I'd be interested in seeing crosstalk at 10/100w into 3-4R and the THD performance of the AFE. Beside that, performance seems to be very reasonable. I'd guess the 10uH filter inductors can be omitted and changed for just 10R.
The possible antenna style noise/EMC pickup by the input coupling caps is something to look at. We once had the situation where noise got coupled back into MKS caps at the input. Regarding 3G/4G (UMTS/LTE) induced burst noise everything larger than MLCC is probe to pickup.
The possible antenna style noise/EMC pickup by the input coupling caps is something to look at. We once had the situation where noise got coupled back into MKS caps at the input. Regarding 3G/4G (UMTS/LTE) induced burst noise everything larger than MLCC is probe to pickup.
That is the reason why I built simple PBTL Monoblocks. Later on I shall have a look at crosstalk measurements taken at the original TI EVM. I agree actual crosstalk is pretty good and any improvements are measurerable but beyond audibility. It is just a sport to push the numbers to the limits😉
I removed the CLC filter between the DC-DC step up and the amp. Also changed out the 2.2uF film caps for all 10uF 35v Elna Silmic II's to get the full bass extension. Listening tests now and it sounds great. The removal of the filter did not hurt the noise spectrum. THD at 5.16Vrms and 6.6ohms is now a bit lower at 0.0023%, still H2 dominant.
Attachments
Meanwhile I located my protocol with measurements of original TI EVM. Hope this is of any help.
Voltwide, X or others,
I have 2 of the TI EVM modules for future projects. I have read some of the threads on the forum which usually involve modding. I assume that if one builds the TI EVM as per the data sheets with no modifications, the sound would satisfy an aging man that has lost some of his upper end hearing. It would be nice to have confirmation that this is so, before commencing the build process. Thanks for the help.
I have 2 of the TI EVM modules for future projects. I have read some of the threads on the forum which usually involve modding. I assume that if one builds the TI EVM as per the data sheets with no modifications, the sound would satisfy an aging man that has lost some of his upper end hearing. It would be nice to have confirmation that this is so, before commencing the build process. Thanks for the help.
Hi Kokanee,
The main mod with this implementation is addition of post filter feedback (PFFB). That reduces THD. I don’t think this amp is lacking any highs. I think you will be happy with the TI EVM. I have listened to this amp for a few days now - and really think it sounds great.
The main mod with this implementation is addition of post filter feedback (PFFB). That reduces THD. I don’t think this amp is lacking any highs. I think you will be happy with the TI EVM. I have listened to this amp for a few days now - and really think it sounds great.
I made a 3.3ohm 32w resistor with 16 series parallel 3.3ohm 2W resistors. I set the amp at 8.12Vrms for 20W power. A moderate level that would correspond to pretty loud music playing.
Here is FFT for 20W into 3.3ohms, about 0.0029% THD and still second harmonic dominant but higher orders 7th and 9th are appearing.:
Here is Crosstalk for driving 8.12Vrms into 4ohms, monitoring 3.3ohm load on un-driven channel with audio input on undriven channel disconnected. About -101dB crosstalk measured:
Here is FFT for 20W into 3.3ohms, about 0.0029% THD and still second harmonic dominant but higher orders 7th and 9th are appearing.:
Here is Crosstalk for driving 8.12Vrms into 4ohms, monitoring 3.3ohm load on un-driven channel with audio input on undriven channel disconnected. About -101dB crosstalk measured:
Attachments
Strange thing is that your H3 level at idle channel is higher than at the active channel. There is room for improving H2, maybe 10dB or so...
Great work @xrk971 This is starting to look very good indeed. Whats the update on the GB timelines please? Also PCBs only and /or PCBs with SMD bits stuffed?
Thank you wineds!
I have to get the Gerbers updated by JPS64 with the latest changes noted from the verification build.
1. Pin 4 on N104 needs to be connected to pin 2 not ground.
2. Change analog input stage from solder pad jumpers to pin switchable PCB jumpers.
3. Add silkscreen labels for polarity of output pins - they are not mirror symmetric, and indicate which is A, B, C, D outputs for SE mode and which is A/B and C/D for BTL.
4. Add silkscreen labels for A, B, C, D input connectors as they are flipped left/right due to TPA3255 being mounted on bottom.
