According to your partner in crime things do change for the better with "other" inductors. Maybe he could answer which inductor to use. After you have measured I can combine all information if you like.
Translation: standard Sanwu TPA3118 boards with certain defined "difficult" load performed better than any other board in his possession. Known good sounding boards trip at 5V output with a 15V 14A PSU while the Sanwu boards (either with original or "other" inductors) reach 11V output and then trip. They sound better with "other" inductors, measurements not available.
My interpretation (when also looking at doctormord measurements): better inductors seem an obvious improvement. Original Sanwu boards seem to be able to output more than twice the voltage of other well designed TPA boards even with their supposedly poor inductors. With better "other" inductors in place still 11V is reached. That could mean a extremely well layed out PCB that deserves to be maxed out for optimal performance.
Translation: standard Sanwu TPA3118 boards with certain defined "difficult" load performed better than any other board in his possession. Known good sounding boards trip at 5V output with a 15V 14A PSU while the Sanwu boards (either with original or "other" inductors) reach 11V output and then trip. They sound better with "other" inductors, measurements not available.
My interpretation (when also looking at doctormord measurements): better inductors seem an obvious improvement. Original Sanwu boards seem to be able to output more than twice the voltage of other well designed TPA boards even with their supposedly poor inductors. With better "other" inductors in place still 11V is reached. That could mean a extremely well layed out PCB that deserves to be maxed out for optimal performance.
Last edited:
I have a big variation of Coilcraft inductors available, so it will be some from MSS series or XAL.
I'll just change the inductors, the rest remains unaltered. Will measure at 2/4R like the stock inductors.
OK Doctor we are on the same frequency. No other variables.
Test situation is: unbranded board with original TPA3118 and unchanged gain factor ? Gain 36 dB ?
* please note that the boards perform objectively best at 20 dB gain, with original or "other" inductors....
Test situation is: unbranded board with original TPA3118 and unchanged gain factor ? Gain 36 dB ?
* please note that the boards perform objectively best at 20 dB gain, with original or "other" inductors....
Last edited:
Here here Jean-Paul. I echo the same sentiments as you. I have subscribed to this thread one others and check on new posts regularly trying to learn and possibly implement some of the choices when I really I have a clear understanding of the info? Not yet so far! I appreciate all the engineering banter about the workings of the board, have a difficult time trying to follow though.
I get the impression that drmor doesn't think this board will improve much with any upgrades? That maybe the amp is no good after all his measurements? I have used it stock in several projects and am having fun and think the outcome is pretty good? If improvements can be made and heard! Easy enough without a major expense and if it was documented in a straight forward way with pics on what to implement and how I would enjoy risking the destruction of a cheap amp board to get the satisfaction that an improved outcome would provide big smiles!
I get the impression that drmor doesn't think this board will improve much with any upgrades? That maybe the amp is no good after all his measurements? I have used it stock in several projects and am having fun and think the outcome is pretty good? If improvements can be made and heard! Easy enough without a major expense and if it was documented in a straight forward way with pics on what to implement and how I would enjoy risking the destruction of a cheap amp board to get the satisfaction that an improved outcome would provide big smiles!
Then there are at least 2 that feel the same 😉 It would be good to document all info in a straight forward way with pics on what to implement for anyone just wanting to improve the boards. Efficient approach in a more nice form for readers specifically those that wish to simply follow what "the knowledgable" have designed/improved.
IMHO it would also be optimal to join all modifications as they enhance each other in the total result. So original Sanwu boards with original TPA3118 chips, 20 dB gain and better inductors that fit nicely on the existing pads. Possibly better input caps too if someone has replaced those ?!
IMHO it would also be optimal to join all modifications as they enhance each other in the total result. So original Sanwu boards with original TPA3118 chips, 20 dB gain and better inductors that fit nicely on the existing pads. Possibly better input caps too if someone has replaced those ?!
Last edited:
And as Irrebel (sp) points out an output filter, please. Like remove this part here and replace with this part with link to exact part number. I get so lost on sites like digikey, mouser, et al! Thanks. I am not lazy or stupid but I am ignorant of the things electronic but am learning slowly.
Who knows what value, quality and dielectric Sanwu has used ?! Changing them for known good X7R or PPD film (when possible) would be nice and it is a small effort. I have 2 of the boards in a box. If I have the time I will see what I can contribute. Till now I only changed gain and connectors for Phoenix ones.
Last edited:
Doctormord is able to design, also PFFB for tpa3118, I surely can't, I can just connect AM1 and use 3251 specified values or wonder why several of my amps switch off when another doesn't and compare layout.
It isn't just distance, the hiampmini is 5x5cm.
I first used the same inductors I used for hiampmini, 8.2uH which inside has coil parallel to pcb, I had does laying. I later used a set of small standard looking 10uH 4 or 4.5A ones I also used for Sure3116 and last one I used were flatwire HCI, all Wurths, because with my limited knowledge I feel safe using Wurths. I switched from first inductor because it was a little uncomfortable to place, it only just fits without touching other components, it is in reality a little bigger than the other Wurths that probably come from other factories. Small standard looking ones might be PD series, don't know. I don't use the Sanwu ampboards for myself, I use 4 Dugs (without rcsplits for avcc/pvcc) for the difficult speakers. But also the 2 per channel Dugs switch off (with easier loads seperate electrostatic panel/subwoofer) earlier than Sanwu.
The Coilcrafts might be highest distortion ones from TI's inductor comparison ? But always superior to chinese supplied with the $2 ampboard.
It isn't just distance, the hiampmini is 5x5cm.
I first used the same inductors I used for hiampmini, 8.2uH which inside has coil parallel to pcb, I had does laying. I later used a set of small standard looking 10uH 4 or 4.5A ones I also used for Sure3116 and last one I used were flatwire HCI, all Wurths, because with my limited knowledge I feel safe using Wurths. I switched from first inductor because it was a little uncomfortable to place, it only just fits without touching other components, it is in reality a little bigger than the other Wurths that probably come from other factories. Small standard looking ones might be PD series, don't know. I don't use the Sanwu ampboards for myself, I use 4 Dugs (without rcsplits for avcc/pvcc) for the difficult speakers. But also the 2 per channel Dugs switch off (with easier loads seperate electrostatic panel/subwoofer) earlier than Sanwu.
The Coilcrafts might be highest distortion ones from TI's inductor comparison ? But always superior to chinese supplied with the $2 ampboard.
My conclusion is the best supply voltage is 24v.
The input power diode should be removed bypassed. It is for diy mistakes.
The output filter is not so critical as I thought.
The generic board layout is perfect or some miracle component behavior not yet understood.
I can rate for 30-35 Watts hi-fi, plenty of ommph headroom before clip.
This is pretty close to best thread ever. Is the most useful for sure.
-bruce
The input power diode should be removed bypassed. It is for diy mistakes.
The output filter is not so critical as I thought.
The generic board layout is perfect or some miracle component behavior not yet understood.
I can rate for 30-35 Watts hi-fi, plenty of ommph headroom before clip.
This is pretty close to best thread ever. Is the most useful for sure.
-bruce
Can't say exactly, something happened in the 80W region before the supply foldback.
Most likely yes. I only plotted up to the first yellow line.
90W into 2R are 6.7Apk, pretty close to 7.5A trip point from datasheet. My supply only can put out 5A continously (lab-supply). Integration time was 100ms.
I never use 24V to listen, I turn down the 24V Meanwells used for Sanwu's, (and the Meanwells require the mute switch on Sanwu's because stuttering on off on off on off on off power off behavior). I didn't know it was because higher 2nd harmonic/better 2nd/3rd harmonic balance, lowering the voltage, I did change/experiment with psu distance from ampboards because it did change sound, again 2nd harmonic or 2nd and 3th harmonic balance, I was told. So there are several ways to influence 2nd harmonic by PSU 🙂
It seems we are again in troubled waters with other boards mentioned and no type number/value of a replacement inductor/coil. Please clear info what brand and type of inductor to use.
You mentioned the Sanwu being way better of difficult loads and now you state that you use Dugs that you mentioned to trip at 5V output ?!?! Sanwu would put out 11V. You speakers seem to be a difficult load so using Sanwu would make sense in a way or not ?! Anyway let's leave other boards and Dugs to be discussed in another thread for now.
Last edited:
Following jfetters experiences:
* Preliminary conclusion rev. 1.1
The original Sanwu TPA3118 mono boards :
- seem quite good PCBs as they deliver more power probably because of optimal layout
- need gain reduction to 20 dB for best results.
- don't trip "early" like any other TPA3118 board (even well designed known good ones)
- sound very good but sound improves considerably with better inductors (which ones ? JP)
- sound with different inductors improves but measurements should prove this
- need remeasuring when supposedly better inductors are used for final conclusion
- need the series diode in the + line to be removed and bridged by a jumper wire
- optimal PSU voltage seems to be 19V or 24V with unclear info regarding wire length
- might benefit from changing input caps, any information is welcome
To do:
- make list of possible inductor replacements as there is no clear recommendation
- remeasuring with chosen replacement inductors, doctormord seems to pick up this task voluntarily 😉
- test different input capacitors
- discover exact "sweet spot" in PSU voltage
- write final conclusion, add pictures and make a PDF of it
* Preliminary conclusion rev. 1.1
The original Sanwu TPA3118 mono boards :
- seem quite good PCBs as they deliver more power probably because of optimal layout
- need gain reduction to 20 dB for best results.
- don't trip "early" like any other TPA3118 board (even well designed known good ones)
- sound very good but sound improves considerably with better inductors (which ones ? JP)
- sound with different inductors improves but measurements should prove this
- need remeasuring when supposedly better inductors are used for final conclusion
- need the series diode in the + line to be removed and bridged by a jumper wire
- optimal PSU voltage seems to be 19V or 24V with unclear info regarding wire length
- might benefit from changing input caps, any information is welcome
To do:
- make list of possible inductor replacements as there is no clear recommendation
- remeasuring with chosen replacement inductors, doctormord seems to pick up this task voluntarily 😉
- test different input capacitors
- discover exact "sweet spot" in PSU voltage
- write final conclusion, add pictures and make a PDF of it
Last edited:
Sanwu functions when other stop functioning. You could conclude it is better, that doesn't mean it might sound more horrible, distort more.
The 30A+ inductors on Dugs together could be glued to bottomside Sanwu, might just fit, wires to pads if you like. I don't like that so limited by 10x10 size Sanwu topside.
I will browse Mouser/Digikey invoices later to look at model numbers other than HCI.
HCI btw was 16uH, those Sanwu's are used with a "normal" Kef speaker. Not ideal, dropping inductancegraph, but much better than standard.
The 30A+ inductors on Dugs together could be glued to bottomside Sanwu, might just fit, wires to pads if you like. I don't like that so limited by 10x10 size Sanwu topside.
I will browse Mouser/Digikey invoices later to look at model numbers other than HCI.
HCI btw was 16uH, those Sanwu's are used with a "normal" Kef speaker. Not ideal, dropping inductancegraph, but much better than standard.