I'm not using a SMPS. I am using my house battery bank (48V 160Ah). And a portable battery bank (48v10ah) when going out with friends. Current capability is, from the wall wirings point of view, 25A continuous. (From the batteries point of view, the number of amps can be overwhelming..)
If you use Lion batteries, when fully charged they have much higher voltages than rated.
You can not ignore the data sheet's max operating voltage, just because you use batteries and it is more convenient for you to have no proper power supply. This is not how the real world outside the internet works.
You can not ignore the data sheet's max operating voltage, just because you use batteries and it is more convenient for you to have no proper power supply. This is not how the real world outside the internet works.
I writted a few posts ago what batteries i use. LiFePO4 with a custom BMS (made by myself). Voltage never passes 54.6v
If Texas Instruments is right, the TPA3255 doesn't work with more than 53.5 Volt. You may know better after testing.
PS and yes, it's got an over voltage protection shutdown.
PS and yes, it's got an over voltage protection shutdown.
I am trying to get a new GB started for this. Look in the thread.
3251 has a little bit better performance on paper than 3255 regarding distortion but can only take 36V .
any reason why use such high voltage (53-55v)? If your TPA3255 shutdown at 50v doesnt mean its fake either, maybe because its to hot especially when you try on cheap chinese board
If you’re using two 4 ohm speaker I would definitely use lower voltage than 48V .
A 36V 10A supply would give more power than any 48V 5A in such case , when using the 3255 in BTL mode ( stereo )
As I see it, using 48-53 V makes most sence if using 8 ohm speakers in BTL or if you use mono coupled tpa 3255 ( PBTL ) that can drive 2 ohm loads.
Last edited:
At 52V, peak of the sinewave will be (taking some losses into account on: MOSFETS, traces, output filters, wiring) around 50V. RMS will be 50 /1.414 = 35.3V. Into 4 ohms, thats 312W. That fits perfectly within specs.
Takin into account that we dont actually listen to a tone generator but to music, the peaks will be, yes, 300W per channel, but the average will be into 60W per channel at a maximum, when really pushed until HiFi ceases to be HiFi and just starts to enter into overdrive. That means roughly 120W total output, and 12 to 20W thermal loss (I think that will be more close to 12w than to 20w, due to the nature of class D).
So, if the chip is 3255, I think it will handle well enough with cooling of a modest 0.1A fan.
Not only that more, but all these will be superfluous if the guys from TI really implemented the overvoltage protection feature (3255 shutting down beyond 53V), that will oblige me to implement a buck, either LDO or SMPS, to make the voltage a bit lower (actually I have to admit that I wanted to do that as an automation anyway, in the distant future, keeping the voltage typically at 19 or 20V when listening at normal audio levels, and rising it only as response when power needed, using the overdrive signal of the 3255 just because it exists (otherwise I could use other means, no problem)
Takin into account that we dont actually listen to a tone generator but to music, the peaks will be, yes, 300W per channel, but the average will be into 60W per channel at a maximum, when really pushed until HiFi ceases to be HiFi and just starts to enter into overdrive. That means roughly 120W total output, and 12 to 20W thermal loss (I think that will be more close to 12w than to 20w, due to the nature of class D).
So, if the chip is 3255, I think it will handle well enough with cooling of a modest 0.1A fan.
Not only that more, but all these will be superfluous if the guys from TI really implemented the overvoltage protection feature (3255 shutting down beyond 53V), that will oblige me to implement a buck, either LDO or SMPS, to make the voltage a bit lower (actually I have to admit that I wanted to do that as an automation anyway, in the distant future, keeping the voltage typically at 19 or 20V when listening at normal audio levels, and rising it only as response when power needed, using the overdrive signal of the 3255 just because it exists (otherwise I could use other means, no problem)
Last edited:
Now you are inventing the first "H" class class-D amp. Nice and useless, as this is only relevant for A/B amps. Any D-amp is a voltage switching device anyhow. Maybe read a little about how it works.
Just get some fat diodes and put two of them in series with the positive feed. The roughly 1.4 Volt reduction should do. If you are still concerned, maybe add a large Zehner diode for burning off any short overvoltage on the amp side. Even this is pointless, as there is no danger for the amp other than switching off for a short period of time.
Last, it may be kind of a hobby to theorize about anything for an endless time, but at some point you should simply build what you have on your mind and try the practical side.
The price for a decend D-amp module like the BDM8-A is around 30€, an quite nice, complete 3255 (even with a basic DSP), the BRU5 is about 40€ from Aliexpress, delivered to your doorstep including tax and customs. This is the risk you have to take.
Just get some fat diodes and put two of them in series with the positive feed. The roughly 1.4 Volt reduction should do. If you are still concerned, maybe add a large Zehner diode for burning off any short overvoltage on the amp side. Even this is pointless, as there is no danger for the amp other than switching off for a short period of time.
Last, it may be kind of a hobby to theorize about anything for an endless time, but at some point you should simply build what you have on your mind and try the practical side.
The price for a decend D-amp module like the BDM8-A is around 30€, an quite nice, complete 3255 (even with a basic DSP), the BRU5 is about 40€ from Aliexpress, delivered to your doorstep including tax and customs. This is the risk you have to take.
Exactly what I suggested earlier and I think is the easiest solution. $2 solution.
https://services.taiwansemi.com/storage/resources/datasheet/MBR2035CT SERIES_M2104.pdf
Interestingly, I found that my 36V 6A SMPS from my A07 sounds (no noise/good power) just as great on the 3E Audio pffb board (outside alloy case).
I also have a generic 36V 5A power supply (no case) that must be emitting RFI. Both 3255 amps are quiet only when placed inside the thick alloy amp case when wired up with this particular SMPS.
Om not 100% convinced that pffb always sounds better with music compared to a good tpa board without it. The measurement often get better with static measurement signals with pffb.
I guess its my scepticism against dual ne5532 often used in the pffb constructions, and the long debate during the -80 (Matti Otala ) about negative feedback making the sound worse if its used to much.
I may be wrong on this - Hypex uses a ton of negative feedback and my ncore amplifier sounds rather good.🙂
I guess its my scepticism against dual ne5532 often used in the pffb constructions, and the long debate during the -80 (Matti Otala ) about negative feedback making the sound worse if its used to much.
I may be wrong on this - Hypex uses a ton of negative feedback and my ncore amplifier sounds rather good.🙂
The 3E board doesn't come with 5532's IIRC. It does sound more refined/detailed than the A07 judging by moderate volume I listen at.
I'll be able to do a quick case swap with the XRK 3255-PFFB when the time comes and compare also. I will be machining a dedicated 1-piece block/heat spreader for XRK's board similar to 3E's design.
I'll be able to do a quick case swap with the XRK 3255-PFFB when the time comes and compare also. I will be machining a dedicated 1-piece block/heat spreader for XRK's board similar to 3E's design.
I actually prefer the sound of no PFFB. It may get -7dB lower THD but at the low levels already intrinsic in the amp not sure if needed.
There’s a slick 50pF or 100pF post filter feedback developed by Bucksbunny that was shown to be even better than conventional PFFB. Basically, take output at speaker out and connect to input pin at chip via a 100pF NP0/C0G cap. It gets rid of overshoot and ringing.
Last edited: