It looks really good, Chris.
Reuse of a cabinet is possible.
thanks you for the roses....
i checked the LRS 150-24V instead of the lrs 350-36 with the yj TPA boards. no differents in normal use. just if you psuh really lod the TPA board get UVP and mut one channel or both.
after the test with the smps the idea with a big cap bank is away from my head. is it possible to calculate the needed cap bank?
smps is abe to handle 6,5Amps or slightly more, but the caps get empty with big bass pulses....
i ask myself if more current is needed with this "weak smps" lrs150-24 what to do? 2 x smsp parallel?
Good and large decoupling capacitors you only find in really expensive amplifiers. From what I have seen, 10.000uF-20.000uF per 100W of serious output power is about the norm. For 4 Ohm loads keep it near the 20.000uF.
The decoupling capacitors are really important with good amplifiers having high current capability because they have an important say on how heavy transients are handled. The SMPS is just used to re-fill the decoupling capacitors. I would say the capacitors are more important for the sound than the SMPS as long as the SMPS is stable, emits no excessive noise and regulates reasonably fast. Evidently, the power of the SMPS decides the steady-state power of the amplifier. I would use the 36V you already have. I use a 31V/5A "brick".
The transient handling depends a lot on the nature of the transients. Bass-drums and church (pibe) organs are among the worst. I use a small 24bit sampled sound-file (FLAC) found on YouTube called "Audiophile Drum". Pure drumming in high quality. At the very end you have three ultra low bass tones. To handle them clean is very demanding.
I do not know of a way to calculate the needs for the capacitor bank.
The decoupling capacitors are really important with good amplifiers having high current capability because they have an important say on how heavy transients are handled. The SMPS is just used to re-fill the decoupling capacitors. I would say the capacitors are more important for the sound than the SMPS as long as the SMPS is stable, emits no excessive noise and regulates reasonably fast. Evidently, the power of the SMPS decides the steady-state power of the amplifier. I would use the 36V you already have. I use a 31V/5A "brick".
The transient handling depends a lot on the nature of the transients. Bass-drums and church (pibe) organs are among the worst. I use a small 24bit sampled sound-file (FLAC) found on YouTube called "Audiophile Drum". Pure drumming in high quality. At the very end you have three ultra low bass tones. To handle them clean is very demanding.
I do not know of a way to calculate the needs for the capacitor bank.
Last edited:
thx
sorry JMF11 i read to fast. i mentoide the YJ TPA3255 board but i gess its the same?
i do the simulation ...
Thank you FF for explanation. i tried with my poor crc 4x4700µF -R=0,05 - 3x4700µF = so its 33mF but as you can see in the previous pages no success.
this crc construction ...post #234
the 36V smps is not possible in this small housing -amp light
i will try with doctors magic caps - Chemicon
EGPD630ELL242MM40H 2400µF -
...post 268This may be true for 50Hz-Transformer supplies, but not for smps. SMPS do primary mains rectification and so the bulky reservoir caps are located in the primary circuitry of the smps. This is a significant difference to old school power supplies often overseen by audio-DIYers.
In that case adding bulky reservoir caps on the secondary side is lost money and will degrade regulation of the smps.
Furthermore they may put additional stress to the smps.
And yes, these 2400uF caps are a proper choice for TPA3255 amps.
Thank you voltwide.
this is the design difference.
if 2 xsMPS of the same size working in parallel -Waht happend???
is there a current flowing between the output terminals dureing different regulations??
chris
You are right, this is "old school" and if the SMPS has a fast regulation, less can do. But, most of the SMPS you buy are designed for rather steady loads, like LED-lights, and they are not particular fast. They still need some kind of reservoir capacitor at the output. 2400uF may be fine depending on the speed of the SMPS regulation. The charger-type "bricks" I use act as a kind of power current generator because they are used to look into a rechargeable battery. They just refill my capacitor banks.
Last edited:
I believe Ireland is still sleeping.
Two SMPS working in parallel will only work if they are designed for load sharing.
If designed to work alone, you risk one will source the other.
That you may avoid by adding diodes at the output. But, then one (the one with the highest output voltage) will typically take the hardest turn (poor load sharing) and the other will have difficulties regulating because it does not understand the regulation of the first SMPS. That is likely to end up in mutual oscillation phenomena.
Load sharing requires particular design like coordinated regulation loops (eventually master-slave) or output "droop" characteristics.
Then, rather one SMPS per amplifier channel (if possible).
If this is true, they won't be light load stable and oscillate, don't they? The needed output capacitance is within the SMPS (+ margin for aging, temperature, tolerances). Ever seen a minimum capacitance (at load site) stated in the datasheet? To my knowledge they only state a maximum capacitance to guarante proper startup and load/line-regulation. Using a LED-light supply isn't possibly the best choice as, for sure, they "might" be slower than others but this is mostly true for cheap chaps.
Having 1,000,000uF at the output with a CC limited slooooow starting supply in front is something that may provide a benefit in bass when sitting next to the amp, but 10,000-50,000uF provide no enhancement in the low end as there is just not enough energy stored. They "may" provide supply support for the mids if you ask for several 10-100W of output but do you do in this range?
Theres a reason why we use these 2400uF caps as close to the chip as possible (within less than 25mm distance) - it's their ESR that counts for higher frequencies (out of audio band). At switching frequency (450-600kHz) their influence/support is also from limited value, due to rising ESR again. This is where those MLCC come into play (4x1uF nominal, more like 4x600nF with DC-bias).
Last edited:
Your first argument I do not understand. Even if they have say 100uF at the output and included as part of the power supply, they may be light load stable if the modulation allows for such low output power operation. Power supplies will hardly be delivered in a configuration where they are instable within the intended operating range. Less output capacitance means higher ripple but that is another matter. The minimum capacitance is already on-board. My point is that using such SMPS, that are not designed for the very dynamic consumption of an audio amplifier, requires enough capacitance at the supply line to supply the energy for signal transients until the SMPS regulation loop reacts. The capacitance may be on the amplifier board, on the SMPS board or on the connection between the two.
We seem to agree that SMPS designed for general purposes, like LED-lighting, may react slowly simply because a slow loop is cheaper to implement and LED-lights is a very steady load. If the total supply line capacitance is too low for the audio amplifier, the voltage will sag too much (from a sudden increase in load current) before the SMPS-loop reacts and increases the capacitor voltage.
We also agree that the more capacitance on the supply line the slower the SMPS regulation loop will react. A trade-off but with more "quick" energy in the decoupling capacitance.
You mention a huge 1.000.000uF capacitor bank which I agree serves no purpose as it is exaggerated. A huge capacitance may even make the SMPS swich off (current overload protection) or oscillate - there the charging adapters have an advantage as they assume a rechargeable battery. The decoupling capacitance needs to match the SMPS reaction speed (regulation loop speed), but rather too much capacitance than too little. You state that 10.000uF - 50.000uF does not help on the bass because there is simply not enough energy stored. There I disagree with reference to the "old school" designs using double rectified net-transformers and where an amplifier was "on its own" in-between the charging pulses appearing every 10ms. In-between these charging pulses and for almost 10ms, the amplifier only had the energy stored in the supply line capacitors to operate from. And, it succeeded including bass.
I hope most SMPS have a reaction speed better than 10ms but the bandwidth needs, for PWM-designs, to be considerably below the SMPS switching frequency. A bandwidth of a couple of KHz perhaps? The needed supply line capacitance is chosen accordingly.
As you seem to be competent and serious, I do not doubt your 2400uF may do a good job in many cases. For the carrier frequency of several hundreds of KHz it is the small foil/ceramic capacitors we put in parallel with the electrolytic's that do the job because they have a lower ESR. The 2400uF is still important for the mid-range and perhaps the upper bass because the SMPS regulation speed is not sufficient.
Do we "do" in the 10W-100W range? I do not because my wife will not allow it. We design for up to 200W+ because the norm has become that more is better. Then, the currents (in transients) increase and the need for energy storage goes up as well.
We seem to agree that SMPS designed for general purposes, like LED-lighting, may react slowly simply because a slow loop is cheaper to implement and LED-lights is a very steady load. If the total supply line capacitance is too low for the audio amplifier, the voltage will sag too much (from a sudden increase in load current) before the SMPS-loop reacts and increases the capacitor voltage.
