Best way to protect my voltage reference

[eurofox]

Hi voltnuts friends,

I would like to get some ideas how to protect the output of an LTC1052 against over voltage from static electricity.

2 weeks ago I killed my voltage reference by connecting it to a DMM.

The pamona cable was connected to the guard of the DMM and of course the DMM is connected to my home ground by the power connector.

There is a red led connected to the power supply and when I connected the cable to my voltage reference the led was shining a fraction of a second brighter .... I suspected something was wrong 

My LTZ1000 based reference voltage aged for one month and was very stable, it was mounted in a Teco box powered with a external very stable and ultra low noise power supply. The power supply was as well checked free of over voltage when powered up.

All this is mounted in an aluminium enclosure well grounded.

Thanks in advance

 

[TiN]

RIP LTZ?

What kind of crappy DMM is it that does not have isolated front-end? 

Best way to protect devices is to limit/avoid the chance or user errors. By using different keyed connectors for power, consistent cable colors/labels, fixed connections, etc.

[rhb]

I lost several hard drives while handling them during an unusually dry period a few months ago.  I mounted a piece of 1/16" x 1/2" aluminum angle to the edge of the desk and grounded it through a 1 meg resistor.

I've got an ESD mat, but those don't help if you're working on something which is not on the mat.

[eurofox]

Any experience with TVS diodes?

[The Soulman]

There is a red led connected to the power supply and when I connected the cable to my voltage reference the led was shining a fraction of a second brighter .... I suspected something was wrong 

I can't imagine that a static discharge would discharge slow enough to notice the increase in brightness of the led,
you sure the voltage regulator hasn't failed?
And how is your grounding scheme, the v-ref negative (nor positive..) should not be connected to any guard or earth connection.

[Brak]

I've done a lot of googling and found almost nothing about output protection for a voltage reference.  It seems almost anything you do affects the output voltage.

I did find an article, "Statistical Voltage Reference" by Mark Hennessy that discusses this a bit.  He used a 100 ohm resistor in series with a JFET current source.  And a zener diode across the output.

See:  http://www.markhennessy.co.uk/dvm_reference/#output_protection

Brak

[eurofox]

There is a red led connected to the power supply and when I connected the cable to my voltage reference the led was shining a fraction of a second brighter .... I suspected something was wrong 

I can't imagine that a static discharge would discharge slow enough to notice the increase in brightness of the led,
you sure the voltage regulator hasn't failed?
And how is your grounding scheme, the v-ref negative (nor positive..) should not be connected to any guard or earth connection.

Power supply is ok and of course none of pos & neg are connected to GNG.

[Andreas]

Hello,

a picture says more than 1000 words:
- cirquit diagram of the reference + the set up?

what do you mean with static electricity?
- static DC-voltage source
- electrostatic discharge?

with best regards

Andreas

[cellularmitosis]

Hi voltnuts friends,

I would like to get some ideas how to protect the output of an LTC1052 against over voltage from static electricity.

2 weeks ago I killed my voltage reference by connecting it to a DMM.

The pamona cable was connected to the guard of the DMM and of course the DMM is connected to my home ground by the power connector.

There is a red led connected to the power supply and when I connected the cable to my voltage reference the led was shining a fraction of a second brighter .... I suspected something was wrong 

My LTZ1000 based reference voltage aged for one month and was very stable, it was mounted in a Teco box powered with a external very stable and ultra low noise power supply. The power supply was as well checked free of over voltage when powered up.

All this is mounted in an aluminium enclosure well grounded.

Thanks in advance

That is terrible!  If this thread comes up with any good suggestions I'm happy to add them to the next round of px-ref board tweaks.

[eurofox]

RIP LTZ?

What kind of crappy DMM is it that does not have isolated front-end? 

Best way to protect devices is to limit/avoid the chance or user errors. By using different keyed connectors for power, consistent cable colors/labels, fixed connections, etc.

I think it is not a DMM issue, for the moment when I touch my car outside I get a painful arc between my car and my fingers, this mean that there is lack of humidity and with isolating shoes we are loaded with high voltage, the same with the car that is isolated on tires and with air movement loaded with high voltage.

It is very important in this season to have a wrist strap connected to ground when working with sensitive electronics.

[orin]

Looking at the Fluke 732B manual, they have an 11V 5% TVS in parallel with a reverse biased 100 PIV 5A diode.  Also a 230V gas discharge tube from their 0V/common terminal to chassis ground.  There is also a 10uF and 0.22uF capacitor in parallel across the output, but I haven't found these capacitors (nor any components in this part of the circuit) in the parts lists.  I'd guess 10uF 20V tantalum and 0.22uF film.

I'm going to add something similar to my references.

[David Hess]

You protect an output the same way that you protect an input.  Add series impedance to limit the current and shunt protection.  In the case of an operational amplifier output, that means protecting both the output and feedback network or connection to the inverting input.

An output capacitor is not a bad idea but consider how the pole formed by the series impedance and output capacitor affects the frequency compensation; that usually means adding another AC feedback network directly from the output to inverting input.

Leakage will not be a problem so normal diodes can be used for shunt protection.  Series current limiting can be implemented with a resistor, current sources like depletion mode FETs, or if you want to madly cackle, with some current sources and a diode bridge as shown below.

For protection against ESD, a shunt capacitor and shunt diodes may be sufficient.

[iMo]

A 15V transil?

https://www.st.com/content/ccc/resource/technical/document/application_note/30/f6/d4/72/7d/74/45/bc/CD00003857.pdf/files/CD00003857.pdf/jcr:content/translations/en.CD00003857.pdf

[Magnificent Bastard]

2 weeks ago I killed my voltage reference by connecting it to a DMM.

Damn!  The dingo done 'et yo' baby!

Try a P6KE12A across the output terminals.

[Nusa]

Climate controlled room, anti-static flooring or floor mat, anti-static work surface, anti-static clothing, anti-static wired wrist strap, and anti-static ankle/heel strap for whenever the wrist strap isn't connected.

And that's probably not even close to the BEST way.

[HighVoltage]

Most important thing for me is to wear anti static shoes in the lab.
Since I do that, I have not killed any electronics from static discharge.

People just look at me strangely, when I take an Ohmmeter along when I go shoe shopping.

[Andreas]

Hello,

to the protection diodes:
there are 2 different voltage specifications.
the 11 V TRANS SUPRESS in the Fluke might most probably be a
SA11A - the 11V specify the "stand off" voltage. The breakdown voltage is more on the 13V side.
A P6KE12A has a stand off voltage of 10V and a breakdown voltage of 12V +/- 5%.

So a SA11A has lower (temperature dependant) leakage current as a P6KE12A at 10V.

In the Fluke 734 handbook the output capacitor is specified as mylar capacitor.
Against ESD it builds a capacitive voltage divider of around 330pF/220nF.

But most probably there is no resistor from output to the negative input of the output buffer.
So together with the input protection diodes between the positive and negative input there is no current limitation for the ESD.

With best regards

Andreas

[eurofox]

Hello,

to the protection diodes:
there are 2 different voltage specifications.
the 11 V TRANS SUPRESS in the Fluke might most probably be a
SA11A - the 11V specify the "stand off" voltage. The breakdown voltage is more on the 13V side.
A P6KE12A has a stand off voltage of 10V and a breakdown voltage of 12V +/- 5%.

So a SA11A has lower (temperature dependant) leakage current as a P6KE12A at 10V.

In the Fluke 734 handbook the output capacitor is specified as mylar capacitor.
Against ESD it builds a capacitive voltage divider of around 330pF/220nF.

But most probably there is no resistor from output to the negative input of the output buffer.
So together with the input protection diodes between the positive and negative input there is no current limitation for the ESD.

With best regards

Andreas

I order this morning 4 5KP11CA transil 5kW 11V bidirectional.

I have a 1.5 KV power supply and will make some test but not on my LTZ1000 reference 

[SvanGool]

With the 5KP11CA clamp voltage of 18.2V and assuming a tolerance of 5%, I don't think your LTC1052 will always survive a test with your 1.5 KV power supply and a random 5KP11CA.
Like Andreas, I would also recommend a lower clamp voltage.

[Andreas]

Hello,

I hope you know that the 5KW are only for a single 1 ms pulse. (5 Joule).

And a unidirectional device would be better since then you have only 1V in forward direction instead of 11V in both directions.
The bi-directional devices are meant for AC.
And be aware of the parasitic capacity of the Zeners.
Some OPs have to be isolated against the capacity in nF range.

with best regards

Andreas

[Andreas]

With the 5KP11CA clamp voltage of 18.2V and assuming a tolerance of 5%, I don't think your LTC1052 will always survive a test with your 1.5 KV power supply and a random 5KP11CA.
Like Andreas, I would also recommend a lower clamp voltage.

The clamp voltage is for 280A current through the device.
ESD is usually below 15A. So the clamping voltage for ESD will be lower.

I have more concern with the leakage current on a sub ppm device.

With best regards

Andreas

[eurofox]

Hello,

I hope you know that the 5KW are only for a single 1 ms pulse. (5 Joule).

Yes of course I know this.

Is the first time I use this kind of component, they are not expensive and will order other type if needed.

[CopperCone]

Well, since its a LTZ1000 and its the cream of the measurement world, an good idea would be to use a over the top protection system that destroys accuracy, like the full heavy RLC filter, Fast Diode, TVS, GDT, Fuse, with a relay or switch across it, so you can connect it then shunt the protection network after you are sure its ok, then reapply the protection before disconnecting.

actually since it has stuff connected to ground it would be better to just disconnect that bit from the circuit.

[TheUnnamedNewbie]

I don't have a great understanding of the noise levels involved, but I would keep in mind noise when you start adding anything that presents itself as a high impedance.

[iMo]

Hello,

to the protection diodes:
there are 2 different voltage specifications.
the 11 V TRANS SUPRESS in the Fluke might most probably be a
SA11A - the 11V specify the "stand off" voltage. The breakdown voltage is more on the 13V side.
A P6KE12A has a stand off voltage of 10V and a breakdown voltage of 12V +/- 5%.

So a SA11A has lower (temperature dependant) leakage current as a P6KE12A at 10V.

In the Fluke 734 handbook the output capacitor is specified as mylar capacitor.
Against ESD it builds a capacitive voltage divider of around 330pF/220nF.

But most probably there is no resistor from output to the negative input of the output buffer.
So together with the input protection diodes between the positive and negative input there is no current limitation for the ESD.

With best regards

Andreas

The transils are not reference-voltage-like devices. The "knee" of the transils is pretty similar to a standard zener diode one (rather "wide", not brick sharp). You have to buy several types (ie. from 11V-15V) and measure them individually..