Hi Anand,Pity. What a travesty.
Unfortunately, you failed to realize that the input stages of the Modulus 86 already runs in "Class A." And honestly, just move on up to the Modulus 186/286/686 series. But since they are SMD, it doesn't allow for "tweaking." See post 4701 for relevant commentary. I guess this diy audio project is certainly a journey for you and not a destination!
Best,
Anand.
Thank you for your comments. I already have mod-686 and mod-86 classic builds. Still prefer the moded one I described over those two.
Best
Tweakers will tweak...
One of my commercial clients has often asked about the various "discrete opamps", including many of those mentioned by Tio. Thus, I've examined their data sheets quite a bit and measured a few samples. I remain skeptical of "discrete opamps" for a number of reasons:
The same can be said for the discrete voltage regulators. I've looked at the Hypex HPR12 and HNR12 ±12 V regulators. They're great regulators and unlike many other vendors Hypex offers a full data sheet that I trust. They're fantastic regulators, but still don't offer anything I can't get with a pair of TPSxxxx regulators from TI at a fraction of the cost. They do have red LEDs, though.
Perhaps TI should take note. Package the OPA1611 with a red LED in a transparent package. Sell it as an audiophile version at 100x the cost.
I truly don't understand the attraction of the JRC MUSE series. Their data sheets are really sparse and the ICs don't offer anything that resembles stellar performance. They offer good performance, but not stellar performance. Yet, they cost 3-4x more than an OPA1611 (which does offer stellar performance) and have an aura of mystique surrounding them. Go figure.
Exactly! However, unlike swapping 6922 equivalents, they're taking considerably more risk when swapping opamps. In many circuits, you can plop in any opamp. But in other circuits you cannot - or at least should not without careful analysis. I trust that Tio tested the amp on a dummy load with an oscilloscope and signal generator before hooking up his speakers.
Well said. I can see how that approach wrecks DIY for some, though. My designs don't lend themselves well to the continuous loop of tweaks. Rather, my designs are in the category of "build it once then find another hobby", which some may find dissatisfying. OK, so maybe you don't have to find another hobby just yet. After all, there are many other components in a stereo system to geek out about, but still...
Tom
One of my commercial clients has often asked about the various "discrete opamps", including many of those mentioned by Tio. Thus, I've examined their data sheets quite a bit and measured a few samples. I remain skeptical of "discrete opamps" for a number of reasons:
- They don't provide any better performance than I can get with an OPA1611/12 (based both on their data sheets and on lab measurements).
- The "discrete opamps" are 30+ dB more expensive than the OPA1611/12 and closer to 40 dB more expensive than the LME49720.
- Many of the "discrete opamps" have a 2nd order loop gain response, i.e. -40 dB/dec rolloff from their dominant poles. They then have a phase compensation network to make them unity gain stable. Such an amp will work well as a buffer, but will oscillate in a gain stage unless you add an external compensation network. They're not plug-n-play replacements for an actual opamp.
The same can be said for the discrete voltage regulators. I've looked at the Hypex HPR12 and HNR12 ±12 V regulators. They're great regulators and unlike many other vendors Hypex offers a full data sheet that I trust. They're fantastic regulators, but still don't offer anything I can't get with a pair of TPSxxxx regulators from TI at a fraction of the cost. They do have red LEDs, though.
Perhaps TI should take note. Package the OPA1611 with a red LED in a transparent package. Sell it as an audiophile version at 100x the cost.
I truly don't understand the attraction of the JRC MUSE series. Their data sheets are really sparse and the ICs don't offer anything that resembles stellar performance. They offer good performance, but not stellar performance. Yet, they cost 3-4x more than an OPA1611 (which does offer stellar performance) and have an aura of mystique surrounding them. Go figure.
Exactly! However, unlike swapping 6922 equivalents, they're taking considerably more risk when swapping opamps. In many circuits, you can plop in any opamp. But in other circuits you cannot - or at least should not without careful analysis. I trust that Tio tested the amp on a dummy load with an oscilloscope and signal generator before hooking up his speakers.
Well said. I can see how that approach wrecks DIY for some, though. My designs don't lend themselves well to the continuous loop of tweaks. Rather, my designs are in the category of "build it once then find another hobby", which some may find dissatisfying. OK, so maybe you don't have to find another hobby just yet. After all, there are many other components in a stereo system to geek out about, but still...
