By the way, on an unrelated note, I was just looking at the Modulus 86 diagram today. I noticed that the outputs of the buffering opamp U2 appear to be AC coupled via C20 and C7 to U3. However, the overall circuit topology provided doesn't seem to reflect that?
Shouldn't those two resistors marked be capacitors instead to indicate AC coupling?
Shouldn't those two resistors marked be capacitors instead to indicate AC coupling?
Future (boards on the way) Modulus 86 builder here. This is a big deal for me as I am a tube amp guy. It will be great to compare.
Regarding a switching power supply for a one-chassis stereo build, I came across this for $28 each at Arrow.
VGS-200C-36 36v 10% adj, I am building for 8 ohm speakers, will need 2, one for +36v, one for -36v.
Any negative comments such as "it audibly whines" or ? If not, I'll probably give them a whirl.
Regarding a switching power supply for a one-chassis stereo build, I came across this for $28 each at Arrow.
VGS-200C-36 36v 10% adj, I am building for 8 ohm speakers, will need 2, one for +36v, one for -36v.
Any negative comments such as "it audibly whines" or ? If not, I'll probably give them a whirl.
The maximum capacitive load on the VGS-200C-36 is specified to 1500 uF. Each Modulus-86 has 1000 uF on each rail, so you run the risk of having the supply go unstable if you use the supply with a stereo pair of MOD86. Also, at 5.9 A output current it's a bit wimpy for a stereo amp. It should work with a mono block, though.
CUI generally makes pretty good stuff. I have no personal experience with that particular supply, though. I know of a couple of builds that were powered by the Mean Well RPS-200-24 with good results. That's 24 V so you'll get lower output power.
The Mean Well RPS-400-27 will work well. You can adjust a pair of them to near ±30 V and get 40-45 W into 8 Ω, 60-65 W into 4 Ω. The RPS-400-36 is the one where about half of builders have reported audible whine. Most were able to solve this by turning the voltage up to 40 V. One case required a heatsink mounted power resistor across the supply rails to bring the power supply out of "whine mode".
The Connex SMPS300REh is good too. It's a further development of the SMPS300RE. If you're thinking to go that route, definitely get the REh version. It comes mounted on an aluminum plate that you mount to the chassis. That allows for heat dissipation and increases the reliability of the power supply.
Tom
CUI generally makes pretty good stuff. I have no personal experience with that particular supply, though. I know of a couple of builds that were powered by the Mean Well RPS-200-24 with good results. That's 24 V so you'll get lower output power.
The Mean Well RPS-400-27 will work well. You can adjust a pair of them to near ±30 V and get 40-45 W into 8 Ω, 60-65 W into 4 Ω. The RPS-400-36 is the one where about half of builders have reported audible whine. Most were able to solve this by turning the voltage up to 40 V. One case required a heatsink mounted power resistor across the supply rails to bring the power supply out of "whine mode".
The Connex SMPS300REh is good too. It's a further development of the SMPS300RE. If you're thinking to go that route, definitely get the REh version. It comes mounted on an aluminum plate that you mount to the chassis. That allows for heat dissipation and increases the reliability of the power supply.
Tom
You may have a little more wiggle room than that. At least with the Mean Well supplies I've used (so mainly the RPS-400-27) I've found that you can expect the supply to provide the rated current, RMS. I.e., if you need to drive 10 A (peak) to the load, you can choose a power supply that can deliver 7.1 A (= 10/sqrt(2)) DC. This may not work in all cases, but it has in the cases I've encountered.
Since all the Modulus amps have balanced inputs, you can also employ another li'l trick: Invert the phase on the input and output of one channel of a stereo amp. That's as simple as swapping pins 2,3 on the input and swapping the speaker output connections. As the left and right channel in stereo material tend to be highly correlated, inverting the phase on one channel will load the SMPS(es) more symmetrically. That prevents bus pumping and adds even more wiggle room.
Take the "Safe-n-Sane" build of the Modulus-686 as an example. A stereo Safe-n-Sane Modulus-686 (i.e., MOD686 powered by ±27 V from a pair of Mean Well RPS-400-27) will need to provide about 12 A peak, per channel, to a 4 Ω speaker. Each Mean Well supply can deliver 14.8 A. 14.8 < 2*12, yet this is enough for the stereo amp to be able to deliver the full output swing into a 4 Ω load on both channels simultaneously.
Tom
Since all the Modulus amps have balanced inputs, you can also employ another li'l trick: Invert the phase on the input and output of one channel of a stereo amp. That's as simple as swapping pins 2,3 on the input and swapping the speaker output connections. As the left and right channel in stereo material tend to be highly correlated, inverting the phase on one channel will load the SMPS(es) more symmetrically. That prevents bus pumping and adds even more wiggle room.
Take the "Safe-n-Sane" build of the Modulus-686 as an example. A stereo Safe-n-Sane Modulus-686 (i.e., MOD686 powered by ±27 V from a pair of Mean Well RPS-400-27) will need to provide about 12 A peak, per channel, to a 4 Ω speaker. Each Mean Well supply can deliver 14.8 A. 14.8 < 2*12, yet this is enough for the stereo amp to be able to deliver the full output swing into a 4 Ω load on both channels simultaneously.
Tom
I see the following stated in this thread:
1. A pair of Mean Well RPS-200-27-C is recommended to power two channels of the MOD86. Its rated current is 5.3A at 27V output.
2. A single SMPS300REh is recommended to power two channels of the MOD86. Its rated current (according to the SMPS300RE data sheet as there is no data sheet for the SMPS300REh) is 4A nominal 5.2A peak at +-30V.
A VGS-200C-36 is rated at 5.9A at 36v. Why would a pair of these to power 2 channels of MOD86 then be considered "a bit wimpy"?
There is also a VGS-350C-36 rated at 9.7A at 36V. It should be plenty! Has anyone tried this power supply?
1. A pair of Mean Well RPS-200-27-C is recommended to power two channels of the MOD86. Its rated current is 5.3A at 27V output.
2. A single SMPS300REh is recommended to power two channels of the MOD86. Its rated current (according to the SMPS300RE data sheet as there is no data sheet for the SMPS300REh) is 4A nominal 5.2A peak at +-30V.
A VGS-200C-36 is rated at 5.9A at 36v. Why would a pair of these to power 2 channels of MOD86 then be considered "a bit wimpy"?
There is also a VGS-350C-36 rated at 9.7A at 36V. It should be plenty! Has anyone tried this power supply?
Heh. I had no idea of that connection.
The name Modulus came from the word modular. My original intent (in 2014) was for people to build multiple boards and connect them together to form various combinations of mono, stereo, bridged, parallel, and bridge-parallel. Decent idea but not super practical and the market steered me in the current direction.Keep in mind that this thread has been going for nearly 10 years now. Opinions and advice shifts over time. Furthermore, note that I do not memorize each and every data sheet out there, so my advice may be based on a quick glance at the headline figures rather than on an engineering analysis. This is DIY after all. The Y in DIY stands for...? Also note that there's some fuzzy in the spec as I assume people will use these amps to reproduce music rather than sine waves. And finally note that I am human. I do make the occasional error.
I aim for a minimum of Iout = Vsupply/(Rload * sqrt(2)), per channel, where Iout is the RMS output current or the DC output current specified for an SMPS. So with a 27 V supply and a 4 Ω load you'll need a supply that can provide 27/(4*sqrt(2)) = 4.8 A for each channel. That's about 130 W. This would push you to a 300 W power supply for a stereo amp (2*130 = 260 W not usually being a standard value).
For a bridged amp like the Modulus-686 the math gets even wilder. A bridged amp will produce 2*Vsupply on the output, so you'll need:
Iout = 2*Vsupply/(Rload*sqrt(2)) for each channel. That's 9.5 A for 27 V, 4 Ω and 12.7 A for 36 V, 4 Ω. Per channel! This pushes you into the territory of really expensive and quite unobtainable power supplies - at least if you're building a stereo amp.
A power supply selected from this math would allow for the reproduction of sine waves.
So this is where the fuzzies begin. Where can we push on the design space to arrive at a solution that allows for good music reproduction with an available (and preferably not too expensive) supply?
The first knob to turn is music vs sine wave. Music has a high peak power but only moderate average power. This is known as the crest factor (CF). With a "linear" power supply (which is not linear at all, but that's another conversation) this fact is often used to reduce the cost, weight, size of the power transformer. Basically the supply is designed for the (much) lower average power. You can play the same game with a switching supply, but you do need to keep in mind that the peak current limit on an SMPS tends to be a hard limit whereas the limit on an unregulated, "linear" supply is more of a guideline. This is precisely why I discontinued the SMPS-86. It was a great li'l supply. 2x60 W. But the hard 2.5 A (or thereabout) current limit meant that some would drive their stereo amps to the point where the supply entered hiccup current limiting mode on the signal peaks. That wasn't quite the music experience they were going for... This is one of the reasons I'm more conservative in my recommendations for SMPSes. I do want the amps to be able to reproduce a 20 Hz bass thump without the power supply turning off.
