EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: MarkKruetzmann on October 28, 2019, 08:53:38 pm
-
Hi!
I recently bought a Tek 465 which is dead and while it isn't even here yet, I was asking myself on how to measure the 15.5kV in that oscilloscope. I didn't think they were that simple, but as it turns out, high voltage probes are basically just a high resistance high voltage resistor with a handle attached. You can see it here: Edit: at 2:33 (starting at a defined point does not work in the embedded player I think)
https://www.youtube.com/watch?v=-_LJXKLgIYc&t=2m33s (https://www.youtube.com/watch?v=-_LJXKLgIYc&t=2m33s)
Wait some seconds as he pulls out the resistor.
As the cheapest probes I can find are still about 80€ or so, I figured I could do that myself quite easy with less than 15 bucks in parts. All I need is a ~1GOhm resistor capable of 20kV or more, maybe add a 10MegOhm or 1MegOhm resistor directly as to form a "worst case" voltage divider and hook it up to some (not even HV) cable and test leads.
Worst case would be the GOhm-range input impedance of my HP3456A if it is accidentally in auto-range. The input capacitance might save it, but I better not count on that. Proper grounding beforehand is mandatory in any case.
Anyway, my question is: Is a DIYed high voltage probe safely usable? I would do the following:
Print a probe from PLA that "wraps around" a 1GOhm resistor in 2 halves. They will later be epoxied to each other.
Make a probe tip out of a brass M4 screw as the resistor has a M4 mount on each side.
Place the resistor away from the grip (as seen in the video).
Add a 1MegOhm resistor (or 1.11MegOhm to account for 10Meg input resistance) to the mentioned 1GOhm resistor and get two leads out or some coax. One with the 1/1000 of the measured voltage and one ground lead.
I figured that PLA has way better dielectric strength than air and therefore it should be just fine.
I found a nice 35kV rated (70kV peak for 5 minutes) resistor with 1 GOhm I would use:
https://highvoltageshop.com/epages/b73088c0-9f9a-4230-9ffc-4fd5c619abc4.sf/sec95a0d6e60c/?ObjectPath=/Shops/b73088c0-9f9a-4230-9ffc-4fd5c619abc4/Products/RES_35kV_1G (https://highvoltageshop.com/epages/b73088c0-9f9a-4230-9ffc-4fd5c619abc4.sf/sec95a0d6e60c/?ObjectPath=/Shops/b73088c0-9f9a-4230-9ffc-4fd5c619abc4/Products/RES_35kV_1G)
This thing certainly will be hard to use for anything but DC and very low frequency AC, but I guess the reasonably priced commercial ones don't have a particularly flat frequency response either. It will therefore be mostly a HV DC probe for meters with 10MOhm input impedance and with an attenuation factor of 1000. Quite a specialized product, but I guess that is just how it is in High voltage world. I will measure the resistance of the HV resistor with my HP3456A to tweak the second resistor or actually measure the attenuation with a 1kV source (Fluke 1653 installation tester. I know this would be kinda crude since this is just for insulation measurements but hey, it's the only more or less trustworthy HV supply I have and I can measure the voltage beforehand.)
My important question is: Did I oversee anything safety-wise?
Performance-wise, I would be happy enough with 5% DC. I opt for better than 1% DC and at least the capability to measure 50Hz High Voltage to say 10%, better would of course be better.
My not so important question is: How to improve frequency response? I guess I have a hell of a lot of stray capacitance, which is really bad with 1GOhm input. Is it possible to just add a capacitor over the 1.1MegOhm and get into kHz range? I think I do not want to try shielding the resistor to a defined input capacitance as this would severely impact safety if I am not cautious. This would also probably be anything but linear as the resistance is spread about 14cm or so of length. So, even with shielding, this sounds like a hard job for the compensation network.
Any input is welcome, as this in its current state is nothing more than a thought project.
-
The devil is in the details. I certainly wouldn't trust a 3D printed shell glued up from halves to provide safety critical insulation. All it takes is a small trace of semi-carbonised degraded plastic that had 'cooked' for too long stuck to the wall of the extruder head to be in a critical part of the print, or for the epoxy to be not quite fully and evenly mixed, and the insulation could break down without any warning.
I would strongly recommend building any such probe inside commercially available clear (to avoid potentially conductive pigments and fillers), thick wall Acrylic or Polycarbonate tubing, thick enough to have a dielectric breakdown voltage at least a factor of five greater than the max working voltage, with 3D printing only used for end fittings and grips.
N.B. Don't treat the low voltage end of the probe as safe - bad things happen if the ground clip comes off or if the HV resistor flashes over e.g. due to surface contamination or humidity.
-
Next best thing would be an acrylic tube with at least 2mm wall strength, but then I would still have the problem with surface degradation if that thing gets dirty. A printed handpiece wouldn't help - assuming worst case again with too hot printed PLA.
Also the use of 20kV or 50kV cabling to the meter would help, with all connections buried deeply in the acrylic. This would make it necessary to put the smaller resistor outside of the probe or using two HV cables.
I could alternatively place a hook at the probe tip and just move it around when the DUT is powered off. That would slow down measurements drastically though.
But I think what I really should do is not trust cobbled-together high voltage probes and either rely on safety-tested commercial options or just omit measurements above 1kV altogether. Hence this post.
Thank you for your input though. I still think that a DIYed probe is "relatively" safe, but as the kids say: "you only live once" and I intend to do that as long as possible.
I didn't think of carbonised plastic, but might go the acrylic tube road and make sure it is clean.
-
It is very questionable to ever need to measure a scopes PDA voltage and when so high it presents several risks.
Far better IMHO to just measure the EHT that the PDA is derived from as it's always less than 3 kV.
Far simpler to make a resistive divider and measure it with a DMM.
-
I don't normally work with enough energy to be lethal at real high voltages so I don't think about human safety too much. I'm more concerned with taking out my DSO's front end. If you use the search you will find a fair bit of info on various home made HV probes I have put together.
This is the one I use most often. I'm using ceramics for the insulator and copper foil to form the capacitor to achieve some decent bandwidth from it. Buying a good Tektronix probe may have been cheaper but not as much fun.
https://www.youtube.com/watch?v=NBCkkv9dwPA&list=PLZSS2ajxhiQCAQ6gIp6s-WoKiIEb1gHPD&index=2&t=0s (https://www.youtube.com/watch?v=NBCkkv9dwPA&list=PLZSS2ajxhiQCAQ6gIp6s-WoKiIEb1gHPD&index=2&t=0s)
-
As the cheapest probes I can find are still about 80€ or so, I figured I could do that myself quite easy with less than 15 bucks in parts. All I need is a ~1GOhm resistor capable of 20kV or more, maybe add a 10MegOhm or 1MegOhm resistor directly as to form a "worst case" voltage divider and hook it up to some (not even HV) cable and test leads.
Worst case would be the GOhm-range input impedance of my HP3456A if it is accidentally in auto-range. The input capacitance might save it, but I better not count on that. Proper grounding beforehand is mandatory in any case.
Anyway, my question is: Is a DIYed high voltage probe safely usable? I would do the following:
Print a probe from PLA that "wraps around" a 1GOhm resistor in 2 halves. They will later be epoxied to each other.
Make a probe tip out of a brass M4 screw as the resistor has a M4 mount on each side.
Place the resistor away from the grip (as seen in the video).
Add a 1MegOhm resistor (or 1.11MegOhm to account for 10Meg input resistance) to the mentioned 1GOhm resistor and get two leads out or some coax. One with the 1/1000 of the measured voltage and one ground lead.
I figured that PLA has way better dielectric strength than air and therefore it should be just fine.
I found a nice 35kV rated (70kV peak for 5 minutes) resistor with 1 GOhm I would use:
https://highvoltageshop.com/epages/b73088c0-9f9a-4230-9ffc-4fd5c619abc4.sf/sec95a0d6e60c/?ObjectPath=/Shops/b73088c0-9f9a-4230-9ffc-4fd5c619abc4/Products/RES_35kV_1G (https://highvoltageshop.com/epages/b73088c0-9f9a-4230-9ffc-4fd5c619abc4.sf/sec95a0d6e60c/?ObjectPath=/Shops/b73088c0-9f9a-4230-9ffc-4fd5c619abc4/Products/RES_35kV_1G)
This thing certainly will be hard to use for anything but DC and very low frequency AC, but I guess the reasonably priced commercial ones don't have a particularly flat frequency response either. It will therefore be mostly a HV DC probe for meters with 10MOhm input impedance and with an attenuation factor of 1000. Quite a specialized product, but I guess that is just how it is in High voltage world. I will measure the resistance of the HV resistor with my HP3456A to tweak the second resistor or actually measure the attenuation with a 1kV source (Fluke 1653 installation tester. I know this would be kinda crude since this is just for insulation measurements but hey, it's the only more or less trustworthy HV supply I have and I can measure the voltage beforehand.)
My important question is: Did I oversee anything safety-wise?
Performance-wise, I would be happy enough with 5% DC. I opt for better than 1% DC and at least the capability to measure 50Hz High Voltage to say 10%, better would of course be better.
My not so important question is: How to improve frequency response? I guess I have a hell of a lot of stray capacitance, which is really bad with 1GOhm input. Is it possible to just add a capacitor over the 1.1MegOhm and get into kHz range? I think I do not want to try shielding the resistor to a defined input capacitance as this would severely impact safety if I am not cautious. This would also probably be anything but linear as the resistance is spread about 14cm or so of length. So, even with shielding, this sounds like a hard job for the compensation network.
Any input is welcome, as this in its current state is nothing more than a thought project.
High voltage Oscope probes are expensive.
A more indepth understanding can be found hereL
http://emcesd.com/1ghzprob.htm (http://emcesd.com/1ghzprob.htm)
I have mad a DIY HV probe, but I don't remember exactly what resistors and capacitor I used.
I had to delete my discription of the one I made because I just checked the last resistance and it was 4 Mohms, not 100k.
So I can't tell you exactly how I made it.
-
Hi
As others have pointed out, the material for the outer handle is critical, hence the high price for commercial units.
Additionally, use multiple resistors in the divider chain, it will give better immunity to HV break down, see the instructables link.
https://www.instructables.com/id/DIY-HIGH-VOLTAGE-MULTIMETER-PROBE/ (https://www.instructables.com/id/DIY-HIGH-VOLTAGE-MULTIMETER-PROBE/)
-
Ok, so I double checked my DIY 100kv probe and found it to be about 1050:1 after putting 100V through it, and measuring the output.
I also used the ucurrent to double check it, and found about 4.3 Gig ohms as R1, while R2 was 4 Meg ohms.
The 4.3 Gigs were multiple resistors in series, and I'm guessing it was 99, 40 Meg ohm resistors, but I not sure because it was sealed but I remember it was a very long line of them and I made sure each resistor could handle the power that would be across each of them.
I used a 2 inch diameter, 3 foot PVC pipe, but I used super corona dope on the resistor chain first.
The whole thing was taped up with electrical tape and the pipe caps were sealed with superglue.
The resistor chain was also centered in the tube by using styrofoam cutouts, so the only point at which it touches the PVC is at the tip where I have a bolt as the probe tip.
I remember using measuring 40kv AC at 10khz, but only DC at about 80kv. I don't think I trusted it too much as I believe I never touch it while the circuit was hot.
Take it for what it's worth, but it's possible to do measurements with a DIY HV probe.
-
@tautech:
Thanks for the advice. the ~2.5kV or so would have been the first HV I'd have tested anyway. This thread is not about the scope repair though, I am just curious for the techology and details of HV probes. It's all for the fun in the end.
@joeqsmith:
Now that's an impressive build! I guess I have to watch your whole series on that puppy. Especially the frequency response is remarkable! I hope theory of operation and compensation techniques are described in the other five videos. Hats off to your accomplishment! It sure was a lot of more fun than to just buy a Tektronix probe. You did very good work here. I might try to copy it to some extent.
@MosherIV:
The guy in the instructables link does claim 15kV, but he totally relies on heatshrink tubing and a used garden hose for insulation. I would not put my life at stake for that. As you pointed out, material is critical, and that just does not look safe.
@sourcecharge:
The 1Ghz probe you linked doesn't exactly help... I am quite sure that in bigger probes compensation gets increasingly difficult. Physics is the same of course, but undesirable effects will show up way quicker with physically bigger probes and many orders of magnitude greater input impedance.
Your description of a HV probe sounds quite promising though, in that even a "simple" resistive divider chain without frequency compensation procedures produce an output usable to some kHz at least. Not fiddling around in a hot circuit is very probably a very good idea in HV measurements anyway.
-
I don't think I'd use PVC. PVC absorbs moisture. I recall long ago seeing in something like 73, CQ, or some other ham radio magazine that coil forms made from PVC tended to arc across the surface and form carbonized tracks.
-
I'm using Ohmite Slim-Mox (https://www.ohmite.com/catalog/slim-mox-series) SM108 (20kV) HV resistors, they are low cost and small. I also use the cheap chinese red HV resistors, like 5GIG 10W 100mm, but their coating is paint thin. No comp cap for DC work.
-
I don't think I'd use PVC. PVC absorbs moisture. I recall long ago seeing in something like 73, CQ, or some other ham radio magazine that coil forms made from PVC tended to arc across the surface and form carbonized tracks.
https://omnexus.specialchem.com/polymer-properties/properties/water-absorption-24-hours#PM-PP
Water Absorption 24 hours
Absorption Properties of Polymer
Water Absorption Values of Several Plastics
Polymer Name-----------------------------------------------------------Min Value (°C) Max Value (°C)
PVC (Polyvinyl Chloride), 20% Glass Fiber-reinforced-----------------------0.01 0.20
PVC, Plasticized----------------------------------------------------------------0.20 1.00
PVC Rigid-----------------------------------------------------------------------0.04 0.40
and as a comparison:
PTFE - Polytetrafluoroethylene------------------------------------------------0.005 0.010
PI - Polyimide------------------------------------------------------------------1.34 1.43
PETG - Polyethylene Terephthalate Glycol------------------------------------0.10 0.10
ABS - Acrylonitrile Butadiene Styrene----------------------------------------0.05 1.80
ABS Flame Retardant----------------------------------------------------------0.10 0.80
ABS High Heat-----------------------------------------------------------------0.10 0.80
ABS High Impact-------------------------------------------------------------- 0.10 0.80
(No PLA listed)
Well the plasticized PVC "Water Absorption Values" seems to be much higher than most plastics, but the rigid PVC look pretty good.
I also used super corona dope on the resistor chain, and electrical tape all around the outside of the PVC pipe.
Unfortunately, it's been so long since I used this at high voltage, I would verify it was working correctly at high voltages before touching it with a hot circuit.
-
Putting resistors in series is the way to go. I would use ones with axial leads, so much easier to deal with than the ones with the leads coming out sideways.
Use physically large enough, brand name resistors so that you can trust their ratings are real. Derate, add your own margin - then put in series to build up the full rating.
PVC heat shrink tubing is your friend. Ones with hot melt glue exist, sealing the connections. Use several layers of tubing to get the physical thickness. Then you can add a 3D printed case for decorative purposes, but I wouldn't trust it as a reliable insulator.
The idea of putting resistors in series is you will create a physically long chain. If you divide your 16kV to, say, the length of 32cm, you'll only have 500V/cm, which is much easier to deal with than, say, several kV per cm you get with optimistically rated single HV resistors. High voltage gradients require much more understanding of insulation materials: avoiding trapped air to avoid corona discharge carbonizing the materials slowly, avoiding materials that absorb moisture, and so on.
Example:
HVA12FA50M0 is 0.37€ at Digikey; 20pcs is 7.40€
https://www.seielect.com/catalog/sei-hva.pdf (https://www.seielect.com/catalog/sei-hva.pdf)
20 in series is 1 GOhm, total 160kV(!) rated working voltage.
Body length is 11mm, voltage per resistor is 16kV/20 = 800V, voltage gradient is 727 V/cm. Creepage distance is actually a bit longer.
Allowing 7mm for soldering the leads, total solution is 360mm in length.
-
...
PVC heat shrink tubing is your friend. Ones with hot melt glue exist, sealing the connections. Use several layers of tubing to get the physical thickness. Then you can add a 3D printed case for decorative purposes, but I wouldn't trust it as a reliable insulator
....
Not necessarily. In my Picoammeter experiments I've found heatshrink to be surprisingly leaky, even at low voltages...
https://www.eevblog.com/forum/projects/picoammeter-design/msg790280/#msg790280 (https://www.eevblog.com/forum/projects/picoammeter-design/msg790280/#msg790280)
At the very least, I would say that not all heatshrinks are created equal and that a sample of the one you intend to use should be specifically tested for resistivity.
-
https://omnexus.specialchem.com/polymer-properties/properties/water-absorption-24-hours#PM-PP
Water Absorption 24 hours
Absorption Properties of Polymer
Water Absorption Values of Several Plastics
Polymer Name-----------------------------------------------------------Min Value (°C) Max Value (°C)
PVC (Polyvinyl Chloride), 20% Glass Fiber-reinforced-----------------------0.01 0.20
PVC, Plasticized----------------------------------------------------------------0.20 1.00
PVC Rigid-----------------------------------------------------------------------0.04 0.40
and as a comparison:
PTFE - Polytetrafluoroethylene------------------------------------------------0.005 0.010
PI - Polyimide------------------------------------------------------------------1.34 1.43
PETG - Polyethylene Terephthalate Glycol------------------------------------0.10 0.10
ABS - Acrylonitrile Butadiene Styrene----------------------------------------0.05 1.80
ABS Flame Retardant----------------------------------------------------------0.10 0.80
ABS High Heat-----------------------------------------------------------------0.10 0.80
ABS High Impact-------------------------------------------------------------- 0.10 0.80
(No PLA listed)
Well the plasticized PVC "Water Absorption Values" seems to be much higher than most plastics, but the rigid PVC look pretty good.
Well, is that surprising, given that we use PVC for water pipes? It wouldn't really work if it absorbed tons of water, I would think.
-
Well, is that surprising, given that we use PVC for water pipes? It wouldn't really work if it absorbed tons of water, I would think.
Not in the least, but I'm pretty sure when I was making this thing, someone in another forum I was posting at may have commented on that very same issue, and I double checked it then.
In any case, the out wall is like 2 to 3 mm thick, and it's not being used as a coil, but as a barrier from the HV on the chain of resistors.
Because the resistor chain was first dipped in super corona dope, (4100V/mil), centered with styrofoam cutouts in the tube, and then the tube was sealed and coated with multiple layers of electrical tape, I was pretty sure of myself then about the probe's ability to withstand the 100kv.
Not sure if I would trust it now without testing it first though.