Tom
sounds exactly like the rolex vs grand seiko debate people will always attribute better performance to more costly parts and in some cases this is true.
but at some point we all have to admit we are all slaves to psycho acoustics and just the modding of a part can be all the brain needs to be persuaded the sound is better.
i dont doubt the mod has brought a improvement but is that a perceived improvement or a measured one?
but at some point we all have to admit we are all slaves to psycho acoustics and just the modding of a part can be all the brain needs to be persuaded the sound is better.
i dont doubt the mod has brought a improvement but is that a perceived improvement or a measured one?
Last edited:
I'm sure it does a wonderful job at capturing the small signal parasitics of the transformer and PCB. Sadly, we're interested in the large signal parasitics. In particular, we're interested in the parasitics of the rectifier diodes when they go through reverse recovery. That'll only be accurately captured by measuring the circuit at its intended working voltage. I also question the need for the additional capacitor in the CRC snubber. To eliminate the ringing, only an RC is needed. I've yet to see a good argument for the added C (Cx) in the CRC snubber.
You may also find Rod Elliott's article on the topic informative: Snubbers For PSUs
Unfortunately for the marketing department, the red LEDs make better bias references.
Tom
I doubt the mod has brought an improvement. If an improvement was perceived, I can come up with several arguments based on cognitive psychology that could explain why that might be the case. Just having blown $200 on discrete opamps would certainly lead one to perceive a difference (expectancy bias).
Some may argue that tweaking is part of the hobby. And that's a valid point.
Tom
The purpose of the second capacitor, "Cx" (across the secondary), is to swamp out the rectifier (and wiring) capacitance so that they become an insignificant portion of the total. Now you can test and tweak the snubber on the lab bench long before you build the complete power supply and hook it up to the AC mains, secure in the knowledge that the rectifiers and wiring will not change the result.
But, as Tom correctly points out, this extra capacitor does cost extra money. The Cx capacitor that I often recommend for chipamp supplies (EPCOS B32529 film cap, 10 nF, 250VDC rating) costs $0.31 in qty=1 at Mouser. It's part number 871-B32529C3103M289 . Bipolar PSUs (positive and negative output voltage) need two of them.
By the way, check out Quasimodo post #142, written by diyAudio member "peufeu" in December 2013. He exactly describes the high pass network for monitoring transformer ringing in situ, that Rod Elliott rediscovered this year.
But, as Tom correctly points out, this extra capacitor does cost extra money. The Cx capacitor that I often recommend for chipamp supplies (EPCOS B32529 film cap, 10 nF, 250VDC rating) costs $0.31 in qty=1 at Mouser. It's part number 871-B32529C3103M289 . Bipolar PSUs (positive and negative output voltage) need two of them.
By the way, check out Quasimodo post #142, written by diyAudio member "peufeu" in December 2013. He exactly describes the high pass network for monitoring transformer ringing in situ, that Rod Elliott rediscovered this year.
Tom, if you think you have it bad with ear lead swappers, you should read the dac threads. There's all sorts of unmeasured junk being thrown inside some of them as mods.
I think it fair to tell you that I modded my bridged 86p monos today. I added a light pipe from the led on my soft start board to the third pin hole on the front panel so I can now tell between unplugged, plugged in, and powered up from the the 3 front panel leds. A mod I'm sure you'll be able to get behind. Cos frankly there's bugger all else to do with them but play music...
I think it fair to tell you that I modded my bridged 86p monos today. I added a light pipe from the led on my soft start board to the third pin hole on the front panel so I can now tell between unplugged, plugged in, and powered up from the the 3 front panel leds. A mod I'm sure you'll be able to get behind. Cos frankly there's bugger all else to do with them but play music...
You need more than 10 nF to swamp out the diode parasitics, though. This brings the validity of your approach into question. Your method also doesn't do anything to address the wiring inductance.
Thankfully, the snubber components are not that critical. As both Rod and I mention, 10 Ω + 100 nF in an RC snubber configuration works very well.
Tom
...most of which utterly wrecks the DAC performance. Yeah... Not my idea of a good time.
GOOD GAWD!!
But did you use audiophile grade light pipes of pure grain-oriented glass drawn from a single glass crystal? Was the crucible heater supplied from a filtered power supply and connected by pure solid silver conductors?
Yeah... I designed them that way. I'm glad you like them.
Tom
The diodes that you posted in 686 #1072 (link) are about 0.2nF, your image copied below.
And the four bridge devices are connected across the secondary (forming a resonant RLC circuit) in a series-parallel configuration of diodes where only two of the four diodes is biased at 0V (the high capacitance regime) when reverse recovery happens. The diodes in series with the high-C devices, are decidedly low-C since they are at about -28V reverse bias.
0.2nF also happens to be, approximately, the zero bias junction capacitance of the diodes in the GBJ four-in-one bridge rectifier package used in the lower current Power-86 PSU.