The second parameter that one can play with is the relative phase of the two channels in a stereo build. Left and right channels tend to be fairly tightly correlated, so by inverting the phase on the input and output of one channel the load of the two channels is spread more evenly across the two supplies in the bipolar (±) supply. So now each supply needs to supply closer to Iout rather than to 2*Iout for a stereo amp. "Closer to". How close? That depends on the correlation between left and right channel, which will vary by the choice of music.
A third parameter is the speaker impedance. One can make life easier for the power supplies (and heat sinks) by optimizing the amp for 8 Ω operation and then accept that with 4 Ω speakers the amp might shut down (or overheat) on the signal peaks. That's usually not my idea of a good time but it can be a perfectly valid approach, for example in an amp intended for use with a specific speaker (LXmini and LX521.4 spring to mind). Also, no "4 Ω" speaker actually measures 4 Ω from DC to daylight, so that's a tricky variable to play with.
There's also the little detail that the LM3886 can't swing all the way to the supply rail, but only goes within 2-3 V of the rail. How close to the rail depends on the load impedance and the supply voltage (as well as process variation, temperature, etc.) So there's a fuzz factor there too. After all, the output current, hence, the supply current is Vout/Rload, not Vsupply/Rload.
Some argue that the amp should be able to produce a brief pulse at full output power and operate continuously at a few watt delivered to the load. "Headroom", they call it. One can design for that with an unregulated, "linear" supply, but with an SMPS the SMPS has to be chosen such that it does not hit its current limit on that brief signal peak at full output power, even if that brief signal peak is a few cycles of 20 Hz. This means that the SMPS will have to be chosen from the peak current requirement and not from the average power spec.
You could also play with the choice of program material. If you only listen to talk radio (Lawrd help you!) or Strauss' "An der schönen, blauen Donau" you could probably skimp more on the power supply than if you like cranking Satriani's "Seven Strings" to 11 on repeat for days.
I have personal experience with the following power supplies: Connex SMPS300RE, SMPS300REh, SMPS800RE; Mean Well RPS-400-27(-C), SE-600-36. Of these I recommend the Connex SMPS300REh, SMPS800RE, and Mean Well RPS-400-27(-C). I know from practical experience that the SMPS300REh is able to power a stereo Modulus-86, -186, and -286 amp, assuming music reproduction. I know that the SMPS800RE is able to power a single channel of Modulus-686 (assuming music, 4 Ω) or a stereo Modulus-686 (assuming music, 8 Ω, ideally channels operated out of phase). I know that the Mean Well RPS-400-27 can power a stereo Modulus-686 (assuming music, 4 Ω, channels operated out of phase).
I have no personal experience with the Mean Well RPS-200-24, but have seen successful builds here.
I hope this gives some insight in to the design space around switching power supplies in audio amps.
Tom
Hello Tom;
On Modulus-86 Rev. 3.0, can i bypass U2 (U2,R3,R8,R9,R13,R14,R15,R16,R19,C4,C11,C15,C17,C19) and connect inputs directly to C2 and C7? What is the purpose of this input buffer. I mean what will be the downside if omit it? Correct me if i am wrong but it doesnt effects the THD+N?
On Modulus-86 Rev. 3.0, can i bypass U2 (U2,R3,R8,R9,R13,R14,R15,R16,R19,C4,C11,C15,C17,C19) and connect inputs directly to C2 and C7? What is the purpose of this input buffer. I mean what will be the downside if omit it? Correct me if i am wrong but it doesnt effects the THD+N?
You can, but you get 2 kΩ input impedance. Few sources can drive that well. You also lose the option to have a gain other than 20 dB.
I've done sine wave testing with the SMPS300RE, which is the same (or very, very similar) circuit as the SMPS300REh. If you run two channels of Modulus-86 on the SMPS300RE at clipping levels, with a sine wave, into a 4 Ω load the supply will overheat after some 20ish minutes. But note that this operating point is the worst case for the power supply. The SMPS300REh has a heat sink (hence the 'h') on the switches and output diodes, so it should be able to last longer. This is why I nudge people towards the SMPS300REh instead of the SMPS300RE.
That said, few listen to sine waves - especially at clipping levels. I have run both the SMPS300RE and the SMPS300REh at clipping levels with a 32-tone test signal with a crest factor of 10 dB. This is the closest I can come to music with a deterministic and periodic test signal. The amp ran at clipping levels on both channels with a 4 Ω load for over an hour at which point I stopped the test.
I've done similar testing on the Modulus-686 Safe-n-Sane configuration.
Tom
