Oscilloscope Probe

Thread Starter

gopalyajur

Joined Jan 3, 2010
100
Hello all,

I recently purchased this oscilloscope from ebay

http://cgi.ebay.de/DSO-2090-100Msa-...182305118&po=&ps=63&clkid=6451951991321486562

I am pretty happy with its performance. I recently made a very basic inverter using arduino and I would like to measure its wave form using my new oscilloscope. The problem is the o/p voltage is above 1 Kv but with very low currents, which I cannot measure with my 10 X probe and oscilloscopes max voltage rating is 30 V p-p.

Can I just use a voltage divider connected to my probes and do the measurements. Or do I need to purchase 100 X probes (seems to pretty expensive).

I would like to hear from the experts before frying my new oscilloscope and my macbook.

Thanks in advance.
 

JDT

Joined Feb 12, 2009
657
You could make your own. But you will have to be very careful if you do not want to fry your new 'scope and computer. Or yourself.

Basically, a X10 probe relies on the fact that the input impedance of most 'scopes is 1MΩ. So the probe is a 9MΩ resistor.

If you built a probe with a 99MΩ resistor you will have a X100 probe. Or 999MΩ for X1000.

But there are a some problems:

These probes rely of the scope's input impedance. If you unplug it, then the full input voltage appears on the connector. Worse, it depends on the ground connection. Open the ground and all the equipment will be at the input voltage.

The input impedance of the 'scope is not just 1MΩ. It also has some parallel capacitance. Therefore the probe needs some parallel capacitance to form a potential divider at high frequencies. In a X10 probe this capacitance will be about 9X the input capacitance. This capacitor must safely block the operating voltage.

A better way:
I think your best solution is to design into the inverter an internal X10 or X100 voltage divider using lower value resistors and provide a "test point" at a safe voltage to connect the 'scope. The input impedance of your 'scope using a X10 probe will be about 10MΩ so design your internal voltage divider so that when loaded with this load it will give the correct result.

Actually, most DVM's also have about 10MΩ input impedance so you could check that the voltage divider is working by measuring with this before connecting the expensive 'scope and PC.
 

tom66

Joined May 9, 2009
2,595
You could make a 1000X probe go into a high speed op-amp or JFET buffer. It would require a very high voltage to damage (>1000V) and the worst case is you damage your probe unless you go as far as to break down the JFET gates, which would be unlikely, assuming you pick sufficient values, and like jpanhalt said, your resistors are rated for high voltages. You could probably build an adapter/probe yourself.
 

BillB3857

Joined Feb 28, 2009
2,570
If you are using a transformer to generate the HV, measure the low voltage side with the scope and multiply by the transformer ratio.
 

marshallf3

Joined Jul 26, 2010
2,358
Most probes have a variable capacitor in them to balance the matching part out.
I don't feel like explaining this, someone else please fill in the holes?
 

Thread Starter

gopalyajur

Joined Jan 3, 2010
100
Thanks everyone for the reply.

I will go through the video on how to make 1000 X probe

If have to use the voltage divider is it correct
that I connect a 323 MΩ resistor in series with the o/p of the transformer for 1000 V. That would mean drop of 970 V across the 323 MΩ resistor and 30 V across my probes having an internal resistance of 10 MΩ.
 
Last edited:

Adjuster

Joined Dec 26, 2010
2,148
Most probes have a variable capacitor in them to balance the matching part out.
I don't feel like explaining this, someone else please fill in the holes?
The problem here is that a "high impedance" oscilloscope input typically behaves like a 1MΩ resistor Rin, in parallel with a capacitance Cin of something like 15pF plus the capacitance of any connecting cable. If a 10X probe simply consisted of a 9MΩ series resistance Rs, the result would be accurate attenuation only from DC up to a few kHz. At higher frequencies, where the reactance of the oscilloscope input capacitance is smaller than its input resistance, the response using such an uncompensated probe would drop off at 6dB/octave.

The remedy is to arrange a small capacitance C1 in parallel with the probe series resistor. If this can be chosen so that Rs*C1 = Rin*Cin, the probe frequency response can be corrected. Unfortunately, Cin may vary between different instruments. To allow for this C1 could be made variable, but this is not so commonly done. More usually, C1 is fixed at the right value to give correct results with the highest expected Cin, and a second trimmer capacitor C2 is placed in parallel with oscilloscope input, so that extra capacitance can be added when using oscilloscopes with lower Cin.

