Annoyed with the never ending and usually ineffective attempts to make a 10$ Chi Fi chip amp even work passably.
I diy'd a Passlabs/ Firstwatt Amp Total costs ~200 $.
The sound is Exquisite.. by Anyone's yardstick.No exaggeration required and it's Absolutely Dead Silent . NO noises whatsoever: on, off, during.
Ears pressed flat to the dustcaps of my tannoy golds... Zero sounds.. period. Volume on full or off.
I had to go back and see if I had powered up the amp.
Best Amp I've ever had or even heard, for that matter.
Move on ? It could be an epiphany.. was for me.😉
Last edited:
Tannoys only need good headphone amplifier 🙂
300uV noise for smallest PassLab, that seems noisier than some of these ampboards and just $8000 more expensive. F4 probably lower noise, but that is not an amplifier 🙂
300uV noise for smallest PassLab, that seems noisier than some of these ampboards and just $8000 more expensive. F4 probably lower noise, but that is not an amplifier 🙂
Trying to squeeze out some more performance on the Sanwu Clone TPA3118 PBTL boards, i measured different thing to see any change.
The boards are unmodded except the output inductors are changed for Coilcraft MSS1278.
Setup:
PVCC: 19V Linear Supply
Gain: 20dB
Load: 4R7 resistive
Source/Testrig: Arta + EMU 0202 + 30dB Attenuator
So to start, here's just the inductors compared.
Stock vs. Coilcraft MSS1278:
No i checked the following things:
- replace local 0603 HF decoupling caps near the chip with Murata 0805 1uF 25V - nope
- replace output filter MLCC for Murata 0805 1uF 25V - nope
- replace input coupling caps for Murata 0805 1uF 25V - nope
- remove all bulk caps (4x330uF NoBrand) - slightly increase in THD
- stuffing 2x330uF Sanyo SEP Polymer for decoupling - nope
- stitching the groundplane for a solid plane - nope
So for this board in particular, the influence of bulk decoupling is negligible - LowGrade/NoBrand, Polymer, NoCap .. doesn't matter. 🙄
Well, does this mean anything for cap-rolling on other boards? You decide. 😉
The performance weakness is hidden somewhere else, haven't found yet. THD is dominated by K3/K5.
1W:
10W:
The boards are unmodded except the output inductors are changed for Coilcraft MSS1278.
Setup:
PVCC: 19V Linear Supply
Gain: 20dB
Load: 4R7 resistive
Source/Testrig: Arta + EMU 0202 + 30dB Attenuator
So to start, here's just the inductors compared.
Stock vs. Coilcraft MSS1278:
No i checked the following things:
- replace local 0603 HF decoupling caps near the chip with Murata 0805 1uF 25V - nope
- replace output filter MLCC for Murata 0805 1uF 25V - nope
- replace input coupling caps for Murata 0805 1uF 25V - nope
- remove all bulk caps (4x330uF NoBrand) - slightly increase in THD
- stuffing 2x330uF Sanyo SEP Polymer for decoupling - nope
- stitching the groundplane for a solid plane - nope
So for this board in particular, the influence of bulk decoupling is negligible - LowGrade/NoBrand, Polymer, NoCap .. doesn't matter. 🙄
Well, does this mean anything for cap-rolling on other boards? You decide. 😉
The performance weakness is hidden somewhere else, haven't found yet. THD is dominated by K3/K5.
1W:
10W:
Attachments
How does the 4 ohm resistor compare to the 14 ohm Rdc, 0,64mH Le speakers I am listening to right now 🙂
Cutting the Sanwu bottomtrace between input and powersupply ground made an audible noise difference, so should be visable too.
So for 550ms one does not need additional filtering/capacity for 1kHz test tone on a small ampboard with distance to your powersupply
Inductor-table Texas Instruments shows 1 watt THD on tpa3251evm for this MSS1278 of: 0.0145114031 I think. How much worse is Sanwuclone ?
Cutting the Sanwu bottomtrace between input and powersupply ground made an audible noise difference, so should be visable too.
So for 550ms one does not need additional filtering/capacity for 1kHz test tone on a small ampboard with distance to your powersupply
Inductor-table Texas Instruments shows 1 watt THD on tpa3251evm for this MSS1278 of: 0.0145114031 I think. How much worse is Sanwuclone ?
I just cut from existing groundplane separation to (long) side of board edge, just below standoff hole input, mute side. No other components/vias end in that trace, I don't think it matters much, this looked "natural to me" 😀
Very audible difference in noise here, maybe not 1000Hz, maybe preamp related ???? Much bigger difference than changing parts, noise becomes darker, less high frequency noise here, compared with switches on mute, instant AB. Funny.
Edit: regarding preamp relation, the Marantz preamp uses transformers on signal output
Edit: regarding preamp relation, the Marantz preamp uses transformers on signal output
Last edited:
So in that graph K2/H2 something like from 0.02% to 0.0025% from groundtrace cut.
With multitone signal (noise) very audible, hard to believe looking at those numbers. And why, how, would second harmonic be affected ???
With multitone signal (noise) very audible, hard to believe looking at those numbers. And why, how, would second harmonic be affected ???
Could outputimpedance signalgenerator and value inputcapacitor form a filter and then also influence phase ??? Could that look similar to the rising distortion with frequency of inductors, up to a point determined by formed filter ?
