1. I attach the schematic here. It uses only single differential (Q1+Q2) fed by CCS below. But it dosn't use CCS anchored VAS, because it uses push-pull VAS (Q4+Q5). This is possible from single differential, because the output of Q2 is level shifted by Q3 so it can perform a push-pull VAS.
My assumption is single differential is better than complementary differential. But Mr.John Curl stated that there is some advantages with push-pull VAS (like the one used here). Mr. NP stated that CCS anchored VAS is good because it anchored the whole system. There is some contradiction about this 2 opinion, I cannot compare these 2 opinion clearly.
From the books, I read that there is another advantage about push-pull VAS, that is the positive and negative slewing is "forced" to be the same number. While in CCS anchored VAS, the positive and negative slew rate will certainly be different (as stated in Mr. Doug Self book).
Maybe this design accomodates all, maybe not. What do you think about this design. What is the merit and drawback of this design?
2. From the paper of Mr.NP and book of Mr.Doug Self, I learn that mosfets devices is good for hot application, like class A with high standing current. But when it comes to design for "cold" class B, is it Bipolars more suitable ? I cannot find any graph about Distortion VS current, volt, etc that stated this. Is it true that bipolars are better used when they are "cold" (while mosfets likes it "hot")?
My assumption is single differential is better than complementary differential. But Mr.John Curl stated that there is some advantages with push-pull VAS (like the one used here). Mr. NP stated that CCS anchored VAS is good because it anchored the whole system. There is some contradiction about this 2 opinion, I cannot compare these 2 opinion clearly.
From the books, I read that there is another advantage about push-pull VAS, that is the positive and negative slewing is "forced" to be the same number. While in CCS anchored VAS, the positive and negative slew rate will certainly be different (as stated in Mr. Doug Self book).
Maybe this design accomodates all, maybe not. What do you think about this design. What is the merit and drawback of this design?
2. From the paper of Mr.NP and book of Mr.Doug Self, I learn that mosfets devices is good for hot application, like class A with high standing current. But when it comes to design for "cold" class B, is it Bipolars more suitable ? I cannot find any graph about Distortion VS current, volt, etc that stated this. Is it true that bipolars are better used when they are "cold" (while mosfets likes it "hot")?
Attachments
Hi,
The only advantage compared to a 'standard 3 stage amp' is some increased openloop gain, but a bit worse phase margin.. I did a simple version of your idea in my sims and it looks good.. How does your idea effect PSRR or CMRR ? Offset seems very low and good controlled with this topology..
Regards,
Thijs
The only advantage compared to a 'standard 3 stage amp' is some increased openloop gain, but a bit worse phase margin.. I did a simple version of your idea in my sims and it looks good.. How does your idea effect PSRR or CMRR ? Offset seems very low and good controlled with this topology..
Regards,
Thijs
Attachments
Memories .... Like the windmills of my mind ...
Looks very much topologically like the old Apt 1 power amp from Tom Holman and Mark Dinsmore. Worked well and sounded good for it's time.
Class AB, bipolar all the way through. Maybe someone can post the entire schematic for review. Jam? I KNOW you have it 🙂
As I recall, there was a good writeup on the design by Tom Holman in "Electronic Design" (maybe 1980-81?). It's worth looking for. There were two audio power amp articles in that issue. One article by Jack Konrath who was at Hitachi espousing the benefits of the soon to be popular lateral power MOSFETs from Hitachi. The article was exciting to me at that time long ago because it was the first place I ever saw the Hafler DH200 schematic (it became my first clone project 🙂 ) as well as discussion on the schematics contained in the Hitachi MOSFET databooks from that time period.
Getting back on topic, the other article was by Tom Holman, who was discussing using the new (at that time) rugged SOA bipolar output transistors from Motorola in his APT 1 power amp. Contained a schematic and some discussion of the features of the topology and their design you may find interesting...
good luck!
mlloyd1
Looks very much topologically like the old Apt 1 power amp from Tom Holman and Mark Dinsmore. Worked well and sounded good for it's time.
Class AB, bipolar all the way through. Maybe someone can post the entire schematic for review. Jam? I KNOW you have it 🙂
As I recall, there was a good writeup on the design by Tom Holman in "Electronic Design" (maybe 1980-81?). It's worth looking for. There were two audio power amp articles in that issue. One article by Jack Konrath who was at Hitachi espousing the benefits of the soon to be popular lateral power MOSFETs from Hitachi. The article was exciting to me at that time long ago because it was the first place I ever saw the Hafler DH200 schematic (it became my first clone project 🙂 ) as well as discussion on the schematics contained in the Hitachi MOSFET databooks from that time period.
Getting back on topic, the other article was by Tom Holman, who was discussing using the new (at that time) rugged SOA bipolar output transistors from Motorola in his APT 1 power amp. Contained a schematic and some discussion of the features of the topology and their design you may find interesting...
good luck!
mlloyd1
lumanauw said:
Thanks, Tschrama for your analysis and newidea.bmp. I just re-read the book of Doug Self and find one important thing about to have exactly splitting the current between the two differential transistor. The exact splitting between Q1 and Q2 is very important and sensitive (specially to distortion), so Doug himself always use current mirror to split equally (so does Randy Slone).
My schematic will not split equal amount between Q1 and Q2, because Q2 have to feed both its rail resistor, and the level shift mechanism.
But in your newidea.bmp, level shifting is done by current mirror, and in my opinion, your newidea.bmp will force the current tobe equally splitted between Q1 and Q2.
But I don't understand about "worse phase margin". First, I dont have any equipment to measure anything about this.
Second, I don't really know what is the criterion of "good phase margin" and "bad phase margin" and also on how to measure them. Could you tell me about this?
My schematic will not split equal amount between Q1 and Q2, because Q2 have to feed both its rail resistor, and the level shift mechanism.
But in your newidea.bmp, level shifting is done by current mirror, and in my opinion, your newidea.bmp will force the current tobe equally splitted between Q1 and Q2.
But I don't understand about "worse phase margin". First, I dont have any equipment to measure anything about this.
Second, I don't really know what is the criterion of "good phase margin" and "bad phase margin" and also on how to measure them. Could you tell me about this?
What is the general guidelines for selecting/calculating the Feedback resistors (voltage divider) between output and differential?
The R to ground varies very much, from 10 0hm to 3kohm, but what is the exact guidance for calculating them?
The R to ground varies very much, from 10 0hm to 3kohm, but what is the exact guidance for calculating them?
I have often wondered the signifiance of resistors in relation to using higher or lower values and I think the main issue would be that higher values are more succeptible to pick up noise from surrounding emi. I've never tested this but it makes sense that a 100ua trace would be more influenced by surrounding emi then a 1ma trace.
So I'll use 330/6.8k instead of 1.2k/22k to get more or less a 20 gain.
I'm going to test the emi issue with exagerated values so I can see how bad (if it's bad) it is.
Happy new year 🙂
So I'll use 330/6.8k instead of 1.2k/22k to get more or less a 20 gain.
I'm going to test the emi issue with exagerated values so I can see how bad (if it's bad) it is.
Happy new year 🙂
Thanks Easyamp, for will do such experiment.
I have an opinion, that it somehow has something to do with the current have to be fed to the basis of the differential (if we use bipolars). The lower the value, the more current will be provided to fed the basis of differential transistors.
If we use Fets for differential, maybe we can use higher order of resistance for voltage divider, because we dont have to fed any current to the gate of fets.
Is this right? Or maybe your opinion of "surrounding Interference" is the correct answer?
I have an opinion, that it somehow has something to do with the current have to be fed to the basis of the differential (if we use bipolars). The lower the value, the more current will be provided to fed the basis of differential transistors.
If we use Fets for differential, maybe we can use higher order of resistance for voltage divider, because we dont have to fed any current to the gate of fets.
Is this right? Or maybe your opinion of "surrounding Interference" is the correct answer?
- Status
- Not open for further replies.