Phantom power protective diodes

[warpie]

I've seen some circuits that suggest the installation of diodes on the output in order to protect the ICs (like the 1646 or the drv134) from accidentally plugging in 48V to these outputs. What diodes are appropriate for this task? For example, THAT recommend the use of 1n4004 but would 1n4007 also be OK? I have a bunch of 4007 and none of 1n4004 so I was wondering whether I could use the 4007 instead.

 

[Brian Roth]

4007 diodes should be fine.

Bri

 

[warpie]

Thank you Brian. I thought so but my OCD kicked in and I thought I'd double-check by asking here.

 

[Ike Zimbel]

I make a little circuit board that mounts a 1646, four diodes (for protection) and a couple of caps and I have used the 1N4007 on it right from day one. Never had any problems with them.

 

[warpie]

I make a little circuit board that mounts a 1646, four diodes (for protection) and a couple of caps and I have used the 1N4007 on it right from day one. Never had any problems with them.

Yeah but have you tried feeding the 1646 with phantom power?

 

[Ike Zimbel]

There's a few dozen of them out there in the wild and I've never had one come back. I forget where most of them went, but 16 of them are direct outputs in a Yamaha M-1516.

 

[Admiral-dk]

I don't know if you will run into Noise problems with the 1N400x series (last digit only has to do with Voltage) ....
But I do know that 1N4148 can be problematic ...!!!!
The best I have used are BAT85 / BAS85 (different house) and a good number of other Small Signal Schootky Diodes in the BAT/BAS Series are really good for reverse protection (48V being higher than the internal Voltage) - having extremely small Reverse Current -> Low Noise.

Per

 

[warpie]

I don't know if you will run into Noise problems with the 1N400x series (last digit only has to do with Voltage) ....
But I do know that 1N4148 can be problematic ...!!!!
The best I have used are BAT85 / BAS85 (different house) and a good number of other Small Signal Schootky Diodes in the BAT/BAS Series are really good for reverse protection (48V being higher than the internal Voltage) - having extremely small Reverse Current -> Low Noise.

Per

Thank you.

That's interesting. So, you mean that the protective diodes can actually inject noise (from the +/-18VDC) to the audio output?

 

[Admiral-dk]

To be completely honnest - I do not see this as a big problem on the Output (as we are discussing here) - though this might be dependent on the actual circuit.
BUT - if you use the same diodes on the Input side, you might be in trouble .... again other things can influence the result.
I have been involved in designing a number of Models for known Guitar Pedal brand, and we found out that you do NOT place 1N4148 on the Input side - unless you don't care about the added noise. Placing either BAT/BAS85, BAT54 (all versions of them) in the same position, adds no extra noise we could measure or hear. ....

And just to be clear - yes we are talking about two diodes in series, with one Cathode connected to the Negative Rail (or on most guitar Pedals - GND), the Midpoint @ Input, and the last Anode connected to the Positive Rail.

No question, that the fact that those Pedals have approx. 1M Ohm Input Impedance and the Instrument has a Source Impedance of many K-Ohms - make the problem bigger than a 600 Ohm Line Level situation.

Per

 

[KA-Electonics.com]

I don't think you're going to have noise issues using 1N400X on a THAT1646/DRV134 output but the 1N400X "GP" glass-passivated are better diodes.

I've recently found phantom input protection using 1N400X diodes with some of the BI-CMOS process mic preamps aren't fast enough to fully protect them.

I haven't tested through-hole options yet but a number of the ultra-fast non-Schottky SMT options have low reverse leakage. The RS1D works well - the lower voltage 100-400V parts are slightly faster with a trr of 150 ns than the >400V parts. The RS1D's Ir is 5 µA but that's at rated voltage: At 15-18V it's not significant and initial tests show them to be quiet.

In through-hole I'm tempted to try the UF400X-series.

 

[Admiral-dk]

The BAT/BAS85 and BAT54 series has a trr of 4nS - hard to beat.

I completely forgot to mention that you need a series Resistor, between the Connector and the Diodes - to protect the Diodes and the rest.

 

[Bo Deadly]

You might consider TVS diodes between signal lines and pin 1 instead. There really is no reason to touch the rails at all. TVS diodes used to be quite high capacitance. But because of DSL modem tech, there are really low capacitance ones now.

For example, a quick search found SP4022 which handles 15A in under 10us, reverse leakage is 0.1 uA and capacitance is 2pF.

But I have to wonder if maybe the matching of a dual package would be good enough to justify using a deliberately high capacitance TVS diode in a high CMMR application like a mic pre. You could literally replace 4 parts with 1 for pennies.

And again, the rails don't have to be routed out to the input. I always thought that was bananas. I really don't see why people insist on running diodes to the rails anymore.

 

[KA-Electonics.com]

The BAT/BAS85 and BAT54 series has a trr of 4nS - hard to beat.

I completely forgot to mention that you need a series Resistor, between the Connector and the Diodes - to protect the Diodes and the rest.

For mic preamp phantom protection they do not appear to have sufficient forward current capability or reverse blocking voltage.

 

[KA-Electonics.com]

And again, the rails don't have to be routed out to the input. I always thought that was bananas. I really don't see why people insist on running diodes to the rails anymore.

TVS have improved but do not fully protect the device when it is powered off and the supply rails are at 0V.
The 48V or other input fault could come from another source while the "protected" device is powered down.
The TVS clamp voltage then exceeds the device supply voltage.
You need diodes to the rails under that fault condition to reduce current flow through the internal input ESD diode and steer it around the device input to the supply pins.

"I really don't see why people insist on running diodes to the rails anymore."
"I always thought that was bananas."
It's called an ESD diode and its internal to most modern devices.
Some fault conditions require external help in parallel.

 

[Bo Deadly]

TVS have improved but do not fully protect the device when it is powered off and the supply rails are at 0V.

My understanding is that a TVS diode would absolutely protect the device when it is powered off. So I would like to know more about what you're saying.

Consider a typical balanced output. A TVS diode would be connected directly between each signal wire and pin 1 directly at the entrance to the device. When a surge on one or both of the signal wires occurs, the TVS diode clamping voltage will be exceeded and direct current to pin 1 of the cable and away from the protected device entirely.

Consider a typical phantom powered balanced mic input. Again, TVS diodes (albeit with a higher clamping voltage to accomodate 48V) would direct surge currents from pins 2 and 3 to pin 1 directly and away from the device.

These diodes should work equally well with the device powered on or off.

What am I missing??

 

[KA-Electonics.com]

When the device is off it's supply pins are at 0V and are decoupled (hopefully) by some electrolytic caps.
When potential is applied to the input current is going to flow through the input, usually to a supply pin, and then to the decoupling caps and supply rail through the internal ESD diodes and input circuitry.
This occurs whether or not the TVS is conducting.

Let's say the TVS clamps at 10V, there's no internal ESD diode in the IC, and the IC has a maximum input voltage that cannot exceed the supply pin voltage by a few hundred mV if any.
When an input fault occurs it can exceed the supply pin absolute maximum by 10V before any protection is offered by the TVS.
There's usually some internal junction that limits it to less than that - and that node is going to have to handle the fault current from the input to the device's unpowered rail.
A diode steers that current around the device and parallels the internal ESD diode when the device has one.

 

[JohnRoberts]

TVS have improved but do not fully protect the device when it is powered off and the supply rails are at 0V.
The 48V or other input fault could come from another source while the "protected" device is powered down.
The TVS clamp voltage then exceeds the device supply voltage.
You need diodes to the rails under that fault condition to reduce current flow through the internal input ESD diode and steer it around the device input to the supply pins.

"I really don't see why people insist on running diodes to the rails anymore."
"I always thought that was bananas."
It's called an ESD diode and its internal to most modern devices.
Some fault conditions require external help in parallel.

Indeed modern chip manufacturers generally include protection incorporated into the silicon.

Speaking about THAT corp, I had some discussions with them about tooling a custom metallization layer to reconfigure one of their I/O chips to use the existing precision resistors to make a high compliance AC current source. The project was dismissed because my proposed new configuration needed one more I/O pin, then they had protection wells already integrated into the existing IC. A metallization layer mask change is relatively cheap compared to starting over from scratch for the whole IC.

IC makers are pretty good at protecting their chips from their customers, but customers keep evolving to find new ways to blow stuff up.

JR

 

[Bo Deadly]

Let's say the TVS clamps at 10V, there's no internal ESD diode in the IC, and

Ok, so are you only talking about circuits that do not have protection diodes like those used in most ICs?

I must admit that I had not fully considered the power off scenario. I will have to think about that more.

But it seems to me that there are two levels of protection. You have pn junction failure caused by over voltage with little to no current. Then you have failures caused by high currents burning things up such as from phantom power being applied incorrectly or lightning.

 

[JohnRoberts]

Over the decades I've used truckloads of protection diodes to PS rails.

Since value customers abhor turn-on/turn-off transients I traded some output source impedance for quieter on/off. These discrete JFET shunts suffered failures from static spikes even with clamp diodes.

Caveat this is all last century experience... So I do not know current practice at my old gig.

JR

 

[Admiral-dk]

OK - I should then mention that some of the new High Quality OpAmps has some very good Build-In protection - like TI's OPA1678 .... Unfortunately it hasn't been in stock since the Chip Chrisis started during Covid

I also has quite a bit of experience with TVS Diodes - most of them NOT good - Here it is Very Important to notice, that this is not the TVS to blame - but the designer, who choose to install ones that where too small.
Later they got changed to some with Much bigger capacity (not capacitance, but Current/Energy) - Easy to see, as they are in a house more than a factor 100 bigger in volume. After that I only saw one I had to change ...!
All where placed across the pins on a XLR DI Output - and I'm pretty sure they died when the House Mixer and the Bass Amp on Stage, where on different Phases (+ likely other faults) in the Mains Installation => 400Vac + Amps ....
About half the time, I saw the OpAmps had survived as the TVS stayed permanently Shorted after the Fact - and the other half had Visible Fault-Finding (ie. burned holes in the plastic over the chip). The major good thing, was that with one exception, the rest of the Bass Amp survived.