D
Deleted member 148505
In the datasheet, SER1512-103MED change in inductance is negligible at 5A and acceptable up to 7A. With rough estimate of 5 amps per channel, the inductor will be linear up to 100W at 4 ohms BTL. (50W per SE ch at 2 ohms). Higher than 100W at 4 ohms BTL, we need some measurements to check if THD performance is still acceptable
TICore260BTL's default inductor is SRP1770C-100M, I will post some THD measurements.
Yes it's compatible with PQ2617BHA-220K and SER2918H-223K.
Though the default soldered inductor will be T130-2 micrometals which is also a linear inductor.
Attachments
I found a new TPA3255 Hi-End board:
RTR TPA3255 Amplificateur de reference classe D | Etsy
Its' very expensive.. but great implementation with coilcraft 2918H... can you do better for less money ?
RTR TPA3255 Amplificateur de reference classe D | Etsy
Its' very expensive.. but great implementation with coilcraft 2918H... can you do better for less money ?
Last edited:
D
Deleted member 148505
I think its only advantage compared to my prototype is the linear post regulator as well as configurability.
I can finish my prototype next week if I receive the parts early. While testing the THD performance, I will be putting 1D17A as output inductor which has a high saturation current and is nearly identical to Sagami 7G17A-100M. I think on TPA3255, putting a larger inductor than 7G17A is overkill, it's like putting a large refrigerator on your bedroom. I am expecting my prototype to perform equally to the EVM board. If LM317 proves to be the bottleneck, I will rerun the test with a better post linear regulator.
Advantage of my implementation is larger output filter capacitor and zobel network. Heatsink profile is compatible with the one used on TPA3255EVM. (ATS-TI10P-519-C1-R3) PCB traces are much simpler and 15V, 12V and 3.3V traces going to the other side of the PCB are of very low current (separated by > 3K ohms resistors on each side with bypass caps) so it won't radiate noise on other components.
On my production version, the PCB can accept balanced signal input, Inductors will be 1D17A, and I will replace the linear post regulator with LM2940, but I'm seriously considering TLS205B0EJV (< 30uV RMS noise) which is only around 1.3USD.
Happy to read that Lester... better regulator make better sound !
I read the TI PDF for LC filter design, interesting... i know how 360customs make is table... he reports the graph value
If the Wurth 744363 is the best tested by TI the Coilcraft SER2918H/Bourns PQ2617BHA was even better with lower DCR = lower core loss ... you think it's not preferable to put a 7µH maybe ?
I read the TI PDF for LC filter design, interesting... i know how 360customs make is table... he reports the graph value
If the Wurth 744363 is the best tested by TI the Coilcraft SER2918H/Bourns PQ2617BHA was even better with lower DCR = lower core loss ... you think it's not preferable to put a 7µH maybe ?
Last edited:
D
Deleted member 148505
Happy to read that Lester... better regulator make better sound !
I read the TI PDF for LC filter design, interesting... i know how 360customs make is table... he reports the graph value
If the Wurth 744363 is the best tested by TI the Coilcraft SER2918H/Bourns PQ2617BHA was even better with lower DCR = lower core loss ... you think it's not preferable to put a 7µH maybe ?
Yes you can use 7uH if you want to squeeze out the remaining 0.001% THD. Although you might consider a PFFB implementation if you care about THD. Though I think pre-filter feedback has a more organic sound.
Use 10uH if you are running it with a battery and has to stay on with long hours of idling.
SER2918 will be better for TPA3255 since it has a more linear behavior at higher current than Wurth 7443631000.
As you can see in the graph Wurth 7443631000 is only linear up to 8A current, while SER2918 is up to 20A.
Reason I chose 7G17A and the likes is that it also has a high saturation current while having a small footprint. In the graph it is linear up to 20A, much like the SER2918.
On 360customs chart they've only tested 7G14A which is a tiny inductor, if they have tested 7G17A the result would have been better.
That's why I'm excited with my production version implementation since it might surpass the TPA3255EVM on THD performance.
Attachments
D
Deleted member 148505
Currently modifying TICore260BTL ProdV1
Presoldered on the top is an LM2940 linear post regulator (U1 on top PCB)
Upgradable low noise regulator on the bottom - TLS205B0 (UO1 on the bottom PCB)
My prototype TICore260BTL V1.0 has 10pcs PCB (post#414), Currently assembling 2pcs and will scrap the others.
Presoldered on the top is an LM2940 linear post regulator (U1 on top PCB)
Upgradable low noise regulator on the bottom - TLS205B0 (UO1 on the bottom PCB)
My prototype TICore260BTL V1.0 has 10pcs PCB (post#414), Currently assembling 2pcs and will scrap the others.
Attachments
D
Deleted member 148505
Yes 7G17B and 7G17D are better, they have a higher Irms rating.
Since we need cheaper inductor, I will try ICE 1D17A-100M for testing, I suppose it's equivalent to 7G17A.
If I will use SER2918/PQ2617BHA, the PCB will become larger, hence manufacturing and shipping cost will become more expensive. Rough estimate is around 10USD per board will be added. I'm not sure if using a larger inductor is justifiable, I need to test the THD of 1D17A first. Otherwise we are just adding bulk and cost to the board.
D
Deleted member 148505
I don't know if you have already read that :
TI TPA3255 EVM
– QuantAsylum
You can directly compare you're version
Another thing, read that concerning OPA2156 (big brother of OPA1656, CMOS type):
OPA2156: HiFi in CMOS
– QuantAsylum
Please consider to replace NE5532 by OPA1656/2156, it's not very expensive...(around 3$)
OPA1612 is already better than 5532 and it's beated by OPA2156:
TI TPA3255 EVM
– QuantAsylum
You can directly compare you're version
Another thing, read that concerning OPA2156 (big brother of OPA1656, CMOS type):
OPA2156: HiFi in CMOS
– QuantAsylum
Please consider to replace NE5532 by OPA1656/2156, it's not very expensive...(around 3$)
OPA1612 is already better than 5532 and it's beated by OPA2156:
Last edited:
D
Deleted member 148505
I don't know if you have already read that :
You can directly compare you're version
Another thing, read that concerning OPA2156 (big brother of OPA1656, CMOS type):
Please consider to replace NE5532 by OPA1656/2156, it's not very expensive...(around 3$)
OPA1612 is already better than 5532 and it's beated by OPA2156:
Looks like their measurement with 10W-70W at 4 ohms load hovers at around 0.005% THD on EVM. Since they've run it in sweep I can't find the actual numbers for each power output.
We only have QA400 and a manual ranger so in order to do that I have to get THD measurement for each output power and plot them in excel sheet. It's tedious so I will only get THD, IMD and noise floor.
NE5532 will not be a bottleneck on THD+N of TPA3255, though I agree that opamps do make different impact on sound, we will solder them by request.
Attached loopback performance of our QA400 and loopback with interface (we also sell the interface
, it's good for diy power amplifier builders )Also attached production version, I retained provision for high power SMT inductors, it is for PA applications that don't require ultralow THD output. (like for videoke music room etc..)
Attachments
Ok QA400 can't perform a sweep... compare the 1W/10W/100W this is sufficient...
This is not for the lowering the THD... this is for the sound indeed !
I have received OPA1656 yesterday (and other OP-Amp) and i compare it with my DIY DAC ES9038Q2M this week-end (listening with headphones or in-ears).
I tell you wich is the best to put him at the demand when ordering
The QA401 perform a little better but is much more expensive than QA400, i think I should buy one with my friend...
Review and Measurements of QuantAsylum QA401 Audio Analyzer | Audio Science Review (ASR) Forum
Review and Measurements of QuantAsylum QA401 Audio Analyzer | Audio Science Review (ASR) Forum
D
Deleted member 148505
You should get one
Yes, so I deleted some messages, have you received my reply?
D
Deleted member 148505
Ok it works at first power up (can't afford to make a mistake). I just need to lessen the brightness of the LED which is currently running at 3mA.
Voltage supervisor is also good, no pops/clicks on startup.
Can't run it at full power though because parts are not complete yet. Substitute heatsink will arrive this week, then ETA of 1D17A-100M is next week.
Voltage supervisor is also good, no pops/clicks on startup.
Can't run it at full power though because parts are not complete yet. Substitute heatsink will arrive this week, then ETA of 1D17A-100M is next week.
Attachments
- Status
- Not open for further replies.