Because there is a clear market demand for a well-designed unregulated supply. Many have requested a soft start board to go with the Power-86. Others have requested a protection board. In reality the two need to be designed together. Now's the time for me to deliver on my promises.
Also, I am not able to find an SMPS that's suited for a stereo build. You'd need four Mean Well RPS-400-36 for a stereo build - two per channel.
No such supply exists.
Exactly!
Tom
BTW: In the first few pages of this thread, we discussed the pros and cons of the isolated vs non-isolated package options of the LM3886. So I replaced the metal backed LM3886T in my prototype with the isolated LM3886TF.
With the (2x) Mean Well SE-600-36 ±36 V supply and a heat sink specified at 0.4 K/W, I'm getting 380 W into 4 Ω at the onset of clipping with the LM3886T and 340 W into 4 Ω at the onset of clipping with the LM3886TF. Both of these measurements were done on a cold heat sink.
The LM3886TF starts to overheat once the heat sink temperature exceeds 62 ºC. The front of the IC package measures about 130 ºC under those conditions. With the LM3886T, the front of the package measures about 95 ºC under similar conditions.
Interestingly, switching to the Power-86 and Antek AN-5225 power supply allows the amp to deliver 130 W into 4 Ω without overheating even with the LM3886TF. The rail voltage droops to ±32 V and this lowers the dissipation in the LM3886es enough that the chips run without overheating.
So the bottom line is that you are able to get a little bit (0.5 dB) more power with the metal back LM3886T than with the isolated LM3886TF. The LM3886T is more finicky to mount. You really need to make sure that you do not have a burr or metal filing shorting one of the ICs to the heat sink. Failing this would likely destroy the board and certainly the LM3886. It's also a little more expensive to use the LM3886T (about $6 in thermal pads and mounting hardware).
My recommendation to use the LM3886TF still stands. If you are using an unregulated supply, you will not benefit from using the LM3886T. To get the full benefit of the LM3886T, you have to use a regulated supply.
I will leave the LM3886T as an advanced option for those who are fully confident in their ability to drill and tap a heat sink without leaving burrs and who are willing to take the additional risk.
Alternatively, those who really want the LM3886T should consider using the mounting bracket I'm having made. The brackets will have studs that allow for the attachment of the LM3886es and standoffs for attaching the board. Then you just bolt the bracket to the heat sink with a coat of thermal grease (or phase change goop) in between. These brackets aren't cheap, but they should be pretty nice.
Note that the 0.4 K/W heat sink I use for testing is a bit smaller than what I've been recommending here. It's the largest heat sink I have...
Tom
With the (2x) Mean Well SE-600-36 ±36 V supply and a heat sink specified at 0.4 K/W, I'm getting 380 W into 4 Ω at the onset of clipping with the LM3886T and 340 W into 4 Ω at the onset of clipping with the LM3886TF. Both of these measurements were done on a cold heat sink.
The LM3886TF starts to overheat once the heat sink temperature exceeds 62 ºC. The front of the IC package measures about 130 ºC under those conditions. With the LM3886T, the front of the package measures about 95 ºC under similar conditions.
Interestingly, switching to the Power-86 and Antek AN-5225 power supply allows the amp to deliver 130 W into 4 Ω without overheating even with the LM3886TF. The rail voltage droops to ±32 V and this lowers the dissipation in the LM3886es enough that the chips run without overheating.
So the bottom line is that you are able to get a little bit (0.5 dB) more power with the metal back LM3886T than with the isolated LM3886TF. The LM3886T is more finicky to mount. You really need to make sure that you do not have a burr or metal filing shorting one of the ICs to the heat sink. Failing this would likely destroy the board and certainly the LM3886. It's also a little more expensive to use the LM3886T (about $6 in thermal pads and mounting hardware).
My recommendation to use the LM3886TF still stands. If you are using an unregulated supply, you will not benefit from using the LM3886T. To get the full benefit of the LM3886T, you have to use a regulated supply.
I will leave the LM3886T as an advanced option for those who are fully confident in their ability to drill and tap a heat sink without leaving burrs and who are willing to take the additional risk.
Alternatively, those who really want the LM3886T should consider using the mounting bracket I'm having made. The brackets will have studs that allow for the attachment of the LM3886es and standoffs for attaching the board. Then you just bolt the bracket to the heat sink with a coat of thermal grease (or phase change goop) in between. These brackets aren't cheap, but they should be pretty nice.
Note that the 0.4 K/W heat sink I use for testing is a bit smaller than what I've been recommending here. It's the largest heat sink I have...
Tom
Attachments
Last edited:
I definitely agree.
What about selling the mounting brackets, maybe pre-mounted to the 3886?
When you have an impedance of 4 ohms each half of the mod 686 will see 2 ohm impedance, each lm3886 will see 6 ohms. For 8 ohms impedance lm3886 will be happy with 37v max each rail(dc) , for 4 ohms lm3886 will be happy with 28v max each rail(dc) so my logic tells me for 6 ohms lm3886 will be happy with max of 32.5v dc each rail. And those 4 ohms will not be always 4 ohms sometimes it could sink alll the way to 2 ohms , most of the 8ohms speaker are sinking to 4ohms , 4 ohms impedance could be actually 2.5 ohms resistance. Is my logic correct?
The 3xlm338 regulator would be a great candidate for an trully DIY regulated power supply, it could be adjusted, max in is 38v max out is 32v , perfect for the lowest impedances.
The idea with the premounted brackets is actually good since so many of us are don't have a press bench for all of those threaded holes.
The 3xlm338 regulator would be a great candidate for an trully DIY regulated power supply, it could be adjusted, max in is 38v max out is 32v , perfect for the lowest impedances.
The idea with the premounted brackets is actually good since so many of us are don't have a press bench for all of those threaded holes.
Last edited:
Hi Tom
Thanks for the clarification.
Could there be a typo in the above statement? 130W at 4ohms? Did you perhaps mean 330W with the linear supply using the TF type package?
Thx
SH
To get the LM3886es mounted at the correct height, the mounting procedure is to drop the LM3886es into the board, attach them and the board to the bracket, and solder the LM3886es. The cutouts make it possible to solder the LM3886es after the board has been attached to the bracket.
What about selling the mounting brackets, maybe pre-mounted to the 3886?
I'll be selling the mounting bracket as an add-on option.
See my thoughts here: LM3886 Output Power
I suggest looking at the amount of power dissipated in the LM338s. You'll likely find that to be the limitation of your design. Unless you're into water cooling, a linear regulator is not likely to lead to happiness. With the Modulus-686 you will not need any regulation on the supply anyway. Just design the supply for ±35-36 V at idle and let it droop under load.
Thank you. The only thing I don't like about them is the price. Perhaps I should ask: How much would you expect to pay for such a mounting bracket? I'm just curious to see if I'm reading the market somewhat right. The question is open to anyone interested in building the Modulus-686.
No. I ran the amp at 130 W. The FTC requires that an amp can run at 1/3 its rated output power for an hour without overheating. 130 W is about 1/3 of the rated output power (a bit higher than 1/3 for the unregulated supply and a bit lower for the SMPS). I'm guessing the FTC used this operating point as it corresponds to a crest factor of 10 dB.
The worst case power dissipation is actually at half the rated output power.
As I also pointed out, I'm getting 380 W out at the onset of clipping with the SMPS. I think the number was about 280-300 W with the unregulated supply. Both with the LM3886T.
Tom
Last edited:
Is this problem?
I mean, if it fits the case (which should be the case with the recommended heatsink dimension which alsos call for a full rack width), is there a problem in running 4 SMPS in a case?
Also, while I am at it, does the enclosed "C" version bring anything in term of shielding compared to the bare one?
No problem .... as long as you're willing to buy four of them.
The caged version is nice in that it'll keep your fingers off of the high voltage points where they don't belong. The cage does constrict airflow a little, so Mean Well specs it to be able to deliver the full output power up to 35 ºC ambient vs 40 ºC for the cage free version. You can find that in the data sheet.
It's not a big difference, but if you really want to crank it, the case will get a bit warm inside.
Tom
Last edited:
8.35 x 1.25 inches. 0.081" thickness. Five #4-40 1/4" length standoffs on the bottom. Six #4-40 concealed head studs 3/8" length on the face. The four mounting holes on the face are 4.2 mm in diameter and will work with M4x0.7 mm or #6-32. All hardware is 300-series stainless steel.
Tom
Tom
Last edited:
All you need is a coat of thermal paste or phase change goop on the back of the bracket (between it and the heat sink).
If you'd like to know whether [whatever material] is better, I suggest you take the dimensions of the face of the bracket (8.35 x 1.25 inches) and the mounting pressure from four M4x0.7 (or #6-32) machine screws equally distributed across the face and work through the math. That's not an undertaking I have time for at the moment. I am plenty busy trying to keep up with my education and the development of the Power-686 and protection board. I suppose I should start writing the Modulus-686 documentation too...
Tom
If you'd like to know whether [whatever material] is better, I suggest you take the dimensions of the face of the bracket (8.35 x 1.25 inches) and the mounting pressure from four M4x0.7 (or #6-32) machine screws equally distributed across the face and work through the math. That's not an undertaking I have time for at the moment. I am plenty busy trying to keep up with my education and the development of the Power-686 and protection board. I suppose I should start writing the Modulus-686 documentation too...
Tom
Last edited:
I used some Bergquist K10 pads. You can see the spec sheet here: SIL-PAD K-10 - Thermal Materials by Henkel
Edit: I poked around a bit. It looks like there are all sorts of exciting phase change materials around. Some of these may be better than the commonly used SilPads. The key is that the thermal pad needs to be electrically insulating - not all thermal pads are electrical insulators. For me to recommend a specific pad, it also needs to be available at common electronic component distributors and reasonably priced.
Tom
Edit: I poked around a bit. It looks like there are all sorts of exciting phase change materials around. Some of these may be better than the commonly used SilPads. The key is that the thermal pad needs to be electrically insulating - not all thermal pads are electrical insulators. For me to recommend a specific pad, it also needs to be available at common electronic component distributors and reasonably priced.
Tom
Last edited: