Tom,
In addition to your recommendation, if someone wants to do a dual-mono build, then a 300VA transformer per channel for the Modulus-686 would be enough?
In addition to your recommendation, if someone wants to do a dual-mono build, then a 300VA transformer per channel for the Modulus-686 would be enough?
I'd go with 300-400 VA for a mono block.
Those who wish to build dual mono amps ought to consider a mains transformer with four secondaries. 4x25 VAC @ 600-800 VA. Then use a Power-686 per channel. Even the theoretical difference in performance between two separate transformers and one transformer with four secondaries is pretty darn small. After all, to transformers will "talk" through the primary connections. The Modulus-686 isn't sensitive to power supply perturbations anyway. The benefit of only having to deal with one transformer is substantial, though.
I'm pretty sure Antek has some options at 4x25 VAC. If not, Toroidy in Poland will happily wind a transformer for you. Their cost is very reasonable.
Tom
Those who wish to build dual mono amps ought to consider a mains transformer with four secondaries. 4x25 VAC @ 600-800 VA. Then use a Power-686 per channel. Even the theoretical difference in performance between two separate transformers and one transformer with four secondaries is pretty darn small. After all, to transformers will "talk" through the primary connections. The Modulus-686 isn't sensitive to power supply perturbations anyway. The benefit of only having to deal with one transformer is substantial, though.
I'm pretty sure Antek has some options at 4x25 VAC. If not, Toroidy in Poland will happily wind a transformer for you. Their cost is very reasonable.
Tom
I run Toroidy 300va 2x25V secondaries, audio grade traffos in my monoblocks, 2x 86P boards, bridged, per monoblock. With 1x rectifier smoother board per amp board, I guess that makes them 486's ? Almost.
They never run out of steam, at least in my modest sized room with ns1000m.
They never run out of steam, at least in my modest sized room with ns1000m.
The main difference between the Power-86 and the Power-686 is that the '686 has twice the capacitance. I designed it for use in a stereo Modulus-686 amp.
With the Power-86 the output voltage of the supply will droop a bit more than for the Power-686, so you won't reach the full 360 W into 4 Ω, but you'll probably get well past 300 W. I honestly wouldn't worry too much about it.
Tom
With the Power-86 the output voltage of the supply will droop a bit more than for the Power-686, so you won't reach the full 360 W into 4 Ω, but you'll probably get well past 300 W. I honestly wouldn't worry too much about it.
Tom
@zman01
You probably recall I used an 800VA Toroidy with quad 25V secondaries. Works perfectly. I have yet to measure the amp to the point at which the voltage rails droop ;-)
If I were to build with dual toroids I would go for full monoblocks in separate enclosures but other than curiosity I have no need at this point.
Best,
Anand.
You probably recall I used an 800VA Toroidy with quad 25V secondaries. Works perfectly. I have yet to measure the amp to the point at which the voltage rails droop ;-)
If I were to build with dual toroids I would go for full monoblocks in separate enclosures but other than curiosity I have no need at this point.
Best,
Anand.
Last edited:
I agree. If you're going through the expense of double everything you might as well use two chassis. ModuShop has some nice mono block chassis for that purpose.If I were to build with dual toroids I would go for full monoblocks in separate enclosures but other than curiosity I have no need at this point.
Tom
It also makes for neater wiring which keeps me up at night. We all have our demons 😈
Best,
Anand.
Best,
Anand.
It's not a disorder if it doesn't cause distress... 😉
Tom
Tom
I should clarify, it's a pair of 86-parallel boards, the old stereo chip boards, per channel as bridged monos. With 2 of the small psu boards per channel, one per board. 300va, 88,000uf.
Parallel-86 to be specific. 🙂 That used the LM4780 which has since then been discontinued by the manufacturer. The modern-day equivalent is the Modulus-286.
If you stay at ±30 V or below you can bridge two Modulus-286 (or Parallel-86) boards even with a 4 Ω load. That gets you close to the Modulus-686 Safe-n-Sane in performance.
Totm
If you stay at ±30 V or below you can bridge two Modulus-286 (or Parallel-86) boards even with a 4 Ω load. That gets you close to the Modulus-686 Safe-n-Sane in performance.
Totm
That's exactly how it's set up Tom, I even have a spare pair of LM4780, just in case. It's been faultless.
The LM3886 is pretty rugged. The LM4780 is basically two LM3886 in one package, so I'd expect it to be rugged too. It's good with a spare, though.
Tom
Tom
Don’t just concentrate on the phase and impedance curve. There is data to show that the EPDR (Equivalent Peak Dissipation Resistance) is even more relevant particularly for Class B(or AB). You can read up on it from articles from Dr. Jack Oclee-Brown (KEF). In essence you have to look at the combination of impedance and phase and simply put, if you have obtuse phase angles along with low impedance, the overall EPDR is lower than the impedance at that frequency. In example for your Revel Studio 2, at 300Hz you are at 3.9 ohms but with a ~(-)15 degree phase angle. That will drop the impedance by a certain magnitude (I don’t know how much). When the phase is closer to 0 degrees there is little to no affect on the impedance. Stereophile, Erin’s Audio Corner and even REW are now calculating EPDR using the phase and impedance curve measurements to give the user a real world look at what a Class AB amplifier has to face. If you have a colleague who has a pair of the Revel model you desire, I suggest you have them measure it using REW. It should be eye opening (particularly when realizing that the speaker’s sensitivity is in the high 80’s).
Best,
Anand.
