Modulus-86 build thread

There should be a gap.
This gap allows for tolerances as I stated and also allows for thermal expansion as I stated.

The PCB as designed deserves a caution statement in the instructions, to ensure the builder does as you did and makes sure the PCB does not project out the back.

If by oversight the PCB did project out beyond, then the disaster is as stated "waiting to happen".


I don't see what purpose an expansion gap serves. What will expand?

Won't SPiKe protect a chip with poor heatsinking? But it would be a disaster to listen to SPiKe instead of music. ;)
 
Yes, this seems a small issue and much ado about nothing. If anything, the pcb resting against the heatsink serves well to lower the mounting forces on the chip's screw.

It does take a little bit of extra prep/thinking about. Any concern of the ground plane on the bottom of the PCB from accidentally making contact with the heatsink? That'd be my concern if/when someone runs the heatsink at VCC (no insulating washer)
 
Any concern of the ground plane on the bottom of the PCB from accidentally making contact with the heatsink? That'd be my concern if/when someone runs the heatsink at VCC (no insulating washer)

The parts list specifies the all-plastic package 3886, but metal back types will work too, with insulator. There is about 1mm gap between the edge of the phenolic board and the ground plane or traces.
 
Thank Richidoo for passing the information for all builders,

I have made the mistake and I was just considering the different options... Give an angle to the board with two washers, or unsold the chip to reposition it with the right angle...

Cheers,

Dominique

Try AndrewT's idea for a extension block under the chip. Copper, or aluminum is easier to find where I live. Thin heatsink grease on both sides. Unsoldering is difficult job, but I have read that ChipQuik desoldering system works well. I don't know if it can do a large IC like this. I never tried it myself.
 
Member
Joined 2014
Paid Member
the metal tab is about 2mm thick. The legs on the 3886 are bent so as to act as springs to absorb any tiny epansion. One the chip is torqued down I can't see how it can be bent up. If worried, use U channel across the body of the 3886 and then it CANT go anywhere. You certainly wont damage the chip, but it wont sound good.

Hopefully Andrew will come up with something to prove me wrong.
 
I'm guilty of pushing mine to the brink. In my 5 day test drive last week I ran it with a couple of efficient 2 way speakers and also chained it up to a pair of vintage power sucking open baffle 5 ways just to see what happened. The little amp did fine and I did notice that when it reached clipping my perception was that it was somehow different in that it was not so much creating harsh harmonics as it was sort of just "giving up" - kinda just stopped making noise so to speak. (ha ha, is there a flowery audio review term for "silent clipping"???)

Maybe I'm used to hearing higher levels of distortion get clipped and that sounds more ugly than when it's just the plain old music?? To be clear, I did not go to this SPL much as even with the power hogs things got pretty loud before then.
 
The attachments of the chip to the heatsink, board to the chassis and heatsink to the chassis require that the chip be soldered perpendicular to the board in order that all of the components align for assembly. The best approach is to use a simple jig that holds the chip perpendicular to the board for soldering. I don't think the issue of whether or not the board projects beyond the chip is relevant as without perpendicular alignment of chip to board the assembly will be under stress when all of the fasteners are tightened.
 
Where on earth have you imagined the 35W from?
That's the official power rating from Tom, into 8ohms. :)

I explored the redline some tonight. The tone remains clear and true up until the SPiKe protection starts cutting in with the clicks and pops. That's what you'll hear instead of traditional clipping. That's why it sounds clean right up until lightning strikes. :D

On my speakers it was playing very loud (90ish from 3 meters away) before occasional clicks started, that's playing uncompressed classical music, on small passive xo 87dB speakers with lots of baffle step correction, flat to 50Hz with a 6.5" woofer, in a large room. The speaker designer recommends 50-200W. It's enough for my purposes, active amping tweeter and mids individually.
 
Member
Joined 2014
Paid Member
That's the official power rating from Tom, into 8ohms. :)

No, Tests were done at that level because Tom was limited by the power supplies available at the time. You can go all the way to the rated 68W if you want and can sort the thermals out. Tom has posted test results at that power level to help with the design.

If your speakers are 87dB/1W then you would have been hitting 100dB peaks unless you have a thermal problem.
 
No, Tests were done at that level because Tom was limited by the power supplies available at the time. You can go all the way to the rated 68W if you want and can sort the thermals out. Tom has posted test results at that power level to help with the design.

If your speakers are 87dB/1W then you would have been hitting 100dB peaks unless you have a thermal problem.

Yeah, it might have been louder dB, I was just guessing at the level. But also I was playing classical music with less compression.

The amp protected most obviously only on percussive peaks on a Naxos classical piano/violin duo, minimal compression. I would have liked the volume even higher on this album but the uncompressed grand piano peaks triggered the protection. But while playing a jazzy singer Karrin Allison album with more compression, the perceived loudness was much higher and I was surprised that there were no protection events. On both of these albums the tonal texture is off the charts, that's the real strength of this amp - microdetail, and the quiet noise floor (same thing?) Then I tried Ben Folds with heavy compression and I played it even louder, enough that I had to walk around the back of the room 16 feet away from speakers to tolerate it, protection happened only occasionally during bass heavy passages. I was worried of hurting the woofers at this level because they have a lot of passive bass boost and were obviously distorting as expected, but the amp did a good job.

The 68W rating is into 4 ohm pure resistive load. Tom's official power rating on his website is 35w@8ohms resistive. Real world 4 ohm speakers will probably need parallel 3886s to play loudly without protection, especially if they play classical music.
 
Member
Joined 2014
Paid Member
Y. Tom's official power rating on his website is 35w@8ohms resistive. Real world 4 ohm speakers will probably need parallel 3886s to play loudly without protection, especially if they play classical music.

er no Tom says '35 W output power (8 Ω) using a ±28 V power supply.' Which is the rating with 28v rails. But if you are clipping on peaks with 35W you will clip with 50W and would probably clip with 100W, so just being my usual pendantic self.

Just to confirm, are you running active or passive at the moment? If you are running with passive XO for testing and swapping to active later then you will have a lot more headroom then (effectively the same as a single140W amp if you are on 28V rails.

BSC is a bitch tho. That's the area I have concern on when I cut the box off my apogees.
 
er no Tom says '35 W output power (8 Ω) using a ±28 V power supply.' Which is the rating with 28v rails. But if you are clipping on peaks with 35W you will clip with 50W and would probably clip with 100W, so just being my usual pendantic self.

If it was voltage clipping, yes I would agree with that, if the rails of these different power amps were all the same the voltage clipping would not change.

But it is not voltage clipping, it is SPiKe protection. It is loud snapping sound indicative of the amp shutting itself off for an instant, not the edgy squarewavey change of tonality with traditional voltage clipping.

I don't know what triggers TI Spike protection. I think it is either overcurrent or maybe it is just internal temperature, or both. I think voltage clipping is allowed, right?

All this talk about protection and clipping is pretty far off topic from the build thread, and people deciding whether to try Modulus86 might be put off by the "not enough power" discussion. The reality is that these 35W amps kicked *** into speakers that are rated for 50-200W and then they automatically protected themselves from damage when I pushed them too hard. The power amp is the size of a thumbnail making solid 35W, not bad. Whatever protection noises I heard can be cured by adding another module per channel as the assembly manual describes.

Just to confirm, are you running active or passive at the moment? If you are running with passive XO for testing and swapping to active later then you will have a lot more headroom then (effectively the same as a single140W amp if you are on 28V rails.

BSC is a bitch tho. That's the area I have concern on when I cut the box off my apogees.

Yes, I'm passive now, auditioning the amps before going back to active. BSC does eat a lot of power. Do you plan to try your Modulus amps on the Apogees?

I have the Sympatico amps to try also, bridged 4780s. It will be interesting to see if they protect or not into this passive speaker. If they don't protect then that means the 3886s are protecting from voltage clipping which I don't think is the case, or even possible? So I should expect 100W Sympatico to protect also, since it is even more sensitive to current and heat in the bridge mode, but it has 2 dies to share. It will see my speakers as 3.5 ohm instead of 7 ohms but playing "half as loud." We'll see.
 
No, Tests were done at that level because Tom was limited by the power supplies available at the time. You can go all the way to the rated 68W.

Actually, the Modulus-86 clips a little bit earlier (a few hundred mV) than the naked LM3886, so on ±28 V rails, the power out is 35 W (8 Ω) and about 65 W (4 Ω). The naked LM3886 will provide 38 W, 68 W respectively under the same conditions. The naked LM3886 will also provide 100x worse THD, but that's another matter... :)

Note that the difference between 35 W and 38 W is 0.35 dB. The acoustic change threshold is commonly accepted to be somewhere around 0.5~1 dB, hence, you won't be able to tell a difference between 35 W and 38 W unless you measure it.

~Tom
 
Last edited: