In the other thread, there is the above statement. I have in mind that idea that the power supply is a critical aspect to the project.
Is it cost and/or performance effective to go to a DIY power supply for the 36V provision? Or is it more effective to go to existing module (Mean Well, connexelectronic, LED brick) with if needed on the amp board improvements to reach the needed quality (more caps, more regulation, more...) ?
guys, 2018 and has no 24/192 it is kind of weird to me.. And actually, I see only one JMF11 who interested in IOS/Adroid app for TAS3251 control )) I asked Partsexpress=lepai=dayton(I checked ebay/amazon for small class D amps and found mostly lepai and dayton amps with external power bricks) they also have no interest in it, and looks pretty happy with what they have. Next few days I gonna have met with some OEM about that again, but looks like everybody happy with what they have and no one wants to change this world ))
Last edited:
some Meanwell SMPS I know pretty closely, it is good and cost-effective devices, only need to use multiplier 2 for the rated power if you gonna use it for audio, and check max current ability for speaker impedance. I mean 36VDC and 2 channels 4ohm needs 18A rated SMPS. Obviously, $4/pcs power bricks for LED is the nightmare. No idea about connexelectronic yet. I remember Hypex always had SMPS rated for double instant power, that's good practice.
PS: I took a look for SMPS2000RxE | Connex Electronic It's looks fishy to me, I see no any safety gaps(all semiconductors fixed on the same aluminum bar without proper UL insulation) or proper EMI fighting things (LLC inductor without shielding, with distributed gap material just impossible to pass conducted EMC tests). Also I see the diodes bridge with no heatsik, just on bare PCB )) Anyone here does believe if it is fine for 2000W? I duno, maybe just the worst example and other products are wonderful..
PS: I took a look for SMPS2000RxE | Connex Electronic It's looks fishy to me, I see no any safety gaps(all semiconductors fixed on the same aluminum bar without proper UL insulation) or proper EMI fighting things (LLC inductor without shielding, with distributed gap material just impossible to pass conducted EMC tests). Also I see the diodes bridge with no heatsik, just on bare PCB )) Anyone here does believe if it is fine for 2000W? I duno, maybe just the worst example and other products are wonderful..
Last edited:
Thanks for the feedback IVX. As I'm a bit rusty, could you clarify how you apply your factor 2 for rated power and instant power? Sorry if it is a really dumb question ;-).
Ex: I intend to have 8R loads, 36V => 4.5A - 2 channels => 9A
The MeanWell LRS-350-36 delivers 9.7A =>OK (https://www.meanwell.com/Upload/PDF/LRS-350/LRS-350-SPEC.PDF)
However, this is peak power, not average power. So shall I take:
- that one ?
- half the power to cope with average power
- twice the power to have huge margins
JMF
I was googling on that topic and reminded about LM3886 chip power amplifier power supply design.
It is a reliable source. It is not for SMPS but it introduces the "crest factor" with a statement that 10dB is a sound value, which means that the RMS power is 10 times less than the peak power.
How to use this in our PS sizing ?
You are right that with 8R load, from TAS 3251 datasheet only 2x80W are usable (even less (70W ?) to stay under 1% THD). But this is I understand this "continuous power".
In parallel, I understand that the P and I ratings of meanwell power supplies are RMS values and not peak values (Am I wrong ?).
So If we consider a crest factor of 10 dB we should be able to reduce the spec on the power supply for real life operation, without degradation of the listening results, except on highly compressed music.
Does it makes sense ? By how much would the power supply requirement be divided (any link would be appreciated) ?
Best regards,
JMF
TI has an application note "System Design Considerations for True Digital Audio Power Amplifiers ": http://www.ti.com/lit/an/slaa117a/slaa117a.pdf
It discusses the power supply in section 3. I had only limited time and I have not fully understood all statements. But I understand what I summarize below (based on my 2x80W above example).
2x80W to be considered as max average power. Need to multiply by 2 and add 25% for 2 channels and to take into account the amplifier efficiency => 200W
(not sure there is not a typo in the app note as the efficiency is 85%: should they only add 15% ? however more margins do not hurt).
The app note proposes a factor 2 to move from max average power to max output power
Ptm => 400W
Then they consider a 1/8 factor because of dynamic nature of music (a peak to average power of 20 dB is normal (factor 100)). It is stated to be a good compromize
=> 50W contituous rate
This is a big gap with the 350W continuous discussed above
Then there is a magic statement "In fact, a good compromise is 1/8 of full power. The dashed horizontal line in Figure 4 is the current corresponding to 1/8 of the maximum power output level (6.25-W output power), but the PSU needs to be capable of delivering the peak power required."
How to you ensure that a continuous rate 50W power supply can deliver 400W in peaks ?
Only decouplig capacitors ?
What is the reasonable compromize between 50W and 350W for a 2x80 amplifier ?
Best regards,
JMF
It discusses the power supply in section 3. I had only limited time and I have not fully understood all statements. But I understand what I summarize below (based on my 2x80W above example).
2x80W to be considered as max average power. Need to multiply by 2 and add 25% for 2 channels and to take into account the amplifier efficiency => 200W
(not sure there is not a typo in the app note as the efficiency is 85%: should they only add 15% ? however more margins do not hurt).
The app note proposes a factor 2 to move from max average power to max output power
Ptm => 400W
Then they consider a 1/8 factor because of dynamic nature of music (a peak to average power of 20 dB is normal (factor 100)). It is stated to be a good compromize
=> 50W contituous rate
This is a big gap with the 350W continuous discussed above
Then there is a magic statement "In fact, a good compromise is 1/8 of full power. The dashed horizontal line in Figure 4 is the current corresponding to 1/8 of the maximum power output level (6.25-W output power), but the PSU needs to be capable of delivering the peak power required."
How to you ensure that a continuous rate 50W power supply can deliver 400W in peaks ?
Only decouplig capacitors ?
What is the reasonable compromize between 50W and 350W for a 2x80 amplifier ?
Best regards,
JMF
JMF11, for me priority #1 is HiFi. High Fidelity stands audible range is 20:20000Hz(or maybe 40:16000Hz for old standards), hence HiFi amp have to be able to play that frequencies with tolerance +/-0.5db because 1db is the threshold of noticeability. My HiFi-TOY SMPS provides enough current for 125W@1kHz@THD=1% and 115W@20Hz@THD=1% i.e. within 0.4db. Due to insane compactness of HiFi-TOY I had no chance to put there >8000uF/35V to rail, according to that, I made invertor's semiconductors and LLC inductor over beefy to be ready double instant power(about 500W). The same time, practically I don't need more than 100W AVG power, thanks for crest-factor(but actually the peak factor of heavy compressed club-music sometimes close to 2!) . I can save some BOM cost and product size on it, and I do. For example, I use much higher current density than suppose to be for 500W for the trafo and LLC inductor winding, that makes those parts smaller and cheaper. However, the product has to reliable, and my thermal protection has to watch if the hottest part isn't overheated. The time before HiFi-TOY off by overtemperature protection I chose for couple minutes. Compromise - yes but with a typical music it should never stop yet.
PS: a huge rail caps tank may help weak SMPS, especially if the speaker not intend to play lower than 100:80Hz. This process very easy to simulate, I highly recommend do some spice simulation before adding 100000uF ))
PS: a huge rail caps tank may help weak SMPS, especially if the speaker not intend to play lower than 100:80Hz. This process very easy to simulate, I highly recommend do some spice simulation before adding 100000uF ))
Thanks IVX,
I understand from your explanation that a smps for our audio applications should have potentially different spec than "steady states ones":
- reasonable (to low) continuous rate power (could be as low as 1/8 of max peak power),
- but high peak power capacity.
I imagine that the Mean Well / Led bricks are all designed for continuous rate operation, so the peak power may be unkown. Connexelectronic and Hypex may be more oriented to higher peak power.
DIY should be the other path to suitable design, but I don't have those design skills, and maybe the parts are expensive compared to off the shelf modules.
JMF
I understand from your explanation that a smps for our audio applications should have potentially different spec than "steady states ones":
- reasonable (to low) continuous rate power (could be as low as 1/8 of max peak power),
- but high peak power capacity.
I imagine that the Mean Well / Led bricks are all designed for continuous rate operation, so the peak power may be unkown. Connexelectronic and Hypex may be more oriented to higher peak power.
DIY should be the other path to suitable design, but I don't have those design skills, and maybe the parts are expensive compared to off the shelf modules.
JMF
Last edited:
regarding meanwell and LED's SMPS all is clear from the specs, they promised max current and OCP behavior as well, and usually, it is CC. Hence, 36V loaded for 4ohm (9A) is fine but 2ohm load will reduce rail voltage twice to keep 9A constantly. So max, and AVG power is almost the same because its rail caps are too small to talk about it, and expressed as 36V*9A=324W. Brute-force approach, without an economy around crest-factor
PS: here is my explanation regarding rail caps simulation Class D Amp Photo Gallery
PS: here is my explanation regarding rail caps simulation Class D Amp Photo Gallery
Last edited:
chermann, if your speakers 2x4ohm, yes. The next LRS350 voltage step is 36V 9.7A it's not enough, and BTW, OCP behavior is "Hiccup mode, recovers automatically after fault condition is removed", so SMPS will simply off for 1-1/2S and start again. CC behavior is preferable because music will not stop.
Thanks IVX, doctor explain me the hiccup mode and the regulation will not be better with caps at the amps outpur voltage connectors, because the regulation will be not getting better.
Thanks JM11 --> 50W each channel at home is enough...speakers @ 86-89db and 26 m2 room
easy thinking:
.. for 8 ohms speakers .... 36V/48V (500VA with more Amp)...not at industrial smps -> better hypex and connex
chris
Thanks JM11 --> 50W each channel at home is enough...speakers @ 86-89db and 26 m2 room
easy thinking:
.. for 8 ohms speakers .... 36V/48V (500VA with more Amp)...not at industrial smps -> better hypex and connex
chris
Hi, I'm not sure about the interpretation of your tests. It seems that the 150W, even with capacitor banks is not doing the job. But what is the result with the 350W one ?
Best regards,
JMF