2
\$\begingroup\$

I am performing some component-level testing to measure the noise in a signal across a PCB. I was able to test the individual components with a low-power (30 V) power supply but I am hoping to test higher voltage versions of these components at ~200 V in a simple circuit. Without going in to too much detail, the circuit will take in and output a set voltage (it is a simple Zener diode circuit to regulate a voltage drop).

The PCB I have made and want to test has a V_in and V_out pad; I am using a Stanford Research Systems SR355 power supply capable of +10 kV. Its output connector is a Kings type 1064-1 HV BNC.

I am unsure of how to couple this into the scope safely; with a 10:1 probe, I am capable of measuring 300 Vrms. Would it be possible to simply have the inner HV pin of the BNC connect to V_in, and have V_out loop to the sheath and be grounded by the PSU? I can then use my 10:1 probe to measure the voltage stability across the circuit. Otherwise, would it be best to build a 10:1 probe box such that I can plug a HV BNC connector to the front and have a standard BNC output to then couple to my oscilloscope?

Unfortunately my experience with testing in this manner is limited to the low voltage regime and I'd rather be cautious and not give myself too big of a shock.

\$\endgroup\$
1
  • \$\begingroup\$ I would build a little measurement adapter for this. Use a off-the-shelf AL-Case with two BNC-Connectors (HV-Types). Inside the box build a simple Voltage divider (use multiple single resistor in series in case one fails short) with Zout <<ZIn of the Scope. Depending on your f-3db requirements you will have to do some thinkering - You can also install a rather "beefy" TVS on the output to protect your scope in case of overvoltage (Be aware of the RC-Circuit that is formed). Make sure to use HV-rated THT resistors and to ground the AL-Case properly (Low Z) to your Source GND and Earth. \$\endgroup\$ Jan 3, 2023 at 19:23

2 Answers 2

1
\$\begingroup\$

If you're trying to measure the noise on top of a high voltage signal, you don't want to divide down the signal -- you want to remove the DC component. See Jim Williams AN118 Fig B12. This makes a 50 ohm signal path, which you'll want to feed into a nice low noise amplifier. You'll want to make sure you're not limited by the noise of your amplifier, too.

\$\endgroup\$
0
\$\begingroup\$

I would go with a external measurement-divider.

Either you can buy a so called "HV-Probe" and be safe. Or you can build your own and be safe.

My assumptions are:

Max VIn of 1KV +/- 10V, Scope RIn of 1Meg||10pF, f-3dB Scope 100MHz, f-3dB required 1MHz, Scale 100:1.

Basic circuit:

schematic

simulate this circuit – Schematic created using CircuitLab

Basic Design:

As R3 will have to drop the majority of the DUT-voltage let's start here. At 1kV max. Input-voltage there are 10V (100:1) across R2. So 990V will be gone across R3. Therefore P_R3 = 10W. Finding 10W and HV rated resistors will be tricky. So you could use 10 9K9 1W type in series. This also provides safety to your scope, in case the single resistor would fail short due to some reason. Now: You could use a higher value R3 - let's say 990K. This requires a R2 of 10K. Still fine to not interfere with your 1M||13pF scope input too much. This will however drop your power requirement to 1W; much more reasonable.

If you increase R3 further to let's say 9M9 you would have to use a 100k R2 - this starts messing with your 1M||13pF scope as your voltage divider is loaded by your scope.

Next thing is: The further you increase your R3 the lower the f-3dB bandwidth of your circuit will be.

So we need to take a different perspective:

schematic

simulate this circuit

This is the circuit from above but reformed (1kV is wrong here - just ignore it). RQ is your source-resistance: R3||R2, RM is your scope resistance so 1 Meg and CM is your stray capacitance: C_par+C1

Let's say you want to include a "feel-good" protection TVS across you divider output. This can add around 100pF. So Cm ~ 150pF (cables and stuff). With R=RQ||RM and F=1MHz R<1/(2piCm*f-3db) R=RQ||RM < 1kOhm for 1MHz.

Lets say 100kHz f-3db is okay for you: R<10kOhm

Basic results:

The smaller your R3 becomes, the bigger your bandwidth gets, but the more power you have to handle. The bigger it gets, the less power but also lower bandwidth. So you will have to do a tradeoff here.

Detailed circuit:

schematic

simulate this circuit

I would recommend this circuit: It will give you ~100kHz f-3db, 100:1 scale and 1 Meg RIn (From DUT side) with extra protection (TVS - you never know).

Detailed Design:

So at 1kV R1 will have to drop 990V. Therefore P=1W. I recommend using a string of 3x 320kOhm HV (>1kV) resistors with a +/-25ppm/°C 1W 1% rating. The "missing" 30kOhm could be a pot. Make sure it can withstand at least 50V. By doing so you can adjust your scale factor and protect your device in case of a single resistor failing short. You could use THT types to avoid designing a PCB. In case of a pot: I would not make it adjustable from the outside.

For R2 I would use any 10k 1% 1/4W +/-25ppm/°C resistor.

I would include the TVS as it can protect your scope in case a single 320kOhm resistor fails short -> 100V on your scope (it will have internal ESD protection, but I am not sure about the continuous rating). It will severely reduce your f-3dB but I would stick with it.

I would select a bidirectional >50V breakdown with low capacitance. By using a bidirectional type you extend your measurement range from 0->1kV to -1kV <-> +1kV. By using a high breakdown voltage you reduce the "idle current" through the diode. The lower this current the better, as it loads your divider. Make sure the package can handle the power of at least one 320kOhm resistor failing Short.

I then would pack all this stuff into a proper sized (leave some room, don't stuff it) AL-casing. Lets say L x W x H of 200 x 50 x 50 with screw-on top lid. A BNC connector (HV rated input, normal output) on each side. After hand-soldering this one-off device I would recommend potting it to prevent parts from bending and shorting out. Use a HV-rated compound (somewhat not cheap, but not terrible). If would also include a 5mm banana-jack to ground the device to the DUT negative (must be protective earth, as scope is PE-referenced). This jack would be connected to the AL housing internally with a "beefy" 1.5mm^2 wire and proper termination (is kind of your/your scopes life insurance).

Careful:

All calculations done are rather rudimentary. You could us some flavor of LTSpice to do a proper design (check thermals, bandwidth, parasitics in cables and connectors, etc.). Also, do a good job selecting the right components. Always check for their voltage rating - never assume. Leave yourself some room in your case and do potting - will prevent nasty shorts and stuff.

And also: If you build it yourself and use it: It's best to connect this thing to PE with as thick a wire you can find. :)

\$\endgroup\$
1
  • \$\begingroup\$ Why are you designing for 1 kV when the requirement is 200 V? Also, a standard DS1054Z is CAT I 300 V RMS right on the input, and the included probe, if indeed it is the RP2200, is CAT II 300V. You can just stick it right on the circuit and AC couple it no problem -- but also the input stage of the scope may swamp the noise. \$\endgroup\$
    – Evan
    Jan 4, 2023 at 2:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.