Products > Test Equipment
Pintek DP-60HS differential probe (low-voltage, high-sensitivity, 60 MHz BW)
<< < (3/6) > >>
IDEngineer:

--- Quote from: CustomEngineerer on October 14, 2017, 12:38:36 am ---Pretty sure the user manual explains exactly that.
--- End quote ---
Hmm, you're exactly right. And I missed that ground reference (if you'll pardon the pun!) when reviewing the user manual prior to purchase. My fault.
IDEngineer:

--- Quote from: David Hess on October 14, 2017, 01:52:19 am ---Differential probes have *three* connections whether the third connection is made available to the user or not.  The third connection is ground and it is *required* for the differential probe to operate correctly.  For probes that lack the ground connection at the probe tip, there is even a warning not to use them with isolated oscilloscope inputs.  Without the ground, the probe's common mode input voltage will drift around and possibly be exceeded or worse.
--- End quote ---
Yes, certainly the front end of a diff probe has something it considers "ground" (for this discussion, I'll refer to that as "common"). But in an isolated probe, common and ground are and must be separate things.

In this case, now that it's clear (to me! :-[) this isn't an isolated probe, some of the behavior is explained. But with respect to common mode input voltage causing the "saturation" behavior, I'm not sure that's the explanation.

In the circuit in question, charging a small capacitance with a constant current (read: high impedance) source brings with it the very likely possibility of spurious charges building up on the cap. So, when the timing circuit is not in use, we actively short both sides of the cap to common (continuing the use of that term for a "ground" in the isolated portion of the system).

Some hard numbers: We charge for ~28uS, and the A/D conversion time is ~14uS. Outside of this ~42uS period, the two sides of the cap are connected to common by ultra-low impedance paths. We perform this measurement every 500mS, so the cap is shorted (42E-06 / 500E-03 =) well over 99% of the time. When the diff probe is connected across this cap, its leads are also shorted >99% of the time. Thus it is unlikely that residual charge is building up *across* the diff probe leads over a time period measured in 10's of seconds. (I realize you didn't suggest that, I'm just covering all the bases.)

As for common mode voltage, that's certainly a possibility. Except in this case, both sides of the cap - and thus both sides of the diff probe - are within the isolated subsystem. So while technically they could be offset by any amount from the diff probe's common, without a return path such an offset is meaningless. There has to be a return path for a voltage potential to exist.

Stated another way: Say we short the diff probe's leads together, then touch them to one terminal on a 12V battery. The other terminal is not connected to anything. Does the probe "see" 12V of common mode? I would argue no, because there is no return path to the other battery terminal. The battery is isolated, and its "12V potential" only exists relative to the other terminal. Likewise this isolated subsystem... it can have various potentials within it, but if the two points to which the diff probe's leads are connected are themselves shorted by a low impedance path, then the diff probe only has a single connection.

Let's take this thought experiment one step further. Let's take another 12V battery and connect its positive terminal to the first battery's negative terminal. Now, measured relative to the second battery's negative terminal, the first battery has an offset of 12V. If we considered the second battery's negative terminal to be "ground", then the first battery has a common mode voltage of 12V; its own 12V has a separate, 12V bias (i.e. the first battery's positive terminal is at 24V relative to the second battery's negative terminal). But that's only relative to the second battery. What will the diff probe see if we connect its shorted leads to just one terminal (either one) of the first battery? Will it experience 12VDC of common mode voltage? I would argue no. It has no reference to the negative terminal of the second battery, so it does not experience the common mode offset.

Then: What will the diff probe see if we separate its leads and connect them to the two terminals of the first battery? Referenced to the negative terminal of the second battery, the first battery's terminals have 12V of potential with a common mode offset of 12V. You and I, visually looking at the setup, would "feel" like the first battery's terminals are at 12V and 24V. But the diff probe has no return path to that negative terminal; all it sees is what appears between its two leads. And the potential between its two leads - the potential across the first battery - is 12V.

Again, I agree that the non-isolated diff probe has a common rail, and my own experiments have confirmed that common rail is connected to the ground on the BNC back to the scope, and that scope's BNC grounds are connected to earth ground. But if that common/BNC/scope/earth ground is isolated from the circuit under test, then anything the leads of the diff probe see within such an isolated circuit is just like them being connected across a battery sitting on the bench: All they see is the potential across the leads, because any common mode that may exist doesn't have a return path that is connected to the diff probe.

