Running Joulescope on high voltage

Hi, new guy here, considering purchasing a Joulescope to measure current in a high voltage line (2kV).
My plan is to use a high voltage USB isolator like this one which I own and has been proven to work.
That means that the Joulescope would ‘float’ on the high voltage (and I will be Very Careful ;-).

Is there any issue that would prevent this to work? The isolator provides a maximum of 500mA, which I guess is sufficient to power the Joulescope.

Jan Didden
Linear Audio

1 Like

Hi Jan, and welcome to the forum! Thank you for considering a Joulescope for your application. I took a quick look at your website. Cool projects!

The Joulescope JS110 is mostly a 1 quadrant device, +current and +voltage, with some measurement range extending into the other three quadrants. The Joulescope JS110’s fast current autoranging only works in the +current direction. It will still autorange on -current, but not quite as fast. Based upon your high-voltage DC regulator work, I am going to assume that you want to measure 2 kV DC currents.

The JS110 should work just fine measuring currents up to 3A sustained. If you want to measure voltage so that you can also measure power and energy, you can select a suitable high-voltage divider circuit. Check out this post for details. If you only want to measure current, connect IN- to IN+ and leave OUT- open.

That Intona USB isolator looks good, and 500 mA is enough to power the Joulescope JS110. The Joulescope JS110 sensor-side is also electrically isolated. For details, see page 27 of the Joulescope JS110 User’s Guide. However, our standoff voltage is further limited by the aluminum case and spacing between the USB chassis ground card rails and the sensor electronics. The JS110 also does not have any safety certification for higher voltages and is only rated for low-voltage ±48V DC. For your work, I definitely recommend a second isolator like you have selected.

Consideration 1
With two isolators, the working voltage will split somehow across them. You likely want to deliberately force Joulescope’s host-side ground to be close to the sensor-side. The JS110 components are not rated for this stand-off voltage, so you really want the majority of the working voltage drop across the Intona isolator. You could add a very high-value, high-voltage rating resistor across the JS110’s isolated plane split. I would guess that the best place would be near the transformer away from the sensitive analog components. Assuming the resistor is not too thick, you could bridge the isolated gap on the backside by scraping away some solder mask and soldering the resistor there. If you order from the Joulescope store and make a note “Send dead bare-board” on your order, I will include a dead bare-board with your order so you can practice. You can also order from our distributors, but I don’t have a way to include a bare-board.

Consideration 2
High dV/dt is also a concern. All isolators allow some coupling, especially for fast transients. We have had reports of JS110’s resetting on ESD events, even on the sensor side, likely due to coupling to the host side. I have not done enough testing to know whether this is through ESD directly leaping to the USB chassis ground or through the isolation components. While USB is a somewhat differential bus, the common-mode voltage range is very small, only 0.8V to 2.5V. While it’s very hard to know definitively, we believe that many of these reset events are caused by coupling that causes ground bounce which violates this constraint causing USB errors and eventually a host-initiated reset. With a higher input voltage range on the sensor side, your setup has the possibility to have higher dV/dt events that may cause this issue. However, if your voltage ramps at reasonable rates and is mostly stable, this should not be an issue.

Consideration 3
Joulescopes use MOSFETs to selected the shunt resistors and disconnect IN+ from OUT+ when the JS110 powers off or you set Current Range to off. If this ever happens in your system, then the JS110 will see your full voltage drop across IN+ and OUT+ which will definitely fry the 10A & 2A selection range MOSFET. I recommend that you add two diodes, one in each direction, across IN+ and OUT+. While these diodes may cause a little bit of unmeasured leakage current, they will help prevent you from blowing up your Joulescope. You want to select diodes that handle your current with minimal leakage at up to 25 mV forward or reverse bias.

:warning:Safety First!

Joulescope’s aluminum enclosure, screws, end panels, and exposed USB chassis ground will all be at or near your operating voltage. Good engineering practice is to limit the chance of accidental contact. You may want to wrap the enclosure or 3D print guards.

The proposed setup has the possibility to inadvertently operate the Joulescope JS110 beyond its rated limits. Please be sure to follow all safety practices including having a quick emergency shutdown and fire extinguisher, just in case.

Does this answer your questions?
What do you think?

Another thought: Do you have an Earth-grounded supply and can you measure the low-side return current? If so, then you can then completely ignore consideration 1 and just use Joulescope directly without an external isolator.

Here’s how:

  1. Connect OUT+ to to your Earth grounded-supply GND and IN+ to the device under test ground.
  2. Connect IN- to IN+.
  3. Leave OUT- disconnected.

This also mostly eliminates consideration 2. However, consideration 3 still applies.

Matt, thank you, appreciate the extensive reply. I fully understand the issues you mentioned, and indeed bridging the internal isolator had occurred to me to force the measurement side to remain within the CM range.
What I didn’t realize though is that basically the Joulescope measures in the first quadrant. My application for this is AC however. So I need to think a bit more on how to attack this.

Jan Didden
Linear Audio

Hi @jan.didden

You can potentially use the existing JS110 for AC current, but it’s not as straightforward as measuring DC. I have a couple of questions:

  1. What is your expected AC frequency?
  2. What is the dynamic current range (RMS A)?

– Matt

The frequency band is basically audio and a bit further, say 10Hz to 50kHz. I would like to measure from say 1uA up to 500mA.

Jan

Hi Jan,

I am going to assuming that your initial 2 kV rating is RMS. Please correct me otherwise! If so, then at 50 kHz you have:

d/dt ((2,000 V RMS) * sqrt(2) V p2p/RMS * sin (ω t))
and ω = 2 π f

so maximum dV / dt is then:
= 2,000 V * sqrt(2) * 2 * π * 50,000 / s
= 888 V / µS = 0.8 V / nS

The Joulescope architecture is essential a low-side ammeter that floats with IN+, which has to follow this dV / dt. That rate of change is on the same order as our data signals traversing the isolation. I just attempted to measure the capacitance between the JS110 IN+ and USB ground using a B&K Precision 880. Although I don’t fully trust this measurement due to the complexity of the circuit, I measured about 50 pF, with ranges from 40 pF to 70 pF depending on positioning. At 50 kHz, this is 1 / (2 π f C) = 63 kΩ => 31 µA RMS of potentially “lost” (unmeasured) current.

Questions:

  1. Any chance you can measure the return current to neutral? Then dV / dt would be nearly zero.
  2. Do you need autoranging? If you are trying to measure the current in different operating modes, then you could possibly use manual ranges. I am pretty sure that the negative side autoranging will introduce a significant amount of error at 50 kHz for your AC signal.

Matt, yes I believe I can measure in the return leg, it will mean some temporary mods to the equipment but that should be doable. This would take care of the high voltage issues.
But how would I be able to measure AC then? And I really would like the autoranging feature as I need to do some swept level measurements.

Jan

Matt, what do you mean with ‘negative side autoranging’? Auto ranging on neg signals or down-ranging on decreasing signal level?

Jan

My primary concern is increasing to a larger, less sensitive range when the signal becomes more negative than the negative limit. While the Joulescope JS110 has dedicated overflow hardware on the +current side, it relies upon the ADC value on the negative side. The ADC path has higher latency.
While the +current side switches in under 1.2 µs (max), the -current side can take up to several µs before triggering the range switch.

Since the switching time is constant, error increases as frequency increases.

The autoranging performance on underflow is equivalent on both - going towards 0 and + going towards zero.

Here are the design targets (not guarantees) for each current range:

Select     Range      Range     Range   Accuracy   Accuracy   Resolution   Resolution
         typical        min       max    typical        max      typical          max
     0    10.4 A    -1.80 A    10.4 A    3.52 mA    7.00 mA       429 µA       816 µA
     1    2.00 A    -470 mA    2.17 A     741 µA    1.49 mA      75.2 µA       158 µA
     2    182 mA   -42.7 mA    197 mA    67.4 µA     135 µA      6.84 µA      14.4 µA
     3   18.0 mA   -4.23 mA   19.5 mA    6.68 µA    13.4 µA       678 nA      1.43 µA
     4   1.80 mA    -423 µA   1.95 mA     667 nA    1.34 µA      67.7 nA       143 nA
     5    180 µA   -42.3 µA    195 µA    66.7 nA     134 nA      6.77 nA      14.3 nA
     6   18.0 µA   -4.23 µA   19.5 µA    6.67 nA    13.4 nA       677 pA      1.43 nA

The negative current range only extends about -22% of the +range for the same setting.

So, I think that Joulescope will be able to measure your current waveform in a low-side configuration, but you still need to add the diodes from Consideration 3 above. It should give great results for your lower frequencies. As the frequency increases, the negative overrange switching error will represent a larger percentage. 50 kHz is 20 µs period, so a few µs is significant.

What do you think?

I think the delay is not a problem, not even at 50kHz as I can run the sweep really slow.
I’m looking at a few other angles.
I would anyway order from one of your distributors here in Europe as direct import from the US gets quite expensive what with import duties and taxes, > 24%!

I have enough info for now, thanks for all your help.

Jan

1 Like

Matt another thought. I have some experience with (audio) autorangers, and to drive the autoranging circuits I use an RMS to DC converter. You can’t run the autoranging directly from the AC signal because it will start switching on the actual waveform depending on frequency and reaction time.
So this may be an issue when trying to use Joulescope with AC signals.

Jan

The Joulescope JS110 will uprange on overflow (like 180 mA to 2 A) very quickly. The downranging on underflow is intentionally much slower to help account for this effect, which happens even with DC signals, such as square waves.

For your 10 Hz signal, the JS110 will definitely downrange at the zero crossings. This is fine, but from an audio perspective, the signal will have higher THD than if you manually set a single fixed range. For your 50 kHz signal, I am not sure that your Joulescope will ever downrange from the 2A range. For some more information, see this old post which prompted us to improved the DSP applied to range switching.

For true, single-frequency AC signals, you would ideally want the instrument to autoranging on the amplitude and ignore the zero crossings. Alternatively, it could detect the lowest frequency component and adjust the downrange time accordingly. The JS110 does not have any features like this, so it will have more range switches that would strictly be necessary.