I bought a new Traco TIS 150-148 for €35 on eBay: http://www.farnell.com/datasheets/1763527.pdf
48V, 3A, 80KHz switching, <50mV noise and ripple, 420000hrs MTBF.
48V, 3A, 80KHz switching, <50mV noise and ripple, 420000hrs MTBF.
Is it worthwhile to attach a capacitor bank of 20000 uf between the MW PS and the 3255 board? The idea is that the caps will provide the short term energy needed for bass reproduction.
Please assume that I use a MW lrs-150-48. Will it be powerful enough to refresh the bank?
No. That's liable to have nasty effects on smps. And "instantaneous power delivery" is a specious concept, as it's not the same here as a linear psu getting refreshed once every fiftieth or sixtieth of a second.
Modest amount is a-okay. Which is conveniently already provided on board. A lc filter around the switching frequency may be more sensible, as loop gain falls off (and psrr with it)
Modest amount is a-okay. Which is conveniently already provided on board. A lc filter around the switching frequency may be more sensible, as loop gain falls off (and psrr with it)
Last edited:
For that SMPS, a cap bank would definitely trip overcurrent protection and go into a restart loop. Or if it is a really huge cap bank you could see smoke from the SMPS.
But think about it...an SMPS is a switching power supply and tpa3255 is a switching amplifier. In one sense, peak transients is how they function! They are both designed to respond much more quickly than any musical transients.
For what you are trying to accomplish, you should buy an SMPS that is rated to deliver the amps needed for those transients. The lrs-150 with its 150W is only able to give you 3.3A (although it can probably peak transients significantly above that). But if you had a 600W SMPS, you'd have 12.5A on hand - instantaneously - with peak transient capability above that.
Important note though: If you don't use enough volts, you'll never reach the amplifier's ability to reproduce transients.
So for you sir, I believe the recipe for success is to go with a high-wattage SMPS (600-750W depending on speakers) and run your amp board at the highest voltage it can take (53.5V for TI EVM, 51V for 3e-audio, YJ unmarked), and you should have musical transients to your heart's content. Be prepared to provide supplemental cooling (i.e., a fan) for the tpa3255.
Cap banks are necessary for linear supplies to both filter noise and counter the effects of voltage sag on the huge coil. And audiophiles, having since the beginning of time followed the "excessive is better" philosophy, like to trot out their big impressive cap banks and 100-lb transformer coils. LOL trust me, I got nothing against that.
But that thinking really doesn't translate over to SMPSes. SMPSes don't have voltage sag. Or I should say when they sag they recover incredibly quickly, depending on the quality of regulation. And they already have the caps needed to compensate.
All this has reminded me of building car stereo systems in the 90s. We used big caps the size of your hand between the battery and amp. It helped, really...because we were driving huge amplifiers from 12V stock battery.
No respectable psu designer cares that I forgot to double the frequency of a full wave rectifier. Especially when juxtaposed to switching frequency of smps.
So how about we don't quote me every time I make a simple math error and completely ignore the salient point. Unless you have some other objective? ;-)
So how about we don't quote me every time I make a simple math error and completely ignore the salient point. Unless you have some other objective? ;-)
It depends on the particular supply - some just have current limiting without foldback. In which case they'll act like a constant current supply until they regain regulation.
How do you know they have peak (transient) ability above their rated current? Suppose you want to have full power capability into 4ohms - from a 50V supply then you'll need 12.5A as you point out. More if your speaker dips below nominal impedance (most do). So then to cope you'll need at least 625W rating for your supply. But the crest factor of audio is such that your average power demands will be comfortably below 100W.
It looks to me that caps are a neat way around this crest factor problem - allowing a much smaller supply still to provide for peaks. Assuming they don't cause start-up problems - which indeed they may.
@DPH - still making unfounded assumptions eh? Your salient point wasn't ignored, just not responded to.
No, I'm not saying that. A small cap bank might not be sufficient to provide excess reserves of current at bass frequencies.
What I reckon is putting an inductor between the PSU and the cap bank. This does tend to ensure that the additional capacitance doesn't cause stability problems with the SMPSU's output regulation. But the size of the cap bank may well need to be very large to provide enough benefit to the short-term (say 10's of mS) current capability of the supply.
Since I haven't seen detailed enough specs from manufacturers about how they implement their current limiting I advise trying say 20,000uF with a series inductor (a few uH, wound around a Sendust toroid should suffice) and checking that the supply still starts up OK.
