Try that "upper limit" or "first try" or "first guess" value of 174 turns.
I am WYSIWYG; snipped-for-privacy@localnet.com .
That's one long set of meter leads!
NT
OK, it seems your measurement techniques are quite good, so we can exclude that as a possible source of error. However, there are still some issues that need to be addressed that could explain the discrepancies.
You seemed to indicate the coil was air core. If so, the core losses may not apply, but the distributed capacitance would be quite large for a
21mH inductor. This could have a significant effect on the readings.Jack R. Smith of Clifton Laboratories did a very interesting review on the AADE, particularly comparing the AADE measurements on inductors with the results using a HP4192A. The entire paper is well worth close reading, but especially the sections on measuring inductance. You can get it here:
Others have run into problems using the AADE on ferrite core inductors. Here's an example on one forum:
< http://dics.voicecontrol.ro/process_mails/arata_discutia/102448/Question_about_my_AADE_L_C_Meter_IIB.html>Note the comment by John Popelish at the end:
"many ferromagnetic materials need some minimum flux swing before they approach their specified permeability. The meter may not drive them to that minimum flux swing."
I believe you indicated your coil was an air core, so this may not be a problem. However, I find it difficult to imagine a 21mH air core inductor. It has to be huge, and would have significant distributed capacity. It would also have very low Q, which would affect the AADE readings as well as the LTspice simulation.
I would recommend to not worry about trying to simulate the oscillator in LTspice until you can get a handle on the losses and distributed capacity. Any simulation you do without these numbers probably has little value.
Obviously the reading from the AADE is wrong. You can figure this out for yourself.
Assuming the Colpitts is not overdriven and has no other problems, it will give an accurate representation of the resonant frequency of the LC tank.
In your circuit, the series capacitors of 0.0108uf and 0.00833uf give
0.004702uF across the inductor. You measured a frequency of 18.18KHz.Assuming a tank Q of 20 (guess for large air core), you can easily calculate all the relevant inductor parameters.
The natural resonant frequency is 18185.68 Hz The effective Series Resistance is 93.063 Ohms The inductor is 16.28 mH
This is a long way from the 21.29mH indicated by the AADE.
You can probably trust the capacitance measurements, but the inductor is way off.
Robert Macy also got a much lower value of inductance than the AADE gave.
In that post you stated:
I assume that is using the Hartley circuit. Again, assuming a Q of 20, the coil inductance is 16.53mH, which is very close to the 16.28 mH from the Colpitts. Notice that neither of these calculations use LTspice.
So the Colpitts and the Hartley both say the same thing. The AADE is giving you bad data.
The question is why? It is distributed capacitance?
HP makes some piece of gear that will show N, surprisingly accurate in this range.
uh, where did 174 come from? 118 to 174 is quite a jump and you didn't say
170, 175, or 180, but 174. What based upon?got it. with 174 turns [my estimate too] the inductance might get up to your measured value, but 22Awg is still going to be like only 10 ohms Rdc. 22Awg is a 'strange' value, sure it's not 24 Awg? Will do a model for both to see what I get.
DEFINITELY NOT OUT OF YOUR LEAGUE!!!
I constntly use it as a 'coil calculator' where I simply put in wire gauge, N, and diameter and out pops Rdc, Ldc, and the two important ones, Reff and Leff at the operating frequency. Works faster and more accurately than all those 'approximation' formulas you find on the web. Plus, you can't believe how handy it is to find coupling factors, which can be really, really difficult. Based on the assumption you would 'like' a basic inductor calculator you can count on, I'll send you the models [for complex drawings, you can also import .dxf drawings] These models will run and show you an answer right now, copy each file to another name, like CoilL_B.fem, and modify away! That way you can explore WITHOUT erasing a starting point. You can't believe how educational femm 4.2 is for teaching about skin effect and those illusive frequency based losses. Plus, I do ALL my work in air core, since it is the only medium I can count on ;) Also femm will plot the currents inside a conductor and you can watch in detail how the skin effect really does damage to your coil's Q. AND how slight modifications can raise your Q.
Will do.
arrrgggg! I hope all the 'bots on the web don't find your email address now!
Will send a set [they're small, like LTspice .asc files AND are in text format, too] Will do later today, after running the 174 turns estimates. oops forgot. when you measure 13.5 inch diamter to WHERE is that? Inside ID, Outside OD, or mid point of the 3/8 inch wad? I've been letting it be to the approx middle, rather arbitrarily, but does make a difference.
For grins, did you ever use just a function generator, scope indicator/voltmeter, freq counter, and resistor to measure the "inductance" ?? By plotting multiple points of the Z magnitude vs freq near the operating range, you can sort out/determine the L, R, and C of a coil pretty accurately. I've not gotten much better than 0.1%, but that's useful if you want to make a 5-7 pole filter out of all these components and have it work the first time.
Nope, femm 4.2 is NOT out of your league. Rather I think you'll find it one powerful 'extra' tool, and best of all, it's free WITH a support group.
Hi, Robert,
I'm designing a PCB that drives a laser. The dI/dT is extreme, so the inductance of the connections will be critical.
We might want someone to do an EM model of various geometries. Would you be interested? Email me if so.
-- John Larkin Highland Technology, Inc jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Absolutely! Sounds challenging, on many fronts.
I have a Federal EIN number so Accounting and IRS keeps happy.
Will email you offline.
The AADE is still giving problems, starting up "not a capacitor", and sometimes switching over from cap meas to "not a capacitor". Took it apart, nothing apparently wrong. Used contact cleaner, modern General Cement Electronics Jif-Action
10-630 bought during the heyday of tubes. (I _did_ say "modern"...) Cleaned up the rosin left on the board during factory assembly. No operational change. The caps read damn close to factory values, and my Protek 6500 also agrees within the 5% cap tolerance. That leaves the inductance measurements suspect.
Well, now that i have FEMM 4.2 it is theoretically possible. And that would give me a good excuse to learn something. Color me ignorant on use, and if you do not mind dumb questions during "progress", i will try to not be too much of a PITA. Call that a qualified "yes".
You are overdriving the oscillator. Increase the emitter resistance. Even with overdriving, I showed the oscillator frequency is still very close to the calculated value. The error is much, much less than the error caused by the incorrect AADE inductance reading.
Yes, this is what we have been saying. Some possible things to check:
It would be interesting to discover what is causing the problem, so please let us know what you find.
No one has mentioned a possibility that the coil has shorted turns. Measuring at ONE frequency would not disclose this, but trying to use the coil in an oscillator would show it. Perhaps, your meter is starting to complain about it, too.
You said the coil was a degaussing coil and you 'shaped' it onto a blank area. You also esitimated 174T yet had fairly low L, Both indications that something is not right here.
I'm going to explore that effect for a bit and will get back to you. But I keep coming back to,"NEVER, EVER measure inductance at ONE frequency!" So what do you get when you measure at a multitude of frequencies? Doing so will instantly show a shorted turn.
got it. I do that too.
Glad to hear it. You're going to enjoy that PC Tool. Even the electrostatic solutions will teach a lot. I sent you a set of models AND some notes describing how to do those two things. Let me guess, localnet.com thinks I'm a Nigerian scammer!
Let me know if you did NOT receive the models AND the $Notes.txt
In the interim, as you define the Problem with DC there is NO skin effect, absolutely DC measurements. Change the frequency to something that seems reasonable, RUN again and you'll see the effective resistance sky rocket. The next exciting educational tool is being able to PLOT those eddy currents inside a conductor! If that plot is 'ratty', your mesh is too coarse. Do that by viewing the answer, honing in on a region of interest, select 'line' and draw in a line, then plot Js and Jeddy current [that's the absolute bottom selection, don't use the one above that in the list something's wrong with it and to my knowledge never been fixed, but is ok, because not much use for it] The plot [which you can export as data] shows the eddy currents and total currents along your line! Absolutely eductional to understand why there is NO advanatage to using 12 parallel conductors instead of a single equivalent cross section wire and why it's actually worse.
There are two ways to play with mesh size. The one most people use [and is not the most efficient for fast solutions] is to select BLOCK and then define the maximum distance between nodes. That works but makes a LOT of nodes. Instead, for eddy currents, I set the maximum nodes ALONG a boundary, like between the conductor and AIR. Makes more sense because all the interesting stuff is happening at a boundary, and you'll notice the mesh is very small along the boundary [conductor's edge] and then becomes large into the material [again makes sense, because homogeneous regions don't change much. That way you can set up the mesh/nodes to be much smaller than skin effect so accurately determine what's going on. For exmaple in the models I sent ou, a small increase in mesh at the boudary changed the resistance almost twice! But no change to inductance. So when looking for nuances, it is important to match the mesh to the problem.
yeah, I noticed your degaussing coil [as measured] would let around 6 Arms through, that's over 600W, so the coiil won't last long. Again supporting that something in the coil is very wonky. Seriously, you HAVE to better characterize your coil over frequency. Is ok to get a set of data in V magnitude, because with enough points octave can sort out the best fit of L and R and C values for you. You do have a copy of free octave, or pay-for-Matlab, don't you? If not, get a copy! It's like the most powerful replacement for Excel programs you'll ever see. Without the little squares.
If all else fails, one thing to try might be to resolder all the connections. Perhaps there is one that was bad from the factory, or maybe has a thin solder bridge. There's not too many connections, so it shouldn't take very long. Remove the battery first, of course.
If you are careful, I don't see how it could hurt anything.
Inductance measurements are always suspect. Grab three or four different L-meters and you'll get different, maybe wildly different, measurements.
A serious L meter needs to have selectable measurement frequency, so you can measure L near where you plan to work. And it has to properly math out the loss components.
The AADE is pretty good for around $100. Especially if you have the surface mount adapter thingie.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Ferrite cores can give problems as discussed earlier.
In this case, the inductor is air core, which should give reasonable accuracy and repeatability. For example, two different oscillator circuits using different tank capacitors gave very similar results with the air core inductor used by Robert Baer:
Colpitts: 16.28 mH Hartley : 16.53 mH
The AADE review by Clifton Labs showed an average error for an ensemble of 18 Boonton standard inductors of 0.19% after correction for distributed capacitance. Five inductors that showed errors up to 4.65% were discussed in detail. Errors on ferrite core inductors were analyzed and compared to the HP4192. The paper is here:
The conclusion is AADE measurements on air core inductors are quite accurate after correction is made the for distributed capacitance.
Ferrite core inductors are a completely different matter. The problems with ferrite core inductors is not limited to the AADE. It affects all LCR meters. As John Popelish mentions:
"many ferromagnetic materials need some minimum flux swing before they approach their specified permeability. The meter may not drive them to that minimum flux swing."
The conclusion in this case is the AADE measurements on the capacitors are fine, but the air core inductor is wrong. The AADE appears to be defective on inductor measurements, and the problem now is to try and figure what is wrong.
The Clifton Labs report shows the AADE measurement frequencies on standard Boonton inductors ran from 38.5KHz to 738.8KHz with an average error of 0.19% after correction for distributed capacitance. So the measurement frequency does not seem to be critical on air core inductors.
As far as the inductor loss, from the formula for resonant frequency,
Fr = (1/2*pi*sqrt(LC))*sqrt(1 - 1/4Q^2),
a Q of 20 will give a correction factor of 0.999687 to the resonant frequency. So as long as the inductor has moderate Q, the loss components will have negligible effect on the resonant frequency.
From all this we conclude the AADE can be an accurate and reliable tool for most work. You just have to be careful with ferrite core inductors.
The AADE web site is closed at the moment due to illness. If anyone wants one, they might consider making their own unit. Neil Heckt (AADE) kindly allowed the circuit and software to be posted to
The BOM does not list the crystal frequency. From other sites, I gather the frequency is 8MHz, but I cannot guarantee that is the correct value.
ARRRGGGG!!! totally forgot! one of the MOST powerful tools for measuring LCR below 90kHz is a really good SoundCard!
On an old WinXP system, I use Creative Labs EMU1212, 2 MATCHED channels of Tx/Rx, 24bit 192kS/s, out to around 89kHz. With a few frequencies, you can curve fit to the measurment and get better than 0.1% and sometimes 0.01%
Or, you could try Bob Mastas' LCR Meter:
DAQARTA v7.60 Data AcQuisition And Real-Time Analysis
If the LCR Meter is not to your liking you still have an incredibly accurate function generator.
Well, explored a possibility of a 174T 24Awg coil with 16 turns removed due to a 'cross-over' short. That cuts the inductance down to 21mH and around 18 ohms, fairly close to measurment. BUT! now I can't get the oscillator to oscillate to see what it does to the ColpittsL.asc circuit!
I'll keep 'playing' with the thing to see if I can get the coupling to be more representative, perhaps THAT will allow the osc to osc.
I stared from scratch and measured "everything". Results follows:
Equipment used:Protek 6500, AADE L/C Meter IIB L1 L2 o--+--uuuuuu-+-uu--+---o--vvvv--o TOP GND | | Rdrive +----||---------+ Cx
L1 degaussing coil, unknown turns R = 17.83 L2 added secondary, 11 turns R = 0.70 Resistance reported is measured value minus 0.02 ohms, the meter lead resistance. Est error 0.01 ohms all values above.
Cx film capacitors, 10% spec unless noted, R drive 5% unless noted, resonance 2% est unless noted.
Cx, uF | | no Cx; drive adjusted as needed as measured by | resonance | Freq |Vtap | Vtop | Rdrive Protek | AADE | 2% est | Rdrive ------+------+----------+-------
--------+-------+-----------+------- 10KC | 14mV | 130mV | 10K 0.1013 |0.1010 | 4.94KC | 10K ------+------+----------+-------
--------+-------+-----------+------- 10KC |135mV | 12V | 1K 0.999 | 1.007 | 1.58KC | 10K ------+------+----------+------- 5% | | 10KC | 1V | 8V | 100
--------+-------+-----------+------- ------+------+----------+------- 1.529 | 1.550 | 1.28KC | 10K 1KC |140mV | 800mV | 100
--------+-------+-----------+------- ------+------+----------+------- 10.25 | 9.54 | 537~ 10% | 10K
--------+-------+-----------+------- 10.25 | 9.54 | 496~ | 1K
--------+-------+-----------+------- no Cx; direct drive at top 10.25 | 9.54 | 502~ | 100 1% Frequency | Vtop | Vtap
--------+-------+-----------+------- ----------+-------+-------- 22.39 | - | 323~ | 100 1% 10KC |5.878V |5.782V 5% | | ----------+-------+--------
--------+-------+-----------+------- 1KC |4.476V |2.219V 42.19 | - | 216.76 | 100 1% ----------+-------+--------
--------+-------+-----------+------- 100~ |4.223V |0.701V 42.19 | - | 219.31 | 1K ----------+-------+--------
--------+-------+-----------+------- (Rdrive=0) - | - | 48.57KC | 100 1% no Cx | |
--------+-------+-----------+------- - | - | 48.55KC | 1K no Cx | |
--------+-------+-----------+------- - | - | 48.36KC | 10K no Cx | |
--------+-------+-----------+-------
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.