ok so ive had a look at the pcb layout.. it seems that to populate the PFFB stuff, id need to use surface mount components..?! ive never even considered hand mounting SMD stuff before. does it require lots of equpment?
Mounting and soldering some SMD stuff is very difficult. Some is impossible without some kind of robotic machine. The zobel across the speaker is a good idea in any case. It can't do harm with the values I mentioned, and the output reconstruction filters are based on a particular impedance.
There's no way for them to predict how high the impedance of a speaker will go, and at what frequencies, so I believe they pretend it's always the nominal value when they design those filters, which it aint once you go above the audio freq. range.
Another thing is this: If the amp is rated at 20watts for example, a 16 ohm load may only get 5 watts, 8 ohms usually gets about 10 watts, and only 4 ohms will get the full rated power of 20 watts. Those are approximate typical numbers. The trouble with paralleling a 16 ohm 10W resistor onto the 16 ohm speaker load is that it will waste half the power. You might consider connecting the speakers in parallel instead, if the amp can handle 4 ohms well. Then you'll get closer to the full power potential of the amp.
I highly recommend checking the operation of any switch mode amp on an oscilloscope with a sinewave generator and an 8ohm (4 - 10ohm) load resistor. I bought 3 different class D amps from Parts Express, each had all these great reviews, and 2 of the 3 looked horrible on the scope. One oscillated at about 3HZ when pushed to about half power. Another had ultra high frequency "bubble" (spurious) oscillations riding on the 1kHZ sine wave when pushed to about half power, and the 3rd one clipped fairly cleanly, which is an indication of stability, and no oscillations at all at any level. The good one used a 3116 chip, but so did one of the bad ones, so I wouldn't just assume that any highly rated switch mode amp must be good because of good reviews or whatever.
Another piece of advice is to get an amp that is rated at twice the power you want, and then run it on a lower power supply voltage than the recommended one (especially of the speaker load will be 4 ohms), so it's less likely to ever over heat and possibly blow your speakers when it blows up. My 3116 chip can run on anything from about 5 volts to 25 volts, so your chip may have that a similar range of power supply voltage flexibility (check the spec for the chip you have). I'm running my good class D amps on four 18650 lithium batteries in series, so about 14.4VDC. The heatsinks on my 3116 boards are so small, I'm afraid to run it at 25VDC. It turns out that about 10 watts is plenty for my little boombox project.
There's no way for them to predict how high the impedance of a speaker will go, and at what frequencies, so I believe they pretend it's always the nominal value when they design those filters, which it aint once you go above the audio freq. range.
Another thing is this: If the amp is rated at 20watts for example, a 16 ohm load may only get 5 watts, 8 ohms usually gets about 10 watts, and only 4 ohms will get the full rated power of 20 watts. Those are approximate typical numbers. The trouble with paralleling a 16 ohm 10W resistor onto the 16 ohm speaker load is that it will waste half the power. You might consider connecting the speakers in parallel instead, if the amp can handle 4 ohms well. Then you'll get closer to the full power potential of the amp.
I highly recommend checking the operation of any switch mode amp on an oscilloscope with a sinewave generator and an 8ohm (4 - 10ohm) load resistor. I bought 3 different class D amps from Parts Express, each had all these great reviews, and 2 of the 3 looked horrible on the scope. One oscillated at about 3HZ when pushed to about half power. Another had ultra high frequency "bubble" (spurious) oscillations riding on the 1kHZ sine wave when pushed to about half power, and the 3rd one clipped fairly cleanly, which is an indication of stability, and no oscillations at all at any level. The good one used a 3116 chip, but so did one of the bad ones, so I wouldn't just assume that any highly rated switch mode amp must be good because of good reviews or whatever.