Thanks very much Anand, this is a very insightful comment. It helps me a lot, even though I can't fully grok the involved concepts. In the Stereophile speaker measurements some additional interpretation is offered, which is good enough guidance for me to base a decision on. The Revel Studio2 measurements are located here: https://www.stereophile.com/content/revel-ultima-studio2-loudspeaker-measurements
I decided to take a step back and go for the F208's (https://www.stereophile.com/content/revel-performa3-f208-loudspeaker-measurements):
The F208's are rated for 50-350 watt, dip a bit below 4 ohm and have some frequencies where relatively low impedence occurs together with relatively high phase angle. As opposed to the Studio2, I think a single stereo Mod-686 at ±36 V will be able to fully control the F208. Or at least enough for me to stop caring about it.
It sounds like you've found this page: https://sound-au.com/project124.htm. I've considering rebuilding my dummy load following that pattern. I've been wanting to change a few things about it anyway, including adding support for remote control so I can perform automated amp testing with the APx555 similar to what I do with the HP-LOAD.
I would definitely not recommend running at ±40 V if you're using an unregulated power supply. With a regulated supply at least you know that you'll get ±40 V even on days where the mains voltage is running a bit high. The drawback is that you get higher heat dissipation and you could potentially engage the protection circuits in the LM3886es a bit earlier than desired. Turning the supply voltage up was preferable to accepting the lost cost of the SMPSes.
Tom
I hadn't seen that dummy load project before, but it's an informative article, thanks! I'm not planning on involving an oil container though. I bought two 200mm Noctua 12 V PWM fans with a fan controller and a Mean Well PSU. Together with 8 load resistors, I'll put it all in a rig with Makerbeam or another T-slot profile. I'll wrap polycarbonate panels around to create a kind of wind tunnel, and I'm looking at 3D printing an internal duct to redirect air over the 8 resistors as much as possible.
I got a whole bunch of nice 4mm banana sockets from Staubli, rated for 32 A. Each resistor wire gets a socket, plus a bunch of extra ones, so I can make different configurations using wires with stackable banana connectors.
For the PSU's I might just get a pair of Connex SMPS800RE's at ±36 V each. With the Connex, at least I know the amp will run good, without whining PSU. The Connex PSU's are typically shipped with a mechanical defect though, but at least I can fix that myself and know that I will have a good amp after that. I don't like the risk of having to up the supply voltage to fix a whining Mean Well PSU. The SMPS800RE description notes:
Does this mean that 800 watt divided by the crest factor of 2.5, the PSU can deliver 320 watt continuous RMS power? Like when running a sine wave through the amp? In that case, two of them will go nicely together with the 20 dB gain on my Mod-686 boards. Their input sensitivity is 4 V RMS, and my DAC has balanced outputs around 4 V RMS too. Not that I'm expecting to go that high playing music.It uses half Bridge resonant topology, capable of delivering 800W continuous and 940W peak power for audio applications, with crest factor of at least 2.5.
@tomchr I would like to load test the Mod-686 at 4 ohm. I can get 4 V RMS from the signal generator on my Neutrik BNC -> XLR plug, but it will only be on one pin. Does this matter when testing the Mod-686? As in a typical situation you'd get +2 on one pin and -2 on another pin from a balanced source.
@TNT On both SMPS300REh that I have, the terminals where you place the faston connectors for your AC and DC wires can become loose. I am assuming the construction decisions for the SMPS800RE were the same/similar. The fastons can be pretty sturdy at times, so you apply quite a bit of force when placing or removing them. You need the terminals to be sturdy as well.
My Connex PSU's have too little solder, the terminals are cut almost flush with the bottom of the PCB, and there's not enough room for enough tin to grab on to enough metal on the bottom. You can fix it by adding some tin from the top of the PCB's. I don't find them particularly easy to solder though, don't remember the temp I used but I needed to raise quite a bit as compared to soldering something like a Modulus-86 PCB.
My Connex PSU's have too little solder, the terminals are cut almost flush with the bottom of the PCB, and there's not enough room for enough tin to grab on to enough metal on the bottom. You can fix it by adding some tin from the top of the PCB's. I don't find them particularly easy to solder though, don't remember the temp I used but I needed to raise quite a bit as compared to soldering something like a Modulus-86 PCB.
I've had that happen as well. I think the flush cutting of the terminals result in cold solder joints during wave soldering. The terminals never get hot enough for the solder to bond to them. A quick touchup with a soldering iron from the top fixes it.My Connex PSU's have too little solder, the terminals are cut almost flush with the bottom of the PCB, and there's not enough room for enough tin to grab on to enough metal on the bottom.
Tom
Nope. The MOD686 only cares about the difference from IN+ to IN-. Both amp halves will be driven even with a single-ended source.I would like to load test the Mod-686 at 4 ohm. I can get 4 V RMS from the signal generator on my Neutrik BNC -> XLR plug, but it will only be on one pin. Does this matter when testing the Mod-686?
Tom
I'm curious why that applies especially to Class B/AB. I would have thought that impedance is impedance. I guess I need to go and dig up those papers.There is data to show that the EPDR (Equivalent Peak Dissipation Resistance) is even more relevant particularly for Class B(or AB).
When the phase angle of the impedance is close to 0º, the impedance is nearly resistive. A pure resistance (which only exists in textbooks) has an impedance of R with a phase angle of 0º. Or R+j0 if you prefer cartesian coordinates.
Tom
- Home
- Vendor's Bazaar
- Modulus-686: 380W (4Ω); 220W (8Ω) Balanced Composite Power Amp with extremely low THD