Series zeners around NE570 in Dimension C

[nickcordle]

Hi all,

I'm getting familiar with the NE570/v571d/etc finally and I notice that in the Boss DC-2 compression stage right after the emphasis filter, there are a few additional parts which aren't on the datasheet circuit.  Taken from the Aion Blueshift schematic for clarity:



And the relevant bit from the Philips NE570 datasheet:



Okay, so they added series-opposing 3.0V zeners and a 1n cap in parallel to the "Rdc" stuff on the datasheet circuit.  So ... shouldn't do much until Vout is > 3.7ish volts away from Vin- / Vref / 1.8V, and only then at somewhat high frequencies, right?  Visually it kinda screams defense, like, maybe the 570 dies if the + and - inputs get separated?  If so, I haven't found it in the datasheets.  Can anyone clarify exactly what requirement is served here?

Happy turkeys

N

P.S. preemptive thanks to Rob Strand and Mark Hammer whose prior posts have helped me much re. 570s already

[Rob Strand]

Okay, so they added series-opposing 3.0V zeners and a 1n cap in parallel to the "Rdc" stuff on the datasheet circuit.  So ... shouldn't do much until Vout is > 3.7ish volts away from Vin- / Vref / 1.8V, and only then at somewhat high frequencies, right?  Visually it kinda screams defense, like, maybe the 570 dies if the + and - inputs get separated?  If so, I haven't found it in the datasheets.  Can anyone clarify exactly what requirement is served here?

I looked at this yesterday.   It's a very odd connection which I haven't seen before.
What I haven't done is drawn out all the other pins to see if the rest of circuit follows the standard connections, like your app-note/datasheet extract.

Should have done that first.   See post below this.


Largely wrong:
The first thing that came to mind was is the PCB really like that?

If we accept the PCB is correct it's a puzzling circuit.  There are two weird connections.   The purpose of both seem to be related and my guess is they intend to solve some specific issue.

The two resistors and cap on pin 5 is a little different to the normal connection of a resistor to ground.   At first glance that seems to reduce the loop gain of the amplifier which could stop oscillations but with such a large cap we might expect the intent of the connection to be more than that.  Perhaps the connection prevent the voltage (AC+DC) on pin 5 from swinging too far, which would occur when the stage overloads.

For the zeners and cap.  Under normal conditions the circuit not going to have much affect.  Perhaps adding a bit of capacitance, which is undesirable if anything.   Under overload conditions it could clamp and spikes that develop on pin 5.

So where the circuit might come into play is under overload conditions.  The problem it solves still isn't clear.    Did Boss find the NE570 opamp gets damaged and goes noisy if the NE570 overloads?   Or does it make the sound of the NE570 less offensive under overload conditions?

Unfortunately you would have to do a lot of testing confirm all these "maybes".

One thing I know is the NE570 doesn't have some undesirable noise issues.   I came up with an odd NE570 connection many years ago which surprisingly reduces the noise.   The mods I did were around Vref and mechanism was that the NE570 would amplify the noise on its own reference voltage.   Bypassing Vref made the noise worse!   There's a weak possibility the odd connection in the DC-2 is related to that but I have a hard time accepting the zeners and caps would help with that issue- they point more to the overload theory mentioned above.

[Rob Strand]

I just drew out the circuit and I've misunderstood the connections.

It seems to clip high frequencies.    Much simpler than what I wrote before.

[nickcordle]

As far as I can tell, it matches the datasheet circuit except for these two diodes + 1n cap.  The two resistors + 10u cap are described as setting the DC point at the output.  My interpretation was that the rectifier + delta-G stuff provides AC feedback while the two resistors + cap provide a DC feedback path, and the 10u cap is in there to eat any AC that comes back that way so it doesn't mess with the signal gain.

But yeah, that much all matches the datasheet circuit.  Boss didn't even change the DC point from the datasheet examples, there's a coupling cap and a new 4.5V reference immediately after this.

But the diodes, still ... to be clipping high frequencies, wouldn't they have to be -crazy- loud high frequencies?  3.6-3.7V threshold on that path.  That output should bias around 3V with the 10k resistors, shouldn't even be possible to get to the threshold on the low side of the signal.  Even with the treble gain in the emphasis filter, starts to seem outside the realm of "normal" signal conditions, doesn't it? 

(If I have to kill a few 570s to find out, I'm not above that at all.  Just won't have any more loose ones for another couple weeks and the curiosity is bugging me.)

[Rob Strand]

As far as I can tell, it matches the datasheet circuit except for these two diodes + 1n cap.  The two resistors + 10u cap are described as setting the DC point at the output.  My interpretation was that the rectifier + delta-G stuff provides AC feedback while the two resistors + cap provide a DC feedback path, and the 10u cap is in there to eat any AC that comes back that way so it doesn't mess with the signal gain.

Yes, I realized that later.  For some reason I though the lower node of the 2xdiode + 1n was ground which totally screwed up my thinking.

I have a vague recollection of one of the compander app notes, it might not even have been the NE57x stuff adding the feedback diodes.   I don't remember the cap, especially a small cap.

But yeah, that much all matches the datasheet circuit.  Boss didn't even change the DC point from the datasheet examples, there's a coupling cap and a new 4.5V reference immediately after this.

I didn't check that, but you do see a few NE57x circuit where the bias is non-optimal.

But the diodes, still ... to be clipping high frequencies, wouldn't they have to be -crazy- loud high frequencies?  3.6-3.7V threshold on that path.  That output should bias around 3V with the 10k resistors, shouldn't even be possible to get to the threshold on the low side of the signal.  Even with the treble gain in the emphasis filter, starts to seem outside the realm of "normal" signal conditions, doesn't it?

The series cap removed any DC voltages so the clip points are entirely AC.   

For guitar the louder low frequencies will help trip the diode threshold by even with that help once the diodes kick in the small cap can only roll-off the highs.  It won't hard-clip the highs.   

For loud high-frequencies the the 2xdiode + cap change what would be hard clipping to soft clipping.  One way to think about it is if the high frequencies clip it generates a lot of high frequency harmonics.   I'm not sure how much that diode set-up will help but the idea/motivation could be that the 2xdiodes+cap try to roll-off the higher harmonics; I'd have to check on spice.    When you have a sampling system strong high harmonics above the Nyquist frequency get folded down into the audible spectra and sound bad.   That's the job of the anti-aliasing filter but simple guitar pedals have fairly relaxed filter slopes since the they don't normally need to worry about harmonics above the Nyquist frequency.

The other motivation is it makes the clipping less offensive.

All speculation and since they are both minor effects it's hard to prove the motivations.  Clearly the circuit will work without the 2xdiodes+cap.

[amz-fx]

The idea of limiting the voltage swing may come from the Signetics applications guide for the NE570.



Boss dropped the zeners from all later 570 compressor circuits.

regards, Jack

[Rob Strand]

The idea of limiting the voltage swing may come from the Signetics applications guide for the NE570.

Thanks I knew it was in there somewhere, just didn't want to plough through all the docs.
That one is fairly aggressive in that it shorts the output.  The feedback version is a little kinder.

[StephenGiles]

Is it a poor man's attempt at the Aphex trick in the Compellor, where the CV is frozen until re-energised to avoid such distortion?