LCR meter tests

[EC8010]

I thought it was satisfyingly quick too. Standard postage.

I can't get a sensible answer for dielectric loss - it's too good. If I do a simple 1kHz C, D measurement after having done short and open circuit calibration, I get a very small negative number for D, and phase angle of fractionally over 90 degrees at low frequencies. That's why I tried fitting a model that included a shunt resistance as well as ESR and ESL, but I wasn't able to get anything there, either. These are far and away the best silvered mica capacitors I've seen. Possibly even the best capacitors I've seen (nothing else has fitted pure capacitance so tightly all the way up to 100kHz).

[mawyatt]

Yeah that's getting into region where <3ns represents 0.001 degrees and easily within the Open/Short Cal and Fixture uncertainty.

We've seen a phase of -89.998 degrees at 1KHz with the 1uF PP types and ventures beyond -90.000 as the test frequency goes up. How does the capacitance hold up as you go from ~100Hz to 1MHz?

Anyway, good news is these Soviet Micas are really good it seems

Best

[TimFox]

I thought it was satisfyingly quick too. Standard postage.

I can't get a sensible answer for dielectric loss - it's too good. If I do a simple 1kHz C, D measurement after having done short and open circuit calibration, I get a very small negative number for D, and phase angle of fractionally over 90 degrees at low frequencies. That's why I tried fitting a model that included a shunt resistance as well as ESR and ESL, but I wasn't able to get anything there, either. These are far and away the best silvered mica capacitors I've seen. Possibly even the best capacitors I've seen (nothing else has fitted pure capacitance so tightly all the way up to 100kHz).

Can you see any change in apparent capacitance at very low frequencies?

[mawyatt]

With our Molded Micas and Fixtures we see a slight rise in capacitance reading as frequency drops which is consistence with both the IM3536 and TH2830 LCR Meters.

Best

[TimFox]

What, roughly, is the frequency where the capacitance change becomes noticeable?

[Hydron]

Ah right - I had that seller on my list too but got caps from elsewhere to get a bigger range of values. I have a few of the exact same type ("SSG-1") coming in 10nF and 4.4nF values, plus some 100/99.5nF "SSG-2" capacitors that appear to be the same style, just in a physically larger can.

Good to hear that they measure well, hopefully mine are similar (they are used rather than new).

RE postage - pretty normal to get next working day via the postal service in the UK, the distances are much less than in the US and it's practical to do via land transport other than from the most far flung areas. Reliability has been a bit hit or miss since covid though!
My issue is that the package is sitting in a warehouse at an airport in the UK somewhere and isn't properly in the postal system yet (though it looks like it should have cleared customs at least) - Ukrainian post took an impressive <2 days to get it out of the country once received (by road/rail - no commercial aviation right now!) but it's been waiting in the UK for nearly 4 days at this point with no progress, which is rather frustrating (would have totally understood a delay on the Ukrainian end of course given the situation there)!

[EC8010]

What, roughly, is the frequency where the capacitance change becomes noticeable?

I've seen that too in my lossy capacitor tests and it's usually below about 300Hz. I've measured quite a few lossy capacitances around the 5pF mark and needed a series CR network across the main capacitor to get a reasonable fit. The discrepancy graph in my earlier post appears to show a different form of dielectric loss above 100kHz.

[TimFox]

The old General Radio decade boxes were available with mica or polystyrene capacitors.
Both have high Q.
G R noted the better tempco of mica, but the increase in capacitance at low frequency.  Did that relate somehow to the water of crystallization in crystalline mica?

[Kleinstein]

Water could contribute to the dielectric losses. However there can be other impurities that can contribute. Mica is a natural material and can thus vary by source.

[d-el]

I have two LCR meter. HP 4285A (75kHz - 30MHz) and Keysight 4980AL (20Hz - 300kHz).
SMD fixture crafted for Hioki axial fixture. It is quite convenient to measure components of size 0402.
In search of standards of inductance and capacitance I look at old Gen Rad standards.
Some photos:



Keysight 4980AL have inside configurable DC bias, 1.5V and 2V.
This allows to measure polar electrolytic aluminum and tantalum capacitors.
But for the future, I will make or buy an adapter for external DC Bias. This would be useful for measuring the capacitance versus voltage dependence for ceramic capacitors. Since the devices are connected to the central server via GPIB, it will not be difficult to script and organize automatic measurement.

P.S. Tell me why the image is blurry in the preview.

[TimFox]

Water could contribute to the dielectric losses. However there can be other impurities that can contribute. Mica is a natural material and can thus vary by source.

I’m not referring to dielectric loss, but a change in dielectric constant, hence capacitance, at low frequencies.

[EC8010]

Water could contribute to the dielectric losses. However there can be other impurities that can contribute. Mica is a natural material and can thus vary by source.

My limited understanding is that stating "mica" is like stating "glass" - there are lots of different types and it's a mix of an awful lot of elements. Electronic wisdom is that Muscovite mica (from Moscow region) has the lowest losses. I am perfectly happy to accept that water could cause variable dielectric losses as I believe that Muscovite mica contains 4.5% water, and that's probably variable.

Thank you TimFox for your comment about dielectric constant; that is consistent with what I have measured. Again, I have a limited understanding of what is going on, but I have been told that the Debye theory predicts a peak in losses at high frequency. Yet I also find that the Debye theory predicts dielectric absorption that can be crudely modelled as a cascade of series CR networks in parallel, which tends to work at low frequencies. Still, for all that, capacitors are far nearer to being perfect reactive components than inductors!

@d-el: I hope you can upload better photographs because what you uploaded was tantalising but unreadable; looks like the autofocus didn't work.

[Hydron]

@d-el: I hope you can upload better photographs because what you uploaded was tantalising but unreadable; looks like the autofocus didn't work.

The blurry pics are links to higher res images - click on them to reveal.

[EC8010]

So they are! Thank you for that.

[mawyatt]

I have two LCR meter. HP 4285A (75kHz - 30MHz) and Keysight 4980AL (20Hz - 300kHz).
SMD fixture crafted for Hioki axial fixture. It is quite convenient to measure components of size 0402.
In search of standards of inductance and capacitance I look at old Gen Rad standards.
Some photos:
(Attachment Link) (Attachment Link) (Attachment Link)
(Attachment Link)

Keysight 4980AL have inside configurable DC bias, 1.5V and 2V.
This allows to measure polar electrolytic aluminum and tantalum capacitors.
But for the future, I will make or buy an adapter for external DC Bias. This would be useful for measuring the capacitance versus voltage dependence for ceramic capacitors. Since the devices are connected to the central server via GPIB, it will not be difficult to script and organize automatic measurement.

P.S. Tell me why the image is blurry in the preview.

Clever idea!!

Here's a link you may find useful wrt a DC Bias Adapter (see towards the end).

https://www.eevblog.com/forum/projects/bias-network-for-lcr-meter/

These type "Plunger" fixtures like you've created and the ones in the link above for SMD components produce good results with high Z components, but degrade with lower Z components. The main reason is lack of uniform DUT contact impedance and that the Force and Sense for Kelvin connections are far removed from the DUT.

This is why we developed the Split-Kelvin type SMD LCR Fixture here.

https://www.eevblog.com/forum/projects/different-type-lcr-smd-fixture/msg5798431/#msg5798431

Here the Force and Sense are connected by the DUT end terminals and this technique has shown very good repeatability with very low Z components.

Anyway, hope this helps!!

Best

[d-el]

I'm also looking at the China Made SMD fixture. Has anyone tested this?
On ebay it costs about $70

[mawyatt]

I'm also looking at the China Made SMD fixture. Has anyone tested this?
On ebay it costs about $70

(Attachment Link)

Sure have. Search for:

cheap-smd-lcr-fixture-the-good-bad-and-ugly

Best

[d-el]

("SSG-1") coming in 10nF and 4.4nF values, plus some 100/99.5nF "SSG-2" capacitors that appear to be the same style, just in a physically larger can.

What is the accuracy of SSG-1 and SSG-2 capacitors that you bought? I see that in the local flea market there is SSG-2 with an accuracy of 0.3% for sale (used, ~2.4$).

[Kleinstein]

Water could contribute to the dielectric losses. However there can be other impurities that can contribute. Mica is a natural material and can thus vary by source.

I’m not referring to dielectric loss, but a change in dielectric constant, hence capacitance, at low frequencies.

Dielectric loss and frequency dependent capacity are linked via the Kramers-Kronig relation. So a lossy capacitor always has a corresponding capacitance that goes up to lower frequency.
Relaxation processes in the dielectic act like an RC series element. If there is only one time constant one gets a broad (about a factor of 10 on the frequency scale) peak in the loss versus frequency and a corresponding step in the capacity. In many cases the time constants are smeared out (a superposition of different relaxation processes) and the loss and capacitance change are over an even longer frequency range, so that no clear peak or steps are visible.

[d-el]

I am found SSG-1 and SSG-2 0.3% for the price of a cup of coffee.
It turned out that the seller was half an hour away from me.

[mawyatt]

Nice!!

How do these large Micas compare to EC8010, do you see a change in apparent capacitance wrt to change in frequency? How does "D" hold up over frequency?

Best

[Hydron]

Just got my batch (ordered before the others, but held up in customs it seems). See pic below.

One of them is well out of spec, might be a candidate for dissection (tbh not unhappy to have a bad one, gives me the excuse). Everything else is well within tolerance, and has Q > 1000 2000 @ 1kHz (limit of the DE-5000). Some of the ones I got are MUCH older than those that others have gotten - (range looks to be late-60s (!!) to late-80s, assuming I'm reading the codes right, they seem to move from month,year to year,month at some point in the 80s).


Edit: @mawyatt - over freq the 10n capacitance seemed to be consistant, the 100n went up slightly at 10k and 100kHz, but take that with a big grain of salt given I'm looking at the least significant digit on my DE-5000!

[TERRA Operative]

Now I am intrigued about buying some of these mica capacitors..

What's the difference in SSG-1, SSG-2 and SSG-3 rating?

[Hydron]

Now I am intrigued about buying some of these mica capacitors..

What's the difference in SSG-1, SSG-2 and SSG-3 rating?

As far as I can tell, just capacitance range.

[d-el]

do you see a change in apparent capacitance wrt to change in frequency? How does "D" hold up over frequency?

I wrote a small script that makes a series of measurements and displays a graph. HP4285A connected by GPIB.
Expandable image:


SSG-1 10000pF


SSG-2 100000pF