i am thinking of placing the input caps off the board since MKP caps are physically large. I am using SE input mode, so i will remove CA1/RA1 (left channel) and CB1/RB1 (right channel) and place them outside outside the board. Will then take the output end of the caps and wire them in to the board pads for these caps that connects RA3 (left channel) and RB3 (right channel).
Just thinking what will be a good value (uf) for these MKP caps.
D
Deleted member 148505
Omg we have 65kHz switching SMPS?So, in the end, isn't it better to use 10uH? I see farnell has 103KL in stock. I know, it is less linear etc. But...
10uH + 1uF has 50kHz freq.
10uH + 1,5uF has 41kHz freq.
Both with usable Q.
I have shielding tape (thin metal foil with adhesive layer, 5cm in width on a 5m long roll) for purposes like that. To create something like half farraday cage over AC parts. Or some 5x5cm cuprextit (proto PCB with full Cu layer without holes) placed between yellow TENTA and Nichicon and grounded. But perhaps it will be useless without full farraday cage. And without oscilloscope or spectral analyser it is only guessing.
What we are avoiding is the peaking at open load. 6.8uH + 0.68uF only peaks at 16dB (6.3x) at open load. Zobel can take care of the peaking.
Forgot about the scope lol. I already ordered chassis, it's 12.12 sale so I got some discount. I can check for optimal layout once it arrives.
I already used SMPS300RS in my THD measurements, it doesn't have mains harmonics if the cabling is good.
6.8uF is the best minimum.
It's funny how I have always thought that the worst part about the SMPS, that needs shielding, is the SMPS itself, or at least its switching part. And now we are fiddling about how to protect the good SMPS from the bad AC
BTW, those "small but good input caps" are hard to find, I know... This is how it ended on my small TPA3116 amp
Clarity cap is the best mkp for the price I have heard, but is very big...
ppppp & Asuslover.... Mouser order placed for the 1uF 50V caps. We shall see how they change the sound.
Pete
commstech: See post #862. Your plan seems like the only way to go if good input caps (read "bigger") are desired to be used. I may have to think about using this plan unless Lester wants to design bigger boards to implement real audiophile input caps.
Pete
If too long, may be. But, the length of the feedback loop from 3255's out through RF4 and SJ4 back to the 3255's input, is some 30cm or 12in. So some 2cm/1inch legs of big cap wont hurt it. And usually the whole board will be shielded in metal case. More serious "problem" is that the wobbly placed big foil cap without stable fixation to the board is prone to microphony and vibrations, resulting in some hum/interference. But that is due to cap itself, not its legs.
Last edited:
They certainly will fit better than the Mundorf MKP caps I’m gonna try to mount..
Pete
Where in the circuit will you use Mundorf and where KEMET? And what values? And will you combine more KEMETs in parallel to get bigger uF value?
I'm certainly interested in the audio results too. And will perhaps order that caps anyway. But your review will be helpful for sure...
ppppp: either Kemets or Mundorfs pre-op amps for now (3e board). If I can get the Mundorfs properly mounted and vibration free AND they sound good) maybe the Kemets will then go into pre-3255 positions (3e board). The pre-3255 position, on the 3e board (unless I maybe alternate one on top then one underneath) is still very crowded and I don’t think I’ll be able to get the Kemets in place.
With Lester’s amp I’m only going to try pre-op amp for either Kemet or Mundorf. I haven’t the electronics savvy to figure out what will need to be changed in the PFFB structure, of Lester’s board, if I fool with the pre-3255 caps’ values.
Pete
With Lester’s amp I’m only going to try pre-op amp for either Kemet or Mundorf. I haven’t the electronics savvy to figure out what will need to be changed in the PFFB structure, of Lester’s board, if I fool with the pre-3255 caps’ values.
Pete
D
Deleted member 148505
I will try 10uF Panasonic FR for pre-3255 position based on this guy's review. Looks compatible with TPA3255's sound signature which is warm
Miniature electrolytic capacitors shootout - Best for Coupling to Headphones?
Miniature electrolytic capacitors shootout - Best for Coupling to Headphones?
That loop is right on top of a ground plane though (probably), not floating in air.
That board is 4-layer? Nice
I dont know how to respond in "realistic" way. The reality is, that absolutely every circuit can pickup at least small portion of EMI/RFI. The other true is also, that audio devices and speakers are full of capacitors with long legs, and work fine. Perhaps it is good idea to put that amp in metal/shielded case, just to be sure. On the other hand, look at my amp in post 875. Those legs surely are little bit longer, and when I place mobile phone next to that amp, the speakers are deadly silent even without shielding...
I dont know how to respond in "realistic" way. The reality is, that absolutely every circuit can pickup at least small portion of EMI/RFI. The other true is also, that audio devices and speakers are full of capacitors with long legs, and work fine. Perhaps it is good idea to put that amp in metal/shielded case, just to be sure. On the other hand, look at my amp in post 875. Those legs surely are little bit longer, and when I place mobile phone next to that amp, the speakers are deadly silent even without shielding...
slaa787 isn't clear about whether I have to bring GND along with the IOM/IOP to the slave boards. I assume I would, otherwise the signaling wouldn't be able to work, but would this end up creating any sort of (harmful) ground loop?
Thoughts?
D
Deleted member 148505
The board is only 2 layers.
I'm listening to my PFFB module with LME49860 and Silmic II on pre-TPA3255 (C1 to C4) right now and I think the veiling sound of Silmic II is not that bad. I think today my ears is sensitive to highs/treble, so I'm good with the sound.
Anyways, just use 10uF Nichicon UES Muse BP on C1 to C4, it adds bloom/ sparkle/engagement in the sound (I will test panasonic FR soon).
Using external caps for the opamps is ok I think, just avoid the switching regulator part of our TPA3255 board by 1 to 2 inches.
Attached my plan layout for my module. The orange divider is what I think it is safe to put the preamp / low level signal components (rca jack, large caps, volume ctrl etc).
I'm listening to my PFFB module with LME49860 and Silmic II on pre-TPA3255 (C1 to C4) right now and I think the veiling sound of Silmic II is not that bad. I think today my ears is sensitive to highs/treble, so I'm good with the sound.
Anyways, just use 10uF Nichicon UES Muse BP on C1 to C4, it adds bloom/ sparkle/engagement in the sound (I will test panasonic FR soon).
Using external caps for the opamps is ok I think, just avoid the switching regulator part of our TPA3255 board by 1 to 2 inches.
Attached my plan layout for my module. The orange divider is what I think it is safe to put the preamp / low level signal components (rca jack, large caps, volume ctrl etc).
Attachments
D
Deleted member 148505
Yes the ground pin is not needed and might cause ground loops if tied together.
Thanks, I will add the clarification on startup manual and remove gnd tap on future modules.
D
Deleted member 148505
Hmm I think it's an unlikely scenario. Ground loop would be more of a problem than the synchronization of the clocks. Long wires between IOM and IOP is a no go. Then long umbilical cords from 1 supply to multiple chassis, that's impractical.
Either you have the modules in the same chassis using the same power supply, or use different chassis with dedicated power supplies.
D
Deleted member 148505
I will add in the startup manual that clipping without load should be avoided with our TPA3255 PFFB module. In clipping scenarios, tpa3255 chip adjusts the oscillating frequency to 1/4th the normal switching frequency. If there's no load, this will make the 220nF+1ohm output zobel dissipate large heat.
If other TPA3255 PFFB modules encounters this scenario without oversized zobel (or other circuitry to prevent it), zobel will be instantly fried. In our module, zobel can handle it for a couple of seconds.
In my IRS2092 modules, if there's an instability from open load, the circuit detects the heat of 5W zobel resistor using SMT PTC, it's a crude solution but it works. Most of the time, it is stable from open load, though there are some scenarios where it is not, so I implemented the temp sensor lol.
If other TPA3255 PFFB modules encounters this scenario without oversized zobel (or other circuitry to prevent it), zobel will be instantly fried. In our module, zobel can handle it for a couple of seconds.
In my IRS2092 modules, if there's an instability from open load, the circuit detects the heat of 5W zobel resistor using SMT PTC, it's a crude solution but it works. Most of the time, it is stable from open load, though there are some scenarios where it is not, so I implemented the temp sensor lol.
Attachments
Last edited by a moderator:
- Status
- Not open for further replies.