You're using the SMD ferrites to protect against RF pick-up at the input? I doubt they'd be very effective as the highest impedance I've ever seen a FB reach is around 1kohm. When the input impedance is of the order of 50k. Maybe FBs in combination with some shunt capacitance (SMT ceramic of course, not foil) but then its important to watch out for resonances.
Then I guess replace the +ve smd coupling caps with some small value R as you have done - 470R iirc? Then off board coupling caps. I seem to remember you added the R on the IC pin but having done that kind of thing myself, it's tricky to do without damaging either the pin or the PCB pad, or both, and from Taita questions, I'm guessing that might be a little daunting. How developed are your skills Taita?
Last edited:
TDA8932 output filter
The initial chokes are replaced by 22uH toroidal chokes. The 680nF ceramic SMD capacitors are replaced by 100nF SMD high-voltage capacitors (bigger / for best HF decoupling). In parallel with the 100nF high-voltage capacitors there are a 470nF foil capacitor (yellow) and a 100nF foil capacitor (blue). In total 470nF+100nF+100nF=670nF. The foil capacitors are arranged external of the board.
The initial chokes are replaced by 22uH toroidal chokes. The 680nF ceramic SMD capacitors are replaced by 100nF SMD high-voltage capacitors (bigger / for best HF decoupling). In parallel with the 100nF high-voltage capacitors there are a 470nF foil capacitor (yellow) and a 100nF foil capacitor (blue). In total 470nF+100nF+100nF=670nF. The foil capacitors are arranged external of the board.
Last edited:
An input (signal) transformer is something I haven't seen for long in solid state amplifiers.
Just to understand your suggestion in detail, the center-tap of the input transformer secondary then connects to the TDA8932 signal reference (pin 12) and each of the other input transformer secondary ends to the two TDA8932 signal inputs (pin 2 and pin 3) without input coupling capacitors?
A transformer doesn't transform DC voltage. That means, you could theoretically leave them out. The input DC could theoretically increase the distortion because it could drive the transformer into saturation but in praxis even on the small audio transformers hardly any source can deliver a high enough current for that this would happen.
Me neither, though plenty of DIYers are using input trafos on their classD amps.
Yes, that is correct. I prefer to increase the decoupling capacitance on pin12 (470uF would be my minimum). I also use 470R resistors between the transformer and the pins to help reduce the effect of any residual RF pick-up.
I recall I got an improvement going up to 1000uF but that was before I discovered the huge benefits from cleaning up the analog rail with its own regulator. Its not so much the corner frequency but the impedance it presents to the CT of the trafo - I prefer to keep that as low as practically possible. But I've not done any sims to see what happens with higher impedance. Recall that with the 100nF in the DS the pin isn't being loaded externally in any way.
I designed mine for a specific purpose (cheaper, smaller size, optimized turns ratio) but you could use something like this : ????600?:10K/15K?????????????????-???
The '600' ohm side connects to the TDA8932 and you drive the 10k tap as your input. Note it has about 4X voltage reduction input to output. Not quite optimal but probably close enough - means you'll get an overall gain from the amp+trafo of 24dB in BTL mode.
The '600' ohm side connects to the TDA8932 and you drive the 10k tap as your input. Note it has about 4X voltage reduction input to output. Not quite optimal but probably close enough - means you'll get an overall gain from the amp+trafo of 24dB in BTL mode.
abraxalito, thanks for this information. I have a pair of Sanwu TDA8932BT boards (and a pair of Sanwu TPA3118D2 boards) which I am soon to play with.
And the output stage of my ES9018 DAC is a pair of Lundahl LL1684 transformers, with a centre tap on the secondaries available, so I'm in a good position to try this out.
A couple of questions, please:
Q1. On my Sanwu board pin 12 goes to GND via a capacitor. I'm guessing this capacitor may be 100nF as per the datasheet. You said to increase the decoupling capacitance to at least 470uF, but does the (newly increased) capacitor remain connected to ground, while I connect my centre-tap to pin 12?
Or do I lift this ground connection, and connect my centre-tap to pin 12 via the 470uF cap?
Apologies if this is a stupid question.
Q2. The TPA3118 has no such input reference connection, correct? If I use the TPA3118 in differential input mode, there is no use for the centre-tap?
Thanks
Yes that's right, the new 'lytic goes in parallel with the old 100nF. The voltage on pin12 is very low (from memory I think no more than 3V) so a 6.3V cap will do fine.
I'm not intimately familiar with the TPA3118 but I think you're correct. I will re-visit the DS though to double check.
Thanks abraxalito.
My transformer output stage has a jumper which configures it either for unbalanced output, or balanced. I have always used the unbalanced configuration to feed my TA2020 amp.
But interestingly, as I check the schematic -
http://www.diyaudio.com/forums/quanghao-audio-design/246579-super-converted-ll1684.html#post3718605
I see that in balanced configuration, the centre-tap is connected to the shield/ground (pin 1) of the XLR output connector. With the TPA3118 in differential input mode, I presume I can connect this to the TPA3118's GND?
My transformer output stage has a jumper which configures it either for unbalanced output, or balanced. I have always used the unbalanced configuration to feed my TA2020 amp.
But interestingly, as I check the schematic -
http://www.diyaudio.com/forums/quanghao-audio-design/246579-super-converted-ll1684.html#post3718605
I see that in balanced configuration, the centre-tap is connected to the shield/ground (pin 1) of the XLR output connector. With the TPA3118 in differential input mode, I presume I can connect this to the TPA3118's GND?
Thanks. You have reinforced what I already understood to be the case regarding transformer balanced outputs.
Nonetheless, I see various references to centre-tap-to-ground, such as here -
Basic Audio
and here -
AMB Laboratories DIY Audio • View topic - Unbalanced to Balanced converter for Beta22
Nonetheless, I see various references to centre-tap-to-ground, such as here -
Basic Audio
and here -
AMB Laboratories DIY Audio • View topic - Unbalanced to Balanced converter for Beta22
Ah, I was always uncertain about this. Back in 2010 I assembled my TA2020 amp board and Paul Hynes regulator board, and just left the DC power GND floating. Now I think I should earth it.
I have one, sounds very good, the only downside is the female voice at full volume when power on or connecting to the amp. I will try to add an pot between Bluetooth and and chip.
Well I just earthed the DC ground of my TA2020 amp's 13.5V power supply. With my old CD player as input, all working fine.
Then I re-connected the inputs to my music server and ES9018 DAC - oops, tons of hum. Clearly ground/earth is the culprit, and I know that my DAC's transformer output has its secondary GND's earthed via its XLR jack-to-chassis mounting screws. Removing these 2 screws disables the chassis earth connection, and the (unbalanced) signal GND is then floating ...
WOW. Not only is the hum gone, the music has opened up, especially the treble.
I'm not sure if this solution is universally applicable, but it was the magic bullet for me; leave source device signal/power GND floating, earth signal/power GND at the amp.