New Modulus-86 Build ?s


While I wait for my Modulus-86 kit, and other parts to arrive, I thought I'd ask a couple questions... I have basically arrived at my build specifications, but just want to be clear on a couple things. Primary use case for this amp will be; near-field listening at low volume, so the amp should not be stressed at all into 8Ohms speakers.

I will be using the following; a) using a very small, unconventional enclosure. Chosen specifically to be a companion mate for my latest Streamer. b) attempting to fit a +-30VDC SMPS in said enclosure, but not sure if everything will fit in the chassis. c) internal, or external heatsinks(HS). [I've already nixed the idea of external HS, due to the fuggly factor.] :) Based on my specs above, I believe the amp will top out at, ~40wpc, so I'm not sure what size HS I will need. My questions are;

1) Instead of trying to fit HSs into my small chassis, I'm considering a heat-spreader attached to the sides of my aluminum case. Would this arrangement work to use the entire enclosure as one big HS and sufficiently keep the amp cool? The size of the intended spreader is; 130 x 45 x 5mm. There will 2 spreaders used, one affixed to each side panel, and the PCB attached to them. I used a similar arrangement for a TPA3255 amp I built, albeit the case was larger, and heat is not an issue at all. Going the spreader route will also free up more room in the enclosure.

2) In the event, I cannot fit the SMPS into my enclosure, I am contemplating plan B. Being, installing the SMPS in a suitable, but smaller enclosure of it's own with an umbilical cord running to the amp enclosure. This would only be about 1 meter in length. This isn't my first choice, but I see no reason why this wouldn't work, since external SMPS power bricks are used all the time to power small amps without issue. It's the convenience factor (or lack of), that makes it a less than ideal solution.

It will be a few weeks before the SMPS arrives, but for fitting and testing, I'll be using a +-36VDC SMPS with the exact same physical form-factor. This will allow me to get everything 'dry' fitted, and to determine whether or not an external enclosure for the PSU will need to be used. Below are some pics and what I've trying to achieve for a cooling solution.

Fuggly Ext HS.jpg

Bare Aluminum Spreader
Bare Aluminum.jpg

Proposed solution
Desired Solution.jpg

That aluminum slab provides some mass, but it likely won't provide enough surface area to serve as a proper heatsink. The "entire case" isn't being used, because there's isn't a good thermal bridge from piece to piece.

If you are truly running it at low volumes, though, then it might be OK. For example, I have an amp (not an x86 variant) with two modules mounted to the base plate, which is a 300x400 aluminum panel. The rest of the case is steel (originally a Modushop Pesante case). It provides enough heat dissipation for occasional use in my bedroom.
I know the side panels are slotted, but I was planning on using thermal adhesive tape to join the two pieces. When I get to that stage, I'll do some tests with just the side panel, and then both sides panel screwed to the case and see what results I get under normal listening conditions. The super structure is cut out of a single sheet of metal, so it will come down to whether, or not, there's enough surface area for the side panels to transfer the heat adequately around to the entire enclosure. I could go with larger aluminum panels too, and that would provide more area to connect to the side panels. I'll have to wait and see. But, I'm hopeful that with my intended usage, heat won't be a real issue. If it was going to to be driven a lot harder, I would have to go with a larger case and traditional heatsinks.
I would definitely not use tape. That's OK for something small like a little heatsink on a Raspberry Pi, but definitely not for a big chunk of aluminum like that. Instead, bolt/screw the two pieces together with thermal paste between them.

If you want the heat to transfer around to the other portions of the amp, use thermal paste on each of those junctions. I'm unsure whether the bracket joining the two would act as a suitable transfer, though. If it's steel, it will not. If it's aluminum, there's a better chance.
Mike said it well. The thermal mass of that chunk of aluminum will prevent the amp from overheating for a while, but if you plan to listen to music for, say, 8 hours/day what you actually want is some surface area.

Compare the surface area, i.e., the sum of all the surface area of your heat sink versus the surface area of that chunk of aluminum. You'll find the heat sink has much greater surface area. Companies like ModuShop take it a step further by having the heat sink fins be serrated. They also anodize the heat sinks, which further increases the surface area.

It looks like a nice chassis. I think it would be a shame to cover the horizontal slots. Who made the chassis? Is it one of your design? The anodization looks very nice.