5. Add silkscreen labels for DIP switch to show SE/BTL modes, Freq select, etc.
6. Add silkscreen label for reset connector.
7. Add silkscreen labels for output error LED status lights (3.3v, OTL/clip, Error, 12v).
8. Add silkscreen labels for jumper pin select modes for SE, BTL, and PBTL.
9. Add silkscreen labels for PFFB resistors and caps.
In meantime, will take pre-orders from my Etsy shop for GB boards. Boards will be $38ea. Shipping will be usual $5 CONUS, $10 CAN, $15 everywhere else for USPS First Class tracked package. Timeline is probably 10 to 14 days from now for me to receive boards, add another 2 to 3 days for me to ship them out.
Populated SMT parts board will take time to get quote from board house for professional assembly. Pricing TBD, but with typical quantities under 25 boards, might be pricey. BOM cost alone is about $140 for a single board and about $80 for 100 units. So there are savings to be had for larger order. If we can get a 50 board order for fully populated boards price might be ok. Let me and JPS64 check on the costs.
Thanks for all of your interest - this is a great sounding amp and a worthwhile project for DIY if you want something that you can’t even get commercially and has the unique harmonic profile of 2nd order harmonic distortion dominant with circa 0.0025% THD.
I have to get the Gerbers updated by JPS64 with the latest changes noted from the verification build.
1. Pin 4 on N104 needs to be connected to pin 2 not ground.
2. Change analog input stage from solder pad jumpers to pin switchable PCB jumpers.
3. Add silkscreen labels for polarity of output pins - they are not mirror symmetric, and indicate which is A, B, C, D outputs for SE mode and which is A/B and C/D for BTL.
4. Add silkscreen labels for A, B, C, D input connectors as they are flipped left/right due to TPA3255 being mounted on bottom.
5. Add silkscreen labels for DIP switch to show SE/BTL modes, Freq select, etc.
6. Add silkscreen label for reset connector.
7. Add silkscreen labels for output error LED status lights (3.3v, OTL/clip, Error, 12v).
8. Add silkscreen labels for jumper pin select modes for SE, BTL, and PBTL.
9. Add silkscreen labels for PFFB resistors and caps.
In meantime, will take pre-orders from my Etsy shop for GB boards. Boards will be $38ea. Shipping will be usual $5 CONUS, $10 CAN, $15 everywhere else for USPS First Class tracked package. Timeline is probably 10 to 14 days from now for me to receive boards, add another 2 to 3 days for me to ship them out.
Populated SMT parts board will take time to get quote from board house for professional assembly. Pricing TBD, but with typical quantities under 25 boards, might be pricey. BOM cost alone is about $140 for a single board and about $80 for 100 units. So there are savings to be had for larger order. If we can get a 50 board order for fully populated boards price might be ok. Let me and JPS64 check on the costs.
Thanks for all of your interest - this is a great sounding amp and a worthwhile project for DIY if you want something that you can’t even get commercially and has the unique harmonic profile of 2nd order harmonic distortion dominant with circa 0.0025% THD.
Last edited:
Thanks. I may order a board form you. One last concern I have is replicating your performance. A 12v battery and step up is not really feasible for me. Are you also going to test with a 48V SMPS (eg meanwell etc)?
I have tested with a 48v SMPS from Connexelectronic and it has quite a bit of added noise. I can hear the hum and the hiss. Amp won't be silent that's for sure. But sure, let me see what I can do for an FFT to show you what I mean.
The battery option is less expensive than the 800w 48v Connexelectronic ($150 w/ shipping). $70 for 35Ahr Gelcell and $15 for DC-DC step up, and $15 1A automatic trickle charger - which as far as I can see, adds no noise if left connected full time. You will have a very powerful portable amp with this option and not deal with any 120vac mains. If left connected with the trickle charger in place, and you never run the amp more than 5 hrs continuously at 120wrms, you should be fine with it acting as if it was a "wall mains powered" amp.
The battery option is less expensive than the 800w 48v Connexelectronic ($150 w/ shipping). $70 for 35Ahr Gelcell and $15 for DC-DC step up, and $15 1A automatic trickle charger - which as far as I can see, adds no noise if left connected full time. You will have a very powerful portable amp with this option and not deal with any 120vac mains. If left connected with the trickle charger in place, and you never run the amp more than 5 hrs continuously at 120wrms, you should be fine with it acting as if it was a "wall mains powered" amp.
Last edited:
Pre-orders for the production version of this amp PCB are now being taken:
TPA3255 Reference Class D amplifier PCB | Etsy
TPA3255 Reference Class D amplifier PCB | Etsy