• These commercial threads are for private transactions. diyAudio.com provides these forums for the convenience of our members, but makes no warranty nor assumes any responsibility. We do not vet any members, use of this facility is at your own risk. Customers can post any issues in those threads as long as it is done in a civil manner. All diyAudio rules about conduct apply and will be enforced.

Modulus-686: 380W (4Ω); 220W (8Ω) Balanced Composite Power Amp with extremely low THD

Thx Tom for your thermal protection case study.

Yes I also would be interested in the actual damping factor, but don’t miss your paper‘s due date over it ;-)

You said earlier that you think improve on the 2ohms thd + n behavior. Do you think about board or circuit modifications and would it be good to wait before ordering in order to get circuits with the improved 2ohms performance?
 
I'm thinking improvements in the test setup and/or build quality. I have several fly-by-wire components on the prototype right now. Some rather heavy caps dangle off of the ends of SMD resistors and such. That creates all sorts of interesting effects, both mechanical (stuff breaks off) and thermal (stuff heats up and expands). It does not take much to degrade the THD+N by a little bit at the extremely low levels I'm dealing with.

I am confident enough in the prototype to move forward with the build, as the prototype is working really, really well in spite of the prototype-related issues (dangling components and such).

Attached is the Modulus-686 MacGyver Style. The prototype has had new additions of duct tape and bailing wire since this picture was taken. Needless to say, I would rather wait another few weeks and get the final builds than spend time making pretty plots of the prototype knowing that I will have to redo all the work when the final boards come in.

I fully expect the professionally assembled boards to show better performance than what I show on the prototype. That's what I was trying to say.

Tom
 

Attachments

  • MOD686_MacGyverStyle.jpg
    MOD686_MacGyverStyle.jpg
    423.3 KB · Views: 811
Last edited:
A side note regarding automated vs hand assembly:

I have extensive experience hand-soldering SMD components, both for my own builds and in the evaluation boards I used to modify for key customers while I worked at National/TI. Even then, the measured performance of the circuits soldered by machine by my local assembly house perform much more consistently both channel-to-channel (for multi-channel builds) and board-to-board than the boards I have soldered by hand.

Cracking of SMD components is pretty common and many of these cracks are invisible to the naked eye (but visible under large magnification). My local assembly house uses a reflow oven with 13 (thirteen) temperature zones to avoid cracking. Many of the overseas outfits use three-zone ovens. My local place also takes steps to control the moisture levels in the parts, baking the parts if necessary.

I've seen some bizarre issues with hand-soldered SMD components that I can only attribute to assembly related issues. These issues are rare, so most build their circuits just fine. For the few unlucky ones, pure misery ensues and finding a cracked component among many others is like finding a needle in a haystack - except more difficult. At least you can see the needle in the haystack with the naked eye. A cracked component looks just like any other component.

This is one of the main reasons I decided to let the pros do the assembly of the MOD686. To deliver consistent high-end performance I need the assembly job to be top notch. That's life at performance levels where such things matter. :)

This is also why I expect the professionally assembled boards to perform better than my MacGyvered prototype.

Tom
 
Last edited:
I noticed you use the LM3886 with the metal casing in your prototype.
Does that improve heat transfer or could one use also the plastic casing in the final built?

Good question. I don't know the answer yet. I will build one MOD686 sample with the LM3886T and another with the LM3886TF and measure the difference (assuming there is a difference).

I would prefer to use the 'TF' (plastic) version. It's just so much easier to work with mechanically. No need to worry about burrs and metal chips shorting the package to the heat sink (and chassis).

The TF has higher thermal resistance that the T version, for obvious reasons. However, once the additional thermal interfaces of the isolating pad are added, I'm not sure how large this difference is in practice. I do intend to find out and make recommendations based on my findings, however.

Tom
 
It all depends on how uncomfortably hot you want the heat sinks to get and which operating conditions you intend to operate the amp at.

Math says:
Ambient temp: 25 ºC
Max. heat sink temp: 60 ºC
Supply voltage: ±35 V
Crest factor, music: 14 dB

8 Ω, sine wave: 0.32 K/W
8 Ω, music: 0.46 K/W
4 Ω, sine wave: 0.17 K/W
4 Ω, music: 0.25 K/W

That says you want a 4U Dissipante chassis if you want the amp to be able to handle playing music continuously into a 4 Ω load with the peaks reaching clipping levels without having the heat sink temperature exceed 60 ºC.
Of course this will mean that under normal operating conditions, the heat sink might reach lukewarm if that. I consider that good thermal design, but some builders seem to be disappointed by cool heat sinks. :)

I chose 60 ºC max as that's really rather hot to touch. It won't burn you instantly, but don't expect to be able to hold your hand on the heat sink for more than a few seconds at a time at that temperature.
In addition to the user's well-being, one should also consider that the SPiKe protection of the LM3886 will start to kick in at the max output power if the heat sink gets much beyond 65-70 ºC.

I linked to the data sheet for the Dissipante heat sinks back in Post #7.

Tom
Tom
The specs for the Dissipante chassis say the rating is for the entire enclosure i.e.
Both heatsinks. I was talking a about a 0.4 K/W heatsink for mono, i.e. For just one channel.
 
What about the damping factor?
Any figures?

See attached. My guess of 400 was pretty close. Measured value at 20 Hz: DF = 354.

The output impedance starts to rise above the DCR of the output inductors as the loop gain of the amp is decreasing with frequency.

Tom
 

Attachments

  • MOD686_PROTO_Zout.png
    MOD686_PROTO_Zout.png
    48.6 KB · Views: 1,041
  • MOD686_PROTO_DF.png
    MOD686_PROTO_DF.png
    48.4 KB · Views: 1,031
Tom
I'm assuming that as with your other amps you will be setting up a BOM with Mouser and Digikey. It would be great if you could do that in advance of the boards becoming available as us early adopters could then order and obtain the parts and be ready to start building as soon as the boards arrive.
 
Absolutely! I just need to decide whether to use the LM3886T or the LM3886TF. If I can get the same performance with the isolated TF version, I'd rather go that route.

I'll decide once I have the first circuit board samples in hand. I am planning to get a couple of samples from the assembly house before pushing 50 boards through. That seems like wise risk mitigation.

Tom
 
Modulus-686: 380W (4Ω); 220W (8Ω) Balanced Composite Power Amp with extremely low THD

Even with the T version, if you add that layer of insulation, doesn’t it add to thermal resistance as well? Wouldn’t it have a total of similar thermal resistance as the TF version?
 
Using a thermal pad between the LM3886T and the heat sink will add thermal resistance. The question is whether it adds more thermal resistance than the plastic of the LM3886TF package. That's what I intend to find out.

With a 60 ºC heat sink temperature, I find 40 W of dissipation in the LM3886TF is the maximum I'm willing to tolerate. Math says 46 W dissipation per LM3886TF in the MOD686 at the worst case dissipation driving a sine wave into a 4 Ω load. This indicates that the TF is probably fine for builds intended for music reproduction but will overheat eventually with sine waves.

A perfectly reasonable solution would be to recommend using the TF version and noting that those who really wish to push the thermals can use the T version.

In the end, the choice becomes a matter of how much mounting hassle the builder is willing to accept and how confident the builder is in his/her ability to drill and tap the heat sink without leaving any burrs and chips behind. You don't want to risk having an LM3886 short to the heat sink.
If I can get, say, 220 W into 8 Ω and 350 W into 4 Ω at clipping with the TF, I'd be perfectly content with that.

In addition to the balls-to-the-walls build using a ±36 V supply, I am planning to recommend using a ±28 V supply for those interested in a more conservative build. Figure 150 W (8 Ω) for the conservative build. That would ease the heat sink requirements as well.

Tom