I've been doing some preliminary tests simply to measure what temperatures I'm getting, using the the aluminum block. Please refer to the images below of how I affixed the LM3886 to the aluminum. I used another small amp I built a few years back using the XY PCB. I simply detached the existing finned external HS, from the that amp and attached the aluminum block to it. I know it's not the Modulus-86 board, or implementation, but it's all I have at the moment to get some temps readings. As can be seen, I used the same thermal pad that was in use on the amp. BTW, the supply PSU was providing +-36VDC. I will be using the exact same PSU, but with lower voltages of +-30VDC in my new Modulus build.

As pictured, I let the amplifier run at a very modest volume for 3 hours today and at the end of that time, the temperatures seemed to have stabilized, and were not changing significantly at all. Measuring 7 various points on the aluminum block I got an average of 28.96 C. At the base of the device, I got 31.0 C. According to the datasheet it seems the built-in thermal protection triggers at 60.0 C.(Tom, correct me if I'm wrong on that number.) So, simply using the aluminum block, affixed to nothing else, it would appear the amp is working 'within' specifications and at a temperature of half of what the design specified. The aluminum block is very warm to the touch, but not hot by any means. I could likely improve the measured temps a little, had I used thermal paste instead of a pad.

I recognize there are a couple variables with my experiment that once in the final build should only improve the cooling situation. 1) the aluminum block in open air, and 2) operating +-36VDC not +-30VDC. The good thing about my case is that it is vented both on the top and bottom, so I should get some air flow through it - even if the sides are covered up.

I have a couple more options I'm exploring to better deal with 'more' adequate cooling. I have a couple smaller HSs coming, so will see how they perform after testing. By then, I will have the Modulus-86 boards in hand and assembled, so will use those instead of my XY implementation PCBs. I will also have the new PSU (+-30VDC), so more combinations to test. In the end, this case may not work out, so I'll have to wait and see. That would be disappointing.



@tomchr, That case is actually a mini-ITX case, design for a computer beings used as an HTPC. Over the past few years, I've tried to expand my skills and move beyond all the traditional looking hi-fi gear enclosures. So I've turned my attention to unique cases from the DIY computer arena(my second passion) :) I've built several Steamers/DACs using computer cases that have turn out fairly well. This approach takes a significant amount of planning, and time to bring a project to life - in a nice looking enclosure - due to all the moding required. There's a lot of 3D printing generally required, as well as some custom metal work. Not everyone likes those kind of challenges! I'm a retired planner from the aerospace industry, so the whole process is right up my ally. Below are a couple pieces I've done in a variety of form-factors


Neat with the ITX chassis.

The maximum junction temperature for the LM3886 is 150 ºC. I'd never run it that hot, though. I recommend no more than 60 ºC heat sink temperature.

So, my junction temperature reading of ~30-32C (+-15%) is well within the design parameters of the chip? With my use case, the amp will never see that level of "ON' state, or loud..ish volume, for more than3-4 hours at a time. Like I said, there are still a couple more ideas I'm pursuing. I'll try and keep the thread informed of my progress.

Junction temperature refers to the semiconductor junction. What you measured was the heat sink temperature. As you can see in the LM3886 data sheet, the thermal resistance from the semiconductor junction to the IC case is 1 K/W for the metal-backed package (LM3886T).
Screenshot 2024-01-30 at 12.01.07.png
The thermal resistance of the insulated package that you're using (LM3886TF) is closer to 2 K/W as I recall. You then need to add the resistance of the thermal pad (which isn't necessary with the insulated package, by the way).

So, if we assume that the LM3886 dissipates 5 W under your conditions and that the resistance of the thermal pad is, say, 0.2 K/W, you're looking at a junction temperature of 32 + 5*(2 + 0.2) = 43 ºC. That's still well below the specified maximum of 150 ºC.

If what I just typed is gibberish to you, I suggest taking a look at my Taming the LM3886 - Thermal Design page:

Not all of it. :) When I took my temperature reading of the 'junction', I aimed my infrared thermometer at the lower, front-side of the IC case, just above the pins, figuring that was as close as I could actually measure the IC without measuring the HS. That's where I got the 32C. When I measured Just to the right of the chip on the HS, I got one reading of ~36.4C. That was included in the my total HS average. In my experiment, I did notice that one side of the chip was running hotter than the other, by about 6 degrees. Not sure if that's a function of the design of the IC itself, or not.

Anyway, if I was going to use the amp in any other listening environment, all of this would be moot, and I'd build it in a bigger chassis with properly sized HS. I figured with my particular use case, I wouldn't have to be too conservative, and it would survive short periods without too much issue. So, I've brought this on myself as a challenge to see if I can pull it off in such a compact chassis. :)

Thanks for the link. I'll take a look at it.

BTW, the boards and parts showed up today. That was super fast for international shipping. When sorting out the resistors, I came across a strip of 2, 100Ohm that don't have a home. There are 100Ohm resistors called out in the BoM, but it's for R19 and R3. That would be 2/PCB, and there is another strip of 4 x 100Ohms resistors, so all is good. So I have 2 mystery resistors. Maybe they got put in the kit by accident.

When I took my temperature reading of the 'junction', I aimed my infrared thermometer at the lower, front-side of the IC case, just above the pins, figuring that was as close as I could actually measure the IC without measuring the HS.
OK. So you measured the case temperature then.

The only way you would have access to the semiconductor junctions of the IC would be if you integrated a temperature sensor on the IC die, and even then you would not measure the temperature of the hottest junction.

"Junction" in this context means the PN junctions on the semiconductor die itself. It's an important distinction as the thermal resistances from the junction to the case and junction to ambient are sizeable.

When sorting out the resistors, I came across a strip of 2, 100Ohm that don't have a home. There are 100Ohm resistors called out in the BoM, but it's for R19 and R3. That would be 2/PCB, and there is another strip of 4 x 100Ohms resistors, so all is good. So I have 2 mystery resistors. Maybe they got put in the kit by accident.
Double-check with an ohmmeter that those are indeed 100 Ω resistors. Many times 100 Ω, 1 kΩ, and 10 kΩ can be pretty difficult to tell apart from the coloured bands alone.

The kits are packed by hand so it is possible that my helper and I miscounted. I guess I'll find out when I try to pack the last kit. As long as you have what you need life is good.

I just rechecked and the colored bands are; BRN-BLK-BLK-BLK-BRN. They measured 101.1 Ohms. I doubled checked my boards and there are no 1/4W resistors placeholders vacant. :) No worries though, everything so far seems to be populated correctly. I just finished up the ceramic caps.

One question... Do the power resistors need to be lifted off the PCB a bit before soldering?
Yeah. It's best to allow some airflow around the power resistors. 4-5 mm is good.

Just started my first round of bench tests on my newly assembled Mod-86 boards. And it wasn't as positive as I'd hoped. Houston we have a problem! :)

Test scenario; I'm using a +-36VDC SMPS (tests only. My +-30VDC supply has not arrived yet). Using single ended RCA as input connection (+in and GND), 8Ohm test speaker. Source is small DAP (Digital Audio Player). No preamp was used.

Pass #1 -
I brought one channel up slowly on the variac to 120VAC without any sparks, small explosions or other fireworks. Didn't see, or smell any magic smoke either. That's always a good sign. :) Measured the voltages at the input of the amp and they checked out fine +-35VDC. All connections seemed nice and tight.

I had the DAP playing at about 50%. As I slowly increased the volume I started to hear music at about 90, but it was very faint and anemic/tinny sounding. I could tell something was off. The volume at 100% just made the same sound a tad louder, but still no real amplification. Okay. I knew the DAP didn't put out much signal level, but it worked fine with other amps, so I thought it would work with the Mod-86. At least some full-throated amplification.

Pass #2 -
Keeping everything the same, I introduced a preamp into the mix - my little Parasound Zpre3. This will at least provide the proper preamplification. Well, no luck. It made it little bit better, but I had to turn the pre all the way to 100%, with still no real amplification, so I knew something was still wrong.

Pass #3 -
I then took the 2-channel board, and did the same first test, with the same symptoms. :(


Power Supply = Good
All Connections = Good
Proper Heatsink = Good

Conclusion -
There's a problem in the boards themselves, and it's likely related to a miss-placed part that needs correcting. And, where there is one miss-placed part, there's another one too. And I thought I was super careful with this build.

I'll defer to the experts on where to start looking.

Using single ended RCA as input connection (+in and GND)
That could be your problem. You need IN- connected to ground along with the GND pin. So the RCA shell goes both to pin 1 and 3 on the input connector. The RCA centre goes to pin 2 (IN+).

Like this:
Screenshot 2024-02-07 at 16.51.24.png

A differential input cares about the voltage difference between IN+ and IN-. So if IN- is floating the input sees practically 0 V difference. So you get no output.

I'm glad you found the issue.

Doing some initial tests with both amps on their heatsink. Playing for about an hour and the heatsink is barely warm to the touch. It's playing beautiful music.