Homebrew HV hiZ scope probe

[josephkk]

Correct, except it is 1000:1; all else is OK.

I am purposely using a larger capacitance as i am not mechanically
inclined; easy to put plastic tube around resistors, centered with thin
sheet, and wrap adhesive copper foil on outside cut for same length as
resistor.
Easy to calculate capacitance of the resulting coax (resistor / air /
inner tube surface / acrylic / outer tube surface = = floating shield.

Whoosh

?-/

 

[josephkk]

That's just nuts!

How squirrely is this going to get?

?-))

 

[Robert Baer]

There's a problem with that. The proposed 1:1000 single-stage probe
requires a parallel compensating capacitance 1/1000 of the total
scope-plus-cable capacitance, which is likely to be in the order of 50pF,
which implies a compensating capacitor of 50 fF, or 0.05 pF. across the
probe series resistor. This is probably impossible to realize.

The Tektronix 1:100 HV probe used a special "leaf-and-collar" capacitor of
a few pF, across the 100M probe resistor, which is taking things about as
far as they can go, and still withstand the voltage gradient across the
resistor. The capacitor dielectric was Freon, rather than air, giving a
higher dielectric strength. Freons have a permittivity of around 2, which
helps as well. The probe actually had a load of 100K at the scope end of
the cable, not just the 1meg/22pF scope input, giving 1:1000 ratio.
Compensation adjustment was done at the scope end of the cable, as is
ubiquitous in higher-end probes today.

Correct, except it is 1000:1; all else is OK.

I am purposely using a larger capacitance as i am not mechanically
inclined; easy to put plastic tube around resistors, centered with thin
sheet, and wrap adhesive copper foil on outside cut for same length as
resistor.
Easy to calculate capacitance of the resulting coax (resistor / air /
inner tube surface / acrylic / outer tube surface = = floating shield.

 

[Robert Baer]

Says so on the schematic in the probe manual.

Probe cables, at least modern ones, use foam dielectric, hence lower
capacitance per foot.

I've got some probes with removable cables, I'll TDR one when I get the
time. I'll have a go at open and shorted measurements, too, and calculate
the complex Zo. With a resistive center, Zo will be significantly complex
at higher frequencies than "regular" coax.

Thought of a way to (perhaps) make resistive coax with element in the
50 ohm per foot region.
Find some Nichrome or other resistive (heater) wire,and use Teflon
sleeving in successive layers or use shrink tubing in successive layers.
Slip braid around that.

 

[Spehro Pefhany]

Acorn't tell you!

Yew guys are knot getting board with these jokes?


("My Tree Puns"... Fred MacMurray?)


Best regards,
Spehro Pefhany

 

[Fred Abse]

Certainly not disagreeing with you Fred. In fact I don't know of any
1000:1 probes that are good at high frequency measurements. But for DC
measurements, you would want a 1Gohm probe for best accuracy measurements.

The discussion is about oscilloscope probes. Thousand meg. DC HV probes
are a dime a dozen. Most for use with 10M DMMs, so there's a 1M load
resistor in those. I have one. Don't use it much.

Also, I think I read somewhere that the Freon 114 used in the 6015 probe
had a permittivity near to 1. There was some discussion of using F-11 as
a replacement but its permittivity is up there and the probe does not
work well with it.

Could be. I just looked up the CFC dielectric properties table in Kaye and
Laby. All there are around 2. It doesn't specifically mention 1.1.4.

There is some good information on the yahoo tekscopes group archives
concerning this probe.

I don't do Yahoo groups.

 

[Fred Abse]

Correct, except it is 1000:1; all else is OK.

Whassadifference?

My usual convention for transfer functions is Vout/Vin.

Equal to 1/1000.
Or 1:1000

BTW, I would never use a high-voltage probe that didn't have an internal
load .

Consider the situation where there is HV applied, and the scope amplifier
is inadvertently switched to GND (which means the input socket will be
open circuit).
Practically the full HV will appear at the amplifier input socket.
The insulation won't stand that.

