Building my first oscillator

[W4AMV]

These are SMALL SIGNAL diodes, tiny fellows. Hot carrier or Schottky diodes are desired, small signal Ge, the old 1n34a would work as well. Follow the design info in ARRL handbook to construct the trifilar xmfr needed and try to match the 4-diodes that form the ring mixer. It is not essential, but if you can, try to match the forward current against the applied V.

[W4AMV]

Yikes, ok SBL1 is not the part number of a mixer diode, but the part number of a completed mixer module from Mini Circuits Labs. They are cheap to buy I believe in small quantities, or try E-bay. However, if you want the thrill of building your own, go for it!

[ssbothwell SWL]

i definitely want to experience the thrill of building my own mixer

i'm gonna be super excited when i pick up my first station on an entirely home brewed receiver.

is the mixer you are talking about Figure 10.22 from chapter 10 (mixers, modulators and demodulators)?
thats the only DBM circuit i saw in the chapter. for whatever reason the book says that it uses a square-wave LO. is that gonna work with a sine-wave LO?

i believe this is the same circuit: http://www.qrp.pops.net/images/2008/2008-larger/big_dbm.gif

[W4AMV]

That is the exact circuit and a very clear diagram. You do not need a square wave. However, it will function with a square wave. Please read the details of how the mixer works in the text. Extreme brief, the LARGE LO signal operates each leg of the diode ring as a commutative switch. Therefore, the LO CHOPs the RF signal at an LO rate. This SWITCH function of the LO on the RF signal gives rise to an IF signal.This process of MULTIPLICATION of the LO and the RF to produce the IF will occur even with a pair of SINE waves. Break out your TRIG book and check it out Good Luck.

[ssbothwell SWL]

i've assembled the trifiliar wound transformers but i need to find some schotky diode.

in the mean time i tried putting together a 2n3866 linear amp stage for the amplifier in LTSpice. first i tried adding a linear amp exactly as i had learned to do in my previous thread about single stage bjt amps:
http://i.imgur.com/xxYwO.png
the two sinewaves displayed are the input and output from the 2n3866 stage. this didn't work at all and completely ruined my nice sinewave output from the buffer stage.

on a hunch i added an rf choke between emtter-ground resistor and base-ground resistor:
http://i.imgur.com/unV3n.png
this cleaned up the waveform massively.

is this circuit producing 480mW into a 50ohm load or am i misreading something? 14V Pk-Pk = 4.9VRMS. (4.9^2)/50 = 0.4802W

[W4AMV]

Your 50 ohm resistor is not returned to ground directly per your schematic, bit via the 1 mH choke. So that calculation is not correct. Is this the arrangement you really wanted?

[ssbothwell SWL]

oh whoops. that was a mistake. i'm not trying to get 480mW i just wanted to see if i had built an amplifier in the right way. i was actually suprised at that output power. when i directly ground the 50ohm resistor i get 200mV Pk-Pk and the sinewave is a little pointy on the positive peaks.

my goal is to get 100mW output to feed into the mixer as you recommended.

[ssbothwell SWL]

i did some adjustments and now i'm getting a much better waveform. its only 160mV Pk-Pk into a 50ohm load. i tried adding another transistor stage but things got really hairy.

http://i.imgur.com/L0McY.png

what kind of load should i be using if i plan to run this thing into a DBM?

[W4AMV]

First, pleased check your DC power dissipation in the 3866. If my calc. are correct, you're going to fry that poor fellow. On the mixer Z, that is a complex question! However, a good estimate is,,, drum roll, 50 ohms.

On the power amp, and the 3866 stage, let me make a reading suggestion. You need to study the idea of impedance matching, load lines, Class-A amplifier, and transforming a series circuit to its parallel equivalent and visa versa. I would simplify the Spice item, drop the oscillator buffer elements and concentrate on the 3866 design. Replace all of that with just a sine wave source whose frequency and voltage level is in line with what you are measuring on the bench. The goal of the 3866 stage will be to deliver between 7-to-20 dBm of power across 50 ohms. Start there. Hint, the 3866 can supply you with 15-20 dB of power gain at HF. Your supply V is fine, perhaps 12 V is a nice selection since voltage regulators are available.

Also, please check your biasing of the 3866, a simple standard 4-R bias is fine, although the emitter R selection is critical as their are tricks for maintaining the gain and achieving the bias stability. One trick there is split the R into 2-pieces, AC bypass one and leave the remaining R not bypassed to improve AC linearity, i.e. lower distortion. 

As time permits, I'll try to put a design flow for the 3866 class A amp-driver, that illustrates the above points.

[W4AMV]

Oh, I should mention on the mixer Z. The reason it is nearly 50 ohms, is partly due to the IF termination, arbitrary selection but convenient, 50 ohms. The longer answer has to do with the nature of the diodes, their impedance operating as switches, and the properties of the transformers. 

[ssbothwell SWL]

thanks for the advice. i definitely need to learn a lot more about linear amp design and impedance matching. z is something in particular that i struggle to understand.

2n3866 ratings are: Ic=.4A , Ib=2.0A, and Pd=5W.

i just did a lot of tinkering with the circuit and got it producing 6V Pk-Pk output with Ic=140mA and Ib=1.3mA.
i started out with the amp in its own spice file with a sinewave signal input and then translated it back into the oscillator file once i got it looks okay.

heres a screenshot showing all three values: http://i.imgur.com/w4EKh.png

does this circuit look okay or am i at least on the right track towards a working circuit?

[W4AMV]

Great! Now the poor transistor will be a bit cooler! I assume you set Spice up running first the DC operating point analysis. Incidentally, when you pass the measure probe over the device you can obtain DC current measure, and if you press CRTL (something, need to check), you get power dissipation of that element. You can check in the help directory.   

[W4AMV]

Ah, in transient analysis, not DC analysis, at least in my version. If you pass over an element as if you were to measure current and press the ALT key, the current probe will turn to a TEMP probe. LT does the math for you and calculates the power. So, in you circuit, if you pass over the load R, you should obtain a time signature of the output power. Do check its math to make sure it makes sense. You might try a simple circuit case just to be sure.

[w3jn]

100 mW is a bit much for a double balanced mixer with schottky diodes.  A better design point would be 10 dBm, or 10 mW.

[W4AMV]

I agree W3JN. However, as he becomes an expert, he may elect to do a high level mixer. In any case, it is easy to dump the power. The design project that I will send coincidentally is ~ 10 mW. Thanks!

Alan

[ssbothwell SWL]

Ah, in transient analysis, not DC analysis, at least in my version. If you pass over an element as if you were to measure current and press the ALT key, the current probe will turn to a TEMP probe. LT does the math for you and calculates the power. So, in you circuit, if you pass over the load R, you should obtain a time signature of the output power. Do check its math to make sure it makes sense. You might try a simple circuit case just to be sure.

oh awesome! i wish i had known i could do this a while ago.

i guess i was doing my power measurements wrong as well. i thought i was getting ~80mW across the load resistor but when i check its power dissipation i get 180/220mW (i get higher voltage on the trough then the peak, how can i fix that?).  http://i.imgur.com/25HlS.png blue is power dissipation in that image.


what are the situations where you would use a high vs low level mixer?

[W4AMV]

Good observation. The power reported appears to be the peak power from the work I looked at earlier. Should check the HELP menu/documentation to be sure. So the true RMS magnitude power is still available as you calulated earlier and measured from the voltage (rms) available across the external load, 50 ohms. A combination of phase shift and some distortion could account for your observation. In the file I put together I note that I do not see this alignment of power with voltage, Peak power aligns with peak voltage. In the final application for your work, it is not an issue.

The high level mixer, will provide the ability to handle larger RF input signals with significantly lower distortion. One method for achieving this is to add additional diodes in each branch of the ring. Thus you need increase operating voltage and increased LO power since the impedance level is still nearly the same. Again, for your work, 10 mW is quite sufficient. Point in fact, there are tricks to reduce the LO power even further by slightly forward biasing the diodes in the ring. You can experiment with that idea upon completion. 

[W4AMV]

I hope this helps

[W4AMV]

I guess I need to post added items, sorry ADMIN.

[W4AMV]

last items, won't do this again.

[ssbothwell SWL]

holy cow thats a lot of great info! this will take me a little bit of time to work through, but thank you so much for writing all this out.

[ssbothwell SWL]

okay so let me make sure i understand what you are doing in the pdf. on the bottom of page2/top of page3  you are calculating the impedance of a hypothetical transistor amp stage right?  and then after that you create an L-Matching circuit to go from 50ohm to the impedance of that hypothetical transistor amp stage?

do i need to create one of these L-Networks between the buffer and the 2n3866 amplifier? if so, would i assume the buffer is providing a 50ohm output or do i want to calculate its impedance and make a matching network based on the impedance of the two transistors?

thanks again for the info.


and just a little update. i put together the 2n3866 amp stage on the breadboard with the oscillator and buffer and i am getting clipped output. in ltspice the waveform looks great so i guess my model is not accurate. http://i.imgur.com/2dHln.jpg

[W4AMV]

In the paper pdf I did not mention finding a load line. So the first part was identifying at least part of a load line. I say part, because I did not identify a current limit just a voltage limit. Recall in the DC calculation, the total emitter current is 25 mA, half the maximum for that transistor. Anyway, if the voltage limit is the only limit than 288 ohms would permit that stated class A power to be realized just prior to clipping. Which brings us to your clipped waveform.  You are clearly being limiting by either the current swing or the voltage swing and to fix the clipping either modify your quiescent bias point or your load impedance. Use your scope, adjust your trace to find the bias point no signals, then see where and why your are limiting and make the appropriate adjustments.

Yes, you could work through an input match as well as you observed. However, if you have sufficient voltage or current gain, additional impedance matching may not be warranted. Since the FET buffer is a source follower, it in effect is a impedance transform network providing a low impedance output for the oscillator signal. Your following stage (3866) has a significantly higher input impedance and therefore at least ALL of the FET output voltage should be available to the 3866 input. Perhaps, that is where the issue of clipping occurs. A key question now, IS THE CLIPPING OCCURRING AT THE INPUT or the OUTPUT of the 3866? You are either providing to large an input voltage or current at the base of the 3866 and/or not providing the appropriate output load for the 3866 collector.       

[W4AMV]

and yes something is a miss in the model setup. The clipping should appear particularly if it is a DC bias error.

[ssbothwell SWL]

i just noticed that when i have the 2n3866 stage connected to the buffer, the buffer output drops down to 2.2VRMS, gets clipped, and is offset above the center line on my scope (is that dc bias?).

here is a photo of the buffer output connected and disconnected from the 2n3866.
http://i.imgur.com/oYFyj.jpg
http://i.imgur.com/HpziR.jpg

heres some questions:

1. can you explain what you mean by "Use your scope, adjust your trace to find the bias point no signals, then see where and why your are limiting and make the appropriate adjustments," and how i would do this?

2. when you say "..providing to large an input voltage or current at the base of the 3866," do you mean too much bias or too much input from the buffer amp?

3. if there is a problem with a mismatched load on the output of the 2n3866, would that be solved by an l-network between load and collector?