As a thread owner/starter, and in in the context of being far more of a speaker guy than an amp guy, I ask since the "load" or speaker is rarely as as well behaved and predictable as a strict dummy load (or resistor load).
A loudspeaker (or amplifier load) is rarely a idealized closed box, or even a Helmholtz function, operating by its own with infinite stiffness material parameters, in a free field with only small signal domain parameters defining its function and load characteristics (a.k.a T/S-parameters).
Adding to this there are extra (as in outside of) T/S parameters to consider on the amplifier side, such as the dynamic, and sometimes non-linear regenerative kinetic energies from diaphragm suspensions, masses and accelerations with the corresponding air load factors, both direct and stemming from the enclosure internal geometry based resonances, or other speakers operating in close proximity and even rebounding close proximity room reflections, and all of this with varying inductance's and resistances stemming partly from temperature dynamics and coil position in the transducer air gap during operation.
This is the actual reality of "the load".
By what means does an amplifier deal with all of this? and how is it taken into account in the design process as per application (see again the two amplifier examples), is it really all down to the DF residing on the right side of 20?
A loudspeaker (or amplifier load) is rarely a idealized closed box, or even a Helmholtz function, operating by its own with infinite stiffness material parameters, in a free field with only small signal domain parameters defining its function and load characteristics (a.k.a T/S-parameters).
Adding to this there are extra (as in outside of) T/S parameters to consider on the amplifier side, such as the dynamic, and sometimes non-linear regenerative kinetic energies from diaphragm suspensions, masses and accelerations with the corresponding air load factors, both direct and stemming from the enclosure internal geometry based resonances, or other speakers operating in close proximity and even rebounding close proximity room reflections, and all of this with varying inductance's and resistances stemming partly from temperature dynamics and coil position in the transducer air gap during operation.
This is the actual reality of "the load".
By what means does an amplifier deal with all of this? and how is it taken into account in the design process as per application (see again the two amplifier examples), is it really all down to the DF residing on the right side of 20?
Last edited:
When the driver acts as a "generator" due perhaps to mass inertia, plus the inductive intertia (the tendency of current in a coil to continue to flow as the flux collapses), PA amps have a diode to the power supply that dumps current outside the voltage of the rails to the rail caps. This keeps voltage on the output transistors from exceeding the max, or even nearing it.
The feedback circuit of the amp senses the "errors" in the speaker voltage, differing from the ideal of the A*v of the input signal, and causes the drivers predrivers etc to provide signal to the output transistors to counterbalance these "errors".
Some creative amps have in the past put sensors on the speaker cone to measure non-linear displacement of the driver, and put that into the feedback loop. LWE III for example (I owned two). L.W. Erath of Texas. The market has determined this is unnecessary.
The feedback circuit of the amp senses the "errors" in the speaker voltage, differing from the ideal of the A*v of the input signal, and causes the drivers predrivers etc to provide signal to the output transistors to counterbalance these "errors".
Some creative amps have in the past put sensors on the speaker cone to measure non-linear displacement of the driver, and put that into the feedback loop. LWE III for example (I owned two). L.W. Erath of Texas. The market has determined this is unnecessary.
Last edited:
I bet it's not. A loudspeaker is such a complex 'machine' that a singe DF or 'fly back' diode will not suffice to deal with all possible occurences. Any analysis about this diode-dumping-in-powersupply found? It must be a magical trick of somesort to eradicate all unexpected emf from the load. DF will surely swallow 'everything' to end ends.
I'm pretty sure that as long the main amp is in class A, most things will behave quiet well. But as soon one side switches off, you're gone. No control whatsoever. Just current dumping and a feedbackloop going beserk.
Class A is not going to be a high wattage amp the OP is asking about.
The "analysis" people do on here with spice, I have to laugh at them. The experienced posters have proposed 3 element speaker model with inductors, but all anybody uses on here in sims is a 4 ohm or 8 ohm resistor. HA! AndrewT did warn us that real speaker currents were sometimes 2.5 times the resistor model predictions. Another thing nobody does here with their expensive scopes is measure the emitter resistor voltage (output transistor current) with real speakers attached, near the power limit or after clipping has started.
Real AB amps, run at way below the power limit, will not use the overvoltage diodes. Which is what the OP is asking about. Whereas real bar band amps are frequently run near the limit, and the ones with those diodes survive the warrenty period better.