Even worse, a 7A13, switched to "near infinite impedance".

Relying on the scope's input resistance is dangerous. We all make mistakes.

I'm sure that UL, CSA, VDE, TüV, et al. would agree.

 

[Robert Baer]

Nicrhome is ok for heater uses, but i prefer constantan for electronic
purposes. Check out the properties differences, it will be worth it.

?-)

I vaguely one of them can be soldered; which one?

 

[Robert Baer]

Nicrhome is ok for heater uses, but i prefer constantan for electronic
purposes. Check out the properties differences, it will be worth it.

?-)

I vaguely remember that one can be soldered; which one is that?

 

[Robert Baer]

Whoosh

?-/

Look at it this way; two coax capacitors in parallel one inside the
other.
First one from resistor as center conductor and inner surface of
plastic tube for "outer" coax and air being the dielectric.
Second one has its center conductor as the inner surface of the
plastic tube, and the outer coax is the outside that has the conductive
foil; the plastic being the dielectric.

In physics, that inner surface would be called a Gaussian surface.

 

[Spehro Pefhany]

I vaguely remember that one can be soldered; which one is that?

Constantan. Type T (Cu-Constantan) can have junctions soft-soldered
without any kind of special flux. Some J (Fe-Constantan) wire has the
iron copper-clad so the same is true.

--sp


Best regards,
Spehro Pefhany

 

[Robert Baer]

You can solder to constantan, for a start!

Cheers

Phil Hobbs

Thanks!
I knew one of them could be soldered..so that is the wire to use if
the right resistivity can be has with a "reasonable" wire gauge (for the
sleeving and outer braid shield).

 

[Fred Abse]

?-/
Look at it this way; two coax capacitors in parallel one inside the
other.
First one from resistor as center conductor and inner surface of
plastic tube for "outer" coax and air being the dielectric.
Second one has its center conductor as the inner surface of the
plastic tube, and the outer coax is the outside that has the conductive
foil; the plastic being the dielectric.

Have you forgotten the stated end-to-end capacitance per Ohmite data?

In the case of MOX-2-13XXXX it's 0.6pF. I suspect that this will be the
dominant parameter.

 

[Robert Baer]

Constantan. Type T (Cu-Constantan) can have junctions soft-soldered
without any kind of special flux. Some J (Fe-Constantan) wire has the
iron copper-clad so the same is true.

--sp


Best regards,
Spehro Pefhany

Tried the BabyBird (GooGull); most hits were Chinese sellers, many do
not say how to buy, those that do say how much do not give resistivity
of the wire (ohms per unit length).
So still have zero idea if any of the wire that might be available is
in the right resistance range.
Nobody seems to have a range of wire sizes.

 

[Fred Abse]

I've got some probes with removable cables, I'll TDR one when I get the
time. I'll have a go at open and shorted measurements, too

The sample was too long to get meaningful open and shorted measurements,
but the TDR showed some unexpected results.

On a 1.2 meter length of cable. transit time was 9.5 nanoseconds, giving
a velocity of 42.11% of c. That corresponds to a dielectric permittivity
of 5.64, which is too high for any flexible dielectric I know of. That
suggests that the inner conductor is a helix. Resistance is 186.66 ohms
per meter. Inductance calculates (from rho at the sending end, and
velocity), to be 1.07 uH per meter, and capacitance 58.6 pF per meter.

The following model corresponds quite closely with measured data:


..model scopecbl ltra (
+ len=1.2
+ R=186.666
+ L=1.07E-006
+ C=5.86E-011)

The following is a good approximation to what the TDR shows. Change the
time (X) axis to "time/2" to show one-way time.


