I think you're looking at it the wrong way, irrebeo. A switch mode amplifier should follow the same design rules as a switch mode power supply in that the filter to switching frequency should follow the golden 1:10 rule which is effectively 1:8-12 really. Therefore it is the switching frequency that sets the optimal filter frequency just with the added requirement that the filter frequency should be outside the audible spectrum usually by at least a factor of 2 to avoid phase issues. What you see as improved THD hides the fact that with 27.7khz filter frequency the filter is only linear to about 13.3khz without adding peaking to the filter. So in effect what TI recommends in the application record is that if you're forced to have a low switching frequency due to for example efficiency requirements that in turn requires you to choose a normally too low filter frequency, you can add peaking to this and have it perform reasonably well.
I built a small regulated linear supply on Sunday and tested it out. The transformer is only a small 12v 40VA unit. I built a voltage doubler circuit with some big diodes and some 3000uF 25v caps, followed by a variable LM317 module and a huge 30,000uF/100v tank cap. I tuned it to a very flat 24v.
It sounds excellent, however the transformer runs out of current once the volume gets high enough. I used a DMM to monitor delivered voltage and it is interesting to watch the voltage fluctuate based on current draw.
At "moderate" to "moderately loud" volume, the voltage is maintained between 18v and 22v. If I go louder, I fairly quickly hit a wall where the voltage starts to plummet and continues till the amp powers off. Without the 30,000uF tank, this happens much quicker, and at lower volume. The sound is also much more dynamic and natural-sounding with the tank cap in place.
My local electronics retailer has a bunch of old, used heavy iron so I plan to dig through it some time soon and find something a bit more stout.
Overall, a very interesting, educational hour or two of playing. These class D chips have incredible efficiency, for sure, but this does not mean that they can be starved of current. I do like the sound better with the linear supply - so I am excited to build a stronger one.
It played for the rest of the day at "moderately loud" volume, and sounded very sweet, while I installed our new sliding screen door.
It sounds excellent, however the transformer runs out of current once the volume gets high enough. I used a DMM to monitor delivered voltage and it is interesting to watch the voltage fluctuate based on current draw.
At "moderate" to "moderately loud" volume, the voltage is maintained between 18v and 22v. If I go louder, I fairly quickly hit a wall where the voltage starts to plummet and continues till the amp powers off. Without the 30,000uF tank, this happens much quicker, and at lower volume. The sound is also much more dynamic and natural-sounding with the tank cap in place.
My local electronics retailer has a bunch of old, used heavy iron so I plan to dig through it some time soon and find something a bit more stout.
Overall, a very interesting, educational hour or two of playing. These class D chips have incredible efficiency, for sure, but this does not mean that they can be starved of current. I do like the sound better with the linear supply - so I am excited to build a stronger one.
It played for the rest of the day at "moderately loud" volume, and sounded very sweet, while I installed our new sliding screen door.
Last edited:
Something I should mention: the amp measures the DC supply voltage just before it enables the amplifier, and if the voltage drops to 3/4 of that value (or below a 2nd fixed threshold at 6V) during operation it shuts down/restarts the amp. Rail measurement is done at a 100Hz rate.
Also, this card (and other class D amps) have a negative resistance. If the input voltage drops, they'll draw more current to compensate.
Also, this card (and other class D amps) have a negative resistance. If the input voltage drops, they'll draw more current to compensate.
Gary,
Mine doesn't do that. Initial supply under load is 22.2v and at moderately high volume it dips to about 18v. If I increase volume a bit more, voltage begins to drop and I can maintain it somewhere between 12v and 16v for a while if I am careful with the volume dial. If I go a bit higher with the volume again, the voltage free-falls until the last reading on the DMM is 4.x volts and the amp quits.
Note: I'm not complaining, and I know this is due to a weak transformer and regulator. Just noting my observations.
Mine doesn't do that. Initial supply under load is 22.2v and at moderately high volume it dips to about 18v. If I increase volume a bit more, voltage begins to drop and I can maintain it somewhere between 12v and 16v for a while if I am careful with the volume dial. If I go a bit higher with the volume again, the voltage free-falls until the last reading on the DMM is 4.x volts and the amp quits.
Note: I'm not complaining, and I know this is due to a weak transformer and regulator. Just noting my observations.
Last edited:
It just occurred to me that the limiting factor in my scenario is in fact the LM317, not the transformer. The LM317 can only pass about 1.5A, so I am guessing it is hitting its upper limit. Not an appropriate use of this regulator, I am sure. On the other hand, it certainly fails gracefully when driven too hard. Also, when the LM317 is used within it's limits, the amp certainly sounds excellent and plays loud enough for most practical home situations. My pre-amp allows me to pre-set a "max volume" setting, so it was easy enough to set that so that I (or others in the house) don't drive the regulator into panic mode. In a way, this is a good thing - it forces me to avoid over-driving my speakers.
So, a more "stout" transformer is not going to do anything for me, unless I plan to omit or replace the regulator. I might have to investigate other methods/devices for linear regulation, if I want the ability to go beyond "18" on my pre-amp volume dial.
Man, I love this stuff!
So, a more "stout" transformer is not going to do anything for me, unless I plan to omit or replace the regulator. I might have to investigate other methods/devices for linear regulation, if I want the ability to go beyond "18" on my pre-amp volume dial.
Man, I love this stuff!
Last edited:
still stuck on that linear reg are you?
if so use a lm350 or equiv or even a simple cap multiplier.
did you try without a linear reg first? to listen for any hum?
i doubt you will hear any hum with an unregulated supply. tpa3118 PSRR is too good.
a voltage doubler is not so good of an idea, it adds more ripple to the supply V from a otherwise regular full wave rectifier.
if so use a lm350 or equiv or even a simple cap multiplier.did you try without a linear reg first? to listen for any hum?
i doubt you will hear any hum with an unregulated supply. tpa3118 PSRR is too good.
a voltage doubler is not so good of an idea, it adds more ripple to the supply V from a otherwise regular full wave rectifier.
Last edited:
On another TPA3118 amp I had great results with an AMB sigma-11 linear regulated supply. It trounced several laptop bricks and a small regulated lab supply that I experimented with. Next I want to try the SMPS that matt_g showed. The Wiener deserves a decent PS. Would be great to see what others are using.
BK
BK
My neck is sore from pick/place of just two boards yesterday. I can only imagine the effort that you put in to hand-build a table full of GB units! Let me know if I can help with your reflow plans. I'd do a couple things differently on my next reflow based on some observations.
BK
still stuck on that linear reg are you?
did you try without a linear reg first? to listen for any hum?
i doubt you will hear any hum with an unregulated supply. tpa3118 PSRR is too good.
a voltage doubler is not so good of an idea, it adds more ripple to the supply V from a otherwise regular full wave rectifier.
Yeah. The thought occurred to me to just take the v out from the transformer, through a full-wave bridge rectifier and then into the 30,000uF tank. Voltage would only end up being around 16v, but I wouldn't have a hard current limit of ~1.5A. I might as well try it and see what it sounds like.
Even with the limitations noted above, I like the sound of the linear supply more than the 24v 5A SMPS that I tried first. Somehow, the SMPS sounds "thinner" or "leaner" to me. Maybe I should try the SMPS and throw in the 30,000uF tank and see what happens.
Last edited:
51 cards hand-built in total
Wasn't that bad, I built them in batches of 6 or 8 at a time. TPA first, then test in the bed of nails jig. Then I'd grab a random SMT part from the box of parts and solder that part on all the cards, then grab the next part, solder that on, rinse, repeat. Once all the SMT was done I'd scrub them in isopropyl alcohol to get the flux residue off, then do the through-hole parts, then program and final test.
Reflow looks really attractive as parts placement would be a lot quicker without the soldering part involved, and it looks like the boards come out of reflow looking clean (is an alcohol wash necessary?)
Any advice you've got I'll happily take.
Yeah, that'll burn your wiener.The v doubler ends up pushing ~34v (no load) so I don't think I'll bother trying that.
If you do burn your wiener, list of casualties will be the TPA (30V), 78L05/oscons/electrolytics (35V), and a 78L05 failure could blow the AVR and the Q1 FET. All repairable/replaceable, though replacing the TPA itself will be a pain in the *** as it's soldered to the PCB.
Have a good 4/20 over in BC, bud
Have a good 4/20 over in BC, bud