We also agree that the more capacitance on the supply line the slower the SMPS regulation loop will react. A trade-off but with more "quick" energy in the decoupling capacitance.
You mention a huge 1.000.000uF capacitor bank which I agree serves no purpose as it is exaggerated. A huge capacitance may even make the SMPS swich off (current overload protection) or oscillate - there the charging adapters have an advantage as they assume a rechargeable battery. The decoupling capacitance needs to match the SMPS reaction speed (regulation loop speed), but rather too much capacitance than too little. You state that 10.000uF - 50.000uF does not help on the bass because there is simply not enough energy stored. There I disagree with reference to the "old school" designs using double rectified net-transformers and where an amplifier was "on its own" in-between the charging pulses appearing every 10ms. In-between these charging pulses and for almost 10ms, the amplifier only had the energy stored in the supply line capacitors to operate from. And, it succeeded including bass.
I hope most SMPS have a reaction speed better than 10ms but the bandwidth needs, for PWM-designs, to be considerably below the SMPS switching frequency. A bandwidth of a couple of KHz perhaps? The needed supply line capacitance is chosen accordingly.
As you seem to be competent and serious, I do not doubt your 2400uF may do a good job in many cases. For the carrier frequency of several hundreds of KHz it is the small foil/ceramic capacitors we put in parallel with the electrolytic's that do the job because they have a lower ESR. The 2400uF is still important for the mid-range and perhaps the upper bass because the SMPS regulation speed is not sufficient.
Do we "do" in the 10W-100W range? I do not because my wife will not allow it. We design for up to 200W+ because the norm has become that more is better. Then, the currents (in transients) increase and the need for energy storage goes up as well.
Last edited:
Be aware that under all circumstances peak current draw is limited by speaker impedance. There is no point in implementing peak current capabilities of some 10amps ore more - these will never be demanded. Once you have defined you upper current requirement, a suitable smps will work under all conditions.
Peak overload capabilities of smps normally degrade with excessive secondary caps.
Peak overload capabilities of smps normally degrade with excessive secondary caps.
Hi FF
Hi Doctor
i can partly follow your discussion
is the time of the "smps reaction" measureable? or is a factor missing in the datasheet. how can i do a test?
my intepretation is=> as fast your caps are (very low ESR) and the capacity to upload the cap is not to high --> the smps will not be affected.
@LRS 150-24V
As my protocol showed at 4 ohms load (#343) that i get 5,75amps. therefore the rated current of 6,5 is not achivable with 24V setup without load.
so for me i will try to trimm the SMPS to e.g. 26V get the max current of 6,5 amps(maybe slightly more) at 4 ohms load.
Hi Doctor
i can partly follow your discussion
is the time of the "smps reaction" measureable? or is a factor missing in the datasheet. how can i do a test?
my intepretation is=> as fast your caps are (very low ESR) and the capacity to upload the cap is not to high --> the smps will not be affected.
@LRS 150-24V
As my protocol showed at 4 ohms load (#343) that i get 5,75amps. therefore the rated current of 6,5 is not achivable with 24V setup without load.
so for me i will try to trimm the SMPS to e.g. 26V get the max current of 6,5 amps(maybe slightly more) at 4 ohms load.
AAAh.. as alway in engineering...
so if your SMPS is not well designed --> the correction is a bad compromise
Hi FF
Hi Doctor
i can partly follow your discussion
is the time of the "smps reaction" measureable? or is a factor missing in the datasheet. how can i do a test?
@LRS 150-24V
As my protocol showed at 4 ohms load (#343) that i get 5,75amps. therefore the rated current of 6,5 is not achivable with 24V setup without load.
so for me i will try to trimm the SMPS to e.g. 26V get the max current of 6,5 amps(maybe slightly more) at 4 ohms load.
The SMPS reaction time is sometimes stated in the datasheet as the step response.
How to measure:
(General) Switch between two different, but realistic, loading levels of the SMPS output. For an SMPS used for an audio amplifier, 200mA <-> 2A is not unrealistic. Check how the output voltage with higher loading for a start sags, then starts increasing and with some oscillation returns to the initial voltage. The response time is the time until the voltage no longer sags.
