EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: reagle on January 26, 2015, 04:21:20 pm
-
Looking for an LCR meter that could handle measuring inductors down to nH, I mostly stumble on things like Agilent/Keysight 4285A at a cool tune of $8k or more.
Are there any other ones the esteemed forum could suggest as slightly cheaper alternatives?
-
How accurately do you want to measure? What form factor do you want to use? SMD?
Assuming these inductors will have DC resistance in few m? range you need to go to about 10MHz to get a reasonable accuracy. So yes, you will probably need an old HP/Agilent to measure them. Additionally, you will need a SMD adaptor and calibration standards. It is not a trivial task.
-
It does require some serious precision instrument to measure nH, as in it must have very low residual inductance. I have been looking for something similar, but it's pretty expensive.
1nH is like 1mm of hookup wire or even less PCB trace.
Agree with owiecc, a 2nd hand HP/Agilent is probably your best bet.
-
The Fluke PM6306 goes down to 10nH
But it is probably in the same price range as the Keysight 4285A
-
Just for fun I thought I would give the DE-5000 a try since it does have 1nH resolution.
I used a piece of 20ga copper which should be 10nH for 1.43cm length at 100kHz.
Did the Cal routine using real Kelvin connections at 100kHz and did the short part of the Cal on the wire with the kelvin clips touching each other for zero wire length. Toggled out of successful Cal and get a reading bobbling between 0 and 1 nH. move the clips apart to 1.43cm and I get 16nH. I realize that is just over a 50% of reading error but that is almost in spec of the 2.5% of reading + 5 lsd tolerance. I was surprised it was that good.
-
Vector network analyzer, HP 8753 or similar, it can resolve picohenrys. (Hint: Its not easy and requires elaborate setups)
-
Good point- we do have RF gear here!
Instek CR-8101G ($6k new) is another option
-
If you happen to find an HP 4815A at a reasonable price, it's much simpler to use than a full VNA.
However, you usually need to measure a larger piece of circuit than just the inductor, in which case the VNA is better.
-
What about the TEGAM Model 3550 http://www.tegam.com/product.asp?modelNumber=3550 (http://www.tegam.com/product.asp?modelNumber=3550)
It's out of production now but still available. Have used this one extensively for characterising SMPS transformers and pulse transformers and it did a good job with that. 8 years ago it was half the price of a comparable Agilent.
-
Very true. The UI on our HP VNA is a royal PITA to get used to
If you happen to find an HP 4815A at a reasonable price, it's much simpler to use than a full VNA.
However, you usually need to measure a larger piece of circuit than just the inductor, in which case the VNA is better.
-
Never heard of them- thanks!
What about the TEGAM Model 3550 http://www.tegam.com/product.asp?modelNumber=3550 (http://www.tegam.com/product.asp?modelNumber=3550)
It's out of production now but still available. Have used this one extensively for characterising SMPS transformers and pulse transformers and it did a good job with that. 8 years ago it was half the price of a comparable Agilent.
-
The IET Labs/ Quadtech 7600(+) does sub 10nH measurements, just don't look at the price.
L: 0000.001 nH to 99.99999 H
Measurement Fast Medium Slow
Accuracy: LCR: ± 0.5%1 ± 0.25%1 ± 0.05%1
DF: ± 0.005 ± 0.0025 ± 0.0005
-
Very true. The UI on our HP VNA is a royal PITA to get used to
If you happen to find an HP 4815A at a reasonable price, it's much simpler to use than a full VNA.
However, you usually need to measure a larger piece of circuit than just the inductor, in which case the VNA is better.
If you think the UI on a HP VNA is bad, try using one from R&S or Anritsu.
-
If you think the UI on a HP VNA is bad, try using one from R&S or Anritsu.
+1 the UI on old R&S VNAs is just appalling.
IMO the HP ones are pretty straight forward for the most used functions. (Advanced stuff and custom CALs another story)
Since you mentioned you have a VNA at your disposal i would really recommend using it.
HP / AgiSight has very good instructions / app notes for their VNAs that show how to measure stuff.
I am certain there is one exactly for this problem that will teach you how to measure your inductor correctly.
edit: fixed wrong quote
-
Guesstimating wire inductance
http://www.thompsonrd.com/Guesstimating_Inductance.pdf (http://www.thompsonrd.com/Guesstimating_Inductance.pdf)
-
If you think the UI on a HP VNA is bad, try using one from R&S or Anritsu.
I did not say this!
I did not suggest that you did, it's just that the quotes got a little muxed ip
-
I'd use a VNA for stuff like this but there are a few caveats...
Depending on the form factor of the inductor under test its inductance will be affected by the test frequency and its proximity to ground. Also it will be affected by how it is soldered or held in a test fixture.
But there are cheaper ways to measure inductors around 10nH with reasonable accuracy if you don't mind testing it at lowish frequencies. eg you could use a scope and a sig gen and a sense resistor (and a few sums).
Or you could measure it with a decent quality resonating capacitor (small and SMD with low package inductance) and cheap and a basic spectrum analyser if it has a tracking gen. eg something like a Rigol 815TG. This way you could test it at various frequencies.
But it gets harder as you try measuring down at maybe 3 or 4nH and the VNA + test fixture is a good option provided you accept that the inductance can be affected when mounted on a PCB or very close to a ground area. So what you measure in the VNA test fixture might not hold once you transfer the inductor to a PCB or some other environment.
-
Just for fun I thought I would give the DE-5000 a try since it does have 1nH resolution.
I used a piece of 20ga copper which should be 10nH for 1.43cm length at 100kHz.
Did the Cal routine using real Kelvin connections at 100kHz and did the short part of the Cal on the wire with the kelvin clips touching each other for zero wire length. Toggled out of successful Cal and get a reading bobbling between 0 and 1 nH. move the clips apart to 1.43cm and I get 16nH. I realize that is just over a 50% of reading error but that is almost in spec of the 2.5% of reading + 5 lsd tolerance. I was surprised it was that good.
It would be interesting to see how the DE-5000 compared to the Agilent or Fluke! Nevertheless, I think the test you did was pretty cool. I have the DE-6000 and have measured some pretty small inductors but never tried that. I just might give it a try and see what I get! ;D
-
Looking for an LCR meter that could handle measuring inductors down to nH, I mostly stumble on things like Agilent/Keysight 4285A at a cool tune of $8k or more.
Are there any other ones the esteemed forum could suggest as slightly cheaper alternatives?
The ancient General Radio 1632-A has 0.1nH resolution, but it can be, um, a challenge to find a null that precisely! The manual recommends using two tuned amplifiers in series when working the last digits. The bridge is rated to 0.1% when calibrated. Normally you use it with a 1000Hz test signal; above 5kHz, residual impedance in the bridge degrades accuracy.
-
Just for fun I thought I would give the DE-5000 a try since it does have 1nH resolution.
I used a piece of 20ga copper which should be 10nH for 1.43cm length at 100kHz.
Did the Cal routine using real Kelvin connections at 100kHz and did the short part of the Cal on the wire with the kelvin clips touching each other for zero wire length. Toggled out of successful Cal and get a reading bobbling between 0 and 1 nH. move the clips apart to 1.43cm and I get 16nH. I realize that is just over a 50% of reading error but that is almost in spec of the 2.5% of reading + 5 lsd tolerance. I was surprised it was that good.
It would be interesting to see how the DE-5000 compared to the Agilent or Fluke! Nevertheless, I think the test you did was pretty cool. I have the DE-6000 and have measured some pretty small inductors but never tried that. I just might give it a try and see what I get! ;D
I am using the clips I made in this thread (https://www.eevblog.com/forum/projects/video-of-making-the-ultimate-kelvin-connection-(major-hacking-of-a-pomona-clip)/msg132596/#msg132596)
-
Vector network analyzer, HP 8753 or similar, it can resolve picohenrys. (Hint: Its not easy and requires elaborate setups)
That's exactly what I do, I thought I was just being a bit cheap not having a gadget that does it off the bat.
It's so rare I need to measure inductance, it's not a big deal to set up and cal a test fixture. My 8753A's battery backup memory has never worked anyway so I'm quite used to recalibrating every time I switch it on!
-
My cheapo B&K 878 can't do anything near that low. However, I can with my antique 8754A VNA. I just use some test boards to solder the parts to that I want to test. The same boards are used to cal the VNA.
-
Another methodological question:
What are you doing that needs accurate values around 10nH? Surely you're dealing with transmission line segments by then? Would it not be better to evaluate the system (filter or whatever) as a whole, not on a per-component basis?
In which case, a frequency sweep, spec + TG, network analyzer or whatever would be the way to go.
Tim
-
Boonton 250A (circa 1954) or hp 250B.
Either does nicely at measuring inductors in this range up to 200 Mhz.
Bernice
-
Looking for an LCR meter that could handle measuring inductors down to nH, I mostly stumble on things like Agilent/Keysight 4285A at a cool tune of $8k or more.
Are there any other ones the esteemed forum could suggest as slightly cheaper alternatives?
The ancient General Radio 1632-A has 0.1nH resolution, but it can be, um, a challenge to find a null that precisely! The manual recommends using two tuned amplifiers in series when working the last digits. The bridge is rated to 0.1% when calibrated. Normally you use it with a 1000Hz test signal; above 5kHz, residual impedance in the bridge degrades accuracy.
I gave this a try on a GR 1632A with a nominal 1uH inductor, and found I could reliably resolve to the nearest 1nH at 1000Hz. There was enough continuous downward drift (due to heating of the DUT, I expect) in the last hundredths of Mhos of conductance that it was just too hard to reliably resolve inductance to 0.1nH. I used a GR 1232A tuned amplifier with an audio band spectrum analyzer to do this. (It was also possible, but harder to do, with a scope bandwidth limited to 1000Hz instead of the SA.)
-
Another methodological question:
What are you doing that needs accurate values around 10nH? Surely you're dealing with transmission line segments by then? Would it not be better to evaluate the system (filter or whatever) as a whole, not on a per-component basis?
In which case, a frequency sweep, spec + TG, network analyzer or whatever would be the way to go.
Tim
In my case, I would use sub 10nH inductors for the design of lumped RF filters up at UHF. Eg at a GHz or so.
I'd normally use SMD parts and model each inductor individually as a 1 or 2 port model and do the same for capacitors and also use an EM simulator to model the PCB layout.
It's possible to get very good agreement between a filter/PCB model and the real filter circuit when these procedures are carried out. So normally I would model a small inductor as a 1 or 2 port s parameter file based on measurement data taken from a VNA across several GHz.
-
I've recently been working on the design of Cauer low pass filters in the HF frequency range. Getting the toroid inductor values correct is proving quite tricky. The usual low cost LC meters use 500 KHz as a test frequency, which is OK for the lower frequencies using Amidon "2" cores, but gets increasingly inaccurate as you go higher in frequency.
The best method of measurement I have found is to wind the inductor based on the formula/chart then use a small 1% capacitor which is estimated to bring the inductor to resonance near the frequency it is to be used, measure it's resonant frequency, then calculate the unknown inductance -however, this is quite tedious. Neither winding the toroids based on a formula, or using a basic "500 KHz" meter, seem very accurate.
Adjusting the turns spacing of a coil while observing the results on a spectrum analyser/tracking generator sometimes leads to poor input matching.
Is there an easier way to get the inductors to the correct value?
-
It's been a while, but I finally ended up getting a Keysight U1733C 100Khz LCR. It works surprisingly well for my purpose- to be able to roughly confirm that the inductor I am about to solder on the prototype is indeed 5nH and not 10nH for example. I tried a few benchtop units (lowish range ones) and ended up getting the high end handheld instead :)
-
Looking for an LCR meter that could handle measuring inductors down to nH
Consider measuring it through an oscillator.
-
I'm interested in conical inductances.
They have useful properties with regard to self-resonance.
-
Here is low cost but very capable solution to measuring low inductance with impressive accuracy and repeatability, the HP 4342A Q-meter. Mine cost about £200 each and have been in regular use over about 15 years.
The direct dial calibration provides a bottom end of 85nH, but by dialing in the maximum frequency of 70MHz and using a manual calculation from the capacitance dial, I can measure down to a couple of nH.
For surface mount I use an FR4 jig as per the photo. Shorted with copper tape, this calibrates to about 14nH and is deducted from the calculated inductance of a component under test. Just measured a 4.7nH 0603 wire wound inductor as 5.2nH, Q 47 at 70 MHz. Also measures Q up to 1000.