D
Deleted member 148505
After a week of delay, already shipped 7pcs TPA3255 PFFB TICore260BTL today, thanks for those who trusted and bought the modules!
Here is the comparison of PFFB vs Non PFFB modules after days of listening,
PFFB: Huge soundstage, will push the vocals and instruments slightly backward. Very smooth, removed some irritating / harshness sound from the regular TPA3255.
Regular Non-PFFB: Very forward sounding vocals and instruments, more like in your face sound. Because of its forward sounding nature, some might prefer this.
Both PFFB and non-PFFB are sweet sounding, very black background, almost 'wire with gain'. TPA3255 has more bass though compared to other amps.
Here is the comparison of PFFB vs Non PFFB modules after days of listening,
PFFB: Huge soundstage, will push the vocals and instruments slightly backward. Very smooth, removed some irritating / harshness sound from the regular TPA3255.
Regular Non-PFFB: Very forward sounding vocals and instruments, more like in your face sound. Because of its forward sounding nature, some might prefer this.
Both PFFB and non-PFFB are sweet sounding, very black background, almost 'wire with gain'. TPA3255 has more bass though compared to other amps.
D
Deleted member 148505
Thanks! I will check with her regarding the sound comparison this weekend.
That's awesome news Lester, thank you. Can you guess how long it will take to deliver the amp to EU in this covid time?
Interesting results. I believe that those differences can be actually a summary result of more parts: TPA's feedback, better regulator TLS205B0, VER2923 etc. Good job!
And the sound is further manageable by changing opamps, all electrolytes, and perhaps also the output caps in lowpass and zobel filters. It will be fun to mod...
BTW, looking into VER2923 datasheet, it's interesting how just the 10uH value has the least linear curve of all values
(The whole Inductance line is constantly slightly decreasing all the way from +5% at 100mA to -5% at 30A; meanwhile the other inductances are almost linear.) But I don't argue that it is some great coil.
---
Yes some good bypass cap can sometimes calm down the treble, it is good idea to have that option on the adapter.
But I read somewhere that LED has its own noise, so I'm not sure it is best idea to solder it right on opamp's power legs.
---
I think what will be better is a matter of taste, and/or combination with other parts of system, which can be sometimes a lottery.
2x Lundahl LL1540 or 2x Dual Burson V6 Vivid, each solution costs the same 139€ in EU. Bursons have good reviews, almost everyone point out their ultimate details and soundstage, without any harshness. But in the same time, I like the idea of getting rid off the "opamp dilemma" alltogether with input electrolytes (and maybe also with input LP filter), and put in one analog part instead (LL1540). Will the sound be better to my hearing than with Bursons or with your cleverly selected default parts? I have no idea
I think the best approach will be to wait the first 200 hours to burn-in the new amp with default parts, rolling only opamps that I already have + maybe some of those mentioned in this thread, like LM4562NA / OPA1692IDR / OPA1656. And then we will see (hear) ...
Interesting results. I believe that those differences can be actually a summary result of more parts: TPA's feedback, better regulator TLS205B0, VER2923 etc. Good job!
And the sound is further manageable by changing opamps, all electrolytes, and perhaps also the output caps in lowpass and zobel filters. It will be fun to mod...
BTW, looking into VER2923 datasheet, it's interesting how just the 10uH value has the least linear curve of all values
(The whole Inductance line is constantly slightly decreasing all the way from +5% at 100mA to -5% at 30A; meanwhile the other inductances are almost linear.) But I don't argue that it is some great coil. ---
Yes some good bypass cap can sometimes calm down the treble, it is good idea to have that option on the adapter.
But I read somewhere that LED has its own noise, so I'm not sure it is best idea to solder it right on opamp's power legs.
---
I think what will be better is a matter of taste, and/or combination with other parts of system, which can be sometimes a lottery.
2x Lundahl LL1540 or 2x Dual Burson V6 Vivid, each solution costs the same 139€ in EU. Bursons have good reviews, almost everyone point out their ultimate details and soundstage, without any harshness. But in the same time, I like the idea of getting rid off the "opamp dilemma" alltogether with input electrolytes (and maybe also with input LP filter), and put in one analog part instead (LL1540). Will the sound be better to my hearing than with Bursons or with your cleverly selected default parts? I have no idea
I think the best approach will be to wait the first 200 hours to burn-in the new amp with default parts, rolling only opamps that I already have + maybe some of those mentioned in this thread, like LM4562NA / OPA1692IDR / OPA1656. And then we will see (hear) ...
D
Deleted member 148505
Thank you and welcome, based on my previous shipments it's around 2 to 6 weeks. Looks like my local courier didn't process my shipments today, I haven't received any tracking number yet, I will follow up with them later.
Interesting observation, maybe it's general behavior for most inductors that's why 7uH inductors have lower distortion on TI's notes.
This comment is for my IRCore350M IRS2092 amp, OPA1656 has more details specially in the highs, then when I tried the OPA1692, it has the same tonalities with the OPA1656 but with reduced treble. I like the OPA1692 for my IRS2092 because of that.
For my TPA3255 PFFB, I think LM4562 has more treble than the OPA1656.
Yes I haven't made the parts under the PCB come as presoldered because some might not like them.
Hmm it's out of reach for me haha, I think the input capacitors are the ones that will affect the sound the most, so if we can remove them it's the ideal scenario. But LL1540 might have its own negative effects on the sound as well like the input caps (just speculation)
Last edited by a moderator:
D
Deleted member 148505
Thanks Alex!
Correct, I bought some cheap adjustable supply and soldered only 2 supply capacitors for testing, the sound is inferior compared to my abletec + complete supply caps.
Attachments
Alex, I'm really glad you mentioned that. I've bought Connex SMPS300RS (I need less power so 300W 36V single version should be ok), and although it looks promising in the datasheet, I was not sure whether it is as good as the manufacturer promises. So thank you.
Speaking about SMPS big caps, this SMPS has output CLC filter with 2x 4700uF noname caps (the 600W 48V version has 2x 6800uF). I want to replace them, but... In the datasheet they writes that this SMPS series is not good for loading big capacitance banks, because the power-on current will blow the overcurrent protection fuse on SMPS. And I also speculate that as they are part of CLC filter, their Z or ESR can be actually a part of design that just works.
So what do you think? Is it good idea to replace them with good Low Z / Low ESR SMPS big caps? What brand and value should be considered as "safe"?
Lester, can you tell if the TPA3255 is especially sensitive to DC offset? More sensitive than LM3886? My DAC has low DC offset so, long time ago, I removed all input DC blocking caps from my LM3886 amp, replaced them with wire, and the sound "alived" so much that I never returned back. And at my listening levels everything worked fine so far.
Indeed. It should have more tube-like character, which can be good or bad, depending of what you want. And as you mentioned, it is not exactly cheap, so that's why I will wait with this input mod. Perhaps your TPA3255 amp will sound excellent with bypassed input Muses...
R.
Last edited:
I appologize, my SMPS300RS actually has 2x 2200uF/50V Nichicon GY (LGY1H222MELZ25), caps suited for industrial PSU and ballasts.
Will there be some benefit for TPA3255, if I change them to something bigger or faster?
e.g.
2x Nichicon GY 6800uF/50V (LGY1H682MELA45)
or
2x Nichicon HE 2700uF/50V (Low Z cap, UHE1H272MHD)
And then on Lester's amp board there will be 4x Nichicon HW 2200uF/63V (Low Z cap).
I hope it is not offtopic... Thanks
R.
Will there be some benefit for TPA3255, if I change them to something bigger or faster?
e.g.
2x Nichicon GY 6800uF/50V (LGY1H682MELA45)
or
2x Nichicon HE 2700uF/50V (Low Z cap, UHE1H272MHD)
And then on Lester's amp board there will be 4x Nichicon HW 2200uF/63V (Low Z cap).
I hope it is not offtopic... Thanks
R.
D
Deleted member 148505
If you got a balanced output source, you can bypass the input caps and opamps altogether.
It's not even meanwell, it's just a generic power supply.
Model is Abletec ALP0400 same as the one in this (thread) I added 2 x 6800uf caps with no problems.
On its manual it says that the OCP might get tripped at power on if you add more output capacitance. You can add but not too much.
I think onboard caps on TPA3255 PFFB board are sufficient.
You may add power resistor - based soft start so that your SMPS won't get stressed on every startup.
You can replace the caps of the smps pppp, i do that at every SMPS from connex I buy...
I change the big on before switching mode and the other in the CLC filter, don't put too big value comparing to the original.
3300µF/50V was a good choice, on SMPS300RS the cap are snap-in pin, don't put standard radial cap, put Snap-in... like Nippon-Chemicon KMH.
Tell me the value/rating of the two model of caps, and the diameter.
But with smps300RS you don't reach the full potential ot the board, the power is restricted and the maximun voltage is not reached...
Actually the best way to reach the full potential of TPA3255 is 600W (min) with 52V (max).
I change the big on before switching mode and the other in the CLC filter, don't put too big value comparing to the original.
3300µF/50V was a good choice, on SMPS300RS the cap are snap-in pin, don't put standard radial cap, put Snap-in... like Nippon-Chemicon KMH.
Tell me the value/rating of the two model of caps, and the diameter.
But with smps300RS you don't reach the full potential ot the board, the power is restricted and the maximun voltage is not reached...
Actually the best way to reach the full potential of TPA3255 is 600W (min) with 52V (max).
D
Deleted member 148505
D
Deleted member 148505
+1
Same results here: depending on personal taste
1/ Non PFFB (TPA3255EVM opamps bypassed)
2/ With PFFB (drMordor TPA3255)
Nice job ...
Enjoy
If you really want the full potential (which is a bit silly for home use), you´ll need a heatsink that takes care of ~80W power dissipation and a rather clever mechanical design to get that heat from that tiny thermal pad.
With the given heatsink you won´t manage much more than 100W (though this is a guess only and assuming sine waves on the input).
Considering that and the crest factor of music you´ll be fine with much less than half of those 600W.
Your sure that the heatsink don't be sufficient ?
On my 2x250W IRS2092 is thiner than that and no problem, on my ICEpower and NCore this is the same, I never see a Class D heathing a lot... Lester can you confirm ?
Concerning the power is not the 600W the problem, is the current output capacity of the SMPS... with the 300W/36V (10.5 peak current) your can achieve the 17A of max current capacity of the TPA3255, with SMPS600RS/48V (15A peak current) your not at maximun but you approach him.
On my 2x250W IRS2092 is thiner than that and no problem, on my ICEpower and NCore this is the same, I never see a Class D heathing a lot... Lester can you confirm ?
Concerning the power is not the 600W the problem, is the current output capacity of the SMPS... with the 300W/36V (10.5 peak current) your can achieve the 17A of max current capacity of the TPA3255, with SMPS600RS/48V (15A peak current) your not at maximun but you approach him.
Last edited:
For this price I buy a Micro Audio Lester, SMPS630S => 110$ (shipping include)
Or SMPS600RS => 100$ (shipping include)
Meanwell is not the good choice for the "level of quality" of your TPA3255...
D
Deleted member 148505
Looks like engineers at TI were told to maximize the profit when they designed TPA3255. The IC is very small compared to power loss at continous rated power output. Power loss is around 80W at 550W output power (see attached screenshot). Even 45W TDP computer CPUs have larger thermal pad area, so they should have provided a larger IC package for the TPA3255 (plus the current IC is difficult to solder, out of reach of regular DIYers). If they have multi level switching technology like the one used in MERUS 12070, then that IC package is sufficient.
With that small of an IC package, amp designer must be creative in providing the heatsinking required in order to dissipate the heat when running the TPA3255 at continuous rated power.
Short answer, TPA3255 is targeted for home use, so the EVM heatsink is sufficient because most music content have high crest factor. With normal music recordings, even at near clipping loudness, the actual power that is being delivered by the amplifier to the speaker is only around 1/8th of the amplifier's rated power.
For compressed music like in the clubs/disco, actual power that's being delivered is around 1/3 of rated output power. So with 200W average power output, ~40W should be dissipated by the heatsink. If we use the EVM heatsink, I think it needs a blower fan for that setup, and a very good contact area between the thermal pad and the heatsink.
New FTC amplifier rule (part 432.3 standard test conditions) requires running the amp at 1/8th rated power for an hour as a precondition, and then applying full rated power for 5 minutes. I bet your 2x250W IRS2092 amp and this TPA3255 with EVM heatsink might not get past 1 minute before shutting down due to overtemp protect.
That's also the reason why I think meanwell EPP-500-48 / RPS-500-48 supply with 550W output power at 3 seconds is sufficient for home use.
With that small of an IC package, amp designer must be creative in providing the heatsinking required in order to dissipate the heat when running the TPA3255 at continuous rated power.
Short answer, TPA3255 is targeted for home use, so the EVM heatsink is sufficient because most music content have high crest factor. With normal music recordings, even at near clipping loudness, the actual power that is being delivered by the amplifier to the speaker is only around 1/8th of the amplifier's rated power.
For compressed music like in the clubs/disco, actual power that's being delivered is around 1/3 of rated output power. So with 200W average power output, ~40W should be dissipated by the heatsink. If we use the EVM heatsink, I think it needs a blower fan for that setup, and a very good contact area between the thermal pad and the heatsink.
New FTC amplifier rule (part 432.3 standard test conditions) requires running the amp at 1/8th rated power for an hour as a precondition, and then applying full rated power for 5 minutes. I bet your 2x250W IRS2092 amp and this TPA3255 with EVM heatsink might not get past 1 minute before shutting down due to overtemp protect.
That's also the reason why I think meanwell EPP-500-48 / RPS-500-48 supply with 550W output power at 3 seconds is sufficient for home use.
D
Deleted member 148505
If we use 36V power supply, near clipping output current (23.6Vrms at 4ohms load) will only be around 5.9A. So for stereo (if it happens that simultaneous 20hz at near clipping is played on your speakers) you need a power supply that can provide ~11A output current at 50ms (20hz)... Only if you are listening to your amp at near clipping loudness (that's very loud already).
If you add capacitors I think it can provide that 'burst'.
- Status
- Not open for further replies.