Entire article, generally not as well as the original referenced article. You have to be honest and say your purposes are not purely educational. Just say'in. For me everything I publish is public domain, i.e. I could care less if anyone manages to make some pin money from my ideas. This is the best strategy since these copyrights of schematics etc. have no legal basis at all.
Last edited:
My parts from Mouser arrived Friday, just waiting for the PCB's now and I'll get cracking on the build.
I have a calibrated mic, but I do not think A/B mic measurements will give a fair comparison. Any reasonable amp will sweep a single frequency fairly accurately. When the complex audio signal is amplified, this is where a GOOD amp shows its self. Noise floor, IMD, ability to process transients etc,etc.
I think the ability to reproduce the low level detail is of utmost importance, this is where I hope Tom's amp will shine.
±28 V. I was using a toroid mains transformer, KBPC 2510 rectifier bridge, and 2x22000 uF, 35 V electrolytics for my testing. Worked well...
As Twest points out, if there was a protection circuit, it would have been included in the block diagram and the verbiage of my website. There are many circuits out there that will provide protection in the event of excessive DC voltage on the amp output. I suggest using one of those. I didn't see any reason to reinvent the wheel. Also, as my boards are set up for easy paralleling, bridging, and bridge/paralleling, the speaker protection circuitry is better addressed as an add-on board.
I'm a bit disturbed by that comment. I have posted countless measurements that show exactly the results you can expect if you buy and build my boards. If I have miscommunicated or if the value proposition of my circuit isn't obvious at this point, please let me know how I can improve my communication.
I've taken my best stab at describing my emotional response when I plugged the amp in the first time. It was overwhelmingly positive. I was pleasantly surprised by the amount of detail, the firmness of the bass, the detail in the midrange, and the timbre of metallic instruments that the Modulus-86 delivered through my Dali 3A speakers. However, the majority of the subjective questions asked have been, "how will the Modulus-86 sound in my system, using my source material, my listening environment, etc." I can certainly project my feelings onto the person asking the question and read them back. For that, see the blurb about the sound quality on my website (or jump back a few lines here). However, I can't predict how you will feel when you turn the Modulus-86 on the first time. The best I can do is to describe my own emotional response and back it up with tons of measurements, as I have done already.
As far as sharing the schematic... This is actually my first time not publishing the schematic. If you look on my website, you will notice that I generally publish my circuits. However, what customers are buying when they buy my boards is in part the intellectual property and engineering that went into the schematic and the layout. I think it's pretty reasonable to try to build a self-sustaining hobby on that. Also note that this thread is actually in the Vendor's forum, so the expectation that I'll give my work away for free is perhaps a tad far fetched.
Almost done with the design documentation. Yay...
Thanks,
~Tom
I actually don't mind the writing. The writing is easy. Getting my thoughts together, drawing the figures, and making sure it all makes sense is the main time consumer. Should my needs change, I'll definitely let you know. Thanks!
~Tom
Hi Andrew, if you are referring to just the schematic, then I'm not sure if fair use would cover it. You would probably need to check with the AES. However it is my understanding that if the schematic is redrawn by someone and that redrawn schematic (not the original copyrighted image) is uploaded then there should not be an issue with that. The copyright laws may differ from country to country though.
Some Mod's with more experience than me on this area may chime in.
Tony.
Yes, that gets fiddly. The intent of copyright law is to control copying; for the limited set of countries I'm familiar with the redrawn schematic would have to clearly add value over the original to not be considered a copyright violation. A better approach might be for someone who has the paper to review DIY Audio's existing grounding article to see if adjustments should be made to it. The article's quite good so I suspect the answer may turn out to be no.
I'd suggest quantifying reasonable and accurate. It's not a common A/B but the few datapoints I'm aware of (2SJ109/2SC2240+2SK2381+2SC3519 discrete amp to LME49811+STD03, ICEPower to Hypex UcD) show broadband level increases of 2dB in the highs (one sample each from dynamic tweeter, magnetostat, and AMT). That's well into the audible range and both the "before" amps are widely considered reasonable and accurate. If level matching isn't done to control this then A/B results between amps will be influenced by changes in speaker voicing. One can argue that's the way the test should be performed. One can also argue equalization should be adjusted to avoid penalizing a more performant amp for having better control of the drivers. I would say it's most useful to measure so the changes are known and either equalization done or sufficient time allowed for subjective adaptation to occur before posting one's results.
I've not found increasing GBP or loop gain above the 49811 to move levels. Chipamp GBPs are low enough it seems plausible a move from a chipamp to a composite chipamp with a fast control amp---55MHz GBP for the 49710 in this case versus around 40MHz for the 49811---could result in level shifts.
