Hi Neel,
Sorry about that... the way I worded it is not quite right.
That is indeed class B, but it's obviously not "optimally biased" while remaining in class B. In other words, performance could be both better or worse than what's shown there while still adhering to the definition of class B. Worse if you reduce the bias further, and better if you increase it right to the edge of 180 degree conduction. Beyond that, and you're into class AB. With this particular amplifier performance will continue to improve well into class AB with the optimal point being about 350-400mA of bias current. This is way beyond the transition between class B and class AB, and you'd actually be operating in class A up to a little over a watt when the output current exceeds the bias current and stops both devices from conducting at 360 degrees.
Regards,
Owen
Sorry about that... the way I worded it is not quite right.
That is indeed class B, but it's obviously not "optimally biased" while remaining in class B. In other words, performance could be both better or worse than what's shown there while still adhering to the definition of class B. Worse if you reduce the bias further, and better if you increase it right to the edge of 180 degree conduction. Beyond that, and you're into class AB. With this particular amplifier performance will continue to improve well into class AB with the optimal point being about 350-400mA of bias current. This is way beyond the transition between class B and class AB, and you'd actually be operating in class A up to a little over a watt when the output current exceeds the bias current and stops both devices from conducting at 360 degrees.
Regards,
Owen
The phrase "optimally biased ClassAB" only applies to a complementary emitter follower BJT output stage.
It can loosely be applied to quasi BJT and common emitter styles of BJT output stages, but the results of "optimum" are not nearly as good. Look at D.Self's graphs.
mosFETs fall into a completely different biasing territory.
They don't have an optimal bias region. The crossover distortion gets less the more the bias is increased. So much so that the crossover distortion does not disappear until they are biased in ClassA.
I'll quote Borbely again.
The total output bias of a mosFET stage should be not be less than 500mA.
That's an ampere of ClassA output current. Equivalent to 4W of ClassA into an 8r0 resistor.
It can loosely be applied to quasi BJT and common emitter styles of BJT output stages, but the results of "optimum" are not nearly as good. Look at D.Self's graphs.
mosFETs fall into a completely different biasing territory.
They don't have an optimal bias region. The crossover distortion gets less the more the bias is increased. So much so that the crossover distortion does not disappear until they are biased in ClassA.
I'll quote Borbely again.
The total output bias of a mosFET stage should be not be less than 500mA.
That's an ampere of ClassA output current. Equivalent to 4W of ClassA into an 8r0 resistor.
Hi OPC..I've found this project really interesting.. could I ask for 2 pcb ?? Do You still have something ? 🙂 Pls contact me.. thanks..😀
Owen I dont want to go too much offtopic here but I was looking at some design points on this national chip and looking at the harmonic spectrum results in the datasheets and then a thought came up.
Have you thought of using TMC compensation on this design ??
Two pole compensation works, so TMC should also. I think it would further improve the design and TMC greatly improves crossover distortion which results in mainly odd order harmonics. When compairing the two schemes in a typical Self circuit youll find that TMC has several advantages over 2 pole.
Have you thought of using TMC compensation on this design ??
Two pole compensation works, so TMC should also. I think it would further improve the design and TMC greatly improves crossover distortion which results in mainly odd order harmonics. When compairing the two schemes in a typical Self circuit youll find that TMC has several advantages over 2 pole.
indeed, it would be interesting to try that on my tweeters as they shouldnt need more than that in a digital XO. also interested in the PSU experiments. Owen posted in the headamp thread that he expects to start accepting payments for the power amp within the next 2 weeks and shipping directly afterwards.
There will be some left, but I won't be dealing with orders for those until after everything has shipped to those who have already ordered. It's just too difficult to keep track of everything if orders are still being added as I try to ship stuff out.
Remaining orders will probably only be for boards and transistor as I've already ordered all the other parts for the kits, and there won't be enough left over. I had to order extra transistors, so I will have many of those around.
I'll be sending out payment requests first thing in January, and things should go very quickly after that.
Cheers,
Owen
Remaining orders will probably only be for boards and transistor as I've already ordered all the other parts for the kits, and there won't be enough left over. I had to order extra transistors, so I will have many of those around.
I'll be sending out payment requests first thing in January, and things should go very quickly after that.
Cheers,
Owen
Power supply question:
What are most people doing for their PS? There was discussion of several SMPS options and tests to be done. Is the thinking converging toward a couple of recommendations?
Happy New Year to All
Neel
What are most people doing for their PS? There was discussion of several SMPS options and tests to be done. Is the thinking converging toward a couple of recommendations?
Happy New Year to All
Neel
we are waiting, many of us i think; for the results of opc's comparative testing of these options, modern switching vs linear. switching sure has its advantages, but many of us havent used the modern solutions and therefor need to be convinced that there arent still issues =)
Like many others, I am eagerly looking forward to getting started building the Wire headphone and power amps. I am definitely a newbie at DIY and electronics with just one simple Bottlehead project under my belt, so I'm trying to absorb as much on this project as I am able!
Newbie question:
I can't quite come to grips with the details of the 2 different supplies, a regulated one for the LME 10V higher than an unregulated one for the MOSFETs. I'm guessing that running the LME at a higher voltage is a compensation for the LME's failure to ouput rail-to-rail coupled with a desire to run the MOSFETs rail-to-rail. It looks to me like trying to precisely match the output of the regulated LME with the unregulated supply of the MOSFETs will work fine for class A or light loads or if the MOSFET PS were itself regulated, but could easily result in clipping under heavy loads in class AB due to PS sag. What am I missing?
Regardless of what I'm missing above, I would prefer to use an off-the-shelf transformer for both cost and bother reasons. I definitely want the advantage of LME regulation to reduce noise, particularly PSU ripple, even though it might cost in efficiency or risk to the speaker drivers! The next question I have is why not just power the regulators from the same high current supply used for the MOSFETs? Regulating the LME rail voltage down 10V from the MOSFET supply should elimate sag as a reason for clipping and all I would have to do is select a transformer with the same wattage but a higher output voltage (20V higher?) to get the LME's output back up. It won't result in going rail-to-rail with the MOSFETs, but should be able to produce nearly the same ouput power. Do you think the approach described is workable? About how much in undesired power losses would I end up with? - don't go to any trouble over this, just a ballpark guess (double, triple?) would be interesting.
Thanks in advance for any feedback!
Newbie question:
I can't quite come to grips with the details of the 2 different supplies, a regulated one for the LME 10V higher than an unregulated one for the MOSFETs. I'm guessing that running the LME at a higher voltage is a compensation for the LME's failure to ouput rail-to-rail coupled with a desire to run the MOSFETs rail-to-rail. It looks to me like trying to precisely match the output of the regulated LME with the unregulated supply of the MOSFETs will work fine for class A or light loads or if the MOSFET PS were itself regulated, but could easily result in clipping under heavy loads in class AB due to PS sag. What am I missing?
Regardless of what I'm missing above, I would prefer to use an off-the-shelf transformer for both cost and bother reasons. I definitely want the advantage of LME regulation to reduce noise, particularly PSU ripple, even though it might cost in efficiency or risk to the speaker drivers! The next question I have is why not just power the regulators from the same high current supply used for the MOSFETs? Regulating the LME rail voltage down 10V from the MOSFET supply should elimate sag as a reason for clipping and all I would have to do is select a transformer with the same wattage but a higher output voltage (20V higher?) to get the LME's output back up. It won't result in going rail-to-rail with the MOSFETs, but should be able to produce nearly the same ouput power. Do you think the approach described is workable? About how much in undesired power losses would I end up with? - don't go to any trouble over this, just a ballpark guess (double, triple?) would be interesting.
Thanks in advance for any feedback!
Newbie questions - part 2
Never mind my 1st newbie question - it finally hit me that only the regulator is running on the supply at higher voltage, the regulator is probably set so the LME is running 10V or so lower, in other words at a regulated voltage approximately equal to the unregulated supply. This would allow the LMEs rail-to-rail shortfall to act as compensation for the unregulated MOSFET PS sag - at least, this is my best current guess at what's involved here and I hope to be corrected if it's way off base. 😎
Still hoping for a response to the 2nd question about using an off-the-shelf transformer. I'm sure several of you immediately considered and dismissed this as another possible PS solution and I'd really appreciate a better idea of what I would be giving up by doing it this way.
Never mind my 1st newbie question - it finally hit me that only the regulator is running on the supply at higher voltage, the regulator is probably set so the LME is running 10V or so lower, in other words at a regulated voltage approximately equal to the unregulated supply. This would allow the LMEs rail-to-rail shortfall to act as compensation for the unregulated MOSFET PS sag - at least, this is my best current guess at what's involved here and I hope to be corrected if it's way off base. 😎
Still hoping for a response to the 2nd question about using an off-the-shelf transformer. I'm sure several of you immediately considered and dismissed this as another possible PS solution and I'd really appreciate a better idea of what I would be giving up by doing it this way.
No. Power supply voltage for the driver stage (LME) is higher for a good reason: the LME should be able to deliver a clean voltage swing when the output stage runs out of power, it is just good engineering.
My Yamaha B2 amplifier has a regulated +/- 80 volt power supply for the driver stage whereas the output stage runs on some +/- 55 volts.
Hi Guys,
Time for an update on the amp and I'll answer a few of the questions posed by rileywc.
1. I still have not received or heard back from AP2 (Robert) about the SMPS PSU. I'm keen to measure it, and could easily pull off a 1-day turnaround on it, but I need the PSU first. I wrote Robert a note the same day he posted about the supply, but never heard back.
2. The design itself is completely finished, and the cost of the kit will need to increase $2 per kit to cover some additional parts that will be required to provide added flexibility to everyone building the kits. ETA on the transistors is this Friday so I'll start accepting payments in the next day or two, and I'll start shipping early next week.
rileywc:
I think there's a bit of a misunderstanding about the approach to the power supply design. The LME49830 benefits a great deal from regulated supplies, but this is secondary to the desire to run the front end on a higher voltage than the output stage.
It works like this:
As the amplifier output swings close to the output stage rails, the voltage between the drain and the source of the mosfet approaches 0V. It never quite gets there due to the Ron of the mosfet and the associated I*R drop across the mosfet, but with higher impedance loads, it can get pretty close. At this point, the gate of the mosfet will need to be driven significantly higher than the source in order to keep the mosfet in the "on" state. That means the driver stage has to be able to swing beyond the rails of the output stage if you want to be able to actually clip the output stage into the output stage rails.
If you use a single high voltage supply and regulate down the driver stage voltage, you'll have to set the regulated voltage at least 10-15V lower than the output stage to ensure it stays regulated under full load. At that point, the driver stage will only be able to swing about 20V short of the main rails, which means at full output power, there will be more than 20V across the mosfet. This results in massive amounts of wasted power (especially with low impedance loads) and large amounts of wasted power at idle. It could also destroy the output devices if you exceed their rated dissipation.
Consider these examples:
AMP 1 - With separate higher driver voltage
Output stage rails = 60V
Driver stage rails = 75V
Peak output volatge into 8R load = ~57V
Peak Power dissipated by device at peak output voltage = 21.5W
Peak Power delivered to load at peak output voltage = 406W
Power at idle with 350mA bias current = 42W
AMP 2 - With driver voltage regulated off output supply
Output stage rails = 75V
Driver stage rails = 65V (regulated down from 75V rails)
Peak output voltage into 8R load = ~57V(accounts for driver not swinging to rails and Vgs required)
Peak Power dissipated by device at peak output voltage = 128W
Peak Power delivered to load at peak output voltage = 406W
Power at idle with 350mA bias current = 53W
If you need to use off the shelf components, you really have two options:
1. Use a single transformer that makes the rails about 10-15V higher than the values listed in the power tables and don't regulate the front end. This will result in higher idle dissipation, and lower possible output voltage for a give rail voltage.
2. Buy two transformers. One will be a large VA unit that feeds the mosfet output stage and the other can be a very small unit that just feeds the driver stage. Something like this costs less than $20 and could power several amplifiers:
Digi-Key - 237-1324-ND (Manufacturer - VPT230-110)
You could hook it up for dual 60VAC windings which will give you roughly +/-75V after a pair of linear regulators. You don't have to be exactly 10V higher than the output stage, but you should be at least 10V higher.
Cheers,
Owen
Time for an update on the amp and I'll answer a few of the questions posed by rileywc.
1. I still have not received or heard back from AP2 (Robert) about the SMPS PSU. I'm keen to measure it, and could easily pull off a 1-day turnaround on it, but I need the PSU first. I wrote Robert a note the same day he posted about the supply, but never heard back.
2. The design itself is completely finished, and the cost of the kit will need to increase $2 per kit to cover some additional parts that will be required to provide added flexibility to everyone building the kits. ETA on the transistors is this Friday so I'll start accepting payments in the next day or two, and I'll start shipping early next week.
rileywc:
I think there's a bit of a misunderstanding about the approach to the power supply design. The LME49830 benefits a great deal from regulated supplies, but this is secondary to the desire to run the front end on a higher voltage than the output stage.
It works like this:
As the amplifier output swings close to the output stage rails, the voltage between the drain and the source of the mosfet approaches 0V. It never quite gets there due to the Ron of the mosfet and the associated I*R drop across the mosfet, but with higher impedance loads, it can get pretty close. At this point, the gate of the mosfet will need to be driven significantly higher than the source in order to keep the mosfet in the "on" state. That means the driver stage has to be able to swing beyond the rails of the output stage if you want to be able to actually clip the output stage into the output stage rails.
If you use a single high voltage supply and regulate down the driver stage voltage, you'll have to set the regulated voltage at least 10-15V lower than the output stage to ensure it stays regulated under full load. At that point, the driver stage will only be able to swing about 20V short of the main rails, which means at full output power, there will be more than 20V across the mosfet. This results in massive amounts of wasted power (especially with low impedance loads) and large amounts of wasted power at idle. It could also destroy the output devices if you exceed their rated dissipation.
Consider these examples:
AMP 1 - With separate higher driver voltage
Output stage rails = 60V
Driver stage rails = 75V
Peak output volatge into 8R load = ~57V
Peak Power dissipated by device at peak output voltage = 21.5W
Peak Power delivered to load at peak output voltage = 406W
Power at idle with 350mA bias current = 42W
AMP 2 - With driver voltage regulated off output supply
Output stage rails = 75V
Driver stage rails = 65V (regulated down from 75V rails)
Peak output voltage into 8R load = ~57V(accounts for driver not swinging to rails and Vgs required)
Peak Power dissipated by device at peak output voltage = 128W
Peak Power delivered to load at peak output voltage = 406W
Power at idle with 350mA bias current = 53W
If you need to use off the shelf components, you really have two options:
1. Use a single transformer that makes the rails about 10-15V higher than the values listed in the power tables and don't regulate the front end. This will result in higher idle dissipation, and lower possible output voltage for a give rail voltage.
2. Buy two transformers. One will be a large VA unit that feeds the mosfet output stage and the other can be a very small unit that just feeds the driver stage. Something like this costs less than $20 and could power several amplifiers:
Digi-Key - 237-1324-ND (Manufacturer - VPT230-110)
You could hook it up for dual 60VAC windings which will give you roughly +/-75V after a pair of linear regulators. You don't have to be exactly 10V higher than the output stage, but you should be at least 10V higher.
Cheers,
Owen
pieter_t,
Thanks for the response! - I was still trying to make sense of your Yamaha amp data when I noticed opc's response.
Owen,
Thanks a lot also for your detailed reply! I didn't expect so much info, but certainly do appreciate it! You are right in that I mistakenly thought the primary reason for the regulated higher voltage supply was the ripple rejection - thanks for straightening me out on that.
I'm still a little confused about the LME rails. I definitely agree with you here:
I see the talk about sagging rails naturally led you to a misconception of my desires with the Wire power amps. I am not really very concerned with sag because my needs are for much lower power. Powering a 50W speaker driver at low volume, rather than blasting a 400W PA speaker or sub should produce somewhat lower losses than your example shows. Still, you have convinced me that your option 2, additional transformers, is the best for me and I'll go that route. Still look forward to seeing your results with the SMPS, even though those appear to be most useful to those requiring more power than I.
Thanks again for your time and effort,
Walter
Thanks for the response! - I was still trying to make sense of your Yamaha amp data when I noticed opc's response.
Owen,
Thanks a lot also for your detailed reply! I didn't expect so much info, but certainly do appreciate it! You are right in that I mistakenly thought the primary reason for the regulated higher voltage supply was the ripple rejection - thanks for straightening me out on that.
I'm still a little confused about the LME rails. I definitely agree with you here:
but still don't understand why you would ever want the Driver stage output swinging wider than the Output stage rails, which I think we agree would cause clipping, or even to swing as wide as the Output stage rails, if you want to allow for those rails to sag without clipping. Whatever I've got wrong makes it hard for me to understand pieter_t's amp because it has voltage differences even more extreme than your Wire. I can be a little dense sometimes.
I see the talk about sagging rails naturally led you to a misconception of my desires with the Wire power amps. I am not really very concerned with sag because my needs are for much lower power. Powering a 50W speaker driver at low volume, rather than blasting a 400W PA speaker or sub should produce somewhat lower losses than your example shows. Still, you have convinced me that your option 2, additional transformers, is the best for me and I'll go that route. Still look forward to seeing your results with the SMPS, even though those appear to be most useful to those requiring more power than I.
Thanks again for your time and effort,
Walter