Rick Campbell's One Transistor Amp Article

[ssbothwell SWL]

hi, this is my first post here so forgive me if i do anything wrong. i'm new to ham radio and don't yet have my license. i've just been really busy lately and haven't gotten around to scheduling a test. i have read the ARRL license study guide and large sections of the ARRL handbook. 

i found the QST article by Rick Campbell KK7B called "Designing and Building Transistor Linear Power Amplifiers" on the ARRL handbook CD-ROM and thought it would be fun to try working through the methodology he presents. i'm not actually broadcasting anything or hooking it up to an antenna so i don't think i am breaking any rules doing this without a license.

i'm using a 2n3866 transistor which i found in a schematic for a low power linear amplifier in the ARRL handbook (page 13.24). i've gotten so far as building the LM317 bias circuit from Figure 4 in Rick Campbell's article and hooking up the transistor as in Figure 3 from the article. I am able to 'turn on' the transistor and i have gotten modest voltage gain on a signal from my function generator.

i am using a 50ohm dummy load as is recommended in the article. In the section titled Load Network (page 3) it says that with a 50ohm load the power output is a product of the supply voltage and that with a 12v supply i can get up to 1.44W of output. i assume that means i should be able to get ~8.5v P-P output across my dummy load (E^2/R=P).

i am sending a 7MHz signal at 100mV from my function generator into the transistor base along with a bias from the LM317 and i am powering the collector with 12V. my collector current and P-P voltage output to the dummy load seem to shift as the transistor heats up. i only have one multimeter so i cant look at bias voltage and collector current at the same time.

in any case the highest P-P voltage output i have achieved is 3.2V which i get when my collector current is ~30mA. the higher the collector current the lower my P-P voltage output. does that sound right?

also, will a heatsink significantly stabilize the collector current from my transistor? i don't have any to-39 heatsinks currently but i have one coming in the mail next week.

i was under the impression that i should be able to get much higher output from one transistor even without a matching network. how do i achieve the 1.44w output mentioned in Rick Campbell's article?

[KA2DZT]

Welcome to the AM Forum.

I'm not familiar with the acticle you're working from, but did want to congratulate you for trying to build something.

Many hams today manage to get a license and then buy everything for a station.  They never build anything.  Most folks you'll find on this forum are builders.

I built my first radio at age 7, that was 58 years ago.  I was given (by a ham) a 1953 ARRL Handbook in 1953.  I still have the book and refer to it often.  I don't think I've mastered everything in the book but I'm going to stay on it until I do.

Keep up the good work.  I'm sure someone on here will come along with more help with your project.

The one thing I will mention,  1.44 watts on 40 meters??  I think you're going to need a bigger set.

And don't worry about doing anything wrong,  there are plenty of experienced hams on here that screw-up all the time (of course, I'm not one of them).

And I will note, that from reading your post, I can see that you have some knowledge of electronics.

Fred, KA2DZT

[ssbothwell SWL]

thanks for the kind words. i'm really excited about getting more into this hobby. i do have a little experience with electronics, mainly audio stuff, but i have always wanted to get into radio. i recently decided to just go for it and have been reading as much as possible. rf electronics is a lot more complex then the kind of stuff i have done in the past!

i admire the the fancy new ham gear that is available for sale, but i really like the idea of building stuff myself. its nice to know exactly what is going on 'under the hood.' 

at some point, likely after i get my license, i plan to try building larger gear but for now i want to try building low power stuff to experiment with on the bench. i would like to know i can properly tune a single transistor stage before trying to add additional ones.

[KF1Z]

Across your 50 ohm load you should see 20 volts Pk-Pk for 1 watt of output. ( with O'scope AFTER the DC blocking cap)

And are you measuring the the output voltage with a multi-meter? or with an oscilloscope?

If you are using a multi-meter, you likely aren't getting an accurate reading.


Are you using figure 3  in the article ?

Also,
The 2n3866 HAS to have a heatsink!
"We" use these devices for 1 watt  driver amplifiers in the SDR transceiver kits I build.

They can get very hot even with the to-5 heatsink on them.
They are good devices, but they are delicate!
\

[ssbothwell SWL]

thanks for the info. i am measuring on an oscilloscope so i am getting Pk-Pk readings.

i have some heatsinks coming in the mail sometime next week and i don't mind burning out some of these transistors testing them before the heatsinks arrive. the main problem i notice with no heatsink is that as the transistor heats up, the collector current changes which causes me to have to constantly adjust the bias voltage to keep a steady power output. at this point its not a big deal because i am just testing out the circuit but hopefully a heatsink will fix this.


yes, i am following figure 3 in the article. the highest voltage i have achieved is 3v Pk-Pk which is what i get with a direct (no resistor) ground shunt from the emitter, and 90-100mA current on the collector.

The only discrepancy i notice between my circuit and that of the articles is the RF Chokes. the article does not say what inductance values to use for the chokes and so i put a 15uH one on the bias input and a homemade one made of 8turns on a FT82-43 (http://toroids.info/FT82-43.php) on the power input (like in the schematic). these are the only inductors i have available, although i have additional FT82-43s and FT37-43s that i could wind if necessary.

when testing a transistor i plan to use for the first stage of amplification, what sort of signal input should i be using? i have a nice HP function generator i got from a swap meet which i have been using. i set it to 100mV sine wave input at whatever frequency. is that the appropriate voltage i should be using to simulate the output from a transmitter's oscillator?

also, if i wanted to drive a transistor i plan to use in a different stage, what would i use? do i just use E=sqrt(P*R) [where P is whatever power i plan to input into the transistor] to determine the voltage of the function generator?

[KB2WIG]

S,

I'd sugest that u go to the homely despot and buy some copper pipe caps. Drill a hole through the cap and shove the transister case through the hole. If u get something of a tite fit, you'll get some enhanced cooling. Or you can place the case upside down in a shallow plate of water and get to chillin this way. Dont spank the silicon.


Whatt kind of biass does this de vice have??

klc

[ssbothwell SWL]

i only have one multimeter so i can't monitor current and bias voltage at the same time. that combined with the fluctuating current due to heating makes it difficult to correlate particular voltages with particular currents but i am running it around 0.6 to 0.7vdc for bias and 12vdc for supply.

i guess when i think about it getting 3v from .1v is a factor of 30. is that as much as i should hope to achieve from one transistor?

[WA1GFZ]

The 2N3866 is a beast but if you want 1 watt output you will need to dissipate over 1 watt in a heatsink. Also look at the transformer output ratio. This will effect the peak to peak output voltage. I've never run them over 1/2 watt output myself. Usually these devices run anroud 17 dB of gain with feedback. They will run higher gain but could oscillate without feedback. They will operate at over 200 MHz.

[ssbothwell SWL]

i just got the heatsinks in the mail. this is the type i got: http://www.westfloridacomponents.com/G280APE12/TO5+TO39+Heat+Sink+Seifert+Electronic+KK-502.html

they are describe as 'snap-on' but they fit very loosely over the transistor. am i supposed to crimp them down?



WA1GFZ, maybe this is exactly my problem. my circuit does not have any transformer at all. heres a drawing of my circuit: http://i.imgur.com/j7Pl7.png

i'm not using any feedback resistors or anything. the bias is coming from an lm317 circuit, eventually i will build a more proper bias circuit but i am using the lm317 as directed in the article i am following.  the 50ohm resistor is my dummy load.

what is the purpose of the transformer you mentioned? is it part of a matching network or something

[ssbothwell SWL]

these heatsinks made a huge difference. i now get a very stable 3v Pk-Pk output from the 2n3866. i'm still not anywhere near the 1w output i want to achieve but at least the heating issues are dealt with.

[AB2EZ]

Hi!

As drawn, if the transistor were an ideal transistor, your circuit would have an RF voltage gain of 50.

That is, the 7MHz current flowing through the 1 ohm emitter resistor would be almost the same as (approximately 101% of) the 7MHz current flowing through the 50 ohm RF load.

As a practical matter, the voltage gain will be somewhat less than 50 because:

1. There is some additional emitter resistance associated with the transistor itself... so the effective emitter resistance is somewhat larger than 1 ohm. Likewise, there is a small effect from the base resistance of the transistor that acts to reduce the 7MHz voltage across the effective emitter resistor (i.e. total of 1 ohm and the transistor's added emitter resistance) v. the applied 7MHz base-to-ground voltage. So, in summary, the rf voltage across the actual 1 ohm emitter resistor will be somewhat less than the applied input rf voltage (measured from base to ground).

2. There is some parasitic capacitance from collector-to-emitter associated with the transistor itself, the wiring, the rf choke, etc. that provides an impedance to ground that is in parallel with the 50 Ohm load. So not all of the 7MHz collector current flows through the 50 Ohm load.

Remember: depending upon the type of probe you are using, your oscilloscope probe, by itself, may have an input capacitance of 10-15 pF. If, instead of using a probe, you run a coaxial cable (or a twisted pair) between the 50 ohm load resistor and the scope... with the scope set to high impedance input... the added parallel capacitance across the 50 ohm load will be around 20pF per foot of cable. At 7MHz, an unterminated 6ft cable (not terminated with 50 Ohms at the scope end) would have an impedance of arround -j189 ohms. You would get a more accurate measurement of the peak-to-peak voltage across the 50 ohm load by placing the 50 ohm load at the scope end of the cable (if you haven't already done so)

Therefore, the actual RF voltage gain will be C x A/B... where A is somewhat less than 50 ohms, and B is somewhat larger than 1 ohm, and C is an input attenuation factor that is somewhat less than 1.

If  C x A/B is 30 (instead of the ideal value of 50), then 100 mV of peak-to-peak rf input will produce 3 volts of peak-to-peak rf output.

Good luck! I'm looking forward to your progress updates

Stu

[ssbothwell SWL]

so what you are saying is that a voltage gain of 30 (100mv to 3v) is perfectly normal and should be expected. is it possible to get greater gain without using additional transistor stages?

if not, what is the best way to go about building additional stages. without any success, i tried running the output of one 2n3866 into a second one, both transistors being setup exactly the same. i the output of the second transistor was only a few hundred mV.

i've done the same thing 2n3904 transistors at audio frequency levels and gotten rather large gain.

[AB2EZ]

There are many ways to obtain more gain. Let's start with using two stages of gain... as you mentioned. [We can then talk about using a 1:2 step up transformer at the output of the 2nd stage, in conjunction with some other changes]

You said that when you placed two identical stages in series, you obtained very little output from the 2nd stage.  

I would suggest that you try it again, keeping the following in mind:

Note: there are a thousand different ways to do this... so let's pick one... and make it work (-:

1. Make sure that the capacitor (C1= 0.1 uF) shown between each stage's collector and its corresponding 50 ohm load resistor is in place in both stages. [I.e., use a 50 load for each stage]

2. In addition, make sure to keep the 0.1uF capacitor at the input of the 2nd stage. This will ensure that the dc biasing of the 2nd stage is not disturbed by the 50 ohm load at the output of the first stage.

3. Use a larger value for the emitter resistor in both stages (try 2.5 Ohms). This should result in a gain of around 13 (maybe a little more) per stage... for a total voltage gain of around 169.

4. If you happen to achieve a total voltage gain that is larger than 200, then you will have to reduce the input peak-to -peak voltage, because your power supply can only support a peak output voltage swing that is around 2 volts less than the power supply voltage. I.e. if your power supply voltage is 12 volts... and if you adjust the rf input voltage level so that the collector voltage of the 2nd stage swings between 22 volts and 2 volts, with rf applied to the input... then that will still leave (at the bottom of the swing) 2 volts from collector to ground. Some of that 2 volts will be across the emitter resistor, and the rest will be needed between collector and emitter to keep the transistor working properly.

5. You will have to adjust the base bias voltage on the 2nd stage to be around 1.7 volts... so that the emitter resistor of the 2nd stage has a voltage across it that swings up and down around a dc value of 1 volt. You should also adjust the base bias of the 1st stage to be around 1 volt, so that there will be around 0.3 volts across the emitter resistor of the 1st stage.

As an aside, in your posted circuit diagram, both the DC voltage sources are drawn with the polarity reversed from what it should be.


Good luck

Stu  

[ssbothwell SWL]

AB2EZ,

i'm wondering what the best way to setup my bias would be. i've been using an lm317 but that doesn't seem like a good long term solution. i just tried a voltage divider but with poor results. i was trying to replicate my current 3v Pk-Pk results before moving on to the suggestions you made.

first i tried 150ohm and 10ohm for R1 and R2 of the voltage divider. i got 200mV output and a burnt finger from an over heating resistor. then i bumped the values up to 15000ohm and 1000ohm. that gave me a 825mV output at 7MHz and a deformed sine wave. at 13MHz i get a normal looking sine wave and a 1v output.

i've read about a couple other bias circuits such as these: http://www.qsl.net/va3iul/Bias/Bias_Circuits_for_RF_Devices.pdf
but haven't found any formulas for generating the proper component values.

how would you recommend i setup the bias?

[AB2EZ]

Getting the bias set and stable (with transistor temperature changes) will be harder with the 1 ohm emitter resistor in the circuit than it will be with a 2 ohm or a 2.5 ohm or a 3 ohm emitter resistor in the circuit. As the transistor heats up, the optimum bias voltage will change a bit.

Try this for the biasing:

[Again... there are a 1000 ways to do this. This approach is not as elegant as some alternative approaches. Resist the temptation of jumping from one approach to another, until you get one approach working. Then, as you begin to understand the details of how the circuit works, you can experiment with the other 999 ways... in order to do some optimizations of: the overall energy efficiency of the amplifier, automatic compensation of the bias voltage for transistor temperature changes, the component count, etc.]

1. Change the emitter resistor from 1 ohm to 3 ohms.

Note: if you use a new emitter resistor value of 3 ohms, you will reduce the single-stage gain from its present value of around 30 to a value of around 30/3 = 10 (probably somewhat more than 10... maybe closer to 13, maybe a bit more than 13). Nevertheless, you should increase the value of the emitter resistor in this particular circuit configuration. After you get a simple 2 stage amplifier working properly... you can consider alternative designs.

2. Add a resistor, from base to ground, whose value is approximately 50 x the value of the new emitter resistor. If the new emitter resistor value is 3 ohms, then the resistor added from base to ground should have a value of 150 ohms.

[Note: looking into the base of the transistor, there is an intrinsic resistance from base to ground... whose value is (approximately) beta x the emitter resistance value. Beta is a number around 100, but according to the 2N3866 data sheet, it could be significantly lower. Beta is temperature dependent. To make the biasing reasonably stable with temperature, the resistor that you add, from base-to-ground, should have a value that is substantially less than: beta x the emitter resistance.]  

3. For the 1st stage (initially the only stage), the resistor from the 12 volt supply to the base should have a value of approximately 12 x the value of the resistor you use from base to ground. If the resistor added between the base and ground is 150 Ohms, then the resistor from the base to the 12 volt supply should have a value of 1800 ohms. The wattage rating of the 1800 ohm resistor should be at least 0.25 watts (not 0.1 watts).

Note: the resistor you add from base to ground will act not only as part of the biasing circuitry. It will also act as an additional rf load on whatever signal source is used to drive the base of the transistor. Therefore, depending upon the characteristics of the 100mV rf source... the rf voltage it produces at the input of the transistor will be somewhat less than before.

4. For the 2nd stage, the resistor from the 12 volt supply to the base should have a value that is around 8 x the value of the resistor that you use from base to ground. If the resistor added between the base and ground is 150 Ohms, then the resistor from the base to the 12 volt supply should have a value of around 1200 ohms. The wattage rating of the 1200 ohm resistor should be at least 0.25 watts (not 0.1 watts).

Note: You are biasing the 2nd stage so that the dc voltage across the emitter resistor (i.e. between the emitter and ground) will be around 0.65 volt. This will allow the emitter-to-ground voltage to swing up to at much as 1.3 volt and down to as little as 0 volts (1.3 volts peak-to-peak of rf emitter voltage, without turning the transistor completely off). The actual peak-to-peak voltage that you will want across the emitter resistor of the 2nd stage, to obtain 20V peak-to-peak output from the 2nd stage will be approximately 1.2V.

With this biasing, the average emitter current will be 0.65V/3 ohms = 217 mA. The heat dissipated in the 2nd stage transistor will be approximately 11.35V x 0.217A = 2.5W. Therefore, only run the amplifier for short periods of time until you determine that the 2nd stage transistor is well enough "heat-sinked".

[WA1GFZ]

The bias needs to be temperature compensated.
I talked about an output transformer because usually a 2:1 stepdown is used to drive 50 ohms with a TO5 device like the 3866.
Consider you want 1 watt of power which is about 7 volts RMS across the 50 ohm load. That is about 20 volts peak to peak so not possible with a 12 volt power source. I like the 2N3375 when I want to make power above 1/2 watt.
It is a stud mounted device that will directly drive 50 ohms.
Now you could make the 2N3866 make that power if you increased the collector voltage or added a 1:2 step up ratio transformer. Either of these options would
really push the maximum ratings. You might consider a push pull pair.

[AB2EZ]

Frank

In the circuit shown, the collector is fed by an inductor (a "collector choke"). Therefore, the peak-to-peak output swing can be as large as 2 x [B+ - 2volts] = 20V.

I agree that the final design should employ temperature compensated biasing of the 2nd stage... but since SWL is apparently in a steep learning mode, I suggest that temperature compensated biasing, etc. be left as embellishments of a simple initial design. For example, running the amplifier in Class A (as per the initial design) is probably not the best approach for the final design. Therefore, as part of the embellishment process, an output filter will need to be added.

Stu

Note that I have made some edits to the suggested values for the biasing resistors in my prior post.

[WA1GFZ]

Stu,
I'm not sure a single ended amplifier would be clean if run outside of class A mode unless the inductor was changed to a tuned tank.

[AB2EZ]

Frank

The inductor will block the fundamental and all of the significant harmonics. The "output filter" that I referred to is equivalent (for the purposes at hand, where we are not looking for a large impedance change between one side of the filter and the other side of the filter) to a "tuned tank" circuit. So I think we are saying the same thing. [Some would say that every LC filter stores energy...and is therefore a collection of tuned "tank" circuits].

I also agree that the filtering will be somewhat easier to accomplish... for a given set of target ratios of: 2nd, 3rd, etc. harmonic power, to the fundamental component's power...  if a balanced design is used to cancel out the even harmonics. But, either way (single ended or balanced), at this output power level, obtaining large amounts of harmonic suppression is less of a concern.

The output filter should be designed to provide a relatively low impedance path to ground for all but the fundamental frequency component of the collector current (just like a tuned tank circuit would). It would probably be a Chebyshev bandpass or lowpass filter. Therefore only the fundamental frequency component of the collector current will flow into the full value of whatever load impedance / resistance is used.

Separately, as you pointed out, a broadband transformer can be used, if needed, to better match the transistor(s) in the output stage to the load.


Stu

[ssbothwell SWL]

sorry i didn't respond sooner. i don't have those R values in 0.25w rating so i am heading over to the electronics shop today to try to find them.  i will report back when i have tested out the circuit.

for the base-ground resistor you mention that it should be significantly lower than Beta*Emitter Resistance. with that in mind, why do you recommend a 150ohm resistor?

[AB2EZ]

You asked:

"for the base-ground resistor you mention that it should be significantly lower than Beta*Emitter Resistance. with that in mind, why do you recommend a 150 ohm resistor?"

If the base-to-ground resistor is too small, then the electrical power consumption in the voltage divider (and the associated heating of the resistors) will be too high. The power consumed in each base-bias voltage divider will be (roughly) 12V x 12V / [the sum of the two resistors that comprise the voltage divider]. The electrical power consumed by the 2nd stage biasing circuit will be 12V x 12V / [9 x 150 ohms] = 0.1 watt. I didn't want to make the power consumption any larger than that, because I don't think it is necessary for this iteration of the design.


I think that 150 ohms will be a good compromise (for the time being) if the emitter resistor value is 3 ohms and if beta is greater than 50 for the specific transistors you have. The 2N3866 specification sheets (from various manufacturers) specify the value of beta as being somewhere between 10 and 200 (depending upon the specific device). That's an enormous range for this parameter. The specified range for a 2N3866A is between 25 and 200. Hopefully the devices you have are on the high side of the range.The earlier results you reported suggest that betas of your devices are on the high side of this range.

After you get the basic design working... your next step will be to transition to a more sophisticated biasing approach that will do a decent job of keeping the bias adjusted properly (as the transistor temperature changes) without wasting a lot of electrical power.

[ssbothwell SWL]

stu,

i just finished testing the setup you recommended and i didn't get the exact results you described.

with the first stage transistor setup i get ~300mv Pk-Pk output from a 100mV signal input. with the second tranasitor setup as well, i got 1.2v Pk-Pk which rapidly declined to ~600mV at which point i turned off the circuit.

i checked the current between the emitter and ground and it was at ~130mA. the current between the collector and the coil was ~120mA.

[AB2EZ]

Okay

Can you produce a complete schematic of what you have now and post it here?

Stu

[WA1GFZ]

I suggest you find yousellf a copy of the ARRL Soild State handbook. It has a great section on broadband amplifiers. Sounds to me the second stage is putting a heavy load on the first. also I have never biased a 3866 over about 65 ma. That poor thing is begging for mercey

[ssbothwell SWL]

heres my schematic:
http://i.imgur.com/Nlh8a.png
i forgot to label all the capacitors but they are all 0.1uF. for some reason i i didn't get any signal if i used only one capacitor on the output of Q1 like this: http://i.imgur.com/1cc8T.png
why is that?

should i be using a different transistor for the second stage? i have a couple of these guys: http://www.fairchildsemi.com/ds/KS/KSE44H.pdf