Does that mean you did not screw the LM’s themselves but only screwed those MAX 08’s in place?
Tom, what is your take on this solution, looks elegant but would it dissipate equally well?
±30 vs ±32-33ish, but the '686 makes twice the swing from that due to the bridging.
Tom
That's generally how those clips are used.
That really depends. If you are planning to push the power envelope on the MOD686 by running the amp at ±36 V, I strongly recommend that you use the metal backed LM3886T with thermal pads and shoulder washers and screw the ICs to the heat sink before you solder them into the board.
If you're looking for a build that approaches the more "safe and sane" end of things, you can consider using the mounting clips. I do recommend that you attach the boards to the chassis and clip the chips onto the heat sink before you solder the chips.
If you solder first and then try to mount the ICs to the heat sink, you'll nearly always have some of the chips that are at a slight angle such that the back of the chip doesn't make good contact with the heat sink. It usually works OK anyway, but isn't ideal.
This is in part why I made the mounting bracket and offer it for sale, basically at cost. You can find the mounting bracket on my website as one of the Modulus-686 options.
Speaking of Safe & Sane. I measured a few things in the lab:
Code:
V[sub]supply[/sub] P[sub]out[/sub] (8 Ω)
±27 V 130 W
±28 V 140 W
±30 V 160 W
±32 V 180 WI really like the operating point around ±27-30 V. It provides a nice balance between thermals and output power.
Tom
Good option Tom. Mounting brackets should make integrating easier. How do you advise soldering the LM3886s to the PCB - after fixing both PCB and LM3886 to the mounting bracket? I see that there are slits in the bracket under the devices - will it be possible to use the soldering iron through these openings?
According to your product page operating within ±28 V will also allow the use of a 3U chassis - nice.
Neurochrome Modulus-686: 240W(8Ω); 360W(4Ω) @ 0.00025% THD+N
And btw, excellent write up on the product page. Lot of things are explained and gives more confidence to the potential builder. Well done Tom!
Last edited:
Yep. That's how I do it. I get all the screws and nuts tight. Then solder the LM3886es. It works quite well.
According to your product page operating within ±28 V will also allow the use of a 3U chassis - nice.
Indeed. The extra 2.3 dB from 140 W to 240 W costs a lot.
I just picked up a pair of Mean Well RPS-400-27 from Mouser. If they can power a pair of MOD686 modules in a stereo amp intended for music reproduction, I'll be stoked. Math says they should be able to, but I'm certainly pushing them a bit on the peak current.
Thank you. I appreciate it.
Tom
Today I took delivery of two Mean Well RPS-400-27 power supplies from Mouser. I connected them in series to form ±27 V. Plugged in two Modulus-686 modules. Man.... That's a nice amp.
130 W into 8 Ω; 200 W into 4 Ω. Plenty of power. I have each module on a 0.4 K/W heat sink, which reaches a bit over 60 ºC after about 20 minutes with the amp delivering 65 W (sine wave) into 4 Ω (1974 FTC test). I would not go smaller on the heat sinks. Coincidentally, the thermal resistance of the heat sinks on the 3U, 300 mm ModuShop Mini Dissipante is 0.4 K/W. How 'bout that.
A pair of MW RPS-400-27 provide enough power for a stereo amp intended for music reproduction. With both channels running in phase, the supplies don't quite have enough oomph to drive the amp to clipping with a 20 Hz sine wave and 4 Ω load. You "only" get 150 W (4 Ω) at 20 Hz with both channels running in phase (worst case test condition). For music or mono operation, I have no reservations.
Spinning Dire Straits, "Brothers in Arms" (which should surprise exactly no one). Man... What an amp! And what a recording.
Tom
130 W into 8 Ω; 200 W into 4 Ω. Plenty of power. I have each module on a 0.4 K/W heat sink, which reaches a bit over 60 ºC after about 20 minutes with the amp delivering 65 W (sine wave) into 4 Ω (1974 FTC test). I would not go smaller on the heat sinks. Coincidentally, the thermal resistance of the heat sinks on the 3U, 300 mm ModuShop Mini Dissipante is 0.4 K/W. How 'bout that.
A pair of MW RPS-400-27 provide enough power for a stereo amp intended for music reproduction. With both channels running in phase, the supplies don't quite have enough oomph to drive the amp to clipping with a 20 Hz sine wave and 4 Ω load. You "only" get 150 W (4 Ω) at 20 Hz with both channels running in phase (worst case test condition). For music or mono operation, I have no reservations.
Spinning Dire Straits, "Brothers in Arms" (which should surprise exactly no one). Man... What an amp! And what a recording.
Tom
Attachments
Last edited:
The only thermal impedance information I have found on the LM3886 is in AN-1192. There it states that the metal-tab version ("T") is 1 C/W and the plastic-insulated-tab ("TF") version is ~2 C/W, with the 'approximately' symbol in the text. I would think any reasonable thermal insulator material applied to the T version would be less than 1 C/W. (note these are values of thermal resistance from junction to the bottom of the tab)
TI has been updating a number of datasheet with thermal models based on more modern modeling techniques. It would be nice if they did so for these power devices where knowledge would be very useful.
TI has been updating a number of datasheet with thermal models based on more modern modeling techniques. It would be nice if they did so for these power devices where knowledge would be very useful.
Last edited:
I agree. In general, it would be nice if the performance of the LM3886 was better characterized versus temperature. It would be nice to have a better handle on when the SPiKe protection kicks in vs temperature, for example.
The recommendation to stay below 40 W dissipation that's mentioned in AN1192 is good, though. I think they even recommend 30 W for the TF version. I generally find that 40 W dissipation is about the limit for both packages, though I haven't gone full circle with the T version and the various thermal pads.
Tom
The recommendation to stay below 40 W dissipation that's mentioned in AN1192 is good, though. I think they even recommend 30 W for the TF version. I generally find that 40 W dissipation is about the limit for both packages, though I haven't gone full circle with the T version and the various thermal pads.
Tom
BTW: The testing of the ISS (Intelligent Soft Start) is going pretty well. I did find one thing that needed to be tweaked, which required a board spin, so that's in progress. I was able to implement the tweak as a hack on the board so I am confident that the next set of boards will work well. I will get the next batch rush processed, so I can have them in time to fill the pre-orders before I leave on vacation (last chance to order before my vacation is June 14th, 9AM).
Thus, I've opened up for pre-orders of the ISS. You can find it here: ISS: Intelligent Soft Start (PREORDER)
Tom
Thus, I've opened up for pre-orders of the ISS. You can find it here: ISS: Intelligent Soft Start (PREORDER)
Tom
Yeah.... I figured there was no reason to show my dirty underwear. Besides, I've been pretty busy measuring and optimizing the circuit. I suspect I'll type up a page on the design and validation of soft starts when I'm done.
I should have pictures by this time next week (cross fingers).
Tom
Besides, I've been pretty busy measuring and optimizing the circuit. I suspect I'll type up a page on the design and validation of soft starts when I'm done.I should have pictures by this time next week (cross fingers).
Tom