Modulus-86 build thread

Looks like Connex can do +/- 32 on SMPS800RE. How would modulus 86 perform at that rail driving 4ohm loads?
I would not run more than two channels on a single SMPS300RE (or SMPS300REh). You could potentially entertain the thought of running six channels on an SMPS800RE, but I don't know if Connex can make them for ±30 V. I do know that they can make a ±36 V version of the SMPS800RE but that limits your amp to 8 Ω operation.

Six channels on a single Power-686 would be alright. Maybe a little light on capacitance compared to many DIY builds, but better than the vast majority of commercial amps. The Power-686 has 4x22000 uF. You might now be able to find a 27000 uF or 33000 uF that'll fit. The board accepts snap-in caps with 10 mm pin pitch and up to 40 mm in specified diameter (so actual max diameter is 41-42 mm).

Tom
x c
 
Looks like Connex can do +/- 32 on SMPS800RE. How would modulus 86 perform at that rail driving 4ohm loads?
That's right on the hairy edge. The LM3886 I measured started to show degraded performance at ±32.5-33 V.

I think the Connex has a trimpot on it for adjustment of the output voltage. You could get a ±32 V version and just turn the voltage down a smidge. You might want to double-check with Connex to make sure this is possible.

Tom
 
hi i need some advice on PSU.

My plan is to assemble caraudio amplifier based on Modulus86 – two or four channels one.
As PowerSupply I’m going to use DC to +-DC converter. I couldn’t find many regulated converters (I’ve found one from CONNEX but with +-30V output) so I’ll probably have to use unregulated one. Problem is that for unregulated converters level of output voltage is tied to input value. For example unit rated +-24V@12Vdc-input will change to ~+-27V@14Vdc-input.

Car’s electric installation voltage is not constant and changes within 12-14,5V range (more or less) and it might change rapidly from 12 to 14V.


The questions are:
a/ will power delivered to speaker change with varying PSU voltage (= volume)??
PSU output voltage dictates available output power from Modulus86 - I’m aware of that.
Gain is fixed [14dB for mod.86 R2.4 with THAT1206 ] and input from DAC is constant.
… so there should be no change ???

b/ will input sensitivity change with varying PSU voltage (clipping threshold will not be constant)

.. or something else?

thanks for any advice
 
The gain of the Modulus-86 does not depend on the power supply voltage. The output voltage does not depend on the power supply voltage either. Basically, as long as you stay below clipping the amp will provide Vout = A*Vin.

The clipping voltage does depend on the power supply voltage. With an 8 Ω load the LM3886 (and thereby the MOD86) will clip once the output voltage approaches about 2.5-3 V of the rail voltage. So if you provide ±24 V to the amp you can rely on around 28 W max output power. ±30 V gets you to the specified 40 W max power. And if you bump the rail voltage up to ±35-36 V you'll get over 50 W. All numbers for 8 Ω load. Also note, as pointed out two posts back, with 4 Ω load the LM3886 will current limit on the peaks with a ±32.5-33 V power supply so stay at or below ±30 V for 4 Ω load.

The input sensitivity is defined as the input voltage required to drive the amp to clipping. So since the max power at clipping depends on the supply voltage, so does the input sensitivity. That's why the input sensitivity is a meaningless spec in my opinion. It's an attempt to express the maximum output power and the gain as one number.

Tom
 
@tomchr Tom, I have a clarification question about Mod86 performance and +/-36VDC supply.
I know the LM3886 will be current limited with +/-36VDC and 4 Ohm load. However, I know I would never use more than 10 watts or so in my normal listening. If the speakers were 4 Ohm (I don't have or use any currently like that), would lower max output affect the performance of the first 10 watts of output? Or is any load at 4 Ohm just not recommended above +/-32 VDC?
 
The first 10 W is probably OK, but you'll have 3-6 dB lower headroom than you would had you used ±30 V.

Hopefully these two pictures are worth a thousand words each:
A_LM3886_ Maximum Output Power vs Supply Voltage (8 ohm, 50 Hz, 20 kHz BW).png

A_LM3886_ Maximum Output Power vs Supply Voltage (4 ohm, 50 Hz, 20 kHz BW).png


Tom
 
So for a given speaker load, a lower supply voltage will lead to lower THD before clipping? Is there a view as to how low this trend would go until reversal (i.e., when a lower voltage would lead to higher THD)? I use Mod 86 to drive headphones (Hifiman HE6 at 60 ohm) and 8ohm single-driver speakers with ~105db sensitivity, so likely no more than 10w. I am powering the amp with a +/- 24V 4.8A linear regulated PS because that's what I have on hand. But I can make the voltage even lower if warranted.

Btw, I previously built my version of the LM4780 "done right" with a linear regulated PS, which everyone (mostly professional classical musicians) that listened to it liked. But your Mod 86 is a clear step up -- much smoother, due to the absence of that hint of upper-frequency harshness in non-composite LM amps you discussed about somewhere. Previously the sound is clean and sharp (if not a bit metallic), now it's clean and liquid-like. Great work!
 
So for a given speaker load, a lower supply voltage will lead to lower THD before clipping?
I would be a bit careful with that conclusion. It's entirely possible that the lower THD at lower supply voltage is a measurement artifact. I wasn't interested in that part of the curve, so I didn't pay too close attention to that. It's very possible that I performed repeated measurements starting at the lowest supply voltage in the plot. This would mean that the heat sink and IC are running hotter at the higher supply voltages. This would lead to higher noise and thereby higher THD+N in the flat part of the curve.

Great work!
Thank you. I'm glad you like it.

Tom
 
  • Like
Reactions: mudihan
Thanks for all the patient explanations on power supplies for the Mod86, @tomchr.

I’ve had such success with my big external linear +/- 24VDC power supply with my class A amps that I want to try it with other amps, too. Is there any reason not to use it with the Mod86? I realize it would be massive overkill for the minimal load, but it‘s low noise and robust. It’s been a while since I used my Mod86 and it’s time to try it out again in the main system as the weather is heating up in my neck of the woods.
A miniscule build without the power supply in the chassis would be great. :idea:
 
I'm not overly concerned about dips in the impedance as long as the nominal impedance is 4 Ω or above. My reasoning is that the narrow dips won't require much energy to fill, so they won't push the amp that hard. That's of course a hand-waving argument, but without know the phase of the impedance and the SOA curve of the devices within the LM3886 IC that's as good as it gets.

Tom
 
The build is coming along…
Replaced few parts, like
  • input impedance resistors to 47k
  • film type cap for Zobel network
  • Panasonic aluminum organic polymer caps for LM3886 decoupling
  • Upped LM317 and 337 regulator caps to 47uF
Now just waiting for the rest parts in the mail…

Super excited to give this one a listen once it’s done!
 

Attachments

  • IMG_1691.jpeg
    IMG_1691.jpeg
    423.1 KB · Views: 159
  • IMG_1689.jpeg
    IMG_1689.jpeg
    406.1 KB · Views: 161
  • Like
Reactions: wyup