Home Brew Oscilloscope Probes

[StudentSA]

Hi,

I`ve bought an old dusty 20Mhz china name oscilloscope on my student budget.

All I got was the scope and power cable. I would like to know if I can make some probes using coax cable. I have bundles of coax cable laying around.
I do not need a professional probe just something that will work from DC to less than 500kHz signals at voltage levels below 12V.

Please could someone point me in the right direction?

Regards,
StudentSA

 

[BrownOut]

If you just use some coax that's lying around, it will probably be low impeadance, which will distort any signal over a few Khz. Better if you can get ahold of 1M ohm coax. Then there are a number of ways to make a probe. Study the construction of existing probes or try google for construction details. I read an article some time ago, it's not hard. You only need to pay attention to how construction affects capacitance. Good luck.

 

[crutschow]

Coax is ok for low frequency signals, as BrownOut noted, but its capacitance will load down high frequency signals (the capacitance can vary from 10pF to over 40pF per foot depending upon the type). The oscilloscope also has perhaps 20-30pF of input capacitance which adds to the cable capacitance. Use the lowest capacitance coax you have for your cable and keep it short.

Don't know what BrownOut means by 1M ohm coax since the highest characteristic impedance cable I've ever seen is about 100 ohms. For high frequencies, a 10:1 probe is usually used which increases the input resistance from 1MΩ to 10MΩ and reduces the probe capacitance to 10pF or less.

 

[Reloadron]

Look at the vertical input channels. You will likely see something like 1 MΩ / 20 pF give or take on the capacitance. That tells you the input impedance is 1 Mega Ohm shunted by 20 pico Farrad so you need to take that into consideration when making a probe. That is typical for most scopes.

This link starts with a good overview.

Also this link has more information. The latter link shows a probe with a compensation adjustment.

I would use RG 58 coax (50 Ohm) for the probe leads. Anyway, I am sure you will get the idea from the links.

Ron

 

[JimB]

The DIY probe in ReloadRons first link will have an awfull high frequency response, it appears to have no frequency compensation whatsoever.

His second link gives a reasonable insight into the frequency response problems.

1M ohm coax?
I am not sure about that, but if you look at the cable of a scope probe, the centre conductor is very fine wire.
The first time you cut one to try and repair it you could be forgiven for thinking that there is no wire in there!
The thin centre conductor makes a cable with low capacitance.

JimB

 

[MikeMl]

The shielded cable used between a car radio and a fender-mounted antenna rod is very low capacitance cable. It has about 1/10 the capacitance per meter than RG58 Coax.

 

[MrAl]

Hi there,


A basic 10x scope probe will have a large series resistor in parallel with a small capacitor.

The idea is to build two voltage dividers into the same circuit that are in parallel, one that is made from the circuit resistances and the other made from the circuit capacitances. As you probably know, the output of a resistive voltage divider is Vo=Vin*R2/(R1+R2) and that is clearly not frequency dependent. Well guess what? The output of a capacitive voltage divider is almost the same: Vo=Vin*C1/(C1+C2), and it's not frequency dependent either even though we are using capacitors, which by themselves are inherently frequency dependent!

The idea then is to take both of these and make a circuit that attenuates by a factor of 10, where the capacitor C1 provides a voltage divider ratio same as the resistive part. The reason we want a factor of 10 is both because it's a convenient divider ratio for scope viewing and also it allows for a nice small probe capacitance that is quite a bit lower than the scope input.

The Details:
The resistive part of the scope is probably 1 Megohm, and so the resistor has to be 9 times that, which is 9 Megohms. Since capacitive reactance is an inverse, C1 must be equal to the scopes input capacitance DIVIDED by 9, so if the scope has 18pf input the probe parallel cap has to be 2pf.
Of course if we take the line capacitance into account, we will have a higher 'input' capacitance so the parallel cap will come out bigger. If the line capacitance is 18pf/foot and we used 3 feet that would add 18pf times 3 to the total input cap, which would come out to 4 times 18pf, so the parallel cap would have to be 8pf. It's usually made variable to make up for irregularities in the line or scope and connectors.
Also, if we could find better line than 18pf/foot that would reduce the parallel cap requirement, so we would end up with less probe capacitance.

Theoretically because in the end analysis all the frequency components go away with this kind of design we end up with a totally flat response:
Vout=Vin/10
and it's as simple as that

To design a nice 1:1 probe, a well chosen unity gain op amp built into the probe body and powered with separate lines through the cable would do the trick really nicely, keeping the distance from the op amp input terminal to the circuit under test under one or two inches.

 

[tvtech]

Hi,

I`ve bought an old dusty 20Mhz china name oscilloscope on my student budget.

All I got was the scope and power cable. I would like to know if I can make some probes using coax cable. I have bundles of coax cable laying around.
I do not need a professional probe just something that will work from DC to less than 500kHz signals at voltage levels below 12V.

Please could someone point me in the right direction?

Regards,
StudentSA

Bugger trying to make one. Seeing you are from South Africa I will send you my other used scope probe that is still good. It has X1 and X10 switching on it. You pay the postage..that's all.

Ex new from Hamrad in CT around year 1998.

PM me with your S.A address details and it's your's.

I feel for Students. I was one once too with buggerall money. I know what it's like.

Cheers

 

[StudentSA]

Hi, Thanks to eveyone for their contributions. I understand now the importance of Impedance matching between probe and scope for maximum power transfer without too much original signal distortion.

The trick is to find a cable with infinite impedance and zero capacitance

I would like to thank tvtech for his personal offer which is much appreciated.

Finally out of interest, can one not "custimise" a co-ax impedance to match that of the scope by adding a calculated capacitor and resistor in series. this will increase resistance but decrease capacitance. Or is that co-ax is too terrible to correct with external components?

Kind Regards,
StudentSA

 

[mneary]

Finally out of interest, can one not "custimise" a co-ax impedance to match that of the scope by adding a calculated capacitor and resistor in series. this will increase resistance but decrease capacitance. Or is that co-ax is too terrible to correct with external components?

I think that's the intent with this link provided by Reloadron.

 

[unclejed613]

if you go to Linear Technology - Linear Home Page, look for an application note,AN-47. it has a chapter in ot that is a Tektronix tutorial on scope probes and how they work, and how they're constructed. i've successfully built compensated X10 circuits for audio dummy loads, and fed into a scope with 50 ohm coax with no problem. i wouldn't use the voltage dividers much above a few hundred khz because of the capacitance of the coax, but they're adequate for monitoring audio dummy loads. once you understand that what you're accomplishing with scope probes is an extension of the scope input right to the circuit under test, you'll appreciate the reasoning behind the construction of scope probes. if you ever dissect a probe with a bad cable, you'll find the secret behind the low capacitance cable. the center wire is VERY thin compared to normal coax. it's usually a single #40 steel strand, and you will see why you NEVER want to make any sharp bends in a scope probe cable...

 

[MrAl]

Hello again,


Here is a drawing which will help understand the basic scope probe.
The analysis of the 10x probe isnt that difficult, and it is interesting that in the final equation all the frequency dependent parts cancel out leaving a simple ratio of input to output voltage of 1/10.
If anyone would like to see this analysis let me know and i'll post details.

Note that the 10x probe lowers the load capacitance of the circuit under test and also raises the load resistance to 10 times the original scope value.

 

[MikeMl]

The big problem is that the capacitance of std. RG58 type coax cable is so high...

Standard Coax cables.

 

[MrAl]

Hi again,


Yes, the cable is part of the problem. Maybe try to keep the cable down to 2 feet.
All of my probes are commercial types.

 

[Roff]

The output of a capacitive voltage divider is almost the same: Vo=Vin*C2/(C1+C2)

I think you meant to say Vo=Vin*C1/(C1+C2).

 

[Dean Huster]

Don't get too cheap on us, here, StudentSA. You got a Chinese scope, get a Chinese probe. It won't cost you any more than a couple of university trips to a fast-food joint. Forego the burger and fries for a couple of nights this week -- it'll do you good, you'll lose a bit of weight, you won't ingest so much fat and you'll end up able to afford a Chinese probe worthy of a Chinese scope. You can never make a probe that will perform as well as a commercially-made probe -- even a Chinese one. If you go to the high end probes, you'll find that there's a heck of a lot of design and engineering that goes into them, including little things like resistance wire for the center conductor, even on a 1X probe. They have to have compensation for the scope's input capacitance or they'll never track the proper attenuation ratio as you go up in input frequency. For instance, the very first link that was cited is for a probe that would be worthless as mammary glands on a boar.

 

[unclejed613]

using Rdiv for the 9Meg divider resistor, and Cc for the compensation cap, Cin for the total input capacitance (including the cable capacitance) and Rin for the input resistance and X being the reactance of the capacitances (so XCin is the reactance of Cin), Vo=Vin*((Rin||XCin)/((Rin||XCin)+(Rdiv||XCc)))

 

[JimB]

For instance, the very first link that was cited is for a probe that would be worthless as mammary glands on a boar.

I totally agree!
JimB

 

[unclejed613]

i agree.... without compensation, the waveform shown on the scope will NOT be an accurate representation of what's being measured. the uncompensated cable capacitance will cause square waves to look more like triangle waves. the compensation capacitance in parallel with the divider resistor causes the phase shift to cancel the phase shift introduced by the cable and input capacitance being in series with the divider resistor. since the reactances of the capacitors remains proportional throughout the bandwidth of the scope, the division ratio remains the same.

you can use regular coax for scope probes, but your compensation cap will have to be proportionally larger. the capacitive loading of the circuit under test will be a lot higher as well

 

[MrAl]

I think you meant to say Vo=Vin*C1/(C1+C2).

Hi again,


Yes, that is why it is 'almost' the same and not 'exactly' the same
Typo, fixed...thanks.