Hello Everyone,

I'm studying frequency response of amplifiers. For example, consider the common emiter amplifier as showed in the picture:

I've simulated this on MULTISIM and the result was similar to the expected theorectal value.

But in the lab I tested the circuit using a protoboard but the result was very diferent. I was expecting high cutoff frequency of about 1.7 MHZ but I got 250KHZ in pratice. I can't explain why... Suposing the circuit is correct, can anyone help me how to see in practice the theorectal value with little error?

Thanks,

 

Did you build it on a protoboard? The medium you use to build it has a butt load of assumptions associated with it.

 

Because you didn't model the parasitics of the interconnecting wiring and the protoboard. Even lengths of straight wire have inductance: http://www.consultrsr.com/resources/eis/induct5.htm
If there are curves in the wire, the inductance goes up a good bit.
The parasitics of protoboards varies widely.
Your components are an unknown. Typical electrolytic capacitors have a considerable amount of parasitics; low-ESR caps are much better.

You might try building a version using the "dead bug" technique. That's where you use a board with copper on one side as a ground plane, glue the parts down with their legs in the air, and do point-to-point wiring in the shortest paths possible. You eliminate most of the external parasitics that way.

 

What are the resistor and capacitor values for your circuit? That mostly determines how the parasitics affect the frequency response.

 

I found one of the circuits I tested. On theory, and Multisim the high cutoff frequency was about 1MHZ.

But in practice, using protoboard I obtained around 200khz. This is a common emiter with a current source using a AMPOP.

The voltage source was 10mV

 

Ok, thanks for replying. I've considered the parasitical capacitances of the transistor model, but I didn't considered the capacitances of the medium and the protoboard.

So the protoboard is not suitable for this kind of test?

I'll try the alternative sugested. Any more tips?

 

Collector-base stray capacitance can be a particularly bad offender. Because of the Miller effect, this usually has worse effects than capacitance from either base or collector to ground. http://en.wikipedia.org/wiki/Miller_effect

One way of getting somewhat better results from a proto-board layout is to place grounded strips between connections likely to suffer from such feedback effects. Do not expect miracles from doing this, but a considerable improvement may be possible.

This web page describes how the coupling between tracks can be reduced this way. http://www.seas.upenn.edu/~ese319/Lab_Handouts/ESE319_Lab_1_FA11.pdf

 

I suspect that if you add a capacitance of few pF of between the base and collector connections of your model to simulate the proto-board stray coupling, you will see a difference.

 

What about the load capacitance? I've read in SEDRA - Microelectronics Circuits, that there's a load capacitance paralel to the load. But the book doesn't show how to calculate this value. I couldn't find good texts on internet about this capacitance. Does anyone know how this load capacitance afects the circuit and how to include this in my model? And when I can disconsider the efect of this capacitance? Supose my load IS NOT CAPACITIVE, just resistive, this LOAD CAPACITANCE still exist? Is it a parameter of the transistor or just the load?

 

Unless the load capacitance is really quite big, it's usually not as dominant as feedback capacitance. This does of course critically depend on the impedance levels and the amount of gain.

Why don't you try adding some capacitances to your model, one at a time, to see what happens. Try a value of say 2pF to begin with.

 

Thanks adjuster,

I supose you are talking about the load capacitance. Ok, I have already added capacitor to the model. But I felt like trial and error. I was expecting something more acurated if possible. For example, how do I know the load capacitance is big or small? How do I know if the circuit is not working quite well due to load capacitance or some other sort of problem?

Ok, but I apreciate the help, and as soon as possible I will try everyone...

 

Try adding a collector-base capacitor of say 2pF first. That circuit of yours has a mid-band voltage gain of what, 100 times? With a 1kohm feed resistance a small feedback capacitance may have a big effect.

A few pF across the collector load probably won't do so much. Try it and see.

 

Thanks, I will try it.

 

You could of course build a circuit to get a least an approximate measurement of your proto-board strip to strip capacitances. (You might want to put a few strips in parallel to stoke the total capacitance up to a more easily measured value.)

Alternatively, the board manufacturer may supply data on the likely capacitance values. Values of a few pF to a few tens of pF are quite possible between adjacent strips

 

I think the capacitance between tracks on a breadboard is 15pF, not just 2pF.
Then you can kiss high audio frequencies goodbye.

A breadboard works well at DC frequencies (sarcasm) if the contacts are not too intermittent.

 

This sort of thing can also rear its head with stripboard. As a youngster I used to build a lot of things on Veroboard. After a bit of practice I got pretty fair results, but I found it really important to avoid long adjacent tracks causing couplings where they are not wanted.

It's rather much of a simplification, but remember the slogan "Miller the killer". Apart from its intentional use in use in filter circuits, oscillators etc. feedback capacitance around a high gain amplifier is usually bad news.

 

Adjuster,

So, the bigger the gain the worst the results i'll obtain?

 

This sort of thing can also rear its head with stripboard. As a youngster I used to build a lot of things on Veroboard. After a bit of practice I got pretty fair results, but I found it really important to avoid long adjacent tracks causing couplings where they are not wanted.

I made many high frequency digital and very high frequency analog circuits on stripboard.
But of course I planned the layout then cut every track to minimum so that the circuit was compact and had little capacitance between tracks.

 

Adjuster,

So, the bigger the gain the worst the results i'll obtain?

You might expect that to be so if the problem is associated with stray feedback. You would need to do some experiments to be sure.

For instance, if the Miller effect is significant, deliberately altering the gain (e.g. by changing the collector current) would have a noticeable effect on the bandwidth.

 

I built an audio amplifier on breadboard before. It was capable of 30kHz -3dB but quite possibly this was due to the LM324 I used, which does not have a superb frequency response.