*As a few others have pointed out, even at a bias voltage of 250V across the resistor, the dissipation is fairly small, 62.5uW. It is therefore safe to assume that the thermal impact is a weak effect on the resistor. Other issues like nonlinearity & leakage may be playing a stronger role here. It we stabilize the resistor temperature, we can isolate the other effects.*
I used a ton of the single channel version of this on automated production test systems for 1000V high voltage high resistance cable leakage. The triaxial test cables were a thousand bucks a pop! Memories...
I am so glad that i can watch and learn from someone that has tremendeous knowledge about electronics like You.Best wishes 🎄.Live long and prosper 🤗 🖖.
The best electronics channel by far.never thought I would need to measure in the G ohms region until I started service and repairing large diaphragm condenser microphones they all use 1 G ohm to bias the diaphragm.
My way of measuring high resistance used to be the otherwise rarely applied special scale of Fluke 8060. My use for it was to sort some diodes by their leakage current - at first 1N914 or its newer clone to pick ones that leaked below 4 nA at the meter 9 V, but then I found it more practical to specify FDH300. Later on, I got a HP 4329A that I have occasionally used for a high voltage DC output. Never needed yet for its actual tera-ohm and beyond capability. As it happens, I also have an auction find, a Keithley electrometer. When cleaning and studying it, I was puzzled by its lack of heater connections on its vacuum tube, but then got a manual for it and found the lack of thermionic function to be fully intentional, or actually the key feature fort the meter. Incredibly smart!
Radiometer IM6 is best high resistance meter I know of. Gives the most consistent readings. TOhm and beyond, accuracy is low, a few % at best, the meter with most consistent readings wins, usually also the meter with best construction/design and care for good isolation, guarding... Also have two of those HP-4349B in the shelf, Radiometer IM6 is the one that gets used when measuring e.g. HV dividers.
250V over 1GOhm gives 62.5µW only, I hope I calculated that correctly (3 times), so that can't be the reason. There's a voltage coefficient for resistors, but I never assumed it being that prominent. Maybe there's a voltage isolation break down of you cables, or in the plastic of your assembly. Strange behavior, and very interesting video, anyway.
You need to clean those resistors. I have build a resistance standard for a client. It in 10 steps went from 10M to 5GOhm. All connectors were mounted in teflon disks for lowest leakage. The resistors were delivered from the distributor in plastic bags and some had fingerprints on them. I mailed with the manufacturer about that and they advised me to clean them with IPA because fingerprints could cause extra leakage current over the outside. For most if the test voltage is high. (This standard is in use in a cal lab and they use voltages up to 5 kV )
That contact checker looks like the same thing you would use for ground loop detection and it could actually also detect a ground loop between shielding on the triax. Since one end of the inner shielding is open, it provides no magnetic shielding at 30khz and all the current is induced in the central conductor, which has a loop so you detect a current if a DUT is connected, but if the inner shield is not left open at the other end and touches the outer shield then you get a ground loop through the shields and no pick up on the central line so a conductor ground loop would also look like an open which is good to know. Would be an interesting and quick experiment to try.
My first ever milliohm meter was just an LM317 constant current source and a multimeter my first ever gigaohm meter was a transimpedance amplifier and a multimeter
there are some great old school keithley electrometers out there for very cheap they can measure small currents, high resistances, and coulombs. i think the max resistance range on mine is 10^13 ohms.
I have an old analog HP 4329A so it goes even higher ( 2-10^16) but only if you have the special cable for it (I do not have that and without that the max I could measure was 10 ^12. I was standing a meter away and the needle moved slowly up and down a bit. Turned out it was the rhythm of my breathing 🙂
I wonder how much of the observed voltage dependence is really due to the heating that was mentioned and how much is due to nonlinearity. High-value resistors often have a specified voltage dependence that is a true deviation from Ohm's law and not just a temperature coefficient.
It is definitely a combination of things, leakage, thermal, non-linearity, etc. If I could stabilize the resistor temperature, we could isolate the non-linear behavior from the thermal one.
@@Thesignalpath 250 V over 1 Gohm is only 62.5 µW, so the temperature rise in this large resistor due to the bias ought to be rather small. A non-quantitative way of separating the effect of heating from the effect of voltage would be to gently blow some hot air on the resistor while it is biased at a lower voltage to see how the resistance changes. Maybe while monitoring the temperature with a thermal camera. If it is takes many degrees of heating to change the resistance the same amount as the 250 V bias did (or if it goes in the other direction), the observed effect is probably not thermal to a significant degree. Other effects may be harder to separate.
I'm surprised it doesn't have an integrated voltage reference output (something like a 50V supply capable of supplying 1mA or so) or at least a sense input to measure the DC voltage used. I guess that means it's technically just a very low current ammeter with built in capability to calculate resistance?
Why not use the sense connections? The lack of an internal voltage source is puzzling. I believe you said the middle shield is not energized. That guard shield should have comparable voltage to the supply to prevent leakage. Your use of this meter seems at odds with what I'm learning about electrometer measurements.
The measurement is degraded by using conductors to connect everything. There is a RUclipsr PhD named Derek who proves that all electron flow is outside the conductor on his "Truth" channel. Unfortunately his PhD is in physics, with a Masters in self promotion, which he got from the university of UUU*. *Unicorn University of Uranus
That was a typo, it is already corrected. At first I had it as 1e14, then I added the 10 but forgot to change the "e" to a "x". Although 1e10^14 would be an insane number! ;)
@@Thesignalpath Eh, just a double exponential. Insane for a resistance, but not so much for a number. Double exponentials are faster growing than pretty much all "real life" functions, but quite slow compared to a lot of functions studied in combinatorics. Harvey Friedman has done a lot of work that happens to involve extremely fast-growing functions. There's also a tiny group of hobbyists who try to define the fastest-growing computable functions possible, and use them to define extremely large finite numbers.
*As a few others have pointed out, even at a bias voltage of 250V across the resistor, the dissipation is fairly small, 62.5uW. It is therefore safe to assume that the thermal impact is a weak effect on the resistor. Other issues like nonlinearity & leakage may be playing a stronger role here. It we stabilize the resistor temperature, we can isolate the other effects.*
I used a ton of the single channel version of this on automated production test systems for 1000V high voltage high resistance cable leakage. The triaxial test cables were a thousand bucks a pop! Memories...
Oh yes... I scored a Keithey electrometer for a great price... only to spend even more money on a damn triax cable so i could use it!
I am so glad that i can watch and learn from someone that has tremendeous knowledge about electronics like You.Best wishes 🎄.Live long and prosper 🤗 🖖.
The best electronics channel by far.never thought I would need to measure in the G ohms region until I started service and repairing large diaphragm condenser microphones they all use 1 G ohm to bias the diaphragm.
My way of measuring high resistance used to be the otherwise rarely applied special scale of Fluke 8060. My use for it was to sort some diodes by their leakage current - at first 1N914 or its newer clone to pick ones that leaked below 4 nA at the meter 9 V, but then I found it more practical to specify FDH300. Later on, I got a HP 4329A that I have occasionally used for a high voltage DC output. Never needed yet for its actual tera-ohm and beyond capability. As it happens, I also have an auction find, a Keithley electrometer. When cleaning and studying it, I was puzzled by its lack of heater connections on its vacuum tube, but then got a manual for it and found the lack of thermionic function to be fully intentional, or actually the key feature fort the meter. Incredibly smart!
Careful, Pooch shows up and bring along plenty of extra charge to disturb these measurements!
Needs more hotdog.
Radiometer IM6 is best high resistance meter I know of.
Gives the most consistent readings. TOhm and beyond, accuracy is low, a few % at best, the meter with most consistent readings wins, usually also the meter with best construction/design and care for good isolation, guarding...
Also have two of those HP-4349B in the shelf, Radiometer IM6 is the one that gets used when measuring e.g. HV dividers.
250V over 1GOhm gives 62.5µW only, I hope I calculated that correctly (3 times), so that can't be the reason. There's a voltage coefficient for resistors, but I never assumed it being that prominent. Maybe there's a voltage isolation break down of you cables, or in the plastic of your assembly. Strange behavior, and very interesting video, anyway.
I tend to agree, the power is pretty small. It is more likely that other effect are dominant. I will add a note to the video.
250^2 / 1,000,000,000 = 0.00000625W That’s 6.25uW
You need to clean those resistors. I have build a resistance standard for a client. It in 10 steps went from 10M to 5GOhm. All connectors were mounted in teflon disks for lowest leakage. The resistors were delivered from the distributor in plastic bags and some had fingerprints on them. I mailed with the manufacturer about that and they advised me to clean them with IPA because fingerprints could cause extra leakage current over the outside. For most if the test voltage is high. (This standard is in use in a cal lab and they use voltages up to 5 kV )
Yeah, another repair vid!
I’ve been watching for one of those on eBay for ages, for whatever reason I never saw that one, damnit.
That contact checker looks like the same thing you would use for ground loop detection and it could actually also detect a ground loop between shielding on the triax. Since one end of the inner shielding is open, it provides no magnetic shielding at 30khz and all the current is induced in the central conductor, which has a loop so you detect a current if a DUT is connected, but if the inner shield is not left open at the other end and touches the outer shield then you get a ground loop through the shields and no pick up on the central line so a conductor ground loop would also look like an open which is good to know. Would be an interesting and quick experiment to try.
I love these kind of repairs. I have repaired several instruments like the HP 3335A that needed the same "repair". :)
My first ever milliohm meter was just an LM317 constant current source and a multimeter
my first ever gigaohm meter was a transimpedance amplifier and a multimeter
there are some great old school keithley electrometers out there for very cheap they can measure small currents, high resistances, and coulombs. i think the max resistance range on mine is 10^13 ohms.
I have an old analog HP 4329A so it goes even higher ( 2-10^16) but only if you have the special cable for it (I do not have that and without that the max I could measure was 10 ^12. I was standing a meter away and the needle moved slowly up and down a bit. Turned out it was the rhythm of my breathing 🙂
I wonder how much of the observed voltage dependence is really due to the heating that was mentioned and how much is due to nonlinearity. High-value resistors often have a specified voltage dependence that is a true deviation from Ohm's law and not just a temperature coefficient.
It is definitely a combination of things, leakage, thermal, non-linearity, etc. If I could stabilize the resistor temperature, we could isolate the non-linear behavior from the thermal one.
@@Thesignalpath 250 V over 1 Gohm is only 62.5 µW, so the temperature rise in this large resistor due to the bias ought to be rather small. A non-quantitative way of separating the effect of heating from the effect of voltage would be to gently blow some hot air on the resistor while it is biased at a lower voltage to see how the resistance changes. Maybe while monitoring the temperature with a thermal camera. If it is takes many degrees of heating to change the resistance the same amount as the 250 V bias did (or if it goes in the other direction), the observed effect is probably not thermal to a significant degree. Other effects may be harder to separate.
I tend to agree, the power is pretty small. It is more likely that other effect are dominant. I will add a note to the video.
I'm surprised it doesn't have an integrated voltage reference output (something like a 50V supply capable of supplying 1mA or so) or at least a sense input to measure the DC voltage used. I guess that means it's technically just a very low current ammeter with built in capability to calculate resistance?
Why not use the sense connections?
The lack of an internal voltage source is puzzling. I believe you said the middle shield is not energized. That guard shield should have comparable voltage to the supply to prevent leakage. Your use of this meter seems at odds with what I'm learning about electrometer measurements.
It is always the connector when you are repairing, I am not that lucky 😅
4339B sells for 3k to 4K USD in Yahoo auction Japan.
what is the application for a 1G Ohm resistor?
If you ever want to convert a really small current to a “measurable” voltage, you would have to essentially pass that current over a resistor. :)
Can you do a video on how to measure with triax and guard and shield.
Click-bait thumbnail got me ... 1e10¹⁴ Ω .😅
The instrument can actually measure up to 1x10^15! But not in my current setup. There was also a typo I fixed. D'oh!
At least you're not playing with LEDs...
So the 4339B is the better model yet the number is lower. Thats odd.
I measure my resistors with a good set of vernier calipers.
All the cool engineers use micrometers.
Meow
The measurement is degraded by using conductors to connect everything. There is a RUclipsr PhD named Derek who proves that all electron flow is outside the conductor on his "Truth" channel.
Unfortunately his PhD is in physics, with a Masters in self promotion, which he got from the university of UUU*.
*Unicorn University of Uranus
And much giga Ohms is the cat ? you should test it !!! ja ja ja ja
Thumbnail is kinda messed up. It's either 1e+15, 10e+14, or 10¹⁵, or 1x10¹⁵, or 10x10¹⁴. But 1e10¹⁴ is just nonsense.
That was a typo, it is already corrected. At first I had it as 1e14, then I added the 10 but forgot to change the "e" to a "x". Although 1e10^14 would be an insane number! ;)
@@Thesignalpath Eh, just a double exponential. Insane for a resistance, but not so much for a number. Double exponentials are faster growing than pretty much all "real life" functions, but quite slow compared to a lot of functions studied in combinatorics. Harvey Friedman has done a lot of work that happens to involve extremely fast-growing functions. There's also a tiny group of hobbyists who try to define the fastest-growing computable functions possible, and use them to define extremely large finite numbers.
@@Thesignalpath Makes me wonder what kind of object would have such a huge resistance. Maybe if you measured across the entire planet?