Modulus-86 build thread

While the listening experience is ultimately important, I suggest that the we stay fact-based here. I do find the correlation between a good listening experience and measurements to be very interesting. That is in fact why I chose a measurements based approach for my business. There is plenty of research to support the argument that good measurements (plural) correlate well with a good perceived sound.

The speculation about what causes you to perceive one amplifier type to be superior over another, while relevant to you, is likely better served in other threads, for example the Solid State vs Tubes thread.

Tom
 
It is very difficult to say one amplifier is better than another in every way. No amplifier like that currently exists. But in the process of improvement, you have to start with speculation based on experience. Generally I follow some steps if I hear a problem:

1. Based on understanding, speculate what the problem might be.
2. If it is possible to do analysis and simulation easier than building a solution concept, then this will be done.
3. Built a test sample and from general observation to see if there is any change.
4. Do measurements to find out what exactly is going on to have better understand what changed and figure out the best way to optimise.

Sorry to say that I am currently at stage 3+ on amplifiers. Currently much of my time is spent on speaker driver and speaker enclosure at this stage because I need better speakers to make amplifier differences more revealing so that additive effects do not lead me in the wrong direction. Current drivers on the market are not meeting my needs.

If I really wanted to compare amplifier performance, I would first look at CSD of speakers and drivers 0.4ms and 4ms range, frequency response and phase within the audio spectrum, then distortion and distortion measurement methods over the audio spectrum. Otherwise, I will not be able to fully understand what I am hearing when I sense a problem.

With electrical circuits, it is even more difficult to conduct good analysis and simulation because of non-realistic models and that your data are always between two points in the system, so at which point does the problem exists? If you can simulate a real layout accurately, then you can start playing around with different things, but that kind of simulation means lots of investment. Whether that investment is going to be appreciated by majority of people in the final product is highly questionable.
 
Last edited:
Sorry to drag ths out, but I am trying to understand RMS power...

It is clear that my method to calculate the RMS power drawn from the supply is wrong. I simply multiplied the RMS load current by the supply voltage: i.e. Ps = IoutRMS * Vcc

Looking at the equation on Tom's website: Ps = (2/pi) * (Voutpk/Rl) * Vcc
This could be rewritten as (2/pi) * (1.414 *VoutRMS/Rl) * Vcc
then 0.9 * (VoutRMS/Rl) * Vcc
so, Ps = 0.9 * IoutRMS * Vcc

That's 0.9 X what seems logically (at least to me) right. Can anyone explain what's gone wrong with my logic?


I think I have found the explanation. The RMS current and average current are different (by the 1:0.9 factor) - see here, in the "Class B" section:
Amplifier Efficiency

I also found the maths for calculating the average here, on pages 10-12:
http://www.seas.upenn.edu/~ese319/Lecture_Notes/Lec_21_Class_B_Amplifier_08.pdf
 
Blue!!

Modulus-286, final rev. (1.1).

I've put a little time-lapse video together of the build process. You can find it on my website: Neurochrome :: Modulus-286 (Click reload or CTRL/CMD-R if you've viewed my site recently).

The inductors were custom wound by forum member 'hiker19'. You can find his website here: http://www.iwindcoils.com I have no relation to him. Observant viewers will recognize the photography style. Yes, I did help him set up his website. 🙂

Tom
 

Attachments

  • MOD286_Stereo_Assy.jpg
    MOD286_Stereo_Assy.jpg
    105.9 KB · Views: 442
Tom,
I like that thing!
The documentation is inviting to read and fact filled as well.

P.S.
Just one note: If I had one, I'd probably use it in parallel mode to an 8 ohm speaker. That's an odd personal preference of mine, but the higher output linearity (of the chip's internalized output transistors) would be maximized. The similar output of a 16 ohm speaker with somewhat 70 pounds of 12ga coils for the crossover is something that I don't need to worry about doing, if using a fair 8 ohm speaker with a parallel amplifier. So, be it a personal preference, but I really do like the parallel amplifiers. It seems to me that a bit more output device linearity in the amplifier does a wee bit of help towards more appreciable (probably more linear) output from the speaker. That's just how I'd use it. Seems that your documentation didn't mention it (parallel mode with the 8 ohm speaker, and that matter of why). But, if you did mention it, I'm sure it would be more efficient than what I just did.
Yeah, there's some praise and some thanks in there somehow.
 
Last edited:
Modulus-286, final rev. (1.1)...............
Nice looking assembly. You do a good design and test before retailing, with adequate support for any builder.

............. If I had one, I'd probably use it in parallel mode to an 8 ohm speaker. ............... Seems that your documentation didn't mention it (parallel mode with the 8 ohm speaker, and that matter of why)...................
I have to agree with Daniel, a good paralleled 3886 (based on Tomchr's PCB) driving a reactive 8ohms speaker load will probably get the best out of an 8ohms speaker.
 
Thanks guys.

Using two LM3886es in parallel to drive one speaker is a nice solution indeed. It's easier on the thermals for each LM3886 as the dissipated power is divided between two ICs. It's easier on the output current as each chip only has to deliver half the load current. Getting the two amp halves to play nicely together is not an easy task and the composite amplifier topology complicates this further. I'm quite happy with the '286.

That said, on a ±28 V power supply, the amp performs very well as a stereo amp. When driving a 4 Ω load, the LM3886 does get a workout, though.

Tom
 
I suggest looking at the LF response (even a quick calculation will do here). I'd re-read the application section of the LM3886 data sheet, in particular the bits about stability. I'd also re-read my article on decoupling:LM3886 chip amp supply decoupling.

While what they have is decent, you can do even better by adding a DC servo as done in the BPA200. Of course, designing a good DC servo requires some skill and know-how. You can find a few good threads on it here.

The circuit, as shown, will have spurious oscillation as the output nears the negative rail. AndrewT and I have hammered on this for years (he longer than I). If they're running ±35 V rails, they might also have stability issues due to poor decoupling. That would depend on the caps chosen, so I can't say if they do or don't have issues.

Tom
 
Last edited:
Nice looking assembly. You do a good design and test before retailing, with adequate support for any builder.

I have to agree with Daniel, a good paralleled 3886 (based on Tomchr's PCB) driving a reactive 8ohms speaker load will probably get the best out of an 8ohms speaker.

A better endorsement than that is not possible.

The one flaw that I spotted in your documentation is in supporting this particular usage. Otherwise, the documentation is really very nice.
 
Yeah. The Modulus-86 was originally intended for use as a module in a PA/BPA style amp, hence the Modulus name. It wasn't until I got to know the LM3886 a lot better that I realized it really wasn't suited for anything but the parallel amp (PA) configuration. Until a few weeks ago, there was still a two-liner mention of bridging in the MOD86 documentation. That's now gone. 🙂

I suppose in theory you could put six (6) LM3886es together in a bridged parallel amp (BPA) and drive a 4 Ω load, but at that point you're looking at a significant BOM cost and really aught to consider whether you're using the right tool for the job.

Tom
 
Yeah. The Modulus-86 was originally intended for use as a module in a PA/BPA style amp, hence the Modulus name. It wasn't until I got to know the LM3886 a lot better that I realized it really wasn't suited for anything but the parallel amp (PA) configuration. Until a few weeks ago, there was still a two-liner mention of bridging in the MOD86 documentation. That's now gone. 🙂

I suppose in theory you could put six (6) LM3886es together in a bridged parallel amp (BPA) and drive a 4 Ω load, but at that point you're looking at a significant BOM cost and really aught to consider whether you're using the right tool for the job.

Tom
Apparently, when it is necessary to succeed with much louder audio, the clear way forth, is bi-amp. That's so you can play both louder And clearer. It is a bit more pleasant that way, and probably more effective for the audience.

Your existing product lineup already has more than enough power output capacity for the mid&treble speaker portion of a high power audio system. So, it seems that you had that problem solved already.

However, the usage is likely to be a capacitive load, so that might need double-checked.
 
I suppose in theory you could put six (6) LM3886es together in a bridged parallel amp (BPA) and drive a 4 Ω load, but at that point you're looking at a significant BOM cost and really aught to consider whether you're using the right tool for the job.
FWIW, Neumann claims pulling 350Watts from a BPA using 4 TDA7293's for the bass driver in their highly reputed KH420 reference studio monitor. MF and HF is spec'd 140W, from two TDA7293's bridged (I don't know if these amps are composite designs, though). Of course not long-term, excursion and thermal limiters kick in to protect drivers, amps and the SPMPS they use to supply the whole thing. They sure do care about reliabilty and BOM cost.
https://www.fidelity-magazin.de/wp-content/uploads/2015/12/1-Neumann-KH-420-Backplate.jpg
 
Last edited:
So , can we parallel 2 of the modulus-86 boards?

With proper ballasting resistors, yes. You won't get quite as good performance as the Modulus-286 in mono mode, but it is possible.

Apparently, when it is necessary to succeed with much louder audio, the clear way forth, is bi-amp.

That is my preferred solution.

However, the usage is likely to be a capacitive load, so that might need double-checked.

The MOD286 is stable to at least 1 uF of load capacitance according to my measurements.

FWIW, Neumann claims pulling 350Watts from a BPA using 4 TDA7293's for the bass driver in their highly reputed KH420 reference studio monitor.

I'm not that excited by the TDA7293, actually. The spec sheet is pretty cooked (100 W into 4 Ω at 10 % THD, for example). They really stretched to get that 100 W. Also note the max dissipation of 50 W. I've looked at various TDA chips and just can't get excited about them. If you really want the power, class D or a discrete or some opamp-discrete hybrid is where it's at.

Tom
 
Last edited:
With proper ballasting resistors, yes. You won't get quite as good performance as the Modulus-286 in mono mode, but it is possible.





Tom

ok thanks for the answer, I asked about the modulus 86 in parallel because I would really want to try that, i had the chance to get the lm4780 in parallel(your boards) but I missed the last groupbuy(btw there are a few guys who hope that you'll have another round). Now, you have the modulus286, but is smd , I wish I could say I am very comfortable to solder smd boards, but I am not.And if modulus86 will work in parallel , that's great!
Could you plz post a sort of drawing for that?
Thanks a lot!