WARNING! Discovery BNC adapter board inputs are NOT differential

[BlacKnight]

Hi Folks

 

Although the Analog Discovery has fully differential analog (scope) inputs, the Discovery BNC adapter board does NOT! For some reason the negative inputs of both analog channels have been connected to the Analog Discovery ground. Therefore, they form a short circuit between any conductors they are connected to.

 

Why is this important? I just found out the hard way, when I connected the scope leads from the Discovery BNC adapter board to parts of a circuit at very different voltages. I was expecting each channel to measure the differential voltage across each probe. Instead, there was a crack, a flash, and a little jet of molten metal, as the clip on one of my scope probes vaporised!

 

IMO this is a safety hazard, because adding the Discovery BNC adapter board to the Analog Discovery changes its electrical behaviour in a significant way (from differential to single ended) but there is no mention of this in the Discovery BNC adapter board reference manual. Users should be warned in BIG LETTERS about this change in behaviour, perhaps printed on the PCB as well. I was lucky, but providing users with a warning about this could save somebody's life one day.

 

I can see that grounding the negative inputs could make sense, in that it makes the analog inputs behave more like a conventional oscilloscope, where the probe shields are usually returned to Earth. However, if you're going to do this you should at least warn users about what you've done.

 

IMO it would be better to provide 4 sockets for scope probes, with each connected to one of the differential inputs, and all the scope shields returned to Analog Discovery ground. Then the Discovery BNC adapter board could be used for fully differential measurements. You could also provide jumpers to select single ended behaviour instead, by shorting the negative inputs to ground. At least that would give the user a choice about whether to operate as differential or single ended inputs.

 

[logansam]

Thank you for bringing this to our attention. I have a reply from our support team:

 

--- from our support team---

 

You are right. We never mention the fact that Discovery BNC adapter board does not have differential analog (scope) inputs. 

We went for this solution as BNC is intended to be used by default single ended (right, we never mentioned this).

 

Finally we apologize for this situation. Thank you for investigating and revealing this. We will consider updating the documentation so that this fact will be highlighted.

[LariSan]

@BlackKnight

Really appreciate you sharing your experience (that helps Digilent make a better product!)

I will make sure to put at least a note on our web-page and put a note on the wiki page for the Analog Discovery. 

[Commanderfranz]

Hi BlackKnight,

I added this note:

Note: The Analog Discovery BNC adapter board does not have differential analog (scope) inputs

to the wiki for the BNC adapter and sent in a webpage change request to add it to the product page.

 

Kaitlyn

[emarte]

Anyone has information on BNC Board Gerber/Schematics ?

I would like to make Board modifications if possible, in order to brake the shorts between grounds.

Edwin

 

 

 

[emarte]

Or are they any workaound for those like me trying to use the BNC board with diferential inputs.?

[attila]

Hello,

The BNC is 'by definition' single ended therefore the negative input of the scope channels are grounded on the Analog Discovery BNC Adapter.
It would make no sense to have the negative scope input on the BNC shielding. The probe compensation and attenuation are for the internal (positive) wiring.
For differential probe you could use twisted wires. In this case make sure to also have GND connection between the AD and circuit under test to respect the common mode limitations. For more information see: https://reference.digilentinc.com/reference/instrumentation/analog-discovery-2/reference-manual

[emarte]

I see  !  Thanks Attila.   

I will have to thibk of somethng else for the time being. (My custom board maybe)

 

[Jeromed]

I'm a newbie to the electronic fields though I had college introduction to electronics decades ago.  I blithely took the statement that the Scope channels 1 & 2 were fully differential.  I cooked a pnp transistor and nearly set a probe on fire until I figure out the grounds where shorted.  In particular, if you have 2 probes, you should really only ground one of them to the circuit for this very reason!!!!!

For someone getting up to speed again, you need to gently guide us.  Statements like AD needs to be grounded in the same was as the circuit doesn't really help.  More explanation is needed.  And why are scope probes single ended (and just exactly what does that mean?).

The major reason I got the probes is that 1) I thought they would be easier to use for testing circuits as they have grabbers and 2) I read somewhere (can't remember where) that the probes have a higher sampling rate.

I agree with the original poster that this is a MAJOR safety issue.

[clf]

On 10/6/2016 at 9:12 AM, attila said:

BNC is 'by definition' single ended

 

Hello @attila,

Could you kindly cite a source for your above statement "BNC is 'by definition' single ended", please?

I offer the Tektronix THS720 oscilloscope [1, page "1-2"] as a particular counterexample, and the Tektronix Application Note "Fundamentals of Floating Measurements and Isolated Input Oscilloscopes Application Note" [2] as a general counterexample.

From my perspective, it's much more desirable to have a versatile instrument that maintains its differential channels all the way through: you can easily reconfigure them to have a common ground by tying their "-" inputs together; taking channels with a common ground and making them differential is a much uglier problem, as highlighted by the other posters in this thread.

Dug into this topic when a Hackaday article [3] elicited a discussion in its comments of oscilloscope channels with common grounds vs. oscilloscope channels with isolated grounds.

Thanks in advance,

Chris

[1] https://www.tek.com/oscilloscope/ths710-manual/ths710-ths720-user-manual

[2] http://in.tek.com/dl/3AW_19134_2_MR_Letter.pdf

[3] https://hackaday.com/2017/10/18/testing-brushless-motors-with-a-scope-or-a-meter/

[JColvin]

Hi @clf,

This is definitely an old thread, but I suspect that the 'by definition' used by @attila wasn't the intended terminology and a word with a better connotation would be "classically".

Note that this is just me putting words in attila's mouth rather than offering a formal explanation, but based on the application note and link to EEVBlog you provided it appears that a large majority of BNC connectors present on oscilloscopes, or in this case an adapter board, the external shield portion of the BNC connector is connected to ground, as is shown as the BNC Adapter schematics.

Thank you,
JColvin

[attila]

Hi @clf

The Analog Discovery Scope inputs are fully differential but not floating. It has certain limitations to the ground of the device, so a ground connection toward the circuit under test is needed.
https://reference.digilentinc.com/reference/instrumentation/analog-discovery-2/reference-manual#scope
Such differential setup requires 3 connections: positive, negative and ground.
The traditional BNC only has two connections (signal and shield), so for the BNC Adapter Board the negative input is tied to ground/shield and the scope input becomes single ended.
I'm not an electric engineer, but as far I know (experienced) differential lines need to be highly symmetric. I think using BNC shielding as negative input and inner line as positive input would be very bad option. Twisting the positive with negative wire should be much better.

The AD is not isolated neither its Scope inputs. The ground of the device connects through USB to the PC.
For isolation USB isolator or battery powered laptop can be used.

[Raphael St-Gelais]

Hi attila, 

I agree that the BNC shield should not be used to carry signal, but why not have two different BNC connectors to carry the In+ and In- of each differential input. This is the way to go for lock-in amplifiers:

The board would have 6 instead of 4 connectors, but that would be much more useful. 

Cheers, 

On 11/1/2017 at 11:50 AM, attila said:

Hi @clf

The Analog Discovery Scope inputs are fully differential but not floating. It has certain limitations to the ground of the device, so a ground connection toward the circuit under test is needed.
https://reference.digilentinc.com/reference/instrumentation/analog-discovery-2/reference-manual#scope
Such differential setup requires 3 connections: positive, negative and ground.
The traditional BNC only has two connections (signal and shield), so for the BNC Adapter Board the negative input is tied to ground/shield and the scope input becomes single ended.
I'm not an electric engineer, but as far I know (experienced) differential lines need to be highly symmetric. I think using BNC shielding as negative input and inner line as positive input would be very bad option. Twisting the positive with negative wire should be much better.

The AD is not isolated neither its Scope inputs. The ground of the device connects through USB to the PC.
For isolation USB isolator or battery powered laptop can be used.

 

[xc6lx45]

On 10/6/2016 at 3:12 PM, attila said:

The BNC is 'by definition' single ended

It's one of those things that's never explicitly stated in textbooks or courses. But, if I can touch it, it's unlikely to be a signal...

Why? Well, for example, if I'd then ground the center pin, the signal would be only on the shield... Wonko the Sane (Douglas Adams) would love this.

There are many instruments on the market that provide differential connectivity on BNC, but always using pairs of connectors.