What I reckon is putting an inductor between the PSU and the cap bank. This does tend to ensure that the additional capacitance doesn't cause stability problems with the SMPSU's output regulation. But the size of the cap bank may well need to be very large to provide enough benefit to the short-term (say 10's of mS) current capability of the supply.
Since I haven't seen detailed enough specs from manufacturers about how they implement their current limiting I advise trying say 20,000uF with a series inductor (a few uH, wound around a Sendust toroid should suffice) and checking that the supply still starts up OK.
Or you can take a sensible engineering approach and ask the more appropriate question of whether, when being recharged at 20-100 kHz (I couldn't find Meanwell's switch frequency, but probably a safe range of assumption), requires anything more than the 4700 uF on-board. And what sorts of signals you might ostensibly need in order to stress the system sufficiently when being supplied by a SMPS properly rated for the task at hand.
I mean I'm all for excess, but that's not where I'd go for it. Placing a series inductive filter to raise the passive rejection of HF hash and switch carrier frequency would be much more up my alley.
In this case, since Sanjeev mentioned the model number, the mean well datasheet helpfully provides the necessary information so we don't have to assume anything. Overcurrent gets tripped between 110-140% of rated power. It is not a constant-current device and has no current limiting circuit, so 3.6-4.2 amps (@48V) sends it into hiccup mode - a restart loop. "Cap bank" implies lots of capacitance that could easily exceed 4.2A.
Sheer, brute-force life experience
Although it should be noted that those transient peaks are of very short duration...far, far shorter than the time it would take to express a musical transient.
It is an easy enough experiment for someone who has both the lrs-150 and a cap bank. It probably won't blow the power supply but I would recommend gradually increasing capacitance until reaching the point of failure.
After some disappointing experiences with Sure amplifiers I am thinking of going the TPA 3255 way. Searching for the blue and black boards I found various options oothandel TPA3255 Gallerij - Koop Goedkope TPA3255 Loten op Aliexpress.com
Is there any difference or should I just buy the cheapest one?
I am thinking of using my Hypex SMPS400A180 for power supply but this one is 2*42V. Can I just use 1*42V for this amplifier?
Is there any difference or should I just buy the cheapest one?
I am thinking of using my Hypex SMPS400A180 for power supply but this one is 2*42V. Can I just use 1*42V for this amplifier?
Would someone explain how gain works in this amplifier (or any class d amplifier for that matter). Here's how I see it, but I could be way off.
Within the chip, The input signal is compared to a sawtooth signal, resulting in a pwm output with a fixed p-p voltage (12 volts perhaps?). The input signal voltage will dictate the duty cycle of the pwm.
This pwm signal turns mosfets on and off, resulting in a fixed p-p voltage equal to the supply voltage p-p through the source/drain pins of the mosfet. The duty cycle of this pwm is the same as the duty cycle out of the comparator.
Therefore, the gain of this amplifier is dictated solely by the supply voltage.
Is that correct?
Mike
Within the chip, The input signal is compared to a sawtooth signal, resulting in a pwm output with a fixed p-p voltage (12 volts perhaps?). The input signal voltage will dictate the duty cycle of the pwm.
This pwm signal turns mosfets on and off, resulting in a fixed p-p voltage equal to the supply voltage p-p through the source/drain pins of the mosfet. The duty cycle of this pwm is the same as the duty cycle out of the comparator.
Therefore, the gain of this amplifier is dictated solely by the supply voltage.
Is that correct?
Mike
I've thought this AN is quite useful for getting started:
https://www.infineon.com/dgdl/an-1071.pdf?fileId=5546d462533600a40153559538eb0ff1
That error amp up front and the feedback network (internal) are going to pin the external gain of the system.
https://www.infineon.com/dgdl/an-1071.pdf?fileId=5546d462533600a40153559538eb0ff1
That error amp up front and the feedback network (internal) are going to pin the external gain of the system.
Last edited:
No, supply voltage (all other things not limiting) dictates maximum power at a specific distortion.
Gain is set with the feedback ration within the IC.
From the data sheet:
G____Inverting voltage Gain____VOUT/VIN_____21.5 dB
Per the TI data sheet, wouldn't you want to increase the value of the caps on the outputs first?
Thanks. I read that article, and this is on page 4
Gain – With Linear amplifiers the gain is constant irrespective of bus voltage variations, however with Class D amplifiers the gain is proportional to the bus voltage. This means that the power supply rejection ratio (PSRR) of a Class D amplifier is 0dB, whereas the PSRR of a linear amplifier is very good. It is common in Class D amplifiers to use feedback to compensate for the bus voltage variations.
Hopefully you can understand my confusion.
Mike