The AM Forum
March 28, 2024, 04:28:44 PM *
Welcome, Guest. Please login or register.

Login with username, password and session length
 
   Home   Help Calendar Links Staff List Gallery Login Register  
Pages: 1 [2] 3 ... 7   Go Down
  Print  
Author Topic: Building my first oscillator  (Read 152144 times)
0 Members and 1 Guest are viewing this topic.
W4AMV
Member

Offline Offline

Posts: 672


« Reply #25 on: August 25, 2011, 05:38:41 AM »

Looks like you are doing very well with your results. The answers to your questions (all) are available in an excellent text by Clarke and Hess, Communication Circuits Analysis and Design. In chapter 8 he treats both the bipolar and FET oscillator which are unique in the way the devices operate TO LIMIT THE FEEDBACK in an oscillator. On the one hand as JN indicates it is important to control the level of positive feedback, on the other how do I limit the feedback. The resulting limiting analysis and the resulting distortion is presented in their book. In the end, you can actually calculate quite exactly the values of the L and C required in the resonator (tank) and the required C's or L's (Colpitts, Hartley) in the feedback network to control the output voltage and distortion of the oscillator. I would search for a re-print of their text. Although the math requires the calculus, the resulting tables and graphs in the text, reduce the process to simple algebra! As time permits, I'll see if I can put their work into a simple example which closes the gap between simulation and calculation.

By trying different C values and their placement (location) in your circuit you are in affect modifying the feedback. Look at the total L and C in the resonator and the feedback, be careful to see if the L's and C's are in series or parallel as you find their total values. Then the operating frequency is APPROX 1/(2*PI* ROOT(LC)). 

Play with that 500 ohm drain R value in LT Spice. You will note that as you vary it in simulation and in real life the frequency will change. Why? Therefore, you may be better off, AC shorting the drain. And remove signal voltage from the source with your buffer.

You can draw a variable C in LT spice by modification of their C using their drawing tool.

 
Logged
w3jn
Johnny Novice
Administrator
Member

Offline Offline

Posts: 4620



« Reply #26 on: August 25, 2011, 11:35:14 AM »

thanks w3jn.

i feel kind of dumb, i had been changing parts around a lot and things got a little confusing. i took the whole circuit apart and put it back together and now the waveform looks great. for whatever reason i can't have more than about 150pF in the tank circuit otherwise it stops oscillating, but i have 100pF in parallel with the variable capacitor (which i now think is closer to 50pF) and the full rotation of the cap is a 1MHz sweep.

is it normal for the voltage to increase or decrease depending on the frequency of the oscillator?


Yes.  That's because the Q of the tank and feedback isn't constant over the frequency range, most likely due to parasitic capacitance and inductance.

It's not oscillating at the lower frequency due to insufficient feedback or too low a Q at that frequency in the tank.  Perhaps your variable cap has vanes that are shorting when it's close to being closed as well.



Logged

FCC:  "The record is devoid of a demonstrated nexus between Morse code proficiency and on-the-air conduct."
W4AMV
Member

Offline Offline

Posts: 672


« Reply #27 on: August 25, 2011, 11:52:45 AM »

Yes, JN is on target. Other issue is the reactance of 150 pf or more is just to low at the frequency of interest and the gain of the oscillator ( so-called loop gain) is too low to sustain oscillation. If you calculate the frequency using the eqn. from prior post you get ~ 2.9 MHz. One point, you have no C across the gate-source, so the feedback (in the Colpitts configuration) is now established by the FET gate-source C. You really want to swamp that out. So you will need to add a C across gate-source and adjust the C source to ground to set the feedback. Furthermore, the source to ground reactance might be best set with a choke in combination with R2. Lower R2 now to increases the open loop gain, since the FET gm will increase with increased current. If you monitor the drain current as you tune the oscillator, you will note that the current will vary. You are seeing the oscillator actually balancing the loop gain, by controlling its gm.  Smiley
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #28 on: August 25, 2011, 12:11:21 PM »

Here are some notes from LT Spice that illustrate some of the prior comments.


* Some OSC stuff.jpg (77.68 KB, 960x720 - viewed 1172 times.)
Logged
ssbothwell SWL
Guest
« Reply #29 on: August 25, 2011, 06:49:49 PM »

thanks for all the great suggestions. i added a 100pF cap between gate and source and things improved quite a bit. i am able to add way more capacitance to the tank circuit which enables me to dial in my frequency range much more precisely. this also seems to have stabilized the amplification quite a bit.

w4amv, the only rf choke i have is a radioshack 100uH one. would that work okay with R2? also, it should go in series right?
i'm going to monitor the drain curent while adjusting R2 later today. i'll report back what i discover.

does this buffer amplifier design look like what i would want to use for my oscillator? http://www.electronics-tutorials.com/amplifiers/buffer-amplifiers.htm
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #30 on: August 25, 2011, 08:17:40 PM »

The RFC value you mention should be fine. For your frequency, say 1-5 MHz, 100 uH has a reasonable reactance. The issue with large chokes eventually is their SRF, self resonant frequency, approach that and the osc. will stop. You can diddle the source R, up and down will change the output voltage ( as observed at the source) and the distortion. I'll look at the buffer, but at these frequencies a source follower should be fine, gate of follower tied to source of oscillator, lightly coupled, then you can add power amplification after that and a LPF. Will turn a blind eye for now to osc noise and a bunch of other complications. However, feel free to experiment. Taking the voltage from the resonator directly, again lightly coupled can be a far better arrangement. 
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #31 on: August 25, 2011, 08:21:05 PM »

Yes, the web site design is fine.
Logged
ssbothwell SWL
Guest
« Reply #32 on: August 25, 2011, 08:48:42 PM »

i included the rfc and dropped r2 down to 500ohms. vrms output jumped up quite a bit and was pretty stable as i swept the variable capacitor but the circuit stopped oscillating at lower frequencies. i went through various resistors between 500 and 1000ohm and found that around 800ohms i still get the full frequency range and vrms is reasonably stable but not perfect.

i'll report back again once i manage to get a buffer amp built. probably later tonight or tomorrow.
Logged
ssbothwell SWL
Guest
« Reply #33 on: August 26, 2011, 12:06:21 AM »

hmm, i dont have any zener diodes or 1mh rf chokes so i cant put this circuit together tonight.

what is the deal with chokes? can i use a little axial inductor like this: http://www.allelectronics.com/make-a-store/item/CC-1000/1000-UH-CHOKE-COIL/1.html

or do i want something else?
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #34 on: August 26, 2011, 10:12:03 AM »

a 1 mH choke is a bit large and the SRF could be an issue. Ideally try to choose a choke value whose reactance is 5x the value of the source-gnd feedback C at the center frequency of the oscillator. So if you use 100 pf for a 3 MHz unit, a 270 uH would be appropriate.

The deal with chokes is the same deal with C's and R's. They all have associated parasitic elements.  So when you buy an L you get a C and and an R at no extra charge Smiley The same is true for C and R.
Logged
ssbothwell SWL
Guest
« Reply #35 on: August 26, 2011, 03:37:44 PM »

sorry i was referring to a choke for the buffer amplifier. it has no feedback and the schematic mentions a 1uH choke.

Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #36 on: August 26, 2011, 03:54:47 PM »

1 uh ok, any thing in the range of 1-100 uH should be fine. 1mH to large, however, try it if it is your junk box. Otherwise, order what was specified 
Logged
ssbothwell SWL
Guest
« Reply #37 on: August 26, 2011, 04:14:37 PM »

oops! sorry again! i keep making things unclear. the buffer amplifier calls for a 1mH choke on the source.

http://www.electronics-tutorials.com/amplifiers/buffer-amplifiers.htm

am i right to assume that the purpose of the choke in this buffer amp circuit is to prevent the ac signal from getting sent to ground?

i'm just gonna go to the surplus shop and get a bunch of different size chokes to experiment with.

in this buffer circuit what would be the symptoms of too much and too little inductance in the choke?
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #38 on: August 26, 2011, 05:59:41 PM »

No problem. The 1 mH is fine and any reasonably large reactance will work well. The goal of the source inductor is to raise the Z of the source to RF AC so the voltage gain of the follower is nearly unity. Ideal, looking back into the source terminal the Z is 1/gm and no less. If less the voltage gain will fall below unity.
Logged
ssbothwell SWL
Guest
« Reply #39 on: August 26, 2011, 09:36:00 PM »

i just got some really nice scores from the scrap yard. i found a 2"x4" ceramic coil form, 1x1.5" ceramic coil form slug tuned, 3/4"x3.5" ceramic coil form, five air-variable capacitors (some the same size and some slightly larger than the one in the photos i have posted here), a mixed bag of silver mica caps, and 10x of 1mH tuning coils and 10x 2.2mH tuning coils. all that for $30.

heres the big coil form: http://i.imgur.com/Irz3C.jpg
the wiring on it is as i found it. i didn't think to get enamel wire in the proper gauge to fit the grooves in it so i am leaving it as is for now. i don't want to mess with recoiling it until i have wire that will fit it properly.


i just tried putting together the buffer amplifier and i'm not sure it is working properly. i used a MPF102 and i was not able to find a zener diode. the output directly from the oscillator is 8.8vrms with a perfect sinewave whereas the output from the buffer amp is 1.3vrms with the top of the waveform clipped off. heres a photo of the buffer amp output: http://i.imgur.com/artqK.jpg

any idea whats going on?
Logged
w3jn
Johnny Novice
Administrator
Member

Offline Offline

Posts: 4620



« Reply #40 on: August 26, 2011, 11:39:17 PM »

Too much coupling from the oscillator to the buffer.  Try a 10 pF or so capacitor between the two.

A zener isn't necessary if your power supply is regulated.  That coil you have is really nice.  It's built for a high power transmitter tank but the size and large wire will make a nice stable oscillator, if you mechanically mount it right.

One thing I noticed is you're using those white plugin proto boards.  Those aren't good for experimenting at RF - been there, done that.  A better way is to get some scrap PC board and engrave some islands with a moto tool, and just solder the components directly to the PC board.
Logged

FCC:  "The record is devoid of a demonstrated nexus between Morse code proficiency and on-the-air conduct."
ssbothwell SWL
Guest
« Reply #41 on: August 27, 2011, 12:32:24 AM »

the coil has nice brackets on the bottom for mounting it. everything is really precarious right now but once i get the whole circuit operational i will make it mechanically stable.

i'm on the fence about what to do with this coil. with it as is i can't get below 2.5MHz. i want to rewrap it with more turns but at the same time it is in really nice shape the way it is. i'm thinking i might try to keep it as is and use something else to make a coil for lower frequency operation. given the sizes (2" diameter) and the number of turn on it currently (10) what frequency band would it work the best on?

well it turns out i had been messing with the buffer circuit to much and had some misplaced parts on my bread board. i rebuilt it and i am getting slightly better results: http://i.imgur.com/7ritk.jpg

i also put together the buffer circuit in ltspice: http://i.imgur.com/EvdMh.png
the blue wave is the gate of q1 and the green wave is the source of q2 (buffer output)

the ltspice results are waaay better than my real life results.
Logged
w3jn
Johnny Novice
Administrator
Member

Offline Offline

Posts: 4620



« Reply #42 on: August 27, 2011, 03:11:51 AM »

Get rid of the breadboard and use short leads on your components.  The parasitic capacitance and inductance is killing you.  That RF choke is way too big and it's lined up in parallel (physically) with your tank coil.  There will be coupling between the two.
Logged

FCC:  "The record is devoid of a demonstrated nexus between Morse code proficiency and on-the-air conduct."
W4AMV
Member

Offline Offline

Posts: 672


« Reply #43 on: August 27, 2011, 03:54:36 AM »

Agree with all of W3JN comments. Pickup some scrap copper pc card, FR4 is fine. If it is only single sided so much the better to reduce capacitance per each pad you cut. A dremel tool with a fine grinder bit will allow you to remove copper and form the tie points.

LT Spice of course is now giving you fantasy results. Welcome to the world of trying to achieve accurate simulation models.
Logged
ssbothwell SWL
Guest
« Reply #44 on: August 27, 2011, 09:08:36 AM »

do you guys really think the breadboard is causing enough stray capacitance and induction to give me the results i am getting? without the buffer amp i get really nice and stable sinewaves.

i do have a bunch of single sided pc board. i usually cut little .25" squares out of it and glue it onto a larger piece which acts as the ground plane. would that be okay or would there be too much capacitance between the little pieces and the large piece they are glued onto?
Logged
w3jn
Johnny Novice
Administrator
Member

Offline Offline

Posts: 4620



« Reply #45 on: August 27, 2011, 11:15:36 AM »

Naw, the squares should be fine so long as you don't try above 15 MHz or so.  Easier though is to just cut islands and traces with a MotoTool. 

And yes, I think parasitics are part of your problem.  Did you reduce the coupling to the buffer amp?  Try a capacitive voltage divider on the input of the buffer amp.  10 pF between the oscilaltor and the amp input, and 200 pF or so to ground.  I think you have a combination of loading down the oscillator with overdriving the buffer amp.
Logged

FCC:  "The record is devoid of a demonstrated nexus between Morse code proficiency and on-the-air conduct."
W4AMV
Member

Offline Offline

Posts: 672


« Reply #46 on: August 27, 2011, 07:33:44 PM »

I agree with JN. From the voltage wave forms you posted, it appears that the gate -source of the FET buffer is driven to hard and you are seeing the gate-source diode junction clamping the waveform. Try a Cap voltage divider or simply use a small C to couple the source point of the osc to the gate of the buffer. You can get all the voltage gain you want from a third stage and low harmonic content with a LPF latter added. The main job of the buffer is to isolate the oscillator section from variations in the external load.
Logged
ssbothwell SWL
Guest
« Reply #47 on: August 28, 2011, 03:05:38 AM »

hi guys,

i figured out the problem and now it is producing pretty much the same results as the ltspice model. i'm just gonna say the problem was really dumb and i am kind of embarrassed. it involves a misplaced jumper cable.

i didn't get a chance to put together a pc board version of the circuit today. but i'll be doing that possibly tomorrow.

other than mechanically securing the oscillator, what do you guys think i should try to do at this point? is it now safe to try hooking up a 2n3866 linear amp like the designs i worked through in my previous thread?

i want to make two of these oscillators so that i can try building a receiver and a qrp transmitter. i guess it makes sense to attempt a receiver first?
Logged
W4AMV
Member

Offline Offline

Posts: 672


« Reply #48 on: August 28, 2011, 04:13:25 AM »

I would do a Rx first. Also, I assume you are going to transfer this work to a nice stable setup, not use the patch board. In any case, if you do the Rx first, it will serve you as a test bed. The Rx will tell you how stable is this thing, how noisy is my osc, is their jitter or a steady drift, etc... Plus its fun to listen. Good Luck.

If you use a high level passive mixer, like a diode ring mixer, a 2N3866 works well. A 100 mW or +20 dBm will suffice and a low pass filter prior to mixer injection is useful. The 3866 will have 15-20 dB gain, run at 13 V and 50 mA. Designs are on the WEB, in the HDBK, etc... or roll your own since you have the tools. 
Logged
ssbothwell SWL
Guest
« Reply #49 on: August 28, 2011, 03:17:19 PM »

okay i'll definitely start with a receiver. it will be fun to pickup signals on a home brewed rig. i started reading the arrl handbook chapter on receivers. i'm going to start with a direct conversion design and then add an IF stage or two afterwards. http://i.imgur.com/5irON.png

just to be clear, in your last post you recommend i amplify the vfo to 100mW and run it through a low pass filter prior to mixing it?

regarding mixers:
all the arrl design examples use sbl1 mixers but they dont specify forward/reverse voltages, and current. i see bunch of variations on sbl1 available from digikey: http://search.digikey.com/scripts/DkSearch/dksus.dll?Cat=1376383&k=sbl1
which of these would you recommend?

also, would this schottky diode work? http://www.allelectronics.com/make-a-store/item/STPS30L40CW/SCHOTTKY-DIODE-40V-30A/1.html
allelectronics is way more convient for me than ordering online so it would be great if i could use this part.
Logged
Pages: 1 [2] 3 ... 7   Go Up
  Print  
 
Jump to:  

AMfone - Dedicated to Amplitude Modulation on the Amateur Radio Bands
 AMfone © 2001-2015
Powered by SMF 1.1.21 | SMF © 2015, Simple Machines
Page created in 0.069 seconds with 19 queries.