Version 4
SHEET 1 880 680
WIRE -160 128 -320 128
WIRE -16 128 -64 128
WIRE -320 272 -320 208
WIRE -160 272 -160 160
WIRE -160 272 -320 272
WIRE -64 272 -64 160
WIRE -64 272 -160 272
WIRE -16 272 -16 160
WIRE -16 272 -64 272
WIRE 32 272 -16 272
WIRE 80 272 80 160
WIRE 80 272 32 272
FLAG 32 272 0
SYMBOL ltline 32 144 R0
SYMATTR InstName O1
SYMATTR Value scopecbl
SYMBOL voltage -320 112 R0
WINDOW 3 -159 -8 Left 2
WINDOW 123 24 132 Left 2
WINDOW 39 24 28 Left 2
SYMATTR Value PULSE(0 1 0 22p 22p 1u 2u 1)
SYMATTR SpiceLine Rser=50
SYMATTR InstName V1
SYMBOL tline -112 144 R0
SYMATTR InstName T1
SYMATTR Value Td=1n Z0=50
TEXT -312 384 Left 2 !.tran 0 100n 0 1p
TEXT -312 336 Left 2 !.opt plotwinsize=0
TEXT -40 336 Left 2 !.model scopecbl ltra (\n+ len=1.2\n+ R=186.666\n+ L=1.07E-006\n+ C=5.86E-011)
TEXT -312 360 Left 2 !.plot v(n001)
TEXT -176 72 Left 2 ;TDR Simulation

 

[Robert Baer]

The sample was too long to get meaningful open and shorted measurements,
but the TDR showed some unexpected results.

On a 1.2 meter length of cable. transit time was 9.5 nanoseconds, giving
a velocity of 42.11% of c. That corresponds to a dielectric permittivity
of 5.64, which is too high for any flexible dielectric I know of. That
suggests that the inner conductor is a helix. Resistance is 186.66 ohms
per meter. Inductance calculates (from rho at the sending end, and
velocity), to be 1.07 uH per meter, and capacitance 58.6 pF per meter.

The following model corresponds quite closely with measured data:


.model scopecbl ltra (
+ len=1.2
+ R=186.666
+ L=1.07E-006
+ C=5.86E-011)

The following is a good approximation to what the TDR shows. Change the
time (X) axis to "time/2" to show one-way time.


Version 4
SHEET 1 880 680
WIRE -160 128 -320 128
WIRE -16 128 -64 128
WIRE -320 272 -320 208
WIRE -160 272 -160 160
WIRE -160 272 -320 272
WIRE -64 272 -64 160
WIRE -64 272 -160 272
WIRE -16 272 -16 160
WIRE -16 272 -64 272
WIRE 32 272 -16 272
WIRE 80 272 80 160
WIRE 80 272 32 272
FLAG 32 272 0
SYMBOL ltline 32 144 R0
SYMATTR InstName O1
SYMATTR Value scopecbl
SYMBOL voltage -320 112 R0
WINDOW 3 -159 -8 Left 2
WINDOW 123 24 132 Left 2
WINDOW 39 24 28 Left 2
SYMATTR Value PULSE(0 1 0 22p 22p 1u 2u 1)
SYMATTR SpiceLine Rser=50
SYMATTR InstName V1
SYMBOL tline -112 144 R0
SYMATTR InstName T1
SYMATTR Value Td=1n Z0=50
TEXT -312 384 Left 2 !.tran 0 100n 0 1p
TEXT -312 336 Left 2 !.opt plotwinsize=0
TEXT -40 336 Left 2 !.model scopecbl ltra (\n+ len=1.2\n+ R=186.666\n+ L=1.07E-006\n+ C=5.86E-011)
TEXT -312 360 Left 2 !.plot v(n001)
TEXT -176 72 Left 2 ;TDR Simulation

Speak of variable results..
Concerning the original Q&D probe where the scope must be DC coupled
ONLY, i have determined that pickup of external signals (hum and related
bazz-fazz), having the floating shield is a bit better than not having
it at all.
The negative of that is, getting pinkies close to that shield does
severely compromise the risetime & undershoot of the probe.

