SPICE VS real life..

Don't those claws make it hard to use a mouse?

--

John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

I thought you said earlier the coil was around 10.5 inch diameter. If true, 118T of 28Awg wire yields approx 19 ohms AND only 11mH, not the 21mH you show. and, the 11mH makes the oscillator operate at? you guessed it, your 21kHz.

Again, your model for the inductance appears to be at fault. Again, I offer to help model, to show how to do it for yourself in the future. Why the delay at supplying information? Are you NOT interested in finding out where the lack of correlation is coming from?

• Estimated 120T+/- 20%, #22 enamel, now roughly round, 13.5 inch dia, "scatter" wound,,cross-section roughly round, and bundled with gobbbbbs of tape (you know how they make degaussing coils, more tape than wire). Actual cross-section "diameter" of the real wire is unknown due to the bundling.

• Exactly 11 turns #22 plastic insulation, Mouser 602-1561-100-02, 10T laid on top, and 1T spiral wound around the mess to keep it together. Cross-section "diameter" of the result about 3/8 inches.

• See above; secondary smack dab on top, some inside pri dia, some outside pri dia might be called "scatter" wound, mixed positioning done to maybe improve coupling; one turn wraps it all together into a fatter mes than before.

• maybe 1/16 inch?

** Taped on a large cardboard surface.

They DO squawk a lot.

I haven't found much use for the Hartley oscillator so I will address your ColpittsC.asc:

The resonant frequency of an LCR circuit can be expressed as

Fn = Fr * Fe

where

Fr = 1 / (2 * pi * sqrt(L * C)) Fe = sqrt(1 - 1/(4Q^2)) Q = XL / Rs

The series combination of C1 and C2 is

(0.0108 * 0.00833) / (0.0108 + 0.00833) = 4.7027E-09

Then

Fr = 1 / (2 * pi * sqrt(L * C)) = 15905.915 Hz

and, ignoring the resistor network across the tank,

Q = XL / Rs = 2127.718 / 1e-3 = 2127718.616

Note this is not achievable with standard inductors.

Then

Fe = sqrt(1 - 1/(4Q^2)) = 0.999999999999972

Note the Fe term only becomes significant for Q < 10

Then

Fn = Fr * Fe = 15905.915 Hz

The LTspice analysis shows Fn = 16.335KHz,averaged over 10 cycles. The discrepancy is

16335 / 15905.915 = 1.026976

Not too bad, but it can be improved.

Please note you are severely overdriving the oscillator. If you look at the voltage at the junction of L1 and C1, you can see the positive peak is clipped as the base-collector junction becomes forward biased.

This has several bad effects. The phase noise is increased dramatically, and the oscillator frequency becomes dependent on the recovery time of the base collector junction, which changes with amplitude, temperature, and the characteristics of the transistor.

Whenever you work with oscillators, you need to check that the base- collector junction does not become forward biased, and the base-emitter junction does not exceed the reverse breakdown voltage. Your can use waveform math in LTspice to obtain these values.

You can reduce the amplitude of oscillation by increasing the resistance of R1, the 10k resistor in the emitter.

If you increase R1 to 200k, the amplitude drops so the base-collector junction is no longer forward biased, and the frequency decreases to

15.932KHz averaged over 10 cycles.

The discrepancy between the calculated and LTspice frequencies is now

15932 / 15905.915 = 1.001639

which is a significant improvement.

We conclude the reason for the discrepancy between the calculated frequency and the LTspice result is overdriving the oscillator, which you never want to do.

The reason for the discrepancy between the calculated frequency and the measured frequency is error in the measurements, probably due to inaccurate equipment.

Also note the inductive reactance of 2127.718 ohms is quite high. You really want to work with values closer to 25 or 50 ohms. This has a number of positive benefits. One benefit is the corresponding tank capacitors become much larger. This swamps the transistor capacitances and makes the oscillator less sensitive to transistor variations.

If you change L1 to 500.290uH, Rs to 1.25 ohms, R1 to 100k, and make C1

calculated resonant frequency is 15.905 KHz. The Q is 40 which is much more realistic.

The frequency discrepancy is now

15.911 / 15.905 = 1.0003772

which is quite good agreement between the calculated value and the LTspice result.

I append the ASC and PLT files for verification.

Version 4 SHEET 1 880 680 WIRE 512 -192 416 -192 WIRE 608 -192 512 -192 WIRE 608 -176 608 -192 WIRE 416 -112 416 -192 WIRE 608 -80 608 -96 WIRE 512 -32 512 -192 WIRE 128 16 112 16 WIRE 240 16 128 16 WIRE 304 16 240 16 WIRE 384 16 368 16 WIRE 416 16 416 -32 WIRE 416 16 384 16 WIRE 448 16 416 16 WIRE 112 48 112 16 WIRE 240 48 240 16 WIRE 240 144 240 112 WIRE 272 144 240 144 WIRE 304 144 272 144 WIRE 480 144 368 144 WIRE 512 144 512 64 WIRE 512 144 480 144 WIRE 112 160 112 128 WIRE 240 160 240 144 WIRE 416 192 416 16 WIRE 512 192 512 144 WIRE 240 240 240 224 WIRE 416 288 416 272 WIRE 512 288 512 272 FLAG 608 -80 0 FLAG 112 160 0 FLAG 240 240 0 FLAG 416 288 0 FLAG 512 288 0 FLAG 384 16 Q1B FLAG 480 144 Q1E FLAG 272 144 C1C2 FLAG 128 16 L1C1 SYMBOL ind 96 32 R0 WINDOW 3 29 108 Left 2

SYMATTR InstName L1 SYMBOL cap 224 48 R0 WINDOW 0 -29 8 Left 2 WINDOW 3 27 1 Left 2 SYMATTR InstName C1

SYMBOL cap 224 160 R0 WINDOW 0 -20 9 Left 2 SYMATTR InstName C2

SYMBOL npn 448 -32 R0 SYMATTR InstName Q1 SYMATTR Value 2N2219A SYMBOL res 496 176 R0 SYMATTR InstName R1 SYMATTR Value 100k SYMBOL res 400 176 R0 SYMATTR InstName R2 SYMATTR Value 100K SYMBOL res 400 -128 R0 SYMATTR InstName R3 SYMATTR Value 100K SYMBOL cap 304 32 R270 WINDOW 0 32 32 VTop 2 WINDOW 3 0 32 VBottom 2 SYMATTR InstName C3

SYMBOL cap 304 160 R270 WINDOW 0 32 32 VTop 2 WINDOW 3 0 32 VBottom 2 SYMATTR InstName C4

SYMBOL voltage 608 -192 R0 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName V1 SYMATTR Value 9V TEXT -80 -168 Left 2 !.tran 0 20m 0 50n startup TEXT -72 -192 Left 2 ;'Colpitts Frequency

Here is the PLT file

[Transient Analysis] { Npanes: 3 Active Pane: 1 { traces: 2 {524290,0,"V(l1c1)"} {268959749,0,"V(q1e)"} X: ('m',2,0.0156,4e-005,0.01608) Y[0]: (' ',0,-8,2,10) Y[1]: ('_',0,1e+308,0,-1e+308) Volts: (' ',0,0,0,-8,2,10) Log: 0 0 0 GridStyle: 1 }, { traces: 1 {524291,0,"V(q1b)-V(q1e)"} X: ('m',2,0.0156,4e-005,0.01608) Y[0]: (' ',1,-5.6,0.7,1.4) Y[1]: ('_',0,1e+308,0,-1e+308) Volts: (' ',0,0,1,-5.4,0.6,1.2) Log: 0 0 0 GridStyle: 1 }, { traces: 1 {524292,0,"-Ie(Q1)"} X: ('m',2,0.0156,4e-005,0.01608) Y[0]: ('m',0,-0.002,0.002,0.016) Y[1]: ('_',0,1e+308,0,-1e+308) Amps: ('m',0,0,0,-0.001,0.001,0.015) Log: 0 0 0 GridStyle: 1 } }

Rats. I forgot to change the inductor series resistance, so all the above has to be recalculated. I'll let you do that since I have run out of time.

But you get the idea. Keep XL low, and don't forward bias the base- collector juntion or break down the base-emitter junction.

You should have much better results with these simple changes.

On Mon, 03 Nov 2014 00:46:06 -0800 in sci.electronics.design, Robert Baer wrote,

Actually that is exactly what it is. You posted them with MIME base64 encoding.

I finally found some time to finish the analysis. The results are

LTspice : 15915.7 Hz Calculated: 15907.45 Hz Ratio: 15915.7 / 15907.45 = 1.000518

The LTspice result depends on how many pixels you can move the mouse to measure the frequency. I know the Measure command would give more accurate results, but this is close enough.

The calculated frequency uses a series resistance of 1.25 ohms for a Q of

1. The Q adjustment equation posted earlier is definitely needed for accuracy. The inductive reactance is about 50 ohms.

This should put to bed any questions of the accuracy of LTspice in this application. Since LTspice includes the effect of the transistor on the oscillation frequency, it has to be considered the most accurate representation available, subject to the accuracy of the input data.

Clearly, the discrepancy between the LTspice result and the measured frequency is due to error in the measured values of the capacitors and the inductor. Since the frequency is the square root of these values, the measurement error has to be quite significant. The inductor is a likely suspect since it is harder to measure accurately than the capacitors.

Here is the ASC file for verification:

Version 4 SHEET 1 880 680 WIRE 512 -192 416 -192 WIRE 608 -192 512 -192 WIRE 608 -176 608 -192 WIRE 416 -112 416 -192 WIRE 608 -80 608 -96 WIRE 512 -32 512 -192 WIRE 128 16 112 16 WIRE 240 16 128 16 WIRE 304 16 240 16 WIRE 384 16 368 16 WIRE 416 16 416 -32 WIRE 416 16 384 16 WIRE 448 16 416 16 WIRE 112 48 112 16 WIRE 240 48 240 16 WIRE 240 144 240 112 WIRE 480 144 240 144 WIRE 512 144 512 64 WIRE 512 144 480 144 WIRE 112 160 112 128 WIRE 240 160 240 144 WIRE 416 192 416 16 WIRE 512 192 512 144 WIRE 240 240 240 224 WIRE 112 256 112 240 WIRE 416 288 416 272 WIRE 512 288 512 272 FLAG 608 -80 0 FLAG 112 256 0 FLAG 240 240 0 FLAG 416 288 0 FLAG 512 288 0 FLAG 384 16 Q1B FLAG 480 144 Q1E FLAG 128 16 L1C1 SYMBOL ind 96 32 R0 WINDOW 3 29 108 Left 2

SYMATTR InstName L1 SYMATTR SpiceLine Rser=0 SYMBOL cap 224 48 R0 WINDOW 0 30 11 Left 2 WINDOW 3 26 54 Left 2 SYMATTR InstName C1

SYMBOL cap 224 160 R0 WINDOW 0 33 12 Left 2 WINDOW 3 36 57 Left 2 SYMATTR InstName C2

SYMBOL npn 448 -32 R0 SYMATTR InstName Q1 SYMATTR Value 2N2219A SYMBOL res 496 176 R0 SYMATTR InstName R1 SYMATTR Value 4.7k SYMBOL res 400 176 R0 SYMATTR InstName R2 SYMATTR Value 100K SYMBOL res 400 -128 R0 SYMATTR InstName R3 SYMATTR Value 100K SYMBOL cap 304 32 R270 WINDOW 0 32 32 VTop 2 WINDOW 3 0 32 VBottom 2 SYMATTR InstName C3

SYMBOL voltage 608 -192 R0 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName V1 SYMATTR Value 9V SYMBOL res 96 144 R0 SYMATTR InstName R4 SYMATTR Value 1.25 TEXT -80 -168 Left 2 !.tran 0 20m 0 50n startup TEXT -72 -192 Left 2 ;'Colpitts Frequency With Low XL TEXT -72 -120 Left 2 ;LTspice : 15915.7 Hz TEXT -72 -96 Left 2 ;Calculated: 15907.45 Hz TEXT -72 -72 Left 2 ;Ratio: 15915.7 / 15907.45 = 1.000518

Thanks for the overdrive analysis.

All i had to start with,was the degaussing coil; re-winding is not an option; there is more tape than wire..

And, pray tell, what controls the encoding of attachments in terrornews?

All i did was drag and drop.

Measurements were made with the AADE L/C Meter IIB, used a few random L and C as reference.

The AADE is a good instrument, primarily for rf work. In this application, you are pushing the bottom end of the operating range. You should measure the inductor close to the frequency you will be using and at the same amplitude.

Inductors, particularly ferrite or iron powder, are tricky to measure accurately. I use a TH2821A LCR meter and have noted large discrepancies on measuring inductors.

You can try putting other inductors in your oscillator and add appropriate capacitors to the tank to set the operating frequency so XL is closer to 50 ohms. Adjust the emitter resistor so the base-collector junction is not forward biased. Calculate how close the operating frequency is to the calculated frequency. If your measurements are accurate, the error should be small. Of course, you should minimize the loading effect of the scope or counter so it doesn't affect the oscillator frequency.

That raises another issue. How accurate is your scope or counter. You may be picking up noise in the connection to the oscillator, or having triggering problems due to low amplitude. You need good grounding, bypassing, and good instrumentation techniques.

Maybe there is some hidden problem that is giving the unwanted results.

Perhaps there is a parasitic oscillation in the Colpitts that is affectng the frequency or the measurement. If the measurements change when you wave your hand over the circuit, that is a pretty good indication of a high frequency parasitic. It may be hard to see on a scope if it has low bandwidth or the HF filter is on.

You can add a 100 ohm resistor right at the base of the transistor with minimum length leads. This should kill any parasitic without affecting the oscillation frequency very much.

But the discrepancy between the measured and calculated results is too large. There has to be some gross error somewhere, or some kind of an instrumentation problem. It is definitely worth tracking down so you don't run into the same problem in the future.

Good Luck!

Which makes an attachment. Next time cut and paste the text itself.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

Robert,

Using femm 4.2 to analyze a coil 118T of 22Awg wire with 13.5 inch diameter; I STILL get a around 11.8mH, which when put into your circuit yields around 21kHz.

The Rdc is around 6 ohms and the operating point reistance is around 19.1 ohms

Something is very wonky here.

Is your localnet.com address valid? I'll send you some *.fem models for you to work with yourself. Installing femm 4.2 is easy and not obtrusive to a system. Did you ever get a copy?

For me, resistance measurements are around 0.1%, Cap measurements around

1% and inductance, well ALL over the place, so...I use several techniques to get there from here. Knowing the 'operating' frequency range, select a resistor that is about equal to the reactance. Using a function generator, drive the resistor inductor combination varying the frequency until see 3dB down. Why relative measurement? because more accurate than an absolute measurement. AC meters are notoriously sensitive to frequency range, and a scope is a 3% accuracy anway. But relative, you can squeeze more out of those two measurement instruments. Ok, now you have the inductance to probably better than 1% accuracy AT the frequency you will be using it. Then selectt a cap that will give near resonant operation put in series with inductor [assuming function generator's output Z is around 50 ohm] you can then read the 'dip' at resonance and calculate the effective series resistance at your operating frequency. [the cap will also confirm inductance value previously determined.]

Now, for a transformer, I 'overdetermine' its parameters using short/open etc, then knowing the individual winding terms use octave to calculate Lpri, Rpri, Cpri, Lsec, Rsec, Csec, a [a is turns ration, not K, coupling factor], and Rcore and Ccore. That usually gets me to better than 1% at operating point.

Again, something is wonky. takes a lot more turns than 118 to get 21mH, and Rdc at 6 ohms is VERY low in comparison to your ohmeter measurement, too. Assumed you used a standard ohmeter, not one that varied frequency.

My conclusion is that the inductor's value is suspect. Very suspect.

let me know about your email address.

I saved them, and they were text.

Here is the ColpittsL.asc file

Version 4 SHEET 1 880 680 WIRE 416 -192 -240 -192 WIRE 512 -192 416 -192 WIRE 416 -112 416 -192 WIRE 512 -32 512 -192 WIRE 240 16 112 16 WIRE 304 16 240 16 WIRE 416 16 416 -32 WIRE 416 16 368 16 WIRE 448 16 416 16 WIRE 112 48 112 16 WIRE 240 48 240 16 WIRE -240 80 -240 -192 WIRE 240 80 240 48 WIRE 112 144 112 128 WIRE 272 144 112 144 WIRE 304 144 272 144 WIRE 384 144 368 144 WIRE 512 144 512 64 WIRE 512 144 384 144 WIRE 112 176 112 144 WIRE 416 192 416 16 WIRE 272 208 272 144 WIRE 288 208 272 208 WIRE 384 208 384 144 WIRE 384 208 368 208 WIRE -240 288 -240 160 WIRE 112 288 112 256 WIRE 112 288 -240 288 WIRE 240 288 240 112 WIRE 240 288 112 288 WIRE 112 432 112 288 WIRE 416 432 416 272 FLAG 112 432 0 FLAG 416 432 0 SYMBOL ind2 96 160 R0 WINDOW 3 -112 84 Left 2 SYMATTR InstName L1 SYMATTR Value 21.29mH SYMATTR Type ind SYMBOL cap 224 48 R0 WINDOW 0 -29 8 Left 2 WINDOW 3 27 1 Left 2 SYMATTR InstName C1

SYMBOL npn 448 -32 R0 SYMATTR InstName Q1 SYMATTR Value 2N2219A SYMBOL res 384 192 R90 WINDOW 0 89 53 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R1 SYMATTR Value 10K SYMBOL res 400 176 R0 SYMATTR InstName R2 SYMATTR Value 100K SYMBOL res 400 -128 R0 SYMATTR InstName R3 SYMATTR Value 100K SYMBOL cap 304 32 R270 WINDOW 0 32 32 VTop 2 WINDOW 3 6 66 VBottom 2 SYMATTR InstName C3

SYMBOL cap 304 160 R270 WINDOW 0 32 32 VTop 2 WINDOW 3 0 32 VBottom 2 SYMATTR InstName C4

SYMBOL voltage -240 64 R0 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName V1 SYMATTR Value 9V SYMBOL ind2 96 32 R0 WINDOW 3 -112 84 Left 2 SYMATTR InstName L2 SYMATTR Value 0.212mH SYMATTR Type ind TEXT -80 -104 Left 2 !.tran 0 20m 0 1u startup TEXT -216 -64 Left 2 ;I count about 15.8 cycles from 10.0mSec to

11.0mSec,\n but measure 22.59Kc in real life TEXT -104 176 Left 2 !K1 L1 L2 1.0 TEXT -40 216 Left 2 ;about 118T TEXT 16 80 Left 2 ;11T

Here is the ColpittsC.asc file

Version 4 SHEET 1 880 680 WIRE 416 -192 -240 -192 WIRE 512 -192 416 -192 WIRE 416 -112 416 -192 WIRE 512 -32 512 -192 WIRE 240 16 112 16 WIRE 304 16 240 16 WIRE 416 16 416 -32 WIRE 416 16 368 16 WIRE 448 16 416 16 WIRE -240 80 -240 -192 WIRE 240 80 240 16 WIRE 112 128 112 16 WIRE 304 144 240 144 WIRE 512 144 512 64 WIRE 512 144 368 144 WIRE 416 192 416 16 WIRE 512 192 512 144 WIRE -240 288 -240 160 WIRE 112 288 112 208 WIRE 112 288 -240 288 WIRE 240 288 240 208 WIRE 240 288 112 288 WIRE 416 400 416 272 WIRE 512 400 512 272 WIRE 512 400 416 400 WIRE 112 432 112 288 WIRE 416 432 416 400 FLAG 112 432 0 FLAG 416 432 0 SYMBOL ind 96 112 R0 WINDOW 3 29 108 Left 2 SYMATTR InstName L1 SYMATTR Value 21.29mH SYMBOL cap 224 80 R0 WINDOW 0 -29 8 Left 2 WINDOW 3 27 1 Left 2 SYMATTR InstName C1

SYMBOL cap 224 144 R0 WINDOW 0 -20 9 Left 2 SYMATTR InstName C2

SYMBOL npn 448 -32 R0 SYMATTR InstName Q1 SYMATTR Value 2N2219A SYMBOL res 496 176 R0 SYMATTR InstName R1 SYMATTR Value 10K SYMBOL res 400 176 R0 SYMATTR InstName R2 SYMATTR Value 100K SYMBOL res 400 -128 R0 SYMATTR InstName R3 SYMATTR Value 100K SYMBOL cap 304 32 R270 WINDOW 0 32 32 VTop 2 WINDOW 3 0 32 VBottom 2 SYMATTR InstName C3

SYMBOL cap 304 160 R270 WINDOW 0 32 32 VTop 2 WINDOW 3 0 32 VBottom 2 SYMATTR InstName C4

SYMBOL voltage -240 64 R0 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName V1 SYMATTR Value 9V TEXT -80 -104 Left 2 !.tran 0 20m 0 1u startup TEXT -216 -64 Left 2 ;I count about 16.3 cycles from 10.0mSec to

11.0mSec,\n but measure 18.18Kc in real life

--
Anyone wanting to run for any political office in the US should have to 
have a DD214, and a honorable discharge.

Ahh....point taken. Thanks.

• The inductor was not Navy, not Army, it was Air Core.

• 10Megs, 5pF.

• At 20V amplitude,noise is not an issue, forget triggering as a Protek

6500 was used to measure frequency at the scope Sig out B trig source (Tek R7704).

• Nope.