The adjustment of the probe compensation should be checked periodically, and always when a probe is used with an instrument for the first time. This can be done by displaying a low-frequency square-wave signal with short rise and fall times. Many oscilloscopes include a calibration source which facilitates this. The displayed waveform must show flat tops and bottoms with no overshoot / undershoot. The trimmer capacitor is adjusted as necessary to achieve this.

Having said all that, it will be seen that is difficult to get a good frequency response using an attenuator made out of resistors of hundreds of megohms. Is this necessary to avoid loading your high voltage circuit unduly?
 

JDT

Joined Feb 12, 2009
657
That would mean drop of 970 V across the 323 MΩ resistor and 30 V across my probes having an internal resistance of 10 MΩ.
This is exactly what you should not do. Because if the ground clip of the scope probe comes disconnected, scope and PC will rise to 1000V.

Better to build into the inverter a resistive divider like in the picture. The bottom resistor of the divider is connected to ground inside the inverter and cannot easily get disconnected. Now, if your probe ground got disconnected the worst voltage you could get is 100V. Not quite so dangerous.

Edit: As someone said, the upper resistor in this circuit has to withstand 900V so needs to carefully selected!
 

Attachments

Adjuster

Joined Dec 26, 2010
2,148
This is exactly what you should not do. Because if the ground clip of the scope probe comes disconnected, scope and PC will rise to 1000V.

Better to build into the inverter a resistive divider like in the picture. The bottom resistor of the divider is connected to ground inside the inverter and cannot easily get disconnected. Now, if your probe ground got disconnected the worst voltage you could get is 100V. Not quite so dangerous.

Edit: As someone said, the upper resistor in this circuit has to withstand 900V so needs to carefully selected!
Dead right, and if you add a compensating capacitance in parallel with the input resistor, that too must be suitably rated. Capacitors being what they are, I would recommend well over 1kV rated.

If you want to add a trimming capacitor to adjust the compensation, put it in parallel with the output where the voltage is lower.

How predictable is that 1kV, by the way? For safety, you need to rate your components to allow for the worst case!
 

Thread Starter

gopalyajur

Joined Jan 3, 2010
100
Thanks for the warnings.

JDT, in your circuit you have indicated 520 K resistor connected to the ground. In doing so, is the input impedance of 10 MΩ from the oscilloscope included. If I dont add that, the voltage across the 520 K would be 110 V.

Anyway, I will have to chose the resistor in such a way that the voltage acorss the lower resistor does not cross 30 V. As I mentioned before my scope upper voltage limit is 30 V.

Adjuster, I am not able to measure the O/P voltage with my multimeter precisely. My meter is not able to follow the high frequency. That's why I wanted to test with a oscilloscope. When I purchased the scope I knew I would not be able to measure high voltages. But, I thought I could do so my employing series of capacitors and resistors which obviously seems to have lot of problems.

In the worst case I would just the input signal as suggested by Bill.
 

Kirtho

Joined Jan 18, 2011
6
Hi, this is my first post here and I'm jumping right in:

JDT's setup charges the chassis with high voltage. I would not do that.

I have an idea that essentially creates a neutral wire. I am guessing that the inverter transformer is being pumped at not much more than 10 kHz. Frequency response shouldn't be a problem.

R1 and R2 are high value resistors, rated for the voltages you are using. R3 and R4 are lower value resistors, and what I have marked as ground might simply be the two resistors soldered together. If R1/R2 are 100 times the resistance of R3/R4, you will be accurate to about 1 percent and if they are a thousand times that resistance, your accuracy will be about .1 percent. Measure the voltage or view the waveform between TP1 and TP2. Doing it this way places a large resistance between either test probe and the source of high voltage and helps prevent stray currents from trying to find other paths. Solder the resistors together.

The best thing to do, for insurance, is to tie the bottoms of R3 and R4 to a confirmed earth ground. Don't count on the grounding prong on the wall outlet or a water pipe unless you have physically verified it. The circuit does not intentionally run even microcurrents to ground. It is grounded for for "leakage" and accidental connections. This is different from circuit ground. Wiring it to circuit ground risks electrical shock and destruction of circuitry.

The bottoms of R3 and R4 must be connected to each other with wire. It isn't drawn that way because I drew it the way that I was taught.

You might want to build this circuit in an insulated case and use it as an adapter for your oscilloscope and then clip it to your test circuit. This may be the safest possible adapter short of an isolation amplifier with an optical connection, but it is still better to clip the divider circuit in while your inverter circuit is not energized, and turn off the power before disconnecting it.

 

Attachments

Top