Another piece of advice is to get an amp that is rated at twice the power you want, and then run it on a lower power supply voltage than the recommended one (especially of the speaker load will be 4 ohms), so it's less likely to ever over heat and possibly blow your speakers when it blows up. My 3116 chip can run on anything from about 5 volts to 25 volts, so your chip may have that a similar range of power supply voltage flexibility (check the spec for the chip you have). I'm running my good class D amps on four 18650 lithium batteries in series, so about 14.4VDC. The heatsinks on my 3116 boards are so small, I'm afraid to run it at 25VDC. It turns out that about 10 watts is plenty for my little boombox project.
Last edited:
Hi Bob, thanks for the tips.
you re right of course, i could just use the two speakers in parallel and deal with a 4 ohm load instead..
the reasons i wanted a 16 ohm load are
a) id been advised that its preferable to do the speakers in series and use an inductor for the lowpass on one, instead of parallel with a capacitor.
b) the amp is seriously overpowered for the speakers.. its rated at 300wpc into 4 ohms, and the speakers are rated for 20w each!
they will be highpassed before the amp at 130 hz.. according to my sims they willl then need 150w before hitting xmax, but obviously thermal limit will be much much lower. i thought at 16 ohms there is slightly less chance of accidentally destroying the speakers.
the amp will be run at 48v-50v. this is necessary as the same psu will run two of these amps, the other will be driving big fat 4 ohm subwoofers with a 600w rms rating.
- again according to my sims, they will only need 150w each to match the max output of the dual fullrangers.. according to the spec sheet, these amps will put out 260w before distortion becomes unreasonable, so ive got a bit of headroom.
i know its not an ideal setup, but a) i wanted a single psu for compactness and b) the amps are highly regarded and were on offer (50% discount)
i would hope that the TI evaluation boards will be rather high quality as class D amp implementations go, so hopefully i wont have weird issues like you mentioned.
thoughts?
p.s. ive been looking at smd soldering.. considering thse are just resistors and capacitors im fairly confident i could do the soldering.. my eyes are quite good, and my hand isnt too shaky!
i will definitely consider the zobel on the speakers in any case.
you re right of course, i could just use the two speakers in parallel and deal with a 4 ohm load instead..
the reasons i wanted a 16 ohm load are
a) id been advised that its preferable to do the speakers in series and use an inductor for the lowpass on one, instead of parallel with a capacitor.
b) the amp is seriously overpowered for the speakers.. its rated at 300wpc into 4 ohms, and the speakers are rated for 20w each!
they will be highpassed before the amp at 130 hz.. according to my sims they willl then need 150w before hitting xmax, but obviously thermal limit will be much much lower. i thought at 16 ohms there is slightly less chance of accidentally destroying the speakers.
the amp will be run at 48v-50v. this is necessary as the same psu will run two of these amps, the other will be driving big fat 4 ohm subwoofers with a 600w rms rating.
- again according to my sims, they will only need 150w each to match the max output of the dual fullrangers.. according to the spec sheet, these amps will put out 260w before distortion becomes unreasonable, so ive got a bit of headroom.
i know its not an ideal setup, but a) i wanted a single psu for compactness and b) the amps are highly regarded and were on offer (50% discount)
i would hope that the TI evaluation boards will be rather high quality as class D amp implementations go, so hopefully i wont have weird issues like you mentioned.
thoughts?
p.s. ive been looking at smd soldering.. considering thse are just resistors and capacitors im fairly confident i could do the soldering.. my eyes are quite good, and my hand isnt too shaky!
i will definitely consider the zobel on the speakers in any case.
Cz and Rz for pffb are a zobelcapacitor and zobelresistor. Their values 1.33uF and 3.3ohm.
Now plot the single 0.2uF and 20 ohm zobel, does that help in a significant way ? Remember next to the 1.33uF 3.3 ohm the ampboard also already has the 10nF 3.3 ohm zobellike thingy on each output. You need 5 RC's with "zobelfunction" per speaker ?
Now plot the single 0.2uF and 20 ohm zobel, does that help in a significant way ? Remember next to the 1.33uF 3.3 ohm the ampboard also already has the 10nF 3.3 ohm zobellike thingy on each output. You need 5 RC's with "zobelfunction" per speaker ?
i have no idea 🙂
im a building stuff guy, not an electronics expert. i can solder, and im good and design and layout..
I "ridicule" with a questionmark.
Well you found what sizes specced in evm manual are worth, just to caution you further
Don't trust footprint/package size dimensions/identification of pffb components for the 3255evm as specced in evm manual (or app note?). Check befor ordering 🙂
Well you found what sizes specced in evm manual are worth, just to caution you further
Don't trust footprint/package size dimensions/identification of pffb components for the 3255evm as specced in evm manual (or app note?). Check befor ordering 🙂
Got a schematic for that arrangement?
If three components, any components, are in a series circuit, the order doesn't really matter, so... How exactly will one fullrange be high-passed, but the other won't? 😕
Sounds (no pun intended) like you're going for a "1.5-way" arrangement (the ".5" being the 1st-order-low-passed "woofer"). I've had one of those on my drawing board since a few years ago, but from the get-go, i had decided on doing that low-passing before the amps.
If three components, any components, are in a series circuit, the order doesn't really matter, so... How exactly will one fullrange be high-passed, but the other won't? 😕
Sounds (no pun intended) like you're going for a "1.5-way" arrangement (the ".5" being the 1st-order-low-passed "woofer"). I've had one of those on my drawing board since a few years ago, but from the get-go, i had decided on doing that low-passing before the amps.
hi im following the advice of Planet10 (David) who sent me the following:
http://www.planet10-hifi.com/downloads/Dual-Driver-Wiring.pdf
it appears i got my description a bit backwards though.
its series with a capacitor, or parallel with an inductor.
http://www.planet10-hifi.com/downloads/Dual-Driver-Wiring.pdf
it appears i got my description a bit backwards though.
its series with a capacitor, or parallel with an inductor.
i guess if i were to run the EVM for the fullrangers in SE mode, i could do a filter before amplification.. would screw up my space-saving modification though.
Technically, neither of those two arrangements has both drivers in series AND a series cap/inductor as well 😉
I was just thinking that, since you're going active anyway, why shell out even more cash for a bulky inductor as well?
PS: The filter doesn't need to be active, y'know 🙂 Since it's a 1st-order only, that can be easily achieved with an appropriately-sized RC low-pass (series resistor, cap to ground).
Low Pass Filter Calculator
I was just thinking that, since you're going active anyway, why shell out even more cash for a bulky inductor as well?
PS: The filter doesn't need to be active, y'know 🙂 Since it's a 1st-order only, that can be easily achieved with an appropriately-sized RC low-pass (series resistor, cap to ground).
Low Pass Filter Calculator
thanks for that.. that RC calculator.. heck even i can understand it.
apart from component cost, is there any reason to do the lowpass before the amp?
i ask because to run the amp in 4 channel mode will require a redesign.. at the moment im removing the large caps on the board which are used only in single ended operation.. without that mod im not 100% sure itll fit in my case!
i imagine one advantage is running the amps each with a nice standard 8 ohm load.
apart from component cost, is there any reason to do the lowpass before the amp?
i ask because to run the amp in 4 channel mode will require a redesign.. at the moment im removing the large caps on the board which are used only in single ended operation.. without that mod im not 100% sure itll fit in my case!
i imagine one advantage is running the amps each with a nice standard 8 ohm load.
just a query.. in all the documentation ive looked at specs and performance are given for SE (4 channel single ended mode) with 2,3 and 4 ohm loads.
Is there any reason why 8 ohm load would not work with the 3255 in SE mode..? id imagine amp output will be quite modest, but that might just be perfect for my speakers.
Is there any reason why 8 ohm load would not work with the 3255 in SE mode..? id imagine amp output will be quite modest, but that might just be perfect for my speakers.
i guess im going to answer my own question now.. i guess its the same issue as running BTL with 16 ohms?
So wait, let me see if i got this straight - practically, you have a pair of 3-way speakers (despite them being split up into two enclosures)?