Note to OP, I have a one transistor pair amp (ST120 mod to AX6), and another 4 transistor pair amp, (CS800s) and at the 1/4 watt to 70w I listen to them in my living room, on the same speakers, there is only one way I can tell the sound apart. The point to point AX6 board I built hums worse. The Dynakit PC15 board (survivor) with djoffe idle bias control mod sounds exactly the same as the CS800s. AX6 & PC15 have input & output speaker caps, and the CS800s doesn't.
Last edited:
Rail Voltages? How many output devices? BJT or FET? Complementary? Sometimes I think you are big teaser... But it is okay, Thanks for the hints and at least making me think that the approach is a good one. Time will tell. And for Dadod sharing so much, He is Phenomenal.
Rail Voltage 90VDC. I think, it is highest voltage for class AB.
It is all BJT with 9 pairs output transistor.
If you want to build amplifier with 77VDC rail voltage, I can send you the schematic. I'm already share it at local DIY community, and they made several version of pcb layout. It is using Blameless topology with TMC compensation, and TEF output.
So basically, you suggest that the moving masses of the cones and coils, within magnetic field, result in the loudspeaker being a power plant. And correctly so.
A simple experiment: Take any woofer and try to move its cone. Easy? Easy.
The woofer-power-plant is generating electricity at "idle", with no load.
Now: Short the terminals of that woofer-power-plant ... and try again. Hard as a Brick? Difficult to move?
A real eye-opener. The woofer-power-plant is now working against a a short circuit, and the power plant is overloaded with the short. This also means that the energy consumption of such woofer-power-plant is very high. That is why it is so difficult to move the membrane, the cone.
...
That is the reason, that the more effective is the "short circuit" (as in: the output impedance of the amplifier, which is essentially a short circuit as seen by the woofer power plant), the lower the output impedance, and the bigger the protest of the cone against any unwanted movement implied externally upon it.
...
But this is just the beginning. Not only do you wish the cone not to move in directions that it should NOT move. You also actually also want it to move in the direction that you WANT it to move.
So basically, the amplifier-power-plant needs to be much more powerful than the woofer-power-plant. And it is CURRENT that counters the unwanted movements of the cone, forcing it to move where you want it to move.
...
IMHO: Damping factor aka low amplifier-output-impedance is just one of the two ingredients that are necessary for good loudspeaker control. The other important thing is the capability of the amplifier to deliver short term peaks of current, that are crazy in value.
Basically, the amplifier should be capable of driving up to five times the current that would at fist glance be necessary, when "listening" to an amplifier driving a resistance dummy load.
Example:
Rail Voltage is 50V.
Resistance dummy load: 8 ohms.
Peak current into resistance: 50V/8ohms = 6,25A.
TRUE NEEDS in terms of peak output current capability: 6,25A * 5 = 32,25A.
In other words, the amplifier should be capable of delivering welding-machine grade currents, as high as 32A, for a limited time span, equal to the duration of a half wave length of a bass note.
At 20 Hz, that would be 1/20Hz/2 = 25 miliseconds.
Typical loudspeakers, of the "well behaved" variety, can have impedance dips down to 3 ohms, at a phase shift of even up to 60 degrees. And, curiously enough, it is not the 3 ohms that is the amp killer. It is the phase shift.
The value of the multiplier "Times Five" is an empirically devised value, a "cover-it-all" real life approximation, that should counter the bad effects of impedance dips, phase shifts, back EMF from the moving mass of the cone, when considering normal music content with normal slew rates as found in a real, but dynamic, music signal.
Hence, a good power amplifier should also serve the purpose of being a welding machine.
In my humble opinion
, which may differ from other opinions.
Last edited:
Send?
Rail Voltage 90VDC. I think, it is highest voltage for class AB.
It is all BJT with 9 pairs output transistor.
If you want to build amplifier with 77VDC rail voltage, I can send you the schematic. I'm already share it at local DIY community, and they made several version of pcb layout. It is using Blameless topology with TMC compensation, and TEF output.
Why not Post it?
Design a house's floor. Add the man, wife, couch, dog, loudspeakers, find the total load. Test some wood to get the strength. Design floor beams to support that load.
NO!! Building Codes require a design load about 2X what any normal occupancy weighs, and de-rates wood about 3X down from test results on small clear specimens. If all goes well a floor is about 6X stronger than it needs to be. Experience shows some occupants are hoarders, and some lumber is knotty/checked, and this much safety margin is needed to prevent collapse.
Audio designers "should" work this way. Speakers do not, because while a 5-pound cone/coil would be hard to destroy, it would also make very little useful sound power. Electronics should have more leeway but a lot of designers design right to the edge.
A saving grace is that we often use 100 Watt amps at 20W peak, which means 90++% of the time we are asking 2W or less. If an amp survives a 100W bench-test, it is unlikely to strain at 2W or even 20W.
This is the schematic. The pcb layout is not mine, so I can not share it.
But it is tested by local audio community.
It use easily to find transistor2 in my country, of course you can make it better by using better transistors.
Q10, Q12, Q14, Q15 on small heat sink.
Transistor bias servo, driver, and output on same heat sink.
Right so!
As posted else on diy, I'm currently designing a bridge amp, running at 10A each side with +/-20V rails (40V/8=5A, 40V/2=20A). End stage can provide 40A if needed, supply can handle 50A. No flyback/protection diodes.
Here you got my attention. I am intrigued. Why do you not use flyback diodes and how can you benefit from NOT using them?
Where is the added value of not having them?
Do they distort the sound, in any way?
You might want to try a pure Class A amp of 25w and see how that sounds. For some reason, a lower power Class A amp sounds like a higher power Class AB amp. It might have to do with fact distortion artifacts on Class A are not as noticeable so power can be turned up higher?
I do know that my 50w Class A sounds like my 100w Class AB in terms of power to drive clean loud bass.
I do know that my 50w Class A sounds like my 100w Class AB in terms of power to drive clean loud bass.
It is difficult to design a class AB amplifier that have distortion almost same at entire level before clipping, and almost same at all audio frequency.
I hope the OP realises how limited the scope of this comparison would be. It can only illustrate how different power ratings of class D, SMPS powered amps affect the perceived sound. It cannot possibly have any bearing on class AB, or output transformer coupled amps.
Having owned a couple of very high power non switching audiophile amps i readily acknowledge that at low power these amps do some things better than a low powered amp possibly could. At other things they were not so great.
Having given it some thought i have come to the rather obvious conclusion that there are two outstanding design features: a very stable PS with a tremendous amount of stored energy and a huge array of parallel output devices. One of these amps had a 2x3kVA power transformers which alone contributes to a particular sonic portrayal just not possible with a 200VA power transformer.
Which does not in any way mean the comparison would not be worthwhile, only that it's scope would be limited to that particular type of amp.
Having owned a couple of very high power non switching audiophile amps i readily acknowledge that at low power these amps do some things better than a low powered amp possibly could. At other things they were not so great.
Having given it some thought i have come to the rather obvious conclusion that there are two outstanding design features: a very stable PS with a tremendous amount of stored energy and a huge array of parallel output devices. One of these amps had a 2x3kVA power transformers which alone contributes to a particular sonic portrayal just not possible with a 200VA power transformer.
Which does not in any way mean the comparison would not be worthwhile, only that it's scope would be limited to that particular type of amp.
Well, as the OP I'm fully aware that these are some very different designs, with very different purposes and vastly different capacities, hence my question given a scenario that both will fulfill, <10W low frequency reproduction (load control) given a bit of a tricky load (back-EMF wise, assumed).
The topic is about if one will perceive a subjective difference in sound quality between the labgruppen IP450 and the labgruppen FP6400 given the conditions described earlier in the tread and in a more simplified way above.
Is this in reference to the two amplifiers I referenced? if so, it is very interesting, could you elaborate the importance of a stable and capable power supply (perhaps extensively so in one of the case of the FP6400) when it comes to load control in the user case as described?
Re-think this. If the final build contains any form of current limiting, the flyback diodes are needed. Any time the current is abruptly limited (either unintentionally by the driving stage running out of gas or intentionally by the activation of a VI limiter) while the load phase angle is lagging (inductive) the output voltage WILL spike and discharge into the supply. If the current flows through the flyback diodes it is harmless. Sounds like $#**, but harmless. If it flows through the transistors on reverse, they die. Oversizing the transistors does not help. They can not and will not handle large reverse currents for any real length of time. Some switching transistors have RBSOA ratings - and the total energy values are in millijoules. There is a lot more energy than that stored in the moving mass of a woofer motor when playing loud enough to heat the coil up. Still don’t wantt the diodes?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.