Sunday, October 28, 2018

Miller effect and gain band width

The voltage multiplier will be used to illustrate the circuit reaction to component changes.
I added a 1p capacitor to see the frequency response. You can see it falls off on the high end.
I changed it to 10p and you can see it falls off faster.
The 20p falls off even sooner. From this we can see the capacitance is our enemy when trying to get high frequency response. Our circuit will have base collector capacitance and it will limit out high frequency. What can we do about it?

A circuit such as this can be resolved to a single vector at a specific angle. In simple terms Xc and Xl cancel each other. One will dominate. So What if we put a coil in our circuit?
 Look at the effect the coil had.

As you can see the coil helped. What else can we do?

I changed the frequency to show it is a frequency response in play.
I reduced the input Z to show the effect.
I varied the gain to show the effect is at a certain frequency. At low frequencies the gain effected the output but a higher frequencies the gain fell.
If you followed this you should see the gain effects the roll off frequency and the input Z effects the roll off  frequency. To get higher gain at high frequencies we need less capacitance. Placing a coil in the circuit can help with this. Using a lower Z will also improve the high frequency response.

Remember the capacitance will be there. You may find a better transistor for high frequencies. Look at the Ft.

A lower Z circuit will have better frequency response. Some RF designers use 50 Ohms for their I/O for this reason. The low Z circuit will have a higher current draw. For this reason a 5 volt amp may have lower power consumption than a 12 volt amp and produce the same gain. You can use this calculator to see the cut off frequency for different R C values.

1/(2*pi*r*c) calculator

Monday, October 22, 2018

Current mirror - observations and final design- ?

I put the current mirror on a board and it does sing. (to well actually)
The test board worked so well I added the jacks, switch and a volume pot.
The problem is it is to sensitive. I put a pair of back to back diodes across the input to limit the drive. With a wire clipped to the input it sounds like a radio between stations. With a tank on the input I could tune it and select a station from the mix. We have a local low power day time station about 12 miles from us. This mourning I turned the amp on and could hear it. Low in the background but  there. With a test lead clipped to the input it is an easy listening level. So it's back to the drawing board. The simplest way to reduce gain is to reduce stages.

This is the amp as built. With the 2.2K resistor in series with the earbud it is still to high gain.
Here is the final version. I bypassed the middle stage. I have two resistor representing the earbuds. The red graph is one and the yellow is across both. Much better but it could still use a gain control.
With a 10K pot adjusted from 100 Ohm to 10K Ohm in 100 Ohm steps, this is what the output looks like.
So I replace the gate resistor for the output fet with a 10k pot and bypass the second stage.

I think a new build is in order. I can use the one I have as a signal tracer. I need to make a probe for it. Another project to add to the list ;).

*** I did the new build with two stages and I'm using 8 Ohm headsets. I connected a DBM and picked up AM1160. It is a low power (1 KW) in Dallas Texas. That's about 600 miles.

Friday, October 19, 2018

A couple of modifications to the current mirror

I put the amp on a board last night and it does sing. It needs an volume control and jacks for the I/O. It was a little unstable, when I clipped a jumper across the input it oscillated. So a closer look at the circuit.
There is current in the battery. This is causing inter-stage modulation.
A little more filtering and the battery current is steady. If you want a high Z output this is a good time to address that.

When adding the filter on J3's drain you can use it to feed a high Z headset. I'm showing a dual output. In this sim we have a headphone on the source and drain of J3. So you can connect it to your favorite crystal set and share the listening experience with a friend.
If you don't have the 2SK2539 you can build the circuit with any Jfet. The biasing may require some adjustment. I substituted J112s and it still function but at a lower gain.
Look at the first set of curves on the last post. You would need a 3.3K Ohm or 4.7K Ohm resistor with that Jfet to get 1 ma. It would function if built using the 1K Ohm but would have a very high current drain on the battery.

Thursday, October 18, 2018

Is it a current mirror?

The circuit above can be used to develop the characteristic curves. The Jfet characteristic curves are divided into two regions. Today I'm looking at the saturation region.
The family of curves you might find in the data sheet. You can use the math to define your circuit parameters are you can find it on the curves. The math may be better for 'real' design work but for home brew the curves will deliver fair results.
For the circuit I'm looking at today all I need to know is where the bias needs to be to create the 'flat line'. The best place to look is the datasheet for the device you intend to use. I will be using the 2SK2539.
Look at the -1.0V curve and you can see it is at about 1 ma. This make the bias point of the circuit very simple to establish. I want about 1 ma so I simply put a 1K Ohm resistor in the source circuit. I need to connect the gate to the bottom of the 1 K resistor. Since there is no current in the circuit I can use any value. I choose 100 K Ohm.
This is my circuit. Looking at J1 you see R1 sets the bias between the source and gate. R2 completes the connection for the gate. Q1 and J1 in series will have the same current ( 1ma ) when Q1 is at it Q point. If Q1 is biased on or off J1 will respond to maintain the 1 ma. As Q1s collector swings up or down C4 couples the change to the input of the second stage. The second stage operates in the same manner.
The gate of J3 will see changes applied thru C3 and its source current will vary. R13 sets the DC bias of J3. C7 couples the AC component to the load. Varying the load resistance will vary the current thru the load. It is much like a transformer. The voltage drop across R13 sets the maximum output so I can use a 10k headset or a 2k headset with the same voltage output. If I increase the load the voltage will drop as the current increases (as in a transformer). I plan to use a 60 Ohm earbud. The earbud is loud enough with 1 mv.

I vary the load from 100 Ohm to 10K Ohm. Once the resistance exceeds 1K Ohm the source resistor draws more (AC) current away from the load.
With 10 uv input I get 17 mv driving my 60 Ohm earbud. This is a must build. I'll get a board and some nails and tack it together.
Here is an AC sweep.

This sweep varies the load. Time for a build.

Wednesday, October 17, 2018

How does noise effect the detector output?

Yet Another Book coming. In the mean time consider this. Here we see a noise limiter operation.
Can the peaks cause the amp to saturate or cut off?
What am I seeing here? Is it internal noise, external noise, or oscillations? Could it be a super regenerative wave form? So many questions and no answers.

 The book is linked here:
Noise in Receivers
 It has a threshold action, could it be breaking into oscillations?

Manhatten style with standoffs

Using large value resistors for standoffs

This works well as long as the resistors are ten times larger than the circuit Z. This is my preferred method because I have a reel of 10 meg ohm resistors. The guy that had them saw no use for them and was willing to let the reel go for $5. The method best for you will be determined by what you have on hand.

Based on Arrow's price I found a deal. Anyway it has been a year since I posted the original and I saw the following article and thought maybe it was time for an update.

There is always more than one way and the best practice for you is determined by the materials on hand. Good luck and happy building.

Friday, October 12, 2018

Beware the counterfit or what do I have?

The MPF102 seems to be the go to for a lot of circuits we find on the web. I'm guessing it was a good reasonably priced device but is getting harder to find. I have K193, K2539, J112, J175 and J176 Jfets on hand so I use them for my circuits. I decided to get some MPF102s so I could compare the circuits. I found some "new old stock". I generally don't pay more than 5 cents for a transistor but wanted the device to compare with my others. I bought a small quantity and set them in their own drawer on the bench. I buy a lot of my stuff on closeout or clearance so get some odd devices. For this reason I have a habit of checking the pinout with my component tester. I gathered the parts to build the 118.5 MHz receiver and prepared to build it. Putting the "MPF102" in the tester told me I had a NPN transistor with Hfe of 450. I tested a K193, J175 and J112 and they all were declared to be Jfets. I put the "MPF102" in my multi-meter transistor test socket and it tested to be a NPN transistor. I had heard these things happen but this a first for me. The shame is I don't know what I have. They appear to be good components just not what I ordered. The shipping was free ;).
I have more parts on hand than I'll ever use. The parts I have will work in the circuit but I will have to adjust bias. They say a lesson bought ....

Thursday, October 11, 2018

decoupling the power supply

This circuit is operating at a low frequency (4 MHz) but it will show the interaction in the power supply.
First the circuit without decoupling.
The battery has RF in it as you can see.
The output doesn't look to bad does it? Well let's take a look at the decoupled circuit.
The decoupling did take the RF out of the battery.
It also lowered the output a little. Is that good or bad? If the coupling was positive feedback we could have been headed towards oscillations.
 Here we see the battery current with the RF bypass. Better but can still be improved.
Now we add a bypass around the battery.
The battery current is  pretty well a straight line with the decoupling. This will prevent the stages coupling through the power supply. Some will say the components are redundant and unnecessary but the proof is in circuit performance. Build the basic circuit and test it . Then add the bypasses and see if you can tell a difference.
The MMIC amp datasheets give some details of the filtering. I posted one a couple of post back.

Wednesday, October 10, 2018

Antenna feeding receiver

The I.E.C. dummy antenna will connect to a receiver and feed the signal. You can just radiate a signal and pick it up. The dummy will feed a specific point. You can feed IF into the antenna terminal or the IF stage input. Anyway this has nothing to do with the question asked for info only. Now to the question.
The length of an antenna determines it radiation pattern and Z. The dipole has a figure eight pattern. "Pointing" the antenna in the wrong direction will cause a null in the signal. This was not the problem either just something to remember. Now the real problem.

The impedance along the half-wave antenna is as shown. When you attach a half-wave antenna to a 50 Ohm amp input your amp loads the 2500 Ohm antenna. Making a center tap and feeding it to get a 73 Ohm feed point would better match you amp. It will be directional and have to be pointed in the right direction.
The vertical with the low Z feed point is also omni-directional so it is what you want for the present testing.
Yet Another Book called for but one more point. The propagation speed in the conductor is slower than free space so the antenna will be about 90% of the calculated value at optimum. It will be close enough for testing but in a transmitter we would make it to the calculated value and trim a little at a time till we find the best length or use an antenna tuner. You can read about that when you get the book.

Why be concerned with component layout - OR - Strays everywhere

Consider a wire in free space. Would it be a resistor? Would it be an inductor? Would it be a capacitor? Any or all?
I'm using an online calculator. This one is on the eeweb site. A search will find more. Anyway, a 3 cm wire has 17.9nH.
The same wire spaced 4 cm from another wire has 9 nH. So moving it close to another conductor decreases the inductance? How can that be? Mutual inductance is in play. If you connect an inductance meter to the primary of a transformer and leave the secondary open you will get a reading. Short the secondary and the reading goes down. A current produces an EMF in the primary and the secondary couples to it. The secondary's induced EMF is in opposition to the force producing it. So the primary feeds power to the load. The primary sees an open circuit with no load and sees the load when applied.
If we continue to bring the wires closer the load overcomes the source and we see a small inductance.
What about capacitance? Using another calculator we see the capacitance of a wire 1 cm from our ground plane.
As you can see moving it farther away will reduce the capacitance. Now put these 2 together. One wire will have inductance and capacitance. The question is will it help or hurt our circuit performance? How to use this knowledge?
 I just clipped this from a text book. Stray capacitance everywhere. Some things are beyond our control. The junction capacitance for our transistor have to be considered. Generally capacitance is the limiting factor for our circuit high end. But the previous data should be applied too.
What about that inductance in our wires? Each little red box represents a source for inductance.
What can we do?
Keep the leads short and close to the ground plane. Ground the input  leads close together. Separate the output lead for each stage. Remember the ground plane can be a source of reactance. In the MMIC circuits they etch the inductors into the circuit board.
Study this oscillator circuit. The foil is the tank. It is tuned by adjusting the foil length and width.
So keep the leads short, keep the ground plane as large as practical, isolate the input and output, and use components with a self resonance to match your circuit.