I understand most of the concepts of the differential pair, level shifting, biasing, and output stages, but I can't seem to figure out how to get a full rail-rail output signal
I've run spice models of an amplifier i've designed and I can get a nice output that clips symmetrically with gain setting negative feedback to the differential pair, but it's limited to half of one of the supply rails. I feel this is due in part because my designs are based off the output of the differential pair, which the DC operating point is half of the positive rail voltage to get symmetrical maximum amplified swing from it.
Basically my design goes as such, differential pair input, common emitter level shift to remove the DC shift of the diff pair, class ab output stage
So if the diff pair is limited like this, as well as common emitter and common collector configurations, how do you get full rail-rail swing in the output section?
I've run spice models of an amplifier i've designed and I can get a nice output that clips symmetrically with gain setting negative feedback to the differential pair, but it's limited to half of one of the supply rails. I feel this is due in part because my designs are based off the output of the differential pair, which the DC operating point is half of the positive rail voltage to get symmetrical maximum amplified swing from it.
Basically my design goes as such, differential pair input, common emitter level shift to remove the DC shift of the diff pair, class ab output stage
So if the diff pair is limited like this, as well as common emitter and common collector configurations, how do you get full rail-rail swing in the output section?
There's a lot of information about this in the op-amp literature.
Basically you need a common emitter OPS.
This complicates the OPS bias, search for "Monticelli" as the basis for several state-of-the-art products.
I looked at this quite a lot* but came to the conclusion that the benefits aren't worth the complications for an audio amp.
Different trade-offs for an op-amp of course, mostly driven by requirements for low power/battery operation.
Best wishes
David
*There's a thread here in DIYaudio that discusses some of the issues.
Basically you need a common emitter OPS.
This complicates the OPS bias, search for "Monticelli" as the basis for several state-of-the-art products.
I looked at this quite a lot* but came to the conclusion that the benefits aren't worth the complications for an audio amp.
Different trade-offs for an op-amp of course, mostly driven by requirements for low power/battery operation.
Best wishes
David
*There's a thread here in DIYaudio that discusses some of the issues.
Last edited:
> limited to half of one of the supply rails.... Basically my design goes as such............
That's not right. Post excatly what you did, with part values and observed voltages.
*Totally* rail-to-rail is difficult. And rarely worthwhile. But is is not hard to get within a few Volts of both rails, not halfway.
That's not right. Post excatly what you did, with part values and observed voltages.
*Totally* rail-to-rail is difficult. And rarely worthwhile. But is is not hard to get within a few Volts of both rails, not halfway.
Somthong like a basic audio amp:
Basic Audio Amplifier
And not using right values RC3 ?
Or C3
Practical power amplifier stages and block diagram. Power amplifier circuit and description.
http://www.circuitstoday.com/wp-content/uploads/2009/08/power-amplifier-circuit.png
Basic Audio Amplifier
And not using right values RC3 ?
Or C3
Practical power amplifier stages and block diagram. Power amplifier circuit and description.
http://www.circuitstoday.com/wp-content/uploads/2009/08/power-amplifier-circuit.png
I suspect what you did but need confirmation, please post schematic.
I guess you are missing one gain stage,commonly calles Vas, which is driven by the differential pair and swings *almost* rail to rail.
Sometimes as much as surpassing one thanks to bootstrapping and within 1V of the other due to saturation, go figure.
After that, you basically have unity gain voltage followers, which do almost the same, except for extra couple volts transistor saturation; so in practice most amps reach within 4V of rails, sometimes even a bit closer.
The problem is with the VAS and output. There are a few options. They address the issue of input bias/ drive voltage.
1. Common emitter/source outputs, aka CFP. This allows the output to come withing a couple volts of the supply rail, ie the saturation voltage of the output devices. This output circuit is not as stable, fast and linear as a follower. It requires a Zobel network to stabilize the CFP feedback.
2. Boosted VAS voltage. By using a boosted VAS power supply voltage, the gate/base of the output stage can be driven beyond the rails, and therefore saturate the outputs.
3. Bootstrap VAS. A bootstrap VAS generates a VAS output that exceeds the rails on signal peaks. Normally this is only done on one side of the VAS but it can be done on both. You can bootstrap the rails of an op-amp in order to get past the voltage limits of the op-amp but it places stress on the input CMRR. Bootstrapping is positive feedback which can be an issue.
4. Offset VAS. The VAS can be driven from an offset so that the cap coupled VAS output exceeds the rail. Combining a positive bootstrapping with a negative offset produces both positive and negative VAS output beyond the rails.
For a quasi-comp output, you can have CFP on the negative side and bootstrap on the positive follower.
Note that saturating the output devices begs problems with rail sticking. And speaker level currents result is significant saturation voltages, especially with "lateral" MOSFETs. Note that the higher "on" resistance of P-channel MOSFETs results in ~twice the saturation voltage of their N-channel complements.
1. Common emitter/source outputs, aka CFP. This allows the output to come withing a couple volts of the supply rail, ie the saturation voltage of the output devices. This output circuit is not as stable, fast and linear as a follower. It requires a Zobel network to stabilize the CFP feedback.
2. Boosted VAS voltage. By using a boosted VAS power supply voltage, the gate/base of the output stage can be driven beyond the rails, and therefore saturate the outputs.
3. Bootstrap VAS. A bootstrap VAS generates a VAS output that exceeds the rails on signal peaks. Normally this is only done on one side of the VAS but it can be done on both. You can bootstrap the rails of an op-amp in order to get past the voltage limits of the op-amp but it places stress on the input CMRR. Bootstrapping is positive feedback which can be an issue.
4. Offset VAS. The VAS can be driven from an offset so that the cap coupled VAS output exceeds the rail. Combining a positive bootstrapping with a negative offset produces both positive and negative VAS output beyond the rails.
For a quasi-comp output, you can have CFP on the negative side and bootstrap on the positive follower.
Note that saturating the output devices begs problems with rail sticking. And speaker level currents result is significant saturation voltages, especially with "lateral" MOSFETs. Note that the higher "on" resistance of P-channel MOSFETs results in ~twice the saturation voltage of their N-channel complements.
Last edited:
If your output is as limited as you say, you need to post your simulation file and then we can see whats wrong with it.
A judicious application of bootstrapping allows a true R-to-R operation without resorting to common emitter (or common source) stages.
This example is particularly optimal, since it does not require emitter resistors:
Old-fashioned amplifiers with a new twist (1)
(in the post, it is said that it has been tested in sim only, but later I tested it IRL)
This one is also R-to-R, but it requires source resistors:
Easy-MOS is a simplistic, efficient and evolutive all-MOS amplifier for beginners
This example is particularly optimal, since it does not require emitter resistors:
Old-fashioned amplifiers with a new twist (1)
(in the post, it is said that it has been tested in sim only, but later I tested it IRL)
This one is also R-to-R, but it requires source resistors:
Easy-MOS is a simplistic, efficient and evolutive all-MOS amplifier for beginners
- Status
- Not open for further replies.