(Amplifier specific) Use the amplifier with the intended (dummy!) load. The test generator (input) must be able to run in "burst"-mode. Adjust the test generator burst amplitude such that the amplifier output is at a defined output power level. Suggested 50% of full output power and 100Hz frequency. Then run no input signal <-> input signal for defined output power and watch how the SMPS output voltage sags every time a signal burst starts. The response time is the time until the voltage no longer sags.
This is a rough simulation of how the supply voltage will behave with heavy transients in the music. By changing the capacitance on the output (within a reasonable range) you can find a reasonable capacitance value. Smaller SMPS output voltage swing is better.
Last edited:
I find the SMPS topic really intersting as:
- it can be counter intuitive, and somehow far from old school approaches,
- many users are not so well educated (I'm in that case),
- it seems that there are not so many off the shelf suitably designed products,
- DIY implementation does not seems so easy because of the above and some specific components like tailor made transformers and inductors (to be confirmed if true)
I reput here the link to that PS reference design from TI, that would perfectly fit the TAS3251: PMP10215 Universal Input to 3.3V, 12V, 36V, 200W Continuous PSU for Class-D Amplifier Reference Design | TI.com
...in case somebody would like to elaborate on that.
JMF
- it can be counter intuitive, and somehow far from old school approaches,
- many users are not so well educated (I'm in that case),
- it seems that there are not so many off the shelf suitably designed products,
- DIY implementation does not seems so easy because of the above and some specific components like tailor made transformers and inductors (to be confirmed if true)
I reput here the link to that PS reference design from TI, that would perfectly fit the TAS3251: PMP10215 Universal Input to 3.3V, 12V, 36V, 200W Continuous PSU for Class-D Amplifier Reference Design | TI.com
...in case somebody would like to elaborate on that.
JMF
The topic is drifting a bit away from class D amplifiers though the SMPS are meant for TPA3255 amplifiers.
I will dare one more posting on the topic on this thread and the moderators may move it if necessary.
My experience is that SMPS were conceived in order to improve efficiency and physical size/weight. With high level of integration on monolithic ICs and a massive production out east, SMPS also became very price competitive.
99% of all SMPS are designed with these main features in mind: efficiency, compact size/weight and price.
Then comes a group of designers having decades of experience with high quality audio amplifiers presenting unusual characteristics of a very dynamic (varying) power consumption. They notice the new power conversion technology and think: "that we can probably use as well". The problem is only that their main wishes are little noise and fast response (in order to maintain the supply voltages constant).
All other qualities like highest efficiency, particular small size/weight, very low price, power levels, safety issues etc. are only secondary concerns or trivial.
The problem for them is then that the bulk production is geared towards the mainstream types and few SMPS designers are particularly knowledgeable about audio amplifier needs. This is where we are today.
I have seen SMPS designed for dedicated audio amplifier use in the sense of output voltage levels (also symmetrical) and power. I did not find mentioning of particular noise/EMI reduction means or any strong attention to dynamic properties.
I know the market is rather limited but we need someone designing dedicated audio amplifier SMPS at a decent price. And, it is not as trivial as it seems.
I will dare one more posting on the topic on this thread and the moderators may move it if necessary.
My experience is that SMPS were conceived in order to improve efficiency and physical size/weight. With high level of integration on monolithic ICs and a massive production out east, SMPS also became very price competitive.
99% of all SMPS are designed with these main features in mind: efficiency, compact size/weight and price.
Then comes a group of designers having decades of experience with high quality audio amplifiers presenting unusual characteristics of a very dynamic (varying) power consumption. They notice the new power conversion technology and think: "that we can probably use as well". The problem is only that their main wishes are little noise and fast response (in order to maintain the supply voltages constant).
All other qualities like highest efficiency, particular small size/weight, very low price, power levels, safety issues etc. are only secondary concerns or trivial.
The problem for them is then that the bulk production is geared towards the mainstream types and few SMPS designers are particularly knowledgeable about audio amplifier needs. This is where we are today.
I have seen SMPS designed for dedicated audio amplifier use in the sense of output voltage levels (also symmetrical) and power. I did not find mentioning of particular noise/EMI reduction means or any strong attention to dynamic properties.
I know the market is rather limited but we need someone designing dedicated audio amplifier SMPS at a decent price. And, it is not as trivial as it seems.
Last edited: