drop away from the original power rail voltages.

But what happens when the voltage on the input gets clamped? Current flows through the clamping diodes into the protected power rails, affecting the rest of the circuit. In addition, when the excess input signal passes the reverse breakdown point of the power rail protection diode, it will flow back to the power supply - something like this - http://tinyurl.com/yycwpxvx

My questions to the Wiggler DIY community are:

- Is this behaviour in the simulation close to reality?
- How significant are any effects on the rest of the circuit?
- Has anyone experienced this "in the wild"?

mskala
Super Deluxe Wiggler
Posts: 2654
Joined: Thu Mar 27, 2014 7:33 am
Location: Toronto
Contact:
Normally, clamping diodes for input protection are used with series resistors between the outside world and everything else, including the clamping diodes. When an overvoltage is clamped, the current has to go through the series resistor before it gets to the clamping diode and then to the power rail. That limits the current through the clamped input. As long as this current is less than what the circuit draws from its power supply anyway (usually an easy constraint to enforce), there will still be some current coming from the power supply too, and so the power rail can't be driven over voltage. The current through the clamped input only reduces the load on the power supply a little.

As for "when the excess input signal passes the reverse breakdown point of the power rail protection diode": that's never gonna happen. Before it did, you would have worse things to worry about. The reverse voltage for a typical power-protection diode like the 1N5818 units I usually use, might be 30 volts or so. In order to apply that to the series protection diode on the module's power feed, the input overvoltage would have to have already brought the module's internal power bus to the reverse voltage plus 12V, or 42V, and it would have to have done that through the series resistor, so the applied voltage might have to be in the hundreds of volts. By then it would already have destroyed all the ICs in the module, and maybe even burned the insulation in the input jacks.

Something to remember is that current flowing through the clamping diodes is not normal circuit operation. The circuit isn't meant to just keep working fine when those diodes kick in hard, nor if you hook it directly to 230V European mains power. The point is only that it ought not to be permanently damaged by a tiny static charge, or some yahoo plugging in a +15V feed from some other modular format or something.
North Coast Synthesis Ltd.
Latest Web log posting: Genome Jazz

guest
Super Deluxe Wiggler
Posts: 4492
Joined: Mon Aug 19, 2013 11:49 am
this is a really good point that i hadnt considered before. those signal input protection diodes arent going to do too much if there are powersupply reverse protection diodes. but, its good practice to put a current limiting resistor in series with signal input protection diodes (i typically use 10k, unless the circuit can allow for something higher). this keeps those diodes from experiencing excessive current and breaking.

for low power circuits, ive seen a similar thing happen, where an input signal overcomes the protection diodes, and starts powering the ICs. this can be used to your advantage in micropower circuits to have them turn on and perform a function at a given input signal level. but, for higher power circuits (even just a few milliamps) the current coming from those forward biased diodes will be small if there is a current limiting resistor, such that the rest of the circuit will be able to sink it.

for example, if you apply 100V to the input, and have a 10k resistor before the diode, you only have 10mA flowing. if your positive rail consumes more than 10mA, your voltage regulator will decrease the amount of current it supplies by 10mA, and everything will continue along about the same. there will be a slight bump in the supply voltage as the regulator tries to compensate.

[EDIT] i put this together, and then noticed mskala had just posted the same thing. but, it was alreayd typed up, so why not. heres some redundant information.
openmusiclabs.com

mskala
Super Deluxe Wiggler
Posts: 2654
Joined: Thu Mar 27, 2014 7:33 am
Location: Toronto
Contact:
guest wrote:i put this together, and then noticed mskala had just posted the same thing. but, it was alreayd typed up, so why not. heres some redundant information.
They do say great minds think alike. Unfortunately, they also say fools seldom differ...
North Coast Synthesis Ltd.
Latest Web log posting: Genome Jazz

Synthiq
Veteran Wiggler
Posts: 590
Joined: Mon Feb 06, 2017 3:10 pm
Location: California
Most, but not all, opamps have have internal protection diodes to positive and negative supply that can handle at least 10mA. So from that perspective the circuit can in theory handle 1000V at the input with a 100kohm input resistor without diodes directly at the input.

In this case with a gain of -1, more than 30V can be applied to the external input before the inverting input would even go outside the supply rails and the internal protection diode would start clamping. So as mskala and guest has already said, already at 30V the supply to the opamp would exceed the maximum voltage rating for many amplifiers and could be destroyed without a current limiting resistor at the input so personally I would avoid these diodes and if I want to protect the input from static electricity I would instead add a bidirectional transient voltage suppressor diode to ground at the input to leave the supply voltages undisturbed.

mskala
Super Deluxe Wiggler
Posts: 2654
Joined: Thu Mar 27, 2014 7:33 am
Location: Toronto
Contact:
Synthiq wrote:Most, but not all, opamps have have internal protection diodes to positive and negative supply that can handle at least 10mA.
That's very common for digital chips, but I think it's uncommon for op amps. The TL074 has no such diodes in its data sheet schematic, and specifies its absolute maximum input range as being from negative supply -0.3V to negative supply+36V - notably, not related to the positive supply at all. I think the range is determined by the reverse-biased junctions of the JFETs at the input, so it can go above positive supply (maybe not usefully, but no damage and no significant current flow) until the point where those junctions break down. The 741 specifies the absolute maximum range for the inputs as equal to the supply voltage range, but it too has no protective diodes shown in the data sheet schematic; those inputs go directly to the bases of NPN transistors.
North Coast Synthesis Ltd.
Latest Web log posting: Genome Jazz

Synthiq
Veteran Wiggler
Posts: 590
Joined: Mon Feb 06, 2017 3:10 pm
Location: California
mskala wrote:
Synthiq wrote:Most, but not all, opamps have have internal protection diodes to positive and negative supply that can handle at least 10mA.
That's very common for digital chips, but I think it's uncommon for op amps. The TL074 has no such diodes in its data sheet schematic, and specifies its absolute maximum input range as being from negative supply -0.3V to negative supply+36V - notably, not related to the positive supply at all. I think the range is determined by the reverse-biased junctions of the JFETs at the input, so it can go above positive supply (maybe not usefully, but no damage and no significant current flow) until the point where those junctions break down. The 741 specifies the absolute maximum range for the inputs as equal to the supply voltage range, but it too has no protective diodes shown in the data sheet schematic; those inputs go directly to the bases of NPN transistors.
It is correct that the TL074 schematic from TI doesn't have any protection diodes, but in the maximum ratings section of the spec it states that the maximum input voltage must never exceed the magnitude of the supply or 15V, whichever is less. Allowing 0.3V outside the supply is typical when protection diodes are in place. It is just hard to believe the input jfets are robust enough to handle 2000V ESD just by themselves.

The MCP600X series from Microchip only have diodes to VSS but there they recommend adding diodes to VDD if needed.

DozenCrows
Common Wiggler
Posts: 95
Joined: Sun Jun 04, 2017 5:28 am
Thank you for both explanations - it's helpful to get two variant explanations!

mskala
Super Deluxe Wiggler
Posts: 2654
Joined: Thu Mar 27, 2014 7:33 am
Location: Toronto
Contact:
Synthiq wrote:It is correct that the TL074 schematic from TI doesn't have any protection diodes, but in the maximum ratings section of the spec it states that the maximum input voltage must never exceed the magnitude of the supply or 15V, whichever is less.
Where? I was looking at the data sheet from http://www.ti.com/product/TL074, which specifies on page 10 the absolute maximum ratings I mentioned: from -0.3V to +36V all referenced to the negative supply, notably not related to the positive supply voltage at all. That's why I think whatever ESD protection it's using is not the clamping-diodes-to-supply-rails method we see on digital logic chips.

The "recommended operating conditions" section on the same page says common mode voltage can be from 4V above negative supply, up to positive supply. That's reasonable because it would be the range over which the circuit can actually work properly. You need the junctions reverse-biased (so input has to stay well above negative power) and you need the differential pair to stay between the power rails so that later stages of the circuit will work. But the "doesn't work, not damaged either" range is wider and seems to go pretty far above the positive power rail when the supplies are +-12V.
North Coast Synthesis Ltd.
Latest Web log posting: Genome Jazz

Synthiq
Veteran Wiggler
Posts: 590
Joined: Mon Feb 06, 2017 3:10 pm
Location: California
mskala wrote:
Synthiq wrote:It is correct that the TL074 schematic from TI doesn't have any protection diodes, but in the maximum ratings section of the spec it states that the maximum input voltage must never exceed the magnitude of the supply or 15V, whichever is less.
Where? I was looking at the data sheet from http://www.ti.com/product/TL074, which specifies on page 10 the absolute maximum ratings I mentioned: from -0.3V to +36V all referenced to the negative supply, notably not related to the positive supply voltage at all. That's why I think whatever ESD protection it's using is not the clamping-diodes-to-supply-rails method we see on digital logic chips.

The "recommended operating conditions" section on the same page says common mode voltage can be from 4V above negative supply, up to positive supply. That's reasonable because it would be the range over which the circuit can actually work properly. You need the junctions reverse-biased (so input has to stay well above negative power) and you need the differential pair to stay between the power rails so that later stages of the circuit will work. But the "doesn't work, not damaged either" range is wider and seems to go pretty far above the positive power rail when the supplies are +-12V.
This is interesting. I looked in a datasheet I downloaded some time ago so it was the previous version (SLOS080M) and the maximum ratings are clearly different between the revisions. I also looked at an old datasheet from ST with the same limits as my old TI datasheet. Either they were very conservative when they first wrote the spec or the design has changed lately. So you might be right here.

mskala
Super Deluxe Wiggler
Posts: 2654
Joined: Thu Mar 27, 2014 7:33 am
Location: Toronto
Contact:
These companies love changing designs behind the scenes without telling the general public (maybe they issue change notices to their direct customers), and second-sourcing is another wildcard. All kinds of variation in the detailed specs of stuff like 78xx voltage regulators depending on exactly which data sheet you read.
North Coast Synthesis Ltd.
Latest Web log posting: Genome Jazz

Synthiq
Veteran Wiggler
Posts: 590
Joined: Mon Feb 06, 2017 3:10 pm
Location: California
It looks we were both partially right and wrong. I just measured a TL074 from TI and a TL064 from ST and they behaved the same. I shorted Vdd to Vss and then measured input current versus voltage and there is definitely a diode to Vss but NOT to Vdd. For positive voltages up to around 40V there was no input current. Above that I saw a spike in input current every time I increased the voltage by another volt up to 45V (my maximum) but each time the current then fell back to a few uA only. This may be a ESD protection that only responds to fast edges or there may be a static level above the voltage I could generate where the signal is clipped but that is just speculations.

mskala
Super Deluxe Wiggler
Posts: 2654
Joined: Thu Mar 27, 2014 7:33 am
Location: Toronto
Contact:
It's interesting to get those test results. I think that diode to Vss is probably the junction in the JFET. In normal operation it's reverse-biased, but if you drive the input below the negative power rail, it'll become forward-biased.
North Coast Synthesis Ltd.
Latest Web log posting: Genome Jazz

Synthiq
Veteran Wiggler
Posts: 590
Joined: Mon Feb 06, 2017 3:10 pm
Location: California
mskala wrote:It's interesting to get those test results. I think that diode to Vss is probably the junction in the JFET. In normal operation it's reverse-biased, but if you drive the input below the negative power rail, it'll become forward-biased.
It is possible, but I would then expect one diode drop from gate to source or drain and at least one other diode drop from source/drain to the supply but only saw 0.9V at 10mA. A good picture of the chip would help to resolve this but I have not been able to find one.

neil.johnson
Super Deluxe Wiggler
Posts: 1404
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

Synthiq
Veteran Wiggler
Posts: 590
Joined: Mon Feb 06, 2017 3:10 pm
Location: California
Thanks for the link to that TL072 die photo. It sure doesn't look like the inputs are connected to anything but the jfet gates. It's never too late to learn new things!