Haven't run the numbers yet, rule of thumb from an old amp design book I got at home is Rzobel = Re, Czobel = (some equation...)
Fingers crossed there's a sensible numerical result 🙂
Low quantities also cost money - if you buy a reel of 2000 TPA3116's straight from TI they're <$2/pop.
that's a speaker zobel , amp zobels are not the same!
all speaker zobels do is conjugate match Le. this usually done as part of a speakers crossover design. (inside the LPF) the breakpoint is the 3 dB point of the midbass rising impedance due to Le.
generally speaker zobels are around 1-3 KHz for small drivers, where amp zobels are waaay above that freq. >> 40KHz and are done for stability and HF loading. some call these snubbers too.
A speakerzobel will be 18 ohm and 1.3 / 2.1 uF, considering the Feastrex d5nf here I would think components selected at Digikey or Mouser most likely not good ??? 33uH Ice and 0.33 mkp2 for additional amplifierfilter maybe?
speaker zobels are speaker components
usually not necessary on full range unless BSC and/or other networks are used.
amplifiers with low feedback/ poor damping factor are another consideration. see full range section for other opinions.
usually not necessary on full range unless BSC and/or other networks are used.
amplifiers with low feedback/ poor damping factor are another consideration. see full range section for other opinions.
how would you come to that conclusion, unless you like coloration eg aliased switching ( more IMD) folded back in-band.
maybe Gmarsh can give some estimated amplifier damping factor in the mean time. class D filters are very low impedance at audio. ratio at 4/8 ohms > and in irribeo's case 16 ohms should NOT be an issue.
Last edited:
OT:
Well, filterless operation usually "sound" better (less THD) than filtered with high enough inductance from the speaker coil. (To my knowledge and what i'd read from whitepapers and lectures and measured at TPA3132D2)
Last edited:
earlier in this thread you argued that the speaker makes a poor class D filter, now you've sited the white papers say otherwise? or maybe increased rf spray and better sound despite and I don't see how "better IMD" w/o any other filters.
Last edited:
The "problem" is excessive peaking of the outputfilter that is not inside the feedback loop here. TI calls 2dB rise at 20khz "excessive" and below 2dB "acceptable". A 4 ohm filter peaks excessive into 6 ohm resistor load, a 6 ohm filter peaks excessive into a 8 ohm resistorload. Filterless there is no peaking of the nonexistant outputfilter. The Feastrex speaker resembles a ~100 ohm load or maybe even Zmax 160 ohm load to the filter you chose, or if you like a 80 ohm load, now make a filter, amplifieroutputfilter
2 dB peaking at 20KHz sounds like a win using a suitable full range speaker with the correct Box / bass filter alignment. IDK seems like a poor excuse to chuck the whole filter away. If the speaker seems bright with a little peaking, it might be a speaker design issue and no amp except a very bad one can fix it..
the other speaker LF impedance rises depend on the T/S and box tunings, it wont be as bad as the data sheet. besides its within the amps feedback at that point.
the other speaker LF impedance rises depend on the T/S and box tunings, it wont be as bad as the data sheet. besides its within the amps feedback at that point.
Last edited:
Depends on how you measure DF. Zload/Zamp?
At DC, due to negative feedback the OUT nodes of the TPA will be zero ohms, the L will be a 9mohm resistor, and the C will be open circuit, so Zamp = 9mohm. The output effectively drives 2 ohms with a 4 ohm load, so 2/.009 = 222.
At higher frequency, the output impedance of the OUTx node will probably increase, topping out at 120mohm once the amp reaches open loop, so something in between that. Add 9mohm to that, add the inductive reactance of the L, and parallel your result with the capacitive reactance of the C, take the absolute value and you should have a Zamp at that frequency. Hard to say if you can come up with a precise number.
I think that largely depends on the output filter design.
Use an inductor with a pronounced L versus I roll-off, and/or a capacitor with a pronounced C versus V roll-off, and you've got an output filter that effectively changes its parameters with voltage/current swing... this will definitely cause intermodulation.
Once you get it up and running, try your ole AM radio as a easy RFI test set (unless you are fortunate enough to have a Spectrum Analyzer). Hopefully your room full of fluorescent lamps do not swamp the test even before it begins. Then you can try the AM avoidance circuitry to see if it makes any difference and you can play with the filters too.
Testing is not as straight forward as a regular amp. If you are going to measure any audio performance parameters I think it is prudent to have a differential scope probe (isolation) and a balanced input on your THD analyzer. Special filtering can be required. TI does not make any mention as to test setups for performance measurements.
Since the pcb's are cheap, I'll take a couple of them. Once/if you post the bom info I can add the parts to other orders that I have pending. I have a lot of passives already. I try to order over $200 cdn at a time, as we get free shipping.
I will compare against my TPA3100D2 amps that has been running solid for a few years now. I used the Murata 12RS333C inductors for that design.
have you sent the pcb's out for fab? posted a final schematic?
Testing is not as straight forward as a regular amp. If you are going to measure any audio performance parameters I think it is prudent to have a differential scope probe (isolation) and a balanced input on your THD analyzer. Special filtering can be required. TI does not make any mention as to test setups for performance measurements.
Since the pcb's are cheap, I'll take a couple of them. Once/if you post the bom info I can add the parts to other orders that I have pending. I have a lot of passives already. I try to order over $200 cdn at a time, as we get free shipping.
I will compare against my TPA3100D2 amps that has been running solid for a few years now. I used the Murata 12RS333C inductors for that design.
have you sent the pcb's out for fab? posted a final schematic?
Last edited:
I'll post the schematic, gotta do some cleanup on it. I'd like a final 'public' review on it before I commit to the design, just to make sure I haven't made any major mistakes, or left anything out that people might want. And of course I want people to have the schematic, for the sake of troubleshooting/tweaking/whatnot.
Nothing's been ordered yet. Need a headcount before I order PCBs 🙂
Nothing's been ordered yet. Need a headcount before I order PCBs 🙂
I don't get the point you are complaining about. Speaker+Ear is the "filter" for fSW (I.e. 400kHz). What I said/meant is, that the speaker alone is a bad/none filter for HF (I.e. 10-120MHz). ( No surprise as is at the end of the "antenna") Maybe your meaning of RF is not the same meaning of HF to me.
Or are you just want to blame me in any way?
Last edited:
no blame I'm just trying to understand the 'No filter advantage' story is all.
will keep an open mind for improved sound data even though it maybe hard to come by.
I'm still looking at your test results so thank you for those.
will keep an open mind for improved sound data even though it maybe hard to come by.
I'm still looking at your test results so thank you for those.
Last edited:
Filterless IMD lower than with LCfilter, but not really relevant because most times this board will use filter, unless speaker is very close eg.
http://www.ti.com/lit/an/sloa023/sloa023.pdf
General filter advise TI
http://www.ti.com/lit/an/sloa119a/sloa119a.pdf
http://www.ti.com/lit/an/sloa023/sloa023.pdf
General filter advise TI
http://www.ti.com/lit/an/sloa119a/sloa119a.pdf
- Status
- Not open for further replies.