Hi Chris,
I may be naive of nature.
Where do you think most of the SMPS in the world are being produced? Right, China. Does that also apply for the "bricks" we all use for our computers? Yes, for most. If you look inside a typical "brick" is it much different than this SMPS? My impression is this board is no worse than any standard SMPS. If you take the 10 EUR and add a cabinet and the in- and output-cables and connectors you end up with the price of a corresponding "brick".
For this board they may have an advantage: it is a DIY product and they may dispense with approvals like for EMI which is an important saving.
With today's SMPS chips most can make a working power supply. The datasheets and application notes guide you in the choice of the components. Respecting the safety distances you are done with the construction.
Should it once go up in smoke, I have only lost 10 EUR.
I have at least 100 SMPS at home and used for different purposes. I can't recall one failing and certainly not causing a safety risk.
Last edited:
I have the 120 and 180W SMPS in 24 and 36 volts. They stay cool under load and sound OK, no noise directly hooked up to D-amps.
They are produced in extreme numbers, so it seems to be a good design. Probably a copy of some non Chinese part.
If you play around with the number of units you order and shipping costs (only use Chinese Ait Mail!), you get them for less than 5 Euros a piece to your doorstep. Just stay under the tax limit, in Europe around 25 € all in. If I need more parts from the same seller, I place several orders a day appart, so you can prevent to get one large parcel and have to pay serious import tax. beware, the Chinese are clever and combine single orders from one day to save costs and dont care about your expenses!
They are produced in extreme numbers, so it seems to be a good design. Probably a copy of some non Chinese part.
If you play around with the number of units you order and shipping costs (only use Chinese Ait Mail!), you get them for less than 5 Euros a piece to your doorstep. Just stay under the tax limit, in Europe around 25 € all in. If I need more parts from the same seller, I place several orders a day appart, so you can prevent to get one large parcel and have to pay serious import tax. beware, the Chinese are clever and combine single orders from one day to save costs and dont care about your expenses!
I thought of using one psu per (mono) amp board, hence the 4-6A version. Space is going to be a problem lol.
Or, budget wise, I should stick to a 8ohm configuration and get 2 stereo boards and 2x smps
Makes sense. Thx
I think the dimension of an amplifier power supply has to be described a bit more for class D and a SMPS.
Even if you play the amp as loud as you can stand, music is never a constant signal. A constant signal would be a light bulb, which permanently consumes 100 watts DC. An 100 W SMPS can deliver such power all day long, if the air flow is OK.
At first sight this would be the near eqivalent to a 2x50 W D-amp with 90% efficency.
But your amp drives a loudspeaker with dynamic sinus waves which means AC. Because of this fact, the capacitors of the 100W SMPS and amp are allways keept fully loaded at nominal level and never brake down on voltage. So overdimensioning has no real advantage. These supplies are even said to have a reserve of 2 A for a short moment, which makes 148 watts.
With "normal" transformer driven power supplies, the more you consume, the lower the voltage. Here a transformer well above the maximum consumption makes (some) sense, as the amps maximum power depends on the PS voltage. Large transformers and large caps improve the sound of amps to some degree, as voltage is keept constant.
This is not right with an SMPS much larger than needed. As long as it is not completely overloaded, the voltage is 99% constant. If the load is exceeded, it simply shuts down. Something no one will see with a 3116 amp and a cheap 100W SMPS. The D-chips features do (should) prevent such massive overload.
One 3116 with a 100W SMPS is perfect and even two driven in PBTL should still work fine if not used as 2 ohm Sub amps driven at 10% distortion.
Even if you play the amp as loud as you can stand, music is never a constant signal. A constant signal would be a light bulb, which permanently consumes 100 watts DC. An 100 W SMPS can deliver such power all day long, if the air flow is OK.
At first sight this would be the near eqivalent to a 2x50 W D-amp with 90% efficency.
But your amp drives a loudspeaker with dynamic sinus waves which means AC. Because of this fact, the capacitors of the 100W SMPS and amp are allways keept fully loaded at nominal level and never brake down on voltage. So overdimensioning has no real advantage. These supplies are even said to have a reserve of 2 A for a short moment, which makes 148 watts.
With "normal" transformer driven power supplies, the more you consume, the lower the voltage. Here a transformer well above the maximum consumption makes (some) sense, as the amps maximum power depends on the PS voltage. Large transformers and large caps improve the sound of amps to some degree, as voltage is keept constant.
This is not right with an SMPS much larger than needed. As long as it is not completely overloaded, the voltage is 99% constant. If the load is exceeded, it simply shuts down. Something no one will see with a 3116 amp and a cheap 100W SMPS. The D-chips features do (should) prevent such massive overload.
One 3116 with a 100W SMPS is perfect and even two driven in PBTL should still work fine if not used as 2 ohm Sub amps driven at 10% distortion.
Last edited:
Oh that's why they state "4-6 amp" !
Given that I am going to work with 4/8ohm speakers I should be safe with one 24v 4(6) Amp SMPS each two amplifiers working in BTL output mode.
With 4 Ohms speakers the maximum output powered by 24V in PBTL can be read from the data sheet. With 1% distortion, which is the most your ears will accept, it should give 2x 60 watts. As this is the highest peak, the average power will stay much lower. This is an easy match for a 4(6) amps SMPS. Typical you will see a power intake of 70 watts for the SMPS while both channels of the amp deliver more than 60 in watt peaks.
This explains why professionel D-amps, rated 2x1000 watt, pull a maximum of well under 1x1000 watt from the mains, even driven into light clipping.
Another factor is the impedance curve of a loudspeaker, that usualy is much higher than its DC resistance. More ohms, less current at the speaker for a given voltage (24 volts).
Have a look at a typical impedance of a loudspeaker, you will not see much of the rated ohms over the frequency range.
To stay save with peaks, I put some larger caps near the D-chip. With the 3116 a 2200- 4700uF 35V cap of good quality (not expensive boutique stuff!) and a 1-2,2 uF film capacitor should give peace of mind. This will keep the amp playing for some seconds when you pull the power cord. An easy test, that tells you there is quite some reserve to help the SMPS under extreme conditions.
This explains why professionel D-amps, rated 2x1000 watt, pull a maximum of well under 1x1000 watt from the mains, even driven into light clipping.
Another factor is the impedance curve of a loudspeaker, that usualy is much higher than its DC resistance. More ohms, less current at the speaker for a given voltage (24 volts).
Have a look at a typical impedance of a loudspeaker, you will not see much of the rated ohms over the frequency range.
To stay save with peaks, I put some larger caps near the D-chip. With the 3116 a 2200- 4700uF 35V cap of good quality (not expensive boutique stuff!) and a 1-2,2 uF film capacitor should give peace of mind. This will keep the amp playing for some seconds when you pull the power cord. An easy test, that tells you there is quite some reserve to help the SMPS under extreme conditions.
Last edited:
I'm ordering soe other stuff from this website, and decided to finally check for some audio components.
Would these 2 be good alternatives for the output filter of this amp?
DTMSS-16/0.022/8.0-V FERYSTER - Inductor: wire DTMSS-16/0.022/8V | TME – Elektronske komponente
MKP01EG468G-B MIFLEX - Capacitor: polypropylene MKP01-0.68U/450 | TME – Elektronske komponente
The capacitor seems OK but the choke not.
Choke value is OK, current rating is acceptable (see curve) but the frequency range is too low. If you look in the datasheet, the choke is meant for EMI suppression up to 300KHz. The carrier frequency for the TPA3116 is normally 400KHZ and may be even higher.
If the choke (core) is not meant for operation at such a high frequency, the inductance will be less than stated (nominal) and the losses high. That will stress the TPA3116, heat up the chip (and choke) and THD increases.
Try looking for another choke that can be used up to 1MHz at least.
Last edited:
Does it matter at all, it I change the R2 and R3 or just remove the R2? Because the resistors are so small I really don't want to solder them back on.
Hi Peroz,
You are right, soldering these small SMD resistors back is a difficult job. I just checked my board and measured (in-circuit) R3 to be 100K. When R2 is removed, R3 serves to pull the pin voltage down to GND. 5.6K is recommended in the datasheet (with R2 open) but 100K may do the job.
Try removing R2 only and if it works, it saves you some struggle.
@peroz: I did this mod;
Use a little solder on the tip when you desolder both ends at the same time (wide flat tip works best) and pick it up with a pair of tweezers or move it with a toothpick.
When soldering them back; use a toothpick to hold the smd resistors down so they do not stick to your soldering iron.
//edit: unscrew the heat sink first to get better access to them and I used a needle to lift and bend out/ shear off leg 16 as it is easier to see then cutting it.
Use a little solder on the tip when you desolder both ends at the same time (wide flat tip works best) and pick it up with a pair of tweezers or move it with a toothpick.
When soldering them back; use a toothpick to hold the smd resistors down so they do not stick to your soldering iron.
//edit: unscrew the heat sink first to get better access to them and I used a needle to lift and bend out/ shear off leg 16 as it is easier to see then cutting it.
Last edited:
I did this mod this morning also. Being a lazy guy, I decided to just remove the R2 and not do all the swap routine 
I also broke the pin 16.
One channel works good, gain is ok, runs cool. The other board's gain is microscopic, though... don't know how to go about it... I didn't try it before, so I don't know if it was faulty before, or I broke it.
Yesterday, however, I managed to start modding my other two boards too fast, and I broke the wrong chip legs. Not proud.
Didn't have any other options than to fix one of the boards today. Soldered a small drop of solder between the chip and the pcb pad. Works a treat, and I have 2 channels again!
Any ideas what might be wrong with the board that can barely be heard?
I also broke the pin 16.
One channel works good, gain is ok, runs cool. The other board's gain is microscopic, though... don't know how to go about it... I didn't try it before, so I don't know if it was faulty before, or I broke it.
Yesterday, however, I managed to start modding my other two boards too fast, and I broke the wrong chip legs. Not proud.
Didn't have any other options than to fix one of the boards today. Soldered a small drop of solder between the chip and the pcb pad. Works a treat, and I have 2 channels again!
Any ideas what might be wrong with the board that can barely be heard?
Then use Table 1 on page 14 of the datasheet for values.
MASTER mode and choose gain. Most choose 20dB for less noise. Small 1/8W resistors can perhaps be used.