High Q smd capacitors

Reading a forum, a member is comparing the Q of an smd capacitor to a good air capacitor. Maybe?, I don't have a reference of the Q of a good High Q capacitor. Range 20pf to 400pf.

For further reference is is used in a ring down Q meter. See last post on this page.

formatting link

The page includes a sales sheet of a 1210 smd in Chinese.

Mikek

Reply to
amdx
Loading thread data ...

Look at (single layer) porcelain capacitors. For example: See charts on Pg 8. Different dielectrics produce radically different Q. For your operating frequency (1MHz AM BCB) such a capacitor might be overkill. From the photos at: the caps sorta, kinda, maybe look like porcelain. Google translate converted the description into: 100B high Q chip capacitor 4, capacitance and test capacitance values are: A-221=220.8pf B-121=128pf C-680=68.2pf D-300=31.4pf, series-parallel capacity in 21.5--448pf

100B is the ATC designation for one type of multi-layer porcelain cap:

What you probably want is the ESR, which is specified for most of the other types of porcelain caps. See tables at: Q can easily be calculated with: Q = Xc / ESR "Q & ESR Explained"

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

I think it would have to be a pretty shit-ass monolithic cap to not have such a high Q at a couple MHz that the Q of the inductor under test doesn't dominate it into irrelevance as to what it is, precisely. Should probably be more concerned with stability wrt temp, humidity, aging, I think

Reply to
bitrex

It would have to be a bad capacitor to have a Q under 1000. The great Wes Hayward reports SMD ceramics with Qs at least 2000 to 5000, see:

piglet

Reply to
piglet

Lots of extrapolation with unknown reality from the graph for a 100pf cap at 1MHz. The graph starts at 0.01 ohms. If I assume the 100pf is

0.01 ohm ESR at 1MHz and it is resonated at 1MHz, the reactance is 1,592 ohms, thus 1592/0.01 = 159200. I find that difficult to believe. I would believe a vacuum variable capacitor at 15,000. (A really good one) But, I'm here to learn, did I mess up the math? Is it over extrapolation? Is the ESR really that good?

Mikek

Reply to
amdx

Cool, I do figure (#) 3. with a 50 ohm sig gen, ~1 ohm source R, and a 'scope. For caps you want to look at the dissipation factor.. it's a resistance, but I think it's different than ESR in polarized caps. COG ceramics are great. I've never had the right coil to get to a Q of 1000.

George H.

Reply to
George Herold

I made a group of coils on 6" styrene pipe couplers with turns from 9 turn per inch to 15 turns per inch. Just to see if one was better than the rest. In my experiment 11 TPI was over all the best, starting at Q=1263 at 500kHz peaking at Q=1476 at 800kHz and dropping to Q=948 at

1600kHz. 13 turns had the highest peak Q=1504, but it dropped off more at the high end. The styrene forms were had a tiny groove cut on a lathe to get the proper spacing on the forms. I used litz wire 660/46. I used a Boonton 260 to measure the Qs. When you are measuring Qs in this range, it is a bit of a guessing game as to the accuracy, I'm comfortable about the coils Qs relative to each other, but the accuracy of the actual Q, I don't know. There have been reports of Q=2000 using bigger litz or two strands of 660/46 litz. This all AM broadcast Band frequencies.

Mikek

Reply to
amdx

I presume you are referring to the RF Perfomance graphs at: The lowest frequency line is 150MHz, which is quite far from 1MHz. Methinks we would do better using the dissipation factor or loss tangent, which are usually specified at lower frequencies: The porcelain caps are made for low loss high power RF and microwave applications, so I doubt there would be a specification supplied.

Interpolating from the graph, my guess(tm) is much lower than 0.01 ohms ESR at 1Mhz. That's roughly the resistance of a PCB trace that's

20mm long which suggests you might have some problems taking advantage of such a low ESR.

With my track record of sloppy arithmetic, you're asking me? Surely you jest. Anyway, the calculations look right but don't make a difference. When you resonate such a capacitor with a typical inductor at 1MHz, the losses in the inductor determine the final Q of the LC combination. You could have a perfect capacitor, but when you put it across a crappy inductor, the combined Q will suck.

The ESR of a cap is much like the series resistance of the inductor. For example, a 100pf and a 2.5mH air wound inductor to resonate at

1MHz. Plugging in 32awg wire, 2.5mH inductance, and 100mm diameter coil, yields 137 turns. Each turn circumference is: Pi * D = 3.14 * 100mm = 314mm Multiply that by the number of turns and the wire length is: 314mm/turn * 137 turns = 43,000mm = 43 meters Looking up the resistance of 43 meters of 32awg copper wire: the resistance is: 43meters * 538ohsm/km / 1000 meters/km = 23 ohms Assuming no other losses, the best Q you can produce with such a loop is 683. If you combine that with the 100pf resonating capacitor with the sky high Q, the result will be: Q = 1 / (1/Q1 * 1/Q2) = 1 / (1/159200 * 1/683) = 682 In other words, increasing the Q of the capacitor beyond about an order of magnitude better than the coil (about 6820) is a wasted effort.
--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

Since the copper skin depth at 1 MHz is about 0.07 mm, so the total cross section is not used and the losses are slightly larger than that. The table lists full cross section utilization for that wire up to 430 kHz only. A Litz wire helps of course.

Reply to
upsidedown

In one of my posts, I mentioned that someone is putting together a ring down Q meter and trying to save a few bucks going with high Q smd vs a good variable air cap. This is crystal radio stuff and Qs can get as high as 1500. So, in order to improve accuracy I expect the best cap at a reasonable price. But I would hope it is 10x higher Q than the highest Q coil to be measured. Even with that, it would still measure the coil Q down my 10%. (I didn't run the numbers, but it's close)

Reply to
amdx

Below 10-20 MHz, it is true that the coils are usually much worse than the caps. But from 100 MHz and above, the Q of the cap must be carefully cons idered. Assume cap Q contributes until evaluated.

In Fig. 3-13 [1] of Rhea, he shows the Q of an AVX 100 pF cap to about 100 at 200 MHz.

In Fig. 8 of Benabe, evaluating a 68 pF 1206 capacitor at about 300 MHz yie lds a Q of about 65.

I often read the "you can safely ignore cap Q" in texts, but work at RF and microwave frequencies has shown me that it does not apply there.

In bandpass filter work, circuit and resonator Q is usually helped by drivi ng the inductances up and the capacitances down, ceteris paribus. As usual, such a concern is often not the most difficult issue in producing a workin g LC filter. There is not much wiggle room for LPF/HPF as the component val ues cannot be much manipulated there.

[1] Rhea, R. W., HF Filter Design and Computer Simulation, Noble Publishing , Atlanta, 1994, (pp.194-198) [2] Benabe, Automated Characterization of Ceramic Multilayer Capacitors, UF L, RPAG-MOT
Reply to
Simon S Aysdie

Sorry Jeff, I guess I didn't read your last paragraph, you did 10x, I did 10x, but it still gives you close 10% error. When you're doing relative measurements to compare one coil to another that's fine, but when you want absolute value then you want a better cap. btw, when you measure 6" air core coils, moving yourself within a foot will affect the Q reading.

Mikek

Reply to
amdx

How does that work? The upper modulation frequency of BCB AM is

10.2KHz yielding an occupied bandwidth of about 20.4KHz. At 1MHz and a Q=1500, the 3dB bandwidth of the LC circuit is: 1MHz / 1500 = 670 Hz That's narrower than the AM occupied bandwidth, so the high frequency audio will not pass. Probably great for CW or 160 meters, but nobody does CW on the broadcast band.

The Q=1500 might be the unloaded Q as the diode detector forward conduction resistance does present a rather low resistance in parallel with at least part of the LC circuit. To get all the audio frequencies through, the maximum loaded Q is: 1MHz / 20.4KHz = 49

Incidentally, I run into the same problem with small loop antennas also known as magnetic loop, where the antenna Q becomes sufficiently high that only the lower audio frequencies are passed. The resulting audio sounds "muffled". It's possible to build very high Q loop antennas that are useless. "Small Transmitting Loop Antennas"

10 times higher Q for the capacitors in the instrument sounds about right, but you could probably survive with less. I'm not familiar with a ring down Q meter, so I don't know exactly what's required.
--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

So this is for a crystal set for receiving AM broadcasts in the 0.5 -

1,5 MHz band ? For AM reception, the detector needs the carrier and at least of one sideband. Assuming 5 kHz required bandwidth, that will required loaded Ql at the low end of the band of 100 and 300 at the top of the band.

On the other hand, the unloaded Qu should be a few times larger than the loaded Ql in order to minimize passband insertion losses. The insertion loss is given by

Loss_dB = 20 log (1/ (1-Ql/Qu) )

Assuming (unrealistically) that Qu remains at 1500 all over the band. Thus at the low end of the band the insertion loss is 0.6 dB and at

1.5 MHz 1.9 dB. A 10 % error in the Qu measurement doesn't affect the insertion loss very much.
Reply to
upsidedown

We had that kind of problem with aviation long-wave beacons (NDB).

The identification should be in amplitude-modulated Morse code with a modulation frequency of 1050 Hz. The Q of a force-tuned

20 meter (70 ft) antennas was too high to pass the modulation sidebands, so the identifier frequency was lowered to 400 Hz (which was the alternate option allowed by ICAO Annex 10).
--

-TV
Reply to
Tauno Voipio

A simple diode detector works quite well with a full carrier and one sideband. Check any analog TV standard. There are a small vestigal opposite sideband, but IIRC it was for phase accuracy of synch pulses. For audio, the is irrelevant.

Does US AM broadcast standards really allow 10.2 MHz modulation frequencies, when he channel spacing is 10 kHz, Apparently only every other channel is used in a region with relatively low transmitter powers (50 kW).

In Europe, the MW channel raster is 9 kHz and every channel is used. Some stations are quite high power (over 1 MW). The modulation frequency i limited to 4.5 kHz. Looking at the waterfall spectral display, many stations utilizes fully the fC+/-4.5 kHz bandwidth.

Shortwave broadcasts are at 5 kHz raster.

You can double loaded Q by slightly detuning and letting through the carrier and one (either upper or lower) sideband. For speech 3.5 kHz (telephone quality) is enough and 5 kHz would more realistic so once again double the loaded Q.

Reply to
upsidedown

Yep, that's just the coil, now add a cap, more loss, through in a diode, not sure how that affects Q, then to extract maximum audio energy, you will end up lowering the total circuit Q by 1/2. But you only need to hear up to a little over 3kHz to understand. People do crystal radio DXing. To high of a Q in a crystal radio is not a problem it is very easy to spoil the Q.

My next antenna build will be a BOG, Beverage on Ground. I had one, but after the hurricane the owner of the land my BOG was on thought they could just remove all the broken trees, tear up the stumps and clear off the land including my antenna! Hrrrmph! Anyway, after clearing, they made a new ditch to replace what was there before. I still have a clear 250ft to 300ft straight shot on the bank of my side of the ditch. I installed ferrite sleaves to lower the velocity factor to make it seem a little longer. It was a great antenna, very quiet, didn't pickup all the lightning off the southern coast, (I'm in the Florida Panhandle). I pointed it at Chicago so it heard to the North well, but very poor on the Boomer 1200 WOAI out of Texas to the West, or even a local station to the East. Mikek

Reply to
amdx

I made the same argument a previous time this came up, but as amdx points out, the earphones provide the additional load. Because they're after the rectifier, it does have strange effects on the received frequencies, but who am I to tell a crystal radio purist what they should like :)

Clifford Heath.

Reply to
Clifford Heath

I was a little surprised with the 100 and 300. But I can see the advantage of having a high Q LC is that you can extra more audio from the recovered signal. Thus on a DX weak signal you can hear it because you started with high Q front end. The DX contest winners aren't #20 wire on a cardboard tube, they go all out to minimize losses.

Reply to
amdx

Just out of curiosity - do you crystal radio folk ever use double tuned (aka stagger tuned) high Q tuned circuits? That may keep the steep slopes of hi-Q but a broader passband to more faithfully pass audio modulation. My last crystal radio was 1967 so I am out of touch.

piglet

Reply to
piglet

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.