iProbes (Homemade Scope Probes) / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

iProbes (Homemade Scope Probes)

Move Thread

LAN_403

Part Scavenger

Sat May 26 2007, 12:12AM

Registered Member #79 Joined: Thu Feb 09 2006, 11:35AM
Location: Arkansas
Posts: 673

Ok, so the white reminided me of my iPod 'buds... :)

See EDIT#3 before you leave, I got them working.

Tired of getting crap in my waveforms, but too cheap to spend 50 bucks on a probe, I determined to make some nice O'Scope probes today whilst waiting on new IGBT's. I needed a custom one for my DRSSTC, which to date has simply been twisted telephone wire wrapped in al foil. Oddly enough, that works pretty well!

Total cost of each probe was less than $3. I am very happy with the 1/1 probes I made, they have a maximum noise level of 5mV when I'm really trying. The frequency response is totally flat as high as my Sig Gen will go (2Mhz). However... I'm not too pleased with the 1/10 or the 1/100 probes...

My new DRSSTC current probe

The 1/1, 1/10, 1/100 probe I spent all morning on. The 1/10 and 1/100 functions are pretty much useless over 1Khz.

Noise level on 1/1 setting. The end is just laying on the table. To make it go up to 5mV you have to wrap it around an AC cord or something. I don't know if this is a really "clean" probe but it's (mathematically) 10x better than the rest of my probes.

5-ish KHz square wave and probe set to 1/10. I didn't figure out how to make a compensation circuit because I didn't figure I'd need at this low Fo! Anybody got a schematic? Although I have a suspicion I need more than just compensation. It resonates somewhere around 30Khz. Quick google search didn't turn up anything. I KNOW I've seen one somewhere.

Frequency response of 1/10 probe swept to 0-20Khz Yuck. Ok, so why is that? My first thought was because I had to fit the resistors in the black box like a folding tape measure, but I would have thought that any capacitance obtained there would present itself in the MHz? Furthermore, the Fq response is identical to a probe I built with a straight divider. The only thing I could come up with would be capacitance in the coax? (RG59U) I can't imagine the resistors could be acting up, at least not yet...? (Xicion 1Mohm and 10Mohm 1kV Carbonized Film). I didn't bring a GND clip out the end of the probe, but rather am using the Ground on the O'scope. Could this be the reason?

Response of the 1/100 probe. Actually there is less distortion in this setting than the 1/10.

All measurements were taken as such: Trace #1 direct from Sig Gen, Trace #2 is 1k resistor directly connected to probe. Can my probes be made useful or is it just not worth it? I still expect it to be plenty handy for DC measurements.

Datasheets:
RG59:

(it's actually RCA cable not Belken)
Resistors:

EDIT:
I found a schematic, I was just searching for the wrong thing. It's simply a 5-20pF variable cap across the 9Mohms.

Is it because I do not have a good impedance match with the load vs cable? Running simulations with PSpice. Don't really know what I'm doing though. Just in case anybody cares, my O'scope has an input of 1Mohm 30pF and my signal generator will output 10V @ 50ohms.

EDIT #2:
Looks like I answered my own question.

"Above is the schematic of the circuit inside a X10 probe. You can see that it is basically a voltage divider. Rp and Cp are selected to form a 10 to 1 divider with the input of the scope. Assume that the scope has 1 Meg-Ohm input resistance and 100 pF of input capacitance. Then Rp is 9 Meg-Ohms and Cp is 9 pF (remember: small C gives big X). Note that Cp is adjustable. That's to allow for adjusting the response of the cable to fast rise times. Often Cp can be adjusted with a small screwdriver. It may be located at the probe end of the cable, or at the end that attaches to the scope."

I did some simulations, and then went down and tried the 5-20pF capacitor... IT WORKS! Quite well I might add. I have to adjust every 20Khz or so, but I don't think that's too bad! I'll post more information later. I'm going to go add the cap into the case...

EDIT#3
Weird... *echoes*, (2 days later) I thought this would be a semi-popular thread.

Guess nobody around here's as cheap as I am; or maybe everyone is just overtaken with awe and ashamed to post... LOL. Anyhow, I thought it was important to note, that once you have the trimmer capacitor "tuned" you do not have to adjust it "every 20Khz or so" and the signal is actually very good. I saw a bit of distortion beginning at 1.9MHz, but I could not tell if that was the probe or the signal generator's fault as it's range is 2Mhz, and it's output was decreasing dramatically, not only that, the generator's signal was distorting as well. The 100x part of the probe needs a smaller capacitor than 5pF to make up it's voltage divider, but I think I will pass on finishing the 100x probe because you can "tune" it simply by sqeezing the divider box.

The 100x will be used for DC measurements only.

Steve Conner

Mon Jul 02 2007, 09:45AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Hey ..., I was browsing the board at random and I came across this thread that I'd never noticed before. Well done! Frequency compensation is a fiddly subject that many people just don't seem to get.