Ie. the subwoofer, the full-range fullrange, and the low-passed fullrange on each side, correct?
If that's the case, is anything preventing you from running both EVMs in the 2.1 BTL + 2xSE mode? You'd only need two of the four DC blocking caps on the outputs (per side / per board), which don't even need to be on the PCB itself. All they do is block DC from reaching the speakers themselves, so you could place them off the board, perhaps nearer to the output connectors on the back of the casing(?).
Ie. the subwoofer, the full-range fullrange, and the low-passed fullrange on each side, correct?
If that's the case, is anything preventing you from running both EVMs in the 2.1 BTL + 2xSE mode? You'd only need two of the four DC blocking caps on the outputs (per side / per board), which don't even need to be on the PCB itself. All they do is block DC from reaching the speakers themselves, so you could place them off the board, perhaps nearer to the output connectors on the back of the casing(?).
hm i think you have a pretty good handle on what im trying to do.. yes two fullrange cabinets with a pair of alpair 7's in each.. im thinking of having one lowpassed in each cab, but i can experiment with what sounds good.
then two subwoofers.
ive got two of the EVM's (or will have when they arrive.. with fedex at the moment)
interesting idea to run both boards in 2.1 mode..
also good to know i can relocate the blocking caps.. might save my compact layout.
is there any reason to run both boards in 2.1 mode, rather than one in btl mode and the other in se mode? i guess its quite typical of highend amps to have seperate boards for left and right.
then two subwoofers.
ive got two of the EVM's (or will have when they arrive.. with fedex at the moment)
interesting idea to run both boards in 2.1 mode..
also good to know i can relocate the blocking caps.. might save my compact layout.
is there any reason to run both boards in 2.1 mode, rather than one in btl mode and the other in se mode? i guess its quite typical of highend amps to have seperate boards for left and right.
My yet-to-be-completed 1.5-way project uses a pair of HiVi B3N's, with one of them low-passed. You want that, on one hand, in order to not have to run a high-shelf cut to compensate for the baffle-step, and on the other hand, so the mids and highs don't end up being comb-filtered due to the distance between the centers of the two drivers (which is that much larger on the Alpairs).
Technical reasons? Perhaps not (although i invite wiser minds to comment), but i personally have a bit of a thing for symmetry, so there's that 🙂
The "high-end amps" argument is arguable - if anything, it would make sense from the point of view of "economies of scale". 1000pcs of one item will likely end up costing you less than 500pcs of two different models 😉 Not to mention the increased ease of assembly & implementation.
Technical reasons? Perhaps not (although i invite wiser minds to comment), but i personally have a bit of a thing for symmetry, so there's that 🙂
The "high-end amps" argument is arguable - if anything, it would make sense from the point of view of "economies of scale". 1000pcs of one item will likely end up costing you less than 500pcs of two different models 😉 Not to mention the increased ease of assembly & implementation.
yes they are overlapping quite heavily now.. i apologise!
anyway, a good reason to do as you say.. 2.1 on each board.. much more even heat distribution between the two amps.
anyway, a good reason to do as you say.. 2.1 on each board.. much more even heat distribution between the two amps.
ok so.. 2.1 seems a good way to go. im sold!
i assume i will need to do the PFFB thing to allow the Single ended outputs to drive 8 ohm speakers? in all the documentation it lists 2-4 ohm speakers for SE operation.
i assume i will need to do the PFFB thing to allow the Single ended outputs to drive 8 ohm speakers? in all the documentation it lists 2-4 ohm speakers for SE operation.
It (only) lists 2-4 ohms because that's where the output power numbers are bigger 😉
Low voltage, high current devices are easier (and thus, cheaper) to make than medium/high-voltage devices, especially in such small and tightly-integrated packages.
Low voltage, high current devices are easier (and thus, cheaper) to make than medium/high-voltage devices, especially in such small and tightly-integrated packages.
- Status
- Not open for further replies.