FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 

Reverse Parallel Diode Question
MUFF WIGGLER Forum Index -> Music Tech DIY  
Author Reverse Parallel Diode Question
So I see this all over Erica synths DIY modules and wondering if someone could explain what this does:

One example is the simple mixer module:

these are protection diodes. if the input voltage goes beyond the rails, the diodes conduct so the opamp (or whatever might follow) doesnt get over-voltaged.
As always, thank you guest.

Are they always recommended? Just as a pre-caution?
cackland wrote:

Are they always recommended? Just as a pre-caution?

They are rarely used. More often they are used to prevent negative voltages (eg. so you can stick an LFO into a gate input), or to protect pedals from wall-warts from output plugs wired the opposite way.
for an inverting amplifier, i rarely use them, as there is typically a large value resistor there anyways. most devices are internally protected as well. so it really depends. the LM339, for example, has phase reversal if the input goes below the negative rail. you can check the datasheet for your parts, and see what sort of voltage/current the inputs can handle, and whether it has protection. some are only protected to one rail, and not the other. both the diodes and series resistor do the protection.

example: you have a cmos device powered off of 0V/+12V and the input can handle 1mA max, and youre using it with a +/-12V system, you should make sure there is a 12k resistor at the input if youre not using protection diodes. if you do put protection diodes in, that resistor can drop to the value of whatever current the protection diodes can handle. but thats only for a mis-patched situation. there is also ESD, which can be in the kV range, but very small charge. adding a bit of input capacitance helps here to absorb the charge. the reverse biased input diodes can act as small capacitors. shottkeys have higher capacitance - 100x, and have a lower forward voltage, so they make good protection diodes. they are extremely leaky, though, so not so great for high impedance circuits, low power designs, or low noise designs.
These are strongly recommended if your input is piezo device.
Keeps the -12-diode drop < signal > +12 +diode drop.
input/output protection, an old and sometimes philosophical discussion ;-)

output diodes: they protect the opamp output if your synth is switched off (+-0V supply instead of +-12V supply) and someone plugs another output to this output. Or the synth is on and someone connects a power amp output, because with normal module levels (+-10V) this diodes will never conduct.
Is this likely ? Is it necessary to protect dumb people from making mistakes without consequences ?

input diodes: in this configuration they start to conduct when all 3 inputs swing about [EDIT] 3x the rails (if power on). Where should these +-35V signals come from ? And even with a 100V signal there is no "burning" current with Rin =100k.

my opinion: forget these diodes
That particular pair seems useless because they're behind blocking capacitors, 100k resistors, and located at the virtual ground of an op amp. In order to exceed the power rails and turn on the diodes at that point, you'd need to apply so much AC voltage to the module inputs that it would destroy the connectors by arcing first. (EDIT: Okay, on further thought maybe not quite that much. But at least about 70V peak to peak, which is a heck of a lot.) In normal operation the diodes will just be reverse-biased, add to the parasitic capacitance on the op amp input, and harm stability (though honestly not enough to worry about).

The pair on the output may be less useless - someone could in theory apply an outside-of-power-rails voltage to the output so that they would have an effect - but I'd still be inclined to leave them out.
MW threads, always educational! Thanks all, I have learned something for the day. Guinness ftw!
Yes, appreciate the explanation from everyone.
If we have a chain of modules with different value caps on their +/-12V supplies, when we power up there is scope for one module's rails to be at +/-12V while the next is only at +/-1V say.

And the DC blocking capacitor on the signal input may not have kicked in, centering the AC.

So I guess these diodes are intended to reduce signal spikes at power-on on a module that has relatively larger power caps than the module feeding it? I know that the chip the Polyvoks used for VCA is very sensitive (I have blown several), so maybe it is a preventative design to prevent spike propagation, even if over-engineering?

(Also, note that the TL072 does not like signal more (negative than) -11V. )

Does that make any sense?
ricko wrote:
So I guess these diodes are intended to reduce signal spikes at power-on on a module that has relatively larger power caps than the module feeding it? I know that the chip the Polyvoks used for VCA is very sensitive (I have blown several), so maybe it is a preventative design to prevent spike propagation, even if over-engineering?

(Also, note that the TL072 does not like signal more (negative than) -11V. )

Does that make any sense?

It's not impossible, but I think it's a stretch.

The idea of power coming up one module at a time is a good one and with some circuits that's a significant issue - or one power rail coming up before another, which has killed a lot of 2164 VCA chips. But for it to happen from different capacitances in the modules, there'd have to also be a lot of resistance on the module's power connection too, in order for other modules to be up and running and putting out substantial voltage before its chips are powered up. Seems like a designer who was worried about that could avoid it in easier ways. Those also would have to be some very sensitive chips to be damaged by +12V when it's on the far side of a 100k resistor, even without also the DC blocking capacitors. Not TL072 op amps, which are so popular partly because they are hard to kill.

The issue with the TL072 and negative input voltages is that it goes into phase inversion for input voltages that are not above the negative power rail by at least about 4V. So it's not "-11V" necessarily - it would be that if the negative rail was at -15V, but on Eurorack power the magic number is -8V. And it goes into phase inversion when the input voltage goes too low, which means it produces the wrong polarity of output. That is not actually damaging to the chip. When the input voltage gets back into the required range it goes back to normal operation. You have to go further, outside the power rails, to do any real damage, and with the 100k resistors, you'd have to go pretty far on the voltage before the current could reach damaging levels.
just checked the 072 datasheet, and it doesnt have built in protection, per se. the input stage is a p-channel diff-pair, so you can take the input as far positive as the Vds breakdown, and in the negative direction it just starts conducting through the gate. Vin can be 36V above the negative rail, which means if you are using +/-12V rails, you can take the input to +24V without damaging it. on the negative side, the Vgs junction can conduct up to 50mA without damage, so its pretty robust on that side.
The way to clamp the -input of the amp is to connect the inverse diode pair across -in and +in. In normal use, the negative feedback from the output via the resistor should pull the -input so that it is always the same voltage as the +input. If the input signal or the amp gain is high enough for the output to reach either supply rail, then the feedback correction stops working and the -input voltage moves away from the +input voltage.

Even if AC coupled, there can be a fast pulse of energy from an accidental high voltage input. Silicon can switch fast, but it can also die fast. A premium approach would be to split the mixer input resistors into two and clamp the middle with a pair of diodes to the power rails for each input. The first half of the input resistor limits the fault current in the clamping diodes, the second half limits the current to the op-amp -input.

I'm not sure if the inversion when overdriven is an actively controlled condition, but simply that the op-amp output stage disconnects to a high impedance when told to go where it cannot and what you see on the output is the input signal bleeding thru around the amp via the feedback resistor. When that happens, it will no longer be inverted.
MUFF WIGGLER Forum Index -> Music Tech DIY  
Page 1 of 1
Powered by phpBB © phpBB Group