FWIW, there is still one reason to buy a manufactured probe, which I'll try to explain now. Coax cable has a certain amount of capacitance per foot of length. Of course it has inductance and impedance and stuff too, but let me ignore that for a minute! :P

Now, one goal when designing a scope probe is to make its impedance as high as possible, so it doesn't load the circuit and alter the very waveform you're probing. That is one reason why we almost always use 10X probes: they have ten times the resistance and one-tenth the shunt capacitance of the scope input alone.

But, the capacitance of the cable adds to the scope input capacitance. It can be considerable. For the popular RG58 coax, it's 100pF per meter. So a 2 meter lead made of RG58 increases your scope's input capacitance by a factor of almost 10! A 10x probe with such a cable will need a compensating capacitor of around 20pF, and hence its input capacitance will be roughly 20pF too. It's still not as bad as if you just used the RG58 with alligator clips on the end, loading the circuit with 230pF, but it's hardly great either.

Commercial probes use custom-made coax with foam dielectric and a hair-thin steel centre conductor, to get that capacitance right down. The only downside is that it breaks very easily, so don't use that probe like a tow rope to pull your Scope-mobile around. :(

Finally, in the light of this discussion, do you think it's better to have the 10:1 divider network in the probe, or at the scope end, or in the middle of the cable like you seem to have put it?

Part Scavenger

Tue Jul 03 2007, 12:56PM

Registered Member #79 Joined: Thu Feb 09 2006, 11:35AM
Location: Arkansas
Posts: 673

Hey thanks!

Finally, in the light of this discussion, do you think it's better to have the 10:1 divider network in the probe, or at the scope end, or in the middle of the cable like you seem to have put it?

Hehe, are you asking or telling? According to my research prior to building, it is better to have the resistor along the entire length of the probe wire. However, that was really hard to accomplish. But, I'm assuming (now that I think about it) that it would be better to put the resistor network at the probe because then the capacitance present between the load and my resistor box now is more likely to oscillate with/load down the item I am testing...

ShawnLG

Tue Jul 03 2007, 02:43PM

Registered Member #286 Joined: Mon Mar 06 2006, 04:52AM
Location:
Posts: 399

Cool, I have been thinking about building an active probe using a MAX4451 IC. This would get rid of the low impedance loading of the coax when probing high impedance circuits.

TheMerovingian

Sun Nov 25 2007, 10:40AM

Registered Member #14 Joined: Thu Feb 02 2006, 01:04PM
Location: Prato/italy
Posts: 383

Before building the probes (10:1) derived a formula to calculate the required compensation capacitance from the generalized voltage divider (using complex impedances). The scope itself puts out a nice 1khz square wave (rich in harmonics) to adjust the compensation. Obviously at higher frequencies the 1:10 probe works better than 1:1 (that is restricted under <<1Mhz)
The formula is:
Ccomp = Rint / Re * (Cint + Ccable) (all measures in pF and Mohms)

Ccomp is the required compensation capacitance
Rin is the scope resistive impedance
Re is the divider resistor (eg 9Mohm with 1Mohm scope capacitance)
Cint is the scope capacitance (20-40pF usually)
Ccable is the cable capacitance (use a pF meter)

I didn't do the calculus for the phase shift but i think that is properly compensated, the fase shift is zero. (using the 1:1 probe and the 10:1 probe in x-y lissajous mode i have noticed a small phase shift probably due to the low pass filter made up of the cable capacitance of the 1:1 probe and the internal impedance of my funcion generator, so the 10:1 is better)

I wanted to develop the divider formula to obtain the required compensation capacitor accuracy to achieve, for eg -+1dB (+-10%) ad a given frequency but it is not very useful. Better by trial and error using square waves

the full generalized formula is:

DVR (divider ratio) = Ri / (1+2*pi*j*f*(Cint+Cc)*Rint)
-------------------------------------------------- ---------------------
Ri / (1+2*pi*j*f*(Cint+Cc)*Rint) + Re / (1+2*pi*j*f*Ccomp*Re)

the only way to make it independant of frequency and zero the phase shift it to equal (1+2*pi*j*f*(Cint+Cc)*Rint) with (1+2*pi*j*f*Ccomp*Re), so it can vanish

1+2*pi*j*f*Ccomp*Re = 1+2*pi*j*f*(Cint+Cc)*Rint
and then
2*pi*j*f*Ccomp*Re = 2*pi*j*f*(Cint+Cc)*Rint
and then
Ccomp*Re = (Cint+Cc)*Rint
so
Ccomp = (Cint+Cc)*Rint/Re

that is the desidered formula

i didn't want to dig in textbooks, and since a was in a train i started to do calculations on paper, better to use the brains

of course you need to remember that your impedance is now 10Mohm and capacitanca slighty less than Ccomp, be careful to not load down your circuit

ps for steve conner: using the divider on the scope end will result in a very high imput capacitance of the probe, loading down the scoped waveform at higher freq, and the probe end you risk oscillations of Lcable and Ccable since the damping factor is very underdamped (some sort of termination would help maybe), in the middle you have both problems lol. I don't understand why they don't make the inner conductor of the coax of a poor conducting material (maybe 100Ohm per metre). The scope probes have a thin conductor as steve previously sayed, with a high distributed resistance. If possile, getting out the inner conductor and replacing it with 30 awg wire would be cool (0,4ohm per metre) but very difficult

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob