To be honest, I'd rather work to bring the cost of the assembled Modulus-286 boards down. That's a better long-term plan for my business.
You can connect two Modulus-86 modules in parallel by adding 0.1 Ω, 3 W in series with each output and taking the "outside" of these resistors to the speaker terminal. Get a resistor with the lowest tempco possible, preferably a metal film or metal element resistor as these add the least amount of THD. I will leave it as an exercise for the reader to carry this out in practice.
Due to TI's decision to discontinue the LM4780, I can no longer support the Parallel-86. Sorry. I can't hinge my business on one location of one (maybe two) distributors sitting on all available inventory and that inventory bouncing in and out of stock. For the last group buy I requested Bill's help getting LM4780s from RS Components in the UK to me, so I could supply everybody (thanks Bill!). Taking on another group buy would be too much of a distraction for me at this time.
Tom
You can connect two Modulus-86 modules in parallel by adding 0.1 Ω, 3 W in series with each output and taking the "outside" of these resistors to the speaker terminal. Get a resistor with the lowest tempco possible, preferably a metal film or metal element resistor as these add the least amount of THD. I will leave it as an exercise for the reader to carry this out in practice.
Due to TI's decision to discontinue the LM4780, I can no longer support the Parallel-86. Sorry. I can't hinge my business on one location of one (maybe two) distributors sitting on all available inventory and that inventory bouncing in and out of stock. For the last group buy I requested Bill's help getting LM4780s from RS Components in the UK to me, so I could supply everybody (thanks Bill!). Taking on another group buy would be too much of a distraction for me at this time.
Tom
I agree, the spec sheets of various linear chips made by ST are sub-standard in general. But never judge a chip by its spec sheet alone. Both the LMs and the TDAs have their following for all good reasons...
I agree that it is possible to eek out good performance from a decent amplifier chip. That's what the Modulus-series is all about. That said, if the manufacturer can't be bothered to write a good spec sheet, I won't bother using their products. I'm one of these weirdos who like to know what I'm buying. 
People's reasons for following aren't always rational. Humans are funny critters that way.
Tom
People's reasons for following aren't always rational. Humans are funny critters that way.
Tom
How much does the position of the LM3886's on a heatsink affect the efficiency? I was thinking that the ideal position of the components would be in the centre of the heat sink but would it reduce the efficiency much/at all if they are mounted more to one end?
Just trying to decide on the size of chassis I can get away with for a monobloc using a Modulus-286.
Just trying to decide on the size of chassis I can get away with for a monobloc using a Modulus-286.
the single point mounting should be well below centre height to get optimum performance from the heatsink. Roughly 35% to 40% up from the bottom of a vertically aligned heatsink.
When two rows of devices are use they should be mounted at roughly 25% up and 60% up from the bottom.
When two devices are mounted along side each other, you want an equal area of back plate dedicated to each interface, i.e. divide the backplate into equal areas with vertical virtual separators and place each device into the centre of their area.
If part of the heatsink is a long distance from the device, then that part will run cooler than parts that are closer. This can take a lot of Rth s-a out of the equation when wide and thin backplates are used.
The manufacturer assumes that the whole backplate is heated to the same temperature when they quote Rth s-a in their specification.
When two rows of devices are use they should be mounted at roughly 25% up and 60% up from the bottom.
When two devices are mounted along side each other, you want an equal area of back plate dedicated to each interface, i.e. divide the backplate into equal areas with vertical virtual separators and place each device into the centre of their area.
If part of the heatsink is a long distance from the device, then that part will run cooler than parts that are closer. This can take a lot of Rth s-a out of the equation when wide and thin backplates are used.
The manufacturer assumes that the whole backplate is heated to the same temperature when they quote Rth s-a in their specification.
Last edited:
0.47 K/W is enough for a mono Modulus-286 - even for sine wave operation. It's plenty for music reproduction.
A stereo Modulus-286 will need a slightly bigger heat sink for sine wave operation into 4 Ω load. 0.47 K/W is enough for music reproduction, however.
I use a crest factor of 14 dB for music signal. Should you prefer heavily compressed music, you'll want a heat sink that's closer to my recommendations for sine wave operation.
In practical terms, the 2U Dissipante series from ModuShop (and DIY Audio Store) is almost enough for sine wave operation in stereo into 4 Ω loads. Mine reached 65 ºC with both channels providing a sine wave at Pout_max/2 into 4 Ω loads. That's too hot for my taste and on the hairy edge of triggering the thermal shutdown of the chip (isolated package, would improve with the non-isolated package though you then have to deal with isolating the package from the heat sink).
The 2U Dissipante heat sink is plenty for mono. Assuming one Modulus-286 board per heat sink.
With the Modulus-286, the distance between LM3886es is 1.5 inches, so your hands are tied there. In case of the 2U Dissipante, I used 19 mm standoffs by the front of the board to secure it to the bottom plate of the chassis along with the two mounting brackets holding the board to the heat sink. The LM3886es ended up around 40 % up on the heat sink.
I would mount the board roughly centred along the length of the heat sink. You probably won't trade off that much thermal resistance by sliding the board left/right a little bit. I think if you start to ask those kinds of questions, you should probably use a bigger heat sink.
Tom
A stereo Modulus-286 will need a slightly bigger heat sink for sine wave operation into 4 Ω load. 0.47 K/W is enough for music reproduction, however.
I use a crest factor of 14 dB for music signal. Should you prefer heavily compressed music, you'll want a heat sink that's closer to my recommendations for sine wave operation.
In practical terms, the 2U Dissipante series from ModuShop (and DIY Audio Store) is almost enough for sine wave operation in stereo into 4 Ω loads. Mine reached 65 ºC with both channels providing a sine wave at Pout_max/2 into 4 Ω loads. That's too hot for my taste and on the hairy edge of triggering the thermal shutdown of the chip (isolated package, would improve with the non-isolated package though you then have to deal with isolating the package from the heat sink).
The 2U Dissipante heat sink is plenty for mono. Assuming one Modulus-286 board per heat sink.
With the Modulus-286, the distance between LM3886es is 1.5 inches, so your hands are tied there. In case of the 2U Dissipante, I used 19 mm standoffs by the front of the board to secure it to the bottom plate of the chassis along with the two mounting brackets holding the board to the heat sink. The LM3886es ended up around 40 % up on the heat sink.
I would mount the board roughly centred along the length of the heat sink. You probably won't trade off that much thermal resistance by sliding the board left/right a little bit. I think if you start to ask those kinds of questions, you should probably use a bigger heat sink.
Tom
Last edited:
Tom,
Since mod-86 has dc servo, does this imply that differential signal with some dc offset is accepted as input for mod-86?
I am wonder if e.g.: I could connect differential signal from es9038 after I/V or raw output from ak4497 directly into mod-86. In those two cases there is some dc offset.
Thanks,
Since mod-86 has dc servo, does this imply that differential signal with some dc offset is accepted as input for mod-86?
I am wonder if e.g.: I could connect differential signal from es9038 after I/V or raw output from ak4497 directly into mod-86. In those two cases there is some dc offset.
Thanks,
I would not recommend connecting any signal with a DC offset to the Modulus-86.
The DC servo is intended to null the internal offsets of the Modulus-86. It will also compensate for reasonable DC voltages (say a few 10s of mV) on the input. However, if you deliberately apply DC to the input and exceed the compensation range of the DC servo, the output of the Modulus-86 will rail and rude noises will be emitted by your speakers (before they catch on fire!).
Tom
The DC servo is intended to null the internal offsets of the Modulus-86. It will also compensate for reasonable DC voltages (say a few 10s of mV) on the input. However, if you deliberately apply DC to the input and exceed the compensation range of the DC servo, the output of the Modulus-86 will rail and rude noises will be emitted by your speakers (before they catch on fire!).
Tom
Last edited:
Exactly. That's what capacitors are for. The Nichicon Muze ES and UES are sonically transparent and I can't measure their impact either when used at line level. Same for NP0/C0G ceramics.
Tom
We are doing too much agreeing of late.
I hope this is not just the honeymoon period !
I guess we've both grown. I'm OK with that.
It helps me out tremendously when you nudge people in the right direction here. Thank you. I, on the other hand, can set up a force field in the direction I consider to be the right direction by selectively agreeing and disagreeing. It seems to be working.
We're all just rats in the same Skinner Box. Here! Have another food pellet.
Tom
It helps me out tremendously when you nudge people in the right direction here. Thank you. I, on the other hand, can set up a force field in the direction I consider to be the right direction by selectively agreeing and disagreeing. It seems to be working.
We're all just rats in the same Skinner Box. Here! Have another food pellet.
Tom
A good question. In planning my build I am considering adding a DC protection circuit, even though Tom and most of the builders here haven't used them.
I personally don't use speaker protection as the LM3886 is quite well protected as it is. But if it makes you sleep better at night, sure...
I should probably make a speaker protection board as there are some fairly easy way to muck up the performance if you do it wrong. The choice of relay matters greatly (as Doug Self writes about in his amp design book).
Tom
I should probably make a speaker protection board as there are some fairly easy way to muck up the performance if you do it wrong. The choice of relay matters greatly (as Doug Self writes about in his amp design book).
Tom