In fact, this thread (and all Carlos' ones!!!) is a vast source of information and research for a doctoral thesis in psychology!
Cheers,
Max.
The problem observed with symmetric amps is exactly this- there are two signal paths from the input to the last gain stage. These signals are summed together at the output of the last gain stage (transconductance stage). These two signals never arrive at the same time. Also there is almost always two current sources one on the plus rail and one on the minus rail that must "balance" to each other. This balance takes time often called settling time. I have seen settling time to 12 bits in nanoseconds and settling time to 16 bits in milliseconds. This ends up with warm up phenomenon and the time difference causes TIM distortion which is very audible. It is far preferable IMHO to have but one signal path through the amplifier. THD will be higher but all these other distortions can be made to completely disappear. THD will still be less than 0.1%...-60dB from the signal and small compared to the transducers.
=SUM
=SUM
But how do you explain a symmetrical design which has much lower TIM levels (this is often the case) if this is what the problem is with symmetrical design compared to a single signal path. Also Symmetrical designs do not always give the best THD performance.
There are a lot of ways to measure non-harmonic distortions. Maybe I should not have used TIM as a name. Here we use "Instantaneous overload distortion" or IOD for short. This includes a lot of things not defined by TIM- my error in using that name as it is a subset of all non harmonic distortions.
An optimized symmetric design should be lower THD than a single ended however any design can have more distortion not optimized. An optimized single path amplifier has the lowest audible distortions of all designs I have worked with and the lowest non-harmonic distortions of all the designs I have worked with. Actually the lowest "total distortions" of any design for audio use.
This thread http://www.diyaudio.com/forums/solid-state/164093-100w-ultimate-fidelity-amplifier-new-post.html has a design similar to what I have found as (and I hate to say this) best sound and lowest non-harmonic distortion for power amplifier.
An optimized symmetric design should be lower THD than a single ended however any design can have more distortion not optimized. An optimized single path amplifier has the lowest audible distortions of all designs I have worked with and the lowest non-harmonic distortions of all the designs I have worked with. Actually the lowest "total distortions" of any design for audio use.
This thread http://www.diyaudio.com/forums/solid-state/164093-100w-ultimate-fidelity-amplifier-new-post.html has a design similar to what I have found as (and I hate to say this) best sound and lowest non-harmonic distortion for power amplifier.
Dooes it apply to the output power stage as well?....
There s more advantage than disadvantages in a symmetrical
differential, just like a push pull output stage vs a single ended
output stage.
Fully complementary amps
All of the negatives described here about full complementary amplifiers are a result of poor engineering. I have seen many schematics of this type where the VAS transistor was chosen poorly, such as a device of too high power rating. A large power transistor may have an ft of 60 mhz, but also have a large miller capacitance and long storage time. Most transistors are tested for speed with forced turn off. As a result I have seen some 4 mhz devices operate much faster in VAS than a 30 mhz device. For VAS, you want the smallest die size that meets the needed soa goal, not gross overkill. Also, you want the + and - signals arriving at the VAS pair at the same time, so transistors must be chosen to match their timing parameters, then further match the timing with small b-c capacitors. Gain is another issue, the various emmiter resistors are not always the same, as so called complements are not really perfect, resistors must be chosen to compensate for that. This type of amp can be the best performer in terms of symetrical slewing and TIM, but it takes real engineering and an AP or D scope to get it right. The schemos I've seen here are often bad copies of bad copies. 
A good symetrical design has excellent PS rejection, and works well with switching PSU. Finally, the people who hate linear amplifiers are the ones who can't design anything outside of spice.
All of the negatives described here about full complementary amplifiers are a result of poor engineering. I have seen many schematics of this type where the VAS transistor was chosen poorly, such as a device of too high power rating. A large power transistor may have an ft of 60 mhz, but also have a large miller capacitance and long storage time. Most transistors are tested for speed with forced turn off. As a result I have seen some 4 mhz devices operate much faster in VAS than a 30 mhz device. For VAS, you want the smallest die size that meets the needed soa goal, not gross overkill. Also, you want the + and - signals arriving at the VAS pair at the same time, so transistors must be chosen to match their timing parameters, then further match the timing with small b-c capacitors. Gain is another issue, the various emmiter resistors are not always the same, as so called complements are not really perfect, resistors must be chosen to compensate for that. This type of amp can be the best performer in terms of symetrical slewing and TIM, but it takes real engineering and an AP or D scope to get it right.
The schemos I've seen here are often bad copies of bad copies. A good symetrical design has excellent PS rejection, and works well with switching PSU. Finally, the people who hate linear amplifiers are the ones who can't design anything outside of spice.
Well I have been using common collector (emitter follower) output stage after the transconductance stage with complementary transistors.
KAZVARGA is right in what he says- timing and die matching leads to the best fully complementary amps possible and bad copies of copies is so true. Still though, my single sided design beats all comers for audio including my own and others carefully designed symmetrical power amps. I believe it is a settling time issue as I have not seen or designed any method to achieve settling time control even close to the single ended version I use. Symmetric IC op amps for A/D and D/A struggled with this for years and finally came up with some schemes to get fast settling. These schemes rely heavily on design methods not available with discrete transistors. Thermal effects of the physically separated devices mess with propagation and settling time in ways that I have no idea how to solve outside of an IC.
Complementary transistors are more of a wishful thought than a reality as P type and N type really function differently in gain curves, reflected impedance, capacitance, thermal effects, and on and on.
As sort of an aside or note, I have found any amp which exhibits warm up phenomenon where the amp does not sound its best until it has been on for some time also has these other issues. Warm up on my amp is well under 1 minute. No other power amp I have ever used or tested took less than an hour to come to most of its "good" sound being pretty scratchy the first moments after turn on.
My single sided amp is dead symmetric slew rate and does not offer the power supply rejection of the fully symmetric designs as KAZVARGA suggest. However, I do not mind a few decent power supplies to get rid of the other problems.
Maybe KAZVARGA would be so kind as to post a schematic of his favorite amp for audio use.
KAZVARGA- if you are close to St. Joseph Missouri we could compare amplifiers in person.
Note, for non-audio I use all kinds of symmetric amps and a great variety of designs which fit the application.
About single ended
Actually, I do agree that excellent single ended amps can be built, they are forgiving of small design mistakes, and with good engineering can be superior. My original post is in reply to the topic of this thread, can symetrical amps be good, and the context appeared to be as used in high power mobile carnival systems. My long answer just said : yes if they are well designed. In the context, they offer better PS rejection (important in mobile use), and often less TIM at screaming volume. No one topology solves all amplifier issues. I do use single ended for hi fi applications up to 200w, and for horn drivers. When used in commercial sound, single ended is more prone to pick ac mains noise, like air conditioners and lighting systems in clubs, or generator noise in mobile. I know a pi filtered PSU could fix that, but at 3kw? Worse yet, some systems demand lighter switching PSU.
I can post a schemo of a typical symetrical amp I like to use, I will need to edit out the SOA limiters as they are proprietary to where I work.
Actually, I do agree that excellent single ended amps can be built, they are forgiving of small design mistakes, and with good engineering can be superior.
My original post is in reply to the topic of this thread, can symetrical amps be good, and the context appeared to be as used in high power mobile carnival systems. My long answer just said : yes if they are well designed. In the context, they offer better PS rejection (important in mobile use), and often less TIM at screaming volume. No one topology solves all amplifier issues. I do use single ended for hi fi applications up to 200w, and for horn drivers. When used in commercial sound, single ended is more prone to pick ac mains noise, like air conditioners and lighting systems in clubs, or generator noise in mobile. I know a pi filtered PSU could fix that, but at 3kw? Worse yet, some systems demand lighter switching PSU. I can post a schemo of a typical symetrical amp I like to use, I will need to edit out the SOA limiters as they are proprietary to where I work.
Leech amp has a very decent SOA limiter of the single slope variety. KAZVARGA is right again which is cool. We do mobile PA and greatly prefer using many smaller amps to less big ones. We also used balanced transformerless output for higher power. Amps over about 400 watts into 8 ohms seem to be a waste by our test for PA. Drivers just go into power compression. We always use symmetric amps for bass as the circuitry is more power efficient than the single ended designs. My single ended is proprietary so will not post that however for the symmetric type the highly copied early BGW seems like the way to go. IC front end into a common base stage to a transconductance stage and on to the drivers and outputs. Simple, with several feedback loops this kind of amp is only second or third place to best there is when properly tuned and as good of bass as any design.
As a note- my single ended amp does not even use current sources. Instead it has very high voltage supplies and large resistors. Shocking tube high voltages!
=SUM
As a note- my single ended amp does not even use current sources. Instead it has very high voltage supplies and large resistors. Shocking tube high voltages!
=SUM
Most of the limiters seen here on DIY are peak current only no matter the output. These I call "the always fail blow up the amplifier" circuit and are truly worse than nothing at all. The Leech amp limits current to a small value when the output voltage is small and increases current capacity as output voltage becomes larger up to the peak current. This loosely tracks the SOA of the transistors, well enough to protect output from ever blowing from over current. So there is the minimum current which is always allowed, the slope with the relationship between output voltage and output current, and then there is the peak current limit which the output voltage is above a certain value.
So possibly I got the name wrong but above is how it works. Please give it a proper name if you like.
So possibly I got the name wrong but above is how it works. Please give it a proper name if you like.
Hi,
Leach discusses his limiter in his Lo Tim paper.
The graph clearly shows a horizontal slope and an inclined slope.
Both of these slopes are controlled by the voltage drop across the emitter resistor.
The resistor to turn this current limiter into a VI limiter must enable monitoring of Vce. One extra resistor gives a VI limiter.
Jens incorporated the extra resistor in his Leach clone PCBs for a true VI limiting action.
I equate SOA limiting to VI limiting in this respect. This is due to the very important Vce dependance for any of the current/power limits implied by the SOA.
Leach discusses his limiter in his Lo Tim paper.
The graph clearly shows a horizontal slope and an inclined slope.
Both of these slopes are controlled by the voltage drop across the emitter resistor.
The resistor to turn this current limiter into a VI limiter must enable monitoring of Vce. One extra resistor gives a VI limiter.
Jens incorporated the extra resistor in his Leach clone PCBs for a true VI limiting action.
I equate SOA limiting to VI limiting in this respect. This is due to the very important Vce dependance for any of the current/power limits implied by the SOA.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.