Concerning the "development" of the two 1G internally terminated
probe, each of those resistors seem to be too long and pickup becomes
intolerable, and a floating shield might have to be referred to ground
(maybe a resistor).
It also looks like a multiple pi-pad model is not the best for
modelling these resistors..

 

[Fred Abse]

Speak of variable results..
Concerning the original Q&D probe where the scope must be DC coupled
ONLY, i have determined that pickup of external signals (hum and related
bazz-fazz), having the floating shield is a bit better than not having it
at all.
The negative of that is, getting pinkies close to that shield does
severely compromise the risetime & undershoot of the probe.

I wouldn't want fingers anywhere near an energized HV probe. The right
place to put compensation adjustment it at the scope end, like most good
commercial probes do.

Concerning the "development" of the two 1G internally terminated
probe, each of those resistors seem to be too long and pickup becomes
intolerable, and a floating shield might have to be referred to ground
(maybe a resistor).

Use a fixed shield, referred to ground, and work with whatever capacitance
you're left with. Minimize capacitance to ground as much as possible, it's
the end-to-end capacitance that *really* matters for compensation.
According to Ohmite, that's 0.6pF. For a 1:1000 probe, that means 600pF at
the scope input.

It also looks like a

multiple pi-pad model is not the best for

modelling these resistors..

I'd agree with that. You need to do some work to determine what they
really look like.

As an aside about modeling, I discovered that LTSpice supports the TXL (Y)
transmission line model, though it's undocumented. It runs faster than the
LTRA model, and also supports full RLCG, whereas LTRA only supports
RLC. G can be a parameter, which suggests that may be a way of modeling
frequency-dependent dielectric loss.

 

[Robert Baer]

I wouldn't want fingers anywhere near an energized HV probe. The right
place to put compensation adjustment it at the scope end, like most good
commercial probes do.


Use a fixed shield, referred to ground, and work with whatever capacitance
you're left with. Minimize capacitance to ground as much as possible, it's
the end-to-end capacitance that *really* matters for compensation.
According to Ohmite, that's 0.6pF. For a 1:1000 probe, that means 600pF at
the scope input.

multiple pi-pad model is not the best for

I'd agree with that. You need to do some work to determine what they
really look like.

As an aside about modeling, I discovered that LTSpice supports the TXL (Y)
transmission line model, though it's undocumented. It runs faster than the
LTRA model, and also supports full RLCG, whereas LTRA only supports
RLC. G can be a parameter, which suggests that may be a way of modeling
frequency-dependent dielectric loss.

I could find no end-to-end capacitance spec for these resistors; i
calculate 0.125pF end-to-end on the basis of endcap spacing and aluminum
oxide insulator.
See http://www.oil4lessllc.org/HV probes/1Gresistor.pdf for
dimensions and calculations.
The cylindrical capacitance info at the bottom of the second page is
reference only for now.

 

[Fred Abse]

I could find no end-to-end capacitance spec for these resistors;

MOX-2-12 series 0.6pF end to end.

Ohmite PDF catalog October 26 2011. "Power resistors, Component Selector
4000J"

Page 93 in PDF version, labeled 85. "Maxi-Mox".

 

[Robert Baer]

MOX-2-12 series 0.6pF end to end.

Ohmite PDF catalog October 26 2011. "Power resistors, Component Selector
4000J"

Page 93 in PDF version, labeled 85. "Maxi-Mox".

Well,i snooped around their site and could find no way to get or find
that catalog.
No way to get that description, etc.
Dead end.
The best i could find, using MOX as a 3-character selector for part
number, thengetting the PDF for the Max-Mox series,was a one page PDF
named res_maximox.pdf; page labeled 85.
Now i see (only since you pointed it out) the 0.60pF .
On the basis of my calculations, i do not see how it can be that large.
I guess i will have to make a rather sensitive capacitance
measurement device, as DVMs tend to imply the low value i calculated.