I want to be able to set the 'offset' of an a buffer. That is, given a signal of 5V and an offset of -4V, the output would be 1V (the offset could be positive as well, in which case the output would increase in voltage.) I could do this with op-amps, but I'm looking for high speed operation, and high-speed op-amps are expensive. I'm only doing this to teach myself about practical discrete amplifiers (as it's something don't know much about.) I've searched the AAC book but found little info.

 

What about the negative side of the pulse, is it -4V?

If so, you'll need a negative power supply, if not, then it is clipping.

 

What about the negative side of the pulse, is it -4V?

If so, you'll need a negative power supply, if not, then it is clipping.

Negative power supply is available. Assume maybe +/- 15V rails.

The offset could be from -5V to +5V.

The input is already buffered and with relatively high impedence (FET buffer.) But the signal has a 3.2V dc component that needs to be removed, along with any other dc component that may be on the input.

 

OK, it is a given you will have a 0.6BE drop, this can not be corrected. I'm going to assume you aren't after an summing circuit, though that may be what your after.

Just pick a bias point that creates the voltage you want.

 

OK, it is a given you will have a 0.6BE drop, this can not be corrected. I'm going to assume you aren't after an summing circuit, though that may be what your after.

BRB, drawing.

A summing circuit could well be what I need. I don't mind having to apply a negative signal to increase the offset or positive signal to decrease it, as long as I can control it with say a microcontroller or a slow op-amp (the slow op-amp inverting a DAC output perhaps.)

 

OK, it is a given you will have a 0.6BE drop, this can not be corrected. I'm going to assume you aren't after an summing circuit, though that may be what your after.

Just pick a bias point that creates the voltage you want.

Thanks for the circuit Bill, but wouldn't that remove all of the dc component? I want to be able to offset it under the control from some external voltage; I'm unsure on how to do this.

Say I have a 1Vp-p sine wave with an offset of 3V, (going from 2.5V to 3.5V.) If I offset it by -3V, then I get effectively a 1Vp-p sine wave with no dc, but if I offset it by -2V I still have 1V dc. It would be easy with op-amps but challenging to find one capable of 100MHz+.

 

Replace C1 with another resistor and you have what passes for a summing circuit. It is drastically inferior to a op amp version, as there is no isolation between the two inputs. With an op amp both inputs go to a virtual ground, but no such animal exists here. I am also merging R2 with the second input that would be summed with the input. There may be better out there, but I'm not aware of them. Want a drawing? I'm warning you in advance though, it sucks.

***************

BTW, this is covered here.

http://www.allaboutcircuits.com/vol_3/chpt_8/8.html

 

I think I know what you mean. Something like what's attached. I will have to think about adding a BJT to it. The signal is being attenuated unfortunately by this summer circuit, will have to look into that.

$ 1 5.0E-6 15.472767971186109 50 5.0 50
j 272 192 304 192 0 -4.0
R 304 176 304 144 0 0 40.0 15.0 0.0 0.0 0.5
r 304 208 304 272 0 10000.0
g 304 272 304 288 0
r 304 208 384 208 0 1000000.0
r 384 208 384 272 0 1000000.0
R 384 272 384 304 0 0 40.0 -3.25 0.0 0.0 0.5
O 384 208 432 208 1
R 272 192 240 192 0 1 40.0 1.0 0.0 0.0 0.5
o 8 64 0 291 2.5 9.765625E-5 0 -1
o 7 64 0 290 1.25 9.765625E-5 1 -1

Using Windows on the college network here, use falstad.com/circuit to simulate the circuit.

 

Figure 1 is the basic model. It assumes you will always have the inputs connected to something, and gives the best results. However, the two signals are coupled with 200Ω between them.

Figure 2 is when you may not have guaranteed inputs, and gives the circuit a guaranteed input when the other two are open. This resistor does interact, and adversely affects the performance. I figure 150KΩ or more.

Figure 3 is for when you don't have a ground, it makes a pseudo ground. Figure 300KΩ for both bias resistors, it has all the same problems as Figure 2.

If the beta of the transistor is 150 then you have an input impedance on the transistor of around 150KΩ, and this does interact with the adder, which is the reason you have such small resistors for the adder inputs.

 

Thanks Bill!

I'm still thinking about how amplification would happen, and I'll post another thread when I'm ready.

Turns out, in AS Electronics (studying now) only MOSFETs are discussed, and AFAIK BJTs are very useful for analog circuits and MOSFETs are really only used as switches, except in rare cases. But maybe we will learn some analog stuff.

 

Tom, MOSFET's are very popular in analog IC design which is why they are taught so much.
Discrete signal MOSFET's aren't common which is where that impression comes from.

 

Tom, MOSFET's are very popular in analog IC design which is why they are taught so much.
Discrete signal MOSFET's aren't common which is where that impression comes from.

Yeah, but in my course description it says "MOSFET as a switch."

But maybe more advanced topics will be discussed in A2 Electronics.

This is only college, so I don't expect it to be too advanced.

We were testing 4069 inverters to see at which point the output transitions. Turns out, with the inverters available (CD4069 IIRC) the transition point is 2.35V, with a 5V supply.

 

JFETs and BJTs are generally used for analog, MOSFETs almost never. Part of the reason is they are not very linear. Linearity is more than a convenience if you are going to use it as such. Their smaller cousins, CMOS, are also used almost exclusively for digital. Wishful thinking doesn't change basic physics.

If you look up the op amps using FETs you will see JFETs.

I have seen some designs using digital CMOS inverters for analog designs, but they have never really taken off because they simply don't work well in that mode.

 

JFETs and BJTs are generally used for analog, MOSFETs almost never. Part of the reason is they are not very linear. Linearity is more than a convenience if you are going to use it as such. Their smaller cousins, CMOS, are also used almost exclusively for digital. Wishful thinking doesn't change basic physics.

If you look up the op amps using FETs you will see JFETs.

I have seen some designs using digital CMOS inverters for analog designs, but they have never really taken off because they simply don't work well in that mode.

Microchip's op-amps use CMOS technology pretty much entirely I believe. That is, MOSFETs all around, not a BJT or JFET in sight.

 

Care to show an example? I looked on their parts list, they don't show schematics of their op amps.


**************

OK, found the application notes AN722.

http://ww1.microchip.com/downloads/en/AppNotes/00722a.pdf

Interesting reading, using CMOS comes with a price...

If the amplifier is designed with NMOS input transistors
as shown in Figure 7b, the input range is restricted near
the negative power supply voltage. In this case, the
input terminals can be taken to a few tenths of a volt
above the positive supply rail, but only to 1.2V above
the negative supply rail.

If an amplifier input stage uses PMOS and NMOS transistors,
it is configured as a composite stage, as shown
in Figure 7c. With this topology, the amplifier effectively
combines the advantages of the PMOS and NMOS
transistors for true rail-to-rail input operation. When the
input terminals of the amplifier are driven towards the
negative rail, the PMOS transistors are turned completely
on and the NMOS transistors are completely off.
Conversely, when the input terminals are driven to the
positive rail, the NMOS transistors are in use while the
PMOS transistors are off.

Although, this style of input stage has rail-to-rail input
operation there are trade-offs. This design topology will
have wide variations in offset voltage. In the region near
ground, the offset error of the PMOS portion of the
input stage is dominant. In the region near the positive
power supply, the input stage offset error is dominated
by the NMOS transistor pair. With this topology, the offset
voltage error can change dramatically in magnitude
and sign as the common mode voltage of the amplifier
inputs extend over their entire range.

The basic topologies shown in Figure 7 can be used
with FET input or Bipolar input amplifiers. In the case of
the FET input amplifier, the offset errors between the
PFET and NFET are consistent with the CMOS errors
with the circuit shown in Figure 7c. With Bipolar input
stages, input offset voltage variations are still a problem,
but input bias current is an additional error that is
introduced. The nano ampere base current of an NPN
transistor comes out of the device, while the nano
ampere base current of a PNP transistor goes into the
device.

 

Care to show an example? I looked on their parts list, they don't show schematics of their op amps.

http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en528727

No schematic, but "The MCP6031 uses Microchip's advanced CMOS technology, which provides low bias current, high-speed operation, high open-loop gain and rail-to-rail input and output swing."

This is a low end chip having only a 10kHz GBWP. But they also make ones up to about 60MHz GBWP.

 

CMOS is made of MOSFETs by definition. A MOSFET does not mean power MOSFET.


CMOS offers lower power dissipation with higher density due to a smaller size with a simpler manufacturing process. Their drawback is lower gain (and hence less possibility for linearity from lots of feedback) and more capacitance, making for smaller bandwidths.
Switched filters, ADCs, DACs, sample & hold, mixed-signal FPGAs... they're very very often CMOS.

Pick up any recent textbook on analog circuit design and they will without a doubt focus on CMOS design.

Some National CMOS products I found in just a few minutes:
http://www.national.com/pf/DA/DAC121S101.html#Overview
http://www.national.com/pf/LM/LMP8358.html#Overview
http://www.national.com/pf/DC/ADC16V130.html#Overview
http://www.national.com/pf/LM/LMP7709.html#Overview
http://www.national.com/mpf/LM/LMC6484.html#Overview

 

Funny thing though, very few op amps have found use at this and similar sites. It may change, but not yet.

 

So, what, are you implying that semiconductor manufacturers and electronics designers don't know what they're doing and design with CMOS analog devices just because they want to?

One of the key advantages of CMOS is low power. Hobbyists usually don't care about microwatts.
CMOS is a modern design process, hence it's used in modern devices. Modern devices are generally SMD. Most hobbyists don't like SMD.

People in general like to work off of what they're used to. The majority of projects online and in magazines etc. are done by guys who started before CMOS was popular, why would they change devices when they don't need to?

There is a million reasons why they're not popularly used on publicly visible projects.

 

So, what, are you implying that semiconductor manufacturers and electronics designers don't know what they're doing and design with CMOS analog devices just because they want to?

One of the key advantages of CMOS is low power. Hobbyists usually don't care about microwatts.
CMOS is a modern design process, hence it's used in modern devices. Modern devices are generally SMD. Most hobbyists don't like SMD.

People in general like to work off of what they're used to. The majority of projects online and in magazines etc. are done by guys who started before CMOS was popular, why would they change devices when they don't need to?

There is a million reasons why they're not popularly used on publicly visible projects.

Agreed. Just like that 10KHz op-amp. Pretty much useless for audio. But maybe it would be used in a multimeter, or a temperature probe, due to its low power consumption and low input current.