Missing clamp diode on DS1054Z schematic?

[ezalys]

I've been trying to figure out Dave's schematic ( http://www.eevblog.com/files/Rigol-DS1054Z-Schematic-FrontEnd.pdf ) for the DS1054Z. The scope box says that it's 300 Vrms tolerant. Imagine a 300 volt 60 Hz signal. Imagine the attenuator is disengaged. The 1n cap looks open (about 3 megohms of reactance at 60 Hz), so the signal goes to the divider. The voltage divider looks like it's a factor of 3... but that's still 100 Vrms to that delicate opamp. The opamp seems to have jfet inputs... if there's no body diodes it'll fry, right? As long as the input stays hi-Z, it'll rise right up to that sqrt(2)/2*100 V. The SSR there appears to pull in and out a capacitor, but that cap provides about 12k of reactance which is pretty insignificant... as I suppose it should be. Why doesn't the TL274 fry?! Are there missing clamp diodes on the schematic? Is this circuit just relying on the part's ESD diodes? Does the scope just fry if you give it 300 volts with the attenuator disengaged? Seems like bad design if you could just add another clamp to prevent something like that.

[Andy Watson]

The datasheet for the TVL274 states a maximum input current of \$\pm\$10mA - which hints at there being some sort of mechanism for clamping the input voltage.

[David Hess]

That is exactly it.  Unless Dave missed some parts, the input protection diodes inside the TVL274 are being used as part of the overload protection.

+/- 300 volts DC is 300 Vrms so with the input circuit shown, 110 volts (I am ignoring the TVL274 supply voltages) in series with 100k is applied to the non-inverting input for 1.1 milliamp of current through one of the input protection diodes depending on polarity.

My old Tektronix 2232 uses the same protection scheme for its TLC271A operational amplifier relying on attenuation, high input resistance, and the input protection diodes.

In theory external diode clamps could be added but if one of them shorted do to an overload, it would still be a hard kill whether the operational amplifier survived or not.

[tautech]

It's not so much of a problem in countries where the mains voltage is only 110VAC and the likelihood of exposing your scope to input values in excess of 300V is somewhat less unless you're doing HV work other than SMPS.
However in countries with 230VAC mains it's a very different proposition, rectified mains is then ~325V and clearly exceeds the max input rating of many scopes if using 1x probes or direct BNC cabling.

Then it's wise to buy a scope with 400V rated inputs.

[David Hess]

The Rigol DS1054Z overload specification may be conservative although my Tektronix 2232 does say 400 volts peak for probably the reason you identify.  Details matter though like the working and maximum voltage of the individual parts which we do not know for the Rigol.

Be careful about x10 probes though.  They can have maximum voltage limits lower than some oscilloscope inputs and when some oscilloscope are set to AC coupling, the 9M probe resistance will happily charge the oscilloscope's input coupling capacitor to the voltage at the probe tip without any attenuation.  This is especially dangerous with x100 probes which have a 100M input resistance and no shunt resistance.

[tautech]

The Rigol DS1054Z overload specification may be conservative although my Tektronix 2232 does say 400 volts peak for probably the reason you identify.  Details matter though like the working and maximum voltage of the individual parts which we do not know for the Rigol.

Be careful about x10 probes though.  They can have maximum voltage limits lower than some oscilloscope inputs and when some oscilloscope are set to AC coupling, the 9M probe resistance will happily charge the oscilloscope's input coupling capacitor to the voltage at the probe tip without any attenuation.  This is especially dangerous with x100 probes which have a 100M input resistance and no shunt resistance.

Not to mention also probe voltage derating with frequency. 

[David Hess]

Not to mention also probe voltage derating with frequency. 

And the oscilloscope input voltage derating with frequency.

[Circlotron]

Not to mention also probe voltage derating with frequency. 

Yeah, when it was brand new I popped my 400V 10:1 Rigol probe with a 450V pulse that had a 1uS rise time.     

[T3sl4co1l]

I once made the mistake of constructing a 100x probe with a 1/2W axial resistor (10 meg, into 100k plus scope).

It worked fantastically, until the ~1.5kV peak I was measuring, jumped the poor resistor* and poofed the thick-film attenuator module in my Tek 475. 

*Average 1/2W resistors are rated for about 300V, maybe 600-1000V surge.

Instead of ordering a replacement attenuator module, I hacked my own, using a few discrete resistors and a trim cap... which didn't quite work the same, having some "hook" that couldn't be adjusted out.

Meanwhile, the other channel's attenuator modules are fuzzy with tin whiskers, so the vertical gain randomly jumps up from nominal, as the machine heats up... give or take if you smack it enough times to momentarily loosen the whiskers. 

So... among other reasons, I don't use my 475 much anymore... but it's good to keep on hand, just in case, anyway...

Tim

[David Hess]

I made the AC coupled x100 probe mistake measuring about -2000 volts but my oscilloscope and probe survived.  Some probes have a built in shunt resistor and this can be determined by checking the probe's input resistance specification which will be lower than the x10, x100, or x1000 multiplier indicates.

Those hybrid attenuators come apart and are pretty easy to clean but with tin whiskers maybe that would not be much help.  I had one with an intermittent lead connection (riveted?) which I resoldered using low temperature silver solder without issues.  I have never seen tin whiskers on them.

I am always very conservative with working and peak voltages on resistors and capacitors and especially so when they will be used in an uncontrolled environment.

[T3sl4co1l]

FYI, the tin whiskers are inside the scope -- on the attenuator hardware itself.  (Seems Tek wasn't so careful about what they plated those parts with... at least, in the intervening four decades during which the whiskers have had time to grow!)

Tim