Repeating my comment from earlier, an LC filter with poorly chosen components can cause intermodulation/THD/etc to happen, due to component values changing with voltage/current and potentially affecting amplitude/phase within the audio band. Meanwhile a ferrite bead filter doesn't have any significant amplitude/phase effect at audio frequencies.
So despite potentially creating EMI hell, yeah, a ferrite filter can perform better acoustically than a poorly implemented LC filter.
Gmarsh, Thank you for your time in designing the "proper" TDA3116 PCB. I like that you have decided to go with 3118 and are taking advantage of GAIN/SLV, AM, SDZ, FAULTS.
It looks like you have provided space for nice TH input film caps. What is the lead spacing here? I've been waiting for is a nice compact 3118 board with the option to use differential input that can easily be switched back to single ended input if needed. If I'm looking at your layout correctly, that would be done by omitting R10 and R22 for differential input and using them for single ended, correct?
If so, put me down for two boards.
It looks like you have provided space for nice TH input film caps. What is the lead spacing here? I've been waiting for is a nice compact 3118 board with the option to use differential input that can easily be switched back to single ended input if needed. If I'm looking at your layout correctly, that would be done by omitting R10 and R22 for differential input and using them for single ended, correct?
If so, put me down for two boards.
Nice!
Gmarsh, Thank you for your time in designing the "proper" TDA3116 PCB. I like that you have decided to go with 3118 and are taking advantage of GAIN/SLV, AM, SDZ, FAULTS.
It looks like you have provided space for nice TH input film caps. What is the lead spacing here? I've been waiting for is a nice compact 3118 board with the option to use differential input that can easily be switched back to single ended input if needed. If I'm looking at your layout correctly, that would be done by omitting R10 and R22 for differential input and using them for single ended, correct?
If so, put me down for two boards.
Gmarsh, Thank you for your time in designing the "proper" TDA3116 PCB. I like that you have decided to go with 3118 and are taking advantage of GAIN/SLV, AM, SDZ, FAULTS.
It looks like you have provided space for nice TH input film caps. What is the lead spacing here? I've been waiting for is a nice compact 3118 board with the option to use differential input that can easily be switched back to single ended input if needed. If I'm looking at your layout correctly, that would be done by omitting R10 and R22 for differential input and using them for single ended, correct?
If so, put me down for two boards.
Lead spacing for the input films is 5mm (or 5.08mm), maximum dimensions 7.5x7.5mm. The BOM currently has a 2.2uF/50V film cap there, but you can put in a big X7R, electrolytic, whatever you want really.
Right now each input pin has a 10K resistor to ground and the 2.2uF film to the INx pin of the TPA. I don't really like this from a reliability standpoint (any input ESD will go straight into the TPA) and will probably add some transient protection.
For single-ended input, just connect the common signal ground to input negative of both channels.
Gmarsh, Okay, I might understand now (new to amplifier design). I didn't notice the resistors to ground on all input pins. Are you saying you have resistors there to protect the chip from ESD, whereas the TI schematic only has input caps without resistors to ground?
Sorry about the double post.
Thanks
Sorry about the double post.
Thanks
Input resistors are there to provide a weak DC path to ground on the input. So if you plug in a portable MP3 player, power this amp off a laptop adapter with a 2 pin plug etc... the 10K resistance to ground keeps everything "grounded together, kinda" but with a high enough resistance that shouldn't cause any significant ground loop current flow.
Schematic attached, added transient suppressors to the audio inputs. Excuse the terminal block labeling..
Member
Joined 2006
Count me in for one completed board (maybe more, but only committing to one at this time).
You might want to start a "formal" group buy thread separate from this one... This one can be the design/discussion thread, and the GB thread could be pure business. Just a thought!
You might want to start a "formal" group buy thread separate from this one... This one can be the design/discussion thread, and the GB thread could be pure business. Just a thought!
analog inputs I'd avoid any connections to the power ground if you don't have to, hey that's why they have diff. inputs
Input protection> standard practice is use a couple of schottky diodes at the inputs tied to the analog section supply / bias rails on chip in this case. verify current limit resistors or limit driving impedance. may only need one SOT-23 per SE input depending on input/source impedance.
limit all inputs to 4-5 Vpp absolute max.
Input protection> standard practice is use a couple of schottky diodes at the inputs tied to the analog section supply / bias rails on chip in this case. verify current limit resistors or limit driving impedance. may only need one SOT-23 per SE input depending on input/source impedance.
limit all inputs to 4-5 Vpp absolute max.
Last edited:
Thanks for the review! Going over your notes:
- Input resistors to ground: This is intended to provide a weak DC 'bleed' for situations where there's no DC connection whatsoever between the amp and the source, eg. if you're running the amp off a class 2 power supply, and/or using a portable MP3 player that's not connected to anything. I'll probably increase the value to 30K, even 100K.
- Ferrites: I'll put those in. Think a series ferrite by itself with no shunt component (eg, 100pF NPO) will provide any meaningful HF rejection? Also, any reason I should put them at the TPA pins instead of at the terminal block?
- PBTL mode ground jumpers: I've always said "buy DUG's card" if you want PBTL, but I'll cave. Changing the output to use a pair of inductors instead of 4 will be a fun adventure, but hey, adding the ground jumpers will at least make changing the input side easier 🙂
- D4: deliberately in place as a shunt for negative voltage on the /STBY pin. Which should never happen, but you never know... It's pennies to add, and I don't think there's any negative consequences having it there.
- D2: provides a fast discharge of C30 if PVCC gets pulled, so if it gets immediately reconnected (eg, a loose power connector situation) the startup/unmute timing sequence happens correctly.
- MUTE/SDZ pullups are tied to PVCC instead of the U2 VCC, as the inputs are compliant to PVCC on the TPA. This shouldn't cause any damage to the TPA in this situation but it feels like the right way to do things 🙂
- Moving R34, inverting polarity on SW1: I'll do that. I'll also find another transorb that can withstand 24V (the trip point of the PESD0603) and shove them on the LED and shutdown pins, and replace the audio input ones with it.
I think I'll just add 100K resistors series resistors on all of the TPA control inputs and call it a day. Means shuffling a few parts to fit everything on the PCB.
Mute/SDZ etc. are TTL level with compliance to PVCC. I were worried about pulling the 393 outputs over VCC.
Output voltage rating on LM393 is 36V regardless of Vcc, according to the OnSemi datasheet... which so far is the only datasheet I can find with a stated output voltage limit in the datasheet.
Currently there's a Fairchild LM393 in the BOM, as it has a higher sink current rating able to handle the 10mA status LED current. I might poke a dumb PNP emitter follower in there to allow use of any LM393.
Currently there's a Fairchild LM393 in the BOM, as it has a higher sink current rating able to handle the 10mA status LED current. I might poke a dumb PNP emitter follower in there to allow use of any LM393.
That's the plan for when the design is finalized and the BOM is fixed.
Right now it's in the "nitpick every last possible detail so everyone's happy with it" stage 🙂 Thanks everyone for the comments/suggestions/etc... I very highly appreciate it.
- Status
- Not open for further replies.