It seems to me that 10nF >> 0.2nF ... is this wrong?
Just ordered a pair of Mod86 boards and waiting on parts to arrive. My music chain will consist of a Topping D70 to the Mod86. The D70 has an XLR output of 4.0v @ 0dBFS. Does this mean that the volume control will be limited in range? In other words, its going to get loud quickly with small increments.
How do you like the SHD? I'm starting to get somewhat serious about revamping our main living room system, and the SHD seems like an almost ideal fit (maybe aside from price) - I think I can live with a 2-way setup with a passive xover on M/T and active to the woofers. I'm 'fairly sure' I could put together a Linux-based system that does much the same thing, but not without a fair bit of work (and when the wife can't watch TV because "my stuff" doesn't work, things aren't good)
Mod-86 content: I have what I think are an old pair of 2.1 boards I never got around to building, mostly because I stumbled across someone selling a completed Mod-86 here on the Swap Meet. My guess is that finding parts to build out a V2.1 set will be basically impossible, but I figured I'd get a more informed opinion before abandoning the idea.
I think a Mod-86 should be sufficient for the M/T section of the planned speakers based on my testing with my current unit, but I'd have to consider a 286 or 686 for future-proofing if I bought new assembled units. If I could build out my boards for 'cheap' though, that might allow me to defer that decision and/or put more $$$ into the rest of the system
I like it. Then again, I designed the analog sections in the SHD for MiniDSP, so maybe I'm biased. The analog sections were designed to get the full performance out of the DACs and ADCs.
I had the MiniDSP 4x10HD before (and still do). Its noise floor is a bit high, so with it as a preamp, I can hear a hiss when I put my ear against the speaker. With the SHD that hiss is gone.
It seems the SHD uses a more powerful DSP than the 4x10, so you have more biquads available for filters. That's nice as the LXmini XO pushed the 4x10 to about the max.
I'm now rocking a pair of KEF R700, so the SHD just acts like a DAC. I rather like it. I have not tried the Dirac room correction, though I hear good things about it.
All that said, I still haven't had time to put the SHD on the bench and measure it. One of these days...
If I was to nitpick (and of course I am
), I'd say that MiniDSP went cheap on the remote. I mean... It does the job just fine, but it is rather lacklustre from a human factors point-of-view. The function you use the most (volume control) receives the least real estate and the buttons are arranged left/right (volume down/up) rather than up/down. That said, the remote does the job.Linux.... Sssss... Yeah... I won't stoke that fire. I've run Reaper and ReaEQ on the 2010 MacBook Pro that's my music machine. That worked very well.
Oy... 2.1. Yeah. You're doomed.
Actually... If you could find a pair of LME49710HA (TO-99 can package), the rest of the BOM is still obtainable. Some of the ceramic caps went obsolete because TDK decided to change the recipe. So you replace FK with FA or FG in the part number and off you go.The real question is whether you can find a genuine LME49710HA for less than it would cost you to replace the boards with the current revision. It may be possible. Just make sure you get a genuine part. The Modulus-86 relies on the characteristics of the LME49710 to work.
For multiway systems optimized for power, I generally recommend running a ±35-36 V power supply. That gives the maximum output power, but you have to select the amplifier modules according to the drivers:
8 Ω drivers can be powered by a Modulus-86 or -186.
4 Ω drivers can be powered by a Modulus-286 (or -686 if you need more power).
If that blows your budget, I recommend running a ±28-30 V supply and using Modulus-86 or -186 everywhere.
Tom
dwk123,
I have a pair of Parallel 86 boards fully built (sans LM4780 which I have here as well), with all genuine parts, etc...and tested. They are ready to be attached to some heatsinks. Give me a holler if interested. I just ordered too many boards from Tom! Let me see, ... I’ve built 16 channels of amplification from Tom’s designs...you can say I am a believer...but I am not the only one
Best,
Anand.
I have a pair of Parallel 86 boards fully built (sans LM4780 which I have here as well), with all genuine parts, etc...and tested. They are ready to be attached to some heatsinks. Give me a holler if interested. I just ordered too many boards from Tom! Let me see, ... I’ve built 16 channels of amplification from Tom’s designs...you can say I am a believer...but I am not the only one
Best,
Anand.
Thanks Tom! Sorry, hopefully this is my last question. I have an SMPS300R +-30v laying around that I'd like to use. Will this be enough for a stereo build or should I get another SPMS300 for a dual mono build? I'll be powering a set of 6 ohm and 4 ohm speakers with this amp. Thanks again.