If I've got this all wrong (and I might!), I'd appreciate an explanation. Thanks!
David Hess:
Depending on exactly how the input protect is arranged on a low voltage differential probe, if the common mode range is exceeded, then the input protection diodes will have the effect of shorting the inputs together through an offset voltage referenced to the probe common.  Usually this will be through a parallel RC protection circuit in series with each input (470k and 1000pF is typical) which limits the current while providing a low AC impedance.  On a high voltage differential probe, the inputs will be shorted together after the input attenuators.

Your description of everything working for 10s of seconds sure sounds like the common mode capacitance across an isolation barrier is being charged while the probe is connected until the common mode input range of the probe is exceeded.
IDEngineer:

--- Quote from: David Hess on October 14, 2017, 06:57:29 pm ---Depending on exactly how the input protect is arranged on a low voltage differential probe, if the common mode range is exceeded, then the input protection diodes will have the effect of shorting the inputs together through an offset voltage referenced to the probe common.
--- End quote ---
Agreed. The question is, how does the probe common get that reference? If the only two connections it has to a truly isolated circuit are its input leads, then probe common can't be the other reference.

This is similar to why a bird sitting on a multi-kilovolt power line doesn't become an LEB (Light Emitting Bird). Its two feet (read: probes) are together raised to a huge "common mode" voltage, but because its body (read: common) has no connection to anything else associated with the power line, it doesn't experience any common mode voltage. Without another reference point, there can be no *difference* in potential. Unless the probe's common is connected to something related to the isolated circuit, there is no common mode voltage - just the difference across its two probes.


--- Quote ---Your description of everything working for 10s of seconds sure sounds like the common mode capacitance across an isolation barrier is being charged while the probe is connected until the common mode input range of the probe is exceeded.
--- End quote ---
I absolutely agree, yet based on the above thought experiment I don't yet understand how the probe can be seeing a common mode potential. More investigation is required!
IDEngineer:

--- Quote from: IDEngineer on October 15, 2017, 01:48:35 am ---I absolutely agree, yet based on the above thought experiment I don't yet understand how the probe can be seeing a common mode potential. More investigation is required!
--- End quote ---
An update....

During all of the initial testing, I moved the power input to a couple of different supplies. The most recent one, and where I'd left the power leads, is a nice big analog power supply. As noted in our discussions above, there pretty much had to be a return path for the diff probe's internal common back to ground, I had a hunch... and sure enough, this supply does not isolate its negative side.  >:(

So I switched the power back to the new Instek GPE-4323 that claims in its documentation to be isolated, and (more importantly) that I have personally confirmed IS isolated. Voila - problem solved. The return path that the diff probe's common found back through the BNC cable, scope, earth ground, power supply, and then the negative side of the supply has been eliminated.

The probe has now been running for 30+ minutes without saturating as before. An interesting side note is that, since this measurement is taking place within the isolated section of the circuitry, the capacitance that was being charged is likely the isolation capacitance of the DC-DC isolation module (spec'd at ~10pF nominal). This proves that a non-isolated differential probe can be used on isolated circuitry if you're excruciatingly careful about return paths to the probe's internal common.

There's still work to do. Connecting the diff probe adds a significant percentage to the capacitance being measured so it strongly affects the circuit behavior; some firmware mods may be able to accommodate that. And the diff probe is still picking up a source of square wave noise, not surprising since a constant current source is a high impedance circuit and thus very susceptible to noise. At this point in getting familiar with the diff probe I'm just using a pair of ~12in long banana-to-micrograbber test leads, which I'm sure are excellent antennas. Despite my careful routing, those leads are also probably contributing to the parasitic capacitance.

I need to figure out a way to dramatically shorten the connection to the diff probe. I'd love to use short shielded wires, or even a pair of 1x scope probes with their shields tied to the diff probe's internal common, but I'm concerned that will add more parasitic capacitance. IIRC Tektronix made/makes a diff probe that has a very short pair of fixed probe tips coming out of its handheld enclosure. I might try to duplicate that with a pair of bare banana plugs and short solid wires.

In any case, my conclusion so far is that the DP-60HS can in fact be used on isolated circuits IF its BNC ground is isolated at some point from the ground feeding into the circuit under test. That could happen with an isolated oscilloscope (my old Phillips PM3214 has isolated inputs), or as in this case at the DC power supply. Going forward my problems appear to be parasitic capacitance and noise, very traditional problems with lots of ways to attack them.

I'll update as I learn more.

Edit: The Tek probe is the old P6046, its inputs look like this, literally on the end of the probe body. I used one of these long ago but had forgotten the specifics. I'm going to experiment with duplicating this arrangement on the DP-60HS.
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod