Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

300MHZ 0 bias AMPLIFIER

Status
Not open for further replies.
A little more detailed explanation of reduced Miller effect in cascodes: The lower transistor has the advantage of its collector driving the very low impedance of the emitter of the top transistor, thus the voltage gain is very small (much less than one), to the capacitive multiplying effects of the collector-to-base capacitance (aka Miller effect) is also very small. The top transistor has the benefit of its base being grounded, thus shunting most of the signal fed back through the collector-to-base junction to ground, similarly lowering the gain that causes the Miller effect.

At hundreds of MHz, parasitic reactances can make the performance of both transistors much less than ideal. Beware the SPICE simulation that dose not include strays and parasitics.
 
stray capacitances and lead inductances begin to get very big above a few Mhz (even above 20khz really) and should be included in any sims of HF circuits. granted the models tend to get rather complex, but without including strays in the circuit, the sim will show an amplifier, but when you get it built, will turn out to be an oscillator. even with audio amps, people tend to forget things like speaker wire capacitances and reactive loads (like speaker crossovers and inductive voice coils) and wonder why their "unconditionally stable" power amp design just became an AM transmitter when they built it. SPICE does include Miller capacitance, but few models include lead inductance or pin capacitances. also, pay attention to peaks in the frequency response sweeps, especially when accompanied by large phase shifts past 180 degrees (or multiples of 180), since these will become oscillation points if the amp gain is above unity.
 
Please tell me there is some point when the bad news stops! I always considered myself as fairly capable however interfacing with you folks gives me an understanding of the expression ”In the country of the blind, the one-eyed man is king”. All I can say is this is a steep hill to climb!
 
it's really not that bad, the model will give you an idea how well it will work, and give you an idea whether the standing currents are within the capabilities of the transistors. you just need to be cautious about the frequency response figures you get, and be aware that even the best models may have some parameters missing or inaccurate, usually because many models are made with more emphasis on some parameters and less emphasis on others. RF transistors for instance will have most of the high frequency parameters very well modeled, but may be less accurate on DC parameters. another caveat is that when you look at a data sheet for a transistor, you will likely see that the beta (current gain) is given as either a minimum value or a range (i.e. "20-80"). variations in beta in a certain part number will usually fall within a wide range, with about 70% of them being somewhere near the value in the "typical" column in a data sheet, so you could see something like this in a data sheet:
___________minimum_________typical__________maximum
beta (Hfe)_____20_____________50_______________80

so the beta would be at least 20, usually around 50, and sometimes as high as 80. if you buy 10 of these transistors selected randomly, you might have 1 or 2 as low as 20, about 6 or 7 of them between 45 and 55, and 1 or 2 of them as high as 80. if they all came from the same lot code, they will usually be within 10% of each other throughout the whole lot code, so you will likely get all 10 of them within the 45-55 range, or within the 65-75 range, etc....
models, on the other hand will most likely be preset at the "typical" value, so all instances of any given transistor in a circuit will all have the same beta. if you wanted to simulate the tolerances in transistors, you would want modified models that either randomize within the beta range, or just make a couple of copies of the "typical" model and create a min and max version and select which one of the 3 models to use at random. resistors, caps. inductors also have tolerances that can be "fudged" at random. it's best to get a circuit working properly with the "ideal" or "typical" values first, then begin "fudging" the tolerances to see if the normal tolerance variations will make the circuit misbehave (or stop working completely).

when i'm not sure what value of a resistor will work best, i run sims and make my changes in a "1-2-5" series, beginning with the most significant digit and fine tuning with progressively less significant digits. for instance i want a resistance that's at least 1k, but no more than 100k, but i'm not really sure where to begin (and probably in too much of a hurry to sit and figure it out on paper, or i'm fine tuning something like distortion levels which are much quicker figured out by the sim). so i start in the approximate middle with 50k and run the sim. then i run the sim with it set at 100k and 20k, and see which one is closer to the desired result. let's say 20k is closer, but not close enough, so i go to 10k. 10k is good, but too far, so i know it's between 10k and 20k, so i go to 15k. closer, but still needs to be higher, but still less than 20k. so i take a "2" step and go to 17k, really good, but can it get better? take an additional "1" step to 18k.... too far, go back to 17k. if i want to narrow it further i go to the next smaller "5" step and go to 17.5k. so i go in big steps of 1-2-5-10, and narrow it down in smaller "1-2-5" steps. some may prefer "1-3-5" or "1-3-7". i actually "borrowed" the "1-2-5" sequence from the range switch sequence of tektronix scopes.

in any case, the sim is really good for "building" the circuit and making sure it will work without "letting out the magic smoke" before i actually put it together. you can see what the device currents will be and compare the currents with the spec sheet of the transistor.
 
Last edited:
in this video circuit, both the transistor's beta (Hfe) and cutoff frequency (Ft) are important parameters, as well as the max collector current (Ic) and collector voltage (Vce, or use the max Vcb if it's lower and make sure the transistor's max voltage is at least 10-20% higher than the supply voltage, just to be safe) you can probably find many suitable transistor part numbers by looking at the schematics for high end SVGA monitors, you might even want to duplicate some of the circuits and see if they will work. when i was repairing SVGA monitors, most of the video amps were good to about 100Mhz or so, some of them may have been 300Mhz amps but there were very few very expensive video cards at the time that could drive those monitors, and since i wrote my own video test pattern software, was limited to the max pixel rates of the video cards i had access to. i worked on some monitors that had video amps driven by ECL logic, but had to rely on the client's own hardware to check their operation. you might see if you can find schematics for the IBM 85xx series monitors, especially the higher end models that used very high horizontal frequencies at 40-50khz and 75hz vertical. (8518? i think?) compaq 461 (not 460) monitors or Mac monitors, or NEC Multisync monitors (later ones after CGA and EGA were dropped). Sony made some very nice monitors that were for Sun Microsystems machines that had sync rates around 75-100khz, and i don't remember the video amps on the CRT board, whether they were discrete or hybrid, but their max pixel rate had to be in the neighborhood of 200-300Mhz.
 
it's really not that bad, .

From your point of view I know that is true. I can relate that sentiment to my field of expertise, software development and integration. It’s like anything, you can read about it and study it but you can’t really implement it properly without years of practical experience.

As it is sometimes said “The devil lies in the details”. With enough determination anyone can cobble a few components together and make something work, but only select few can make it work correctly. I understand that, and that is why I entered this forum; to find out the details, to understand the practical effects of the not so obvious Miller effects, cross conduction and the like. I certainly have gained a tremendous amount of knowledge from the articulate writings provided by both yourself and Dick Cappels and thank you both for that!

Moving forward I plan to study a few schematics, as recommended, getting a feel for different design concepts and component usage then creating a working model using the SPICE (my new favorite tool). A viable option for this application would be to duplicate the hybrid with a slight modification; adding a current sense resistor in the sink transistor collector circuit. The ability of current sensing is what forces the redesign the existing hybrid amplifier. When it is working in SPICE I’ll post for review before starting the board layout phase.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top