The other main area of amp-speaker interaction I've encountered is cone breakup. That'll show in measurements---in my limited experience, better control of the driver seems be associated with higher Q resonances---but usually is enough dB down it's hard to track unless the tweeter's disconnected (passive XO) or turned off (active XO). Close to 100% of multiway speakers are crossed high enough for power handling their woofer cones go into breakup at SPL levels well above the 50 or 60dB down from the tweeter needed for inaudibility. The most common case of a two way with passive LR2 XO is particularly subject to this.
Except for Class-D with LC-filter after the feedback loop I've never seen nor could imagine any decent amp with global feedback to show 2dB (25%!!!) gain peaking from load changes. Something must be severly wrong here.
Broadband here means starting from a few kHz and continuing to where I stopped measuring at 20kHz. If that's a peaking phenomena it's incredibly low Q. I should have the comparative data saved out of HOLMImpulse somewhere but that was three computers ago and I'm failing to turn it up.
I don't think anything's severely wrong. Knowledge of amp-driver interactions is surprisingly limited.
I don't think anything's severely wrong. Knowledge of amp-driver interactions is surprisingly limited.
I have not seen the AES document.
However it is my understanding that if the schematic is redrawn by someone and that redrawn schematic (not the original copyrighted image) is uploaded then there should not be an issue with that. The copyright laws may differ from country to country though.
Some Mod's with more experience than me on this area may chime in.
Tony.
I suspect that only posting the schematic may not explain what the document probably describes.
At the moment I see no point in asking for the document to be posted, since the Forum may simply expunge it due to copyright rules and protecting the Forum's integrity.
Could an AES Member approach AES and ask for their permission to use the document?
Hmm, how often do you do A/B SPL measurements between amps? In 10 years of DIY audio I've had opportunity to do it twice. Stig Erik has, I think, done it once. That may be the entire extent of the dataset. So even a couple more data points would substantially expand the size of the working set.
What I can say is I've gone so far to throw together a custom version of Spice to model this stuff. So long as one treats the speaker (and room) as having infinite return loss there's little variation in behavior with amplifier output impedance. Once the return loss is reduced to 40dB (quite good) or a more typical 10-20dB there's substantially more divergence. Modeling back EMF developed due to drivers acting as microphones is complex enough I wouldn't consider this a quantitative result, just a qualitative one. But it does yield a testable hypothesis.
Specifically, since dipoles have no enclosed volume and hence higher acoustic return loss, they should be less sensitive to the output impedance/damping factor provided by the amp and speaker cabling. Cabling changes and such require a lot of apparatus to properly ABX test but, to the extent I've been able to AB it, this seems to be the case. I can be surprising to discover just how small a bump in the acoustic impulse response one can hear. In one case I thought I'd found an audible difference between use of the LME49710 and LME49990 as the control amp as the measurements were showing a single pixel in the trace moving by one pixel on the display. Turned out to be a cold solder joint.
The Modulus-86 is neat as it puts these sort of experiences within reach of more folks. Quite curious to see how it turns out.
Last edited:
What I was trying to say very poorly was, it is very difficult to measure in room as room acoustics swamp even a 2db variation in an amps performance. My room is well treated with bass traps in all corners and ceiling and wall treatment, I still have 10db peaks in the room response due to room shape and (WAF).
What I am looking for in my system is imaging. I heard a system a few months ago and the imaging was better than anything I had ever heard. It was holographic. Damn experience cost me $5k as I bought a Rega RP8 to replace my Linn. How do you measure an amps ability to image? I don't think this has much to do with a flat FR.
Yes, nothing to do with FR ... but everything to do with distortion. Most systems for a variety of reasons introduce too much, audible, low level distortion, which kills the imaging - cleaning up this detreitus is not a trivial thing to do, but is essential for a holographic presentation.What I am looking for in my system is imaging. I heard a system a few months ago and the imaging was better than anything I had ever heard. It was holographic. Damn experience cost me $5k as I bought a Rega RP8 to replace my Linn. How do you measure an amps ability to image? I don't think this has much to do with a flat FR.
Measuring? Tricky, most in the game don't even believe in this sort of stuff, so not much has been done to establish the parameters that count ...
Fortunately that's easy to control for; leave the speaker, mic, and amps in place when swapping amps and evaluate the difference in SPLs. If you're really worried about it you can AB the measurements over cable placement as well.
Trying to decide what it means to EQ a speaker flat is, indeed, a hard problem. Fortunately it doesn't have to be solved to assess the interaction between the amp and speaker.
Do you have details or a reference? Nonlinear distortion mechanisms within an amplifier don't produce arrival time differences which are significant for Doppler position detection; propagation delays around the control loop are hundreds of nanoseconds, meaning one's hearing would need to go to MHz or tens of MHz to pick that up. I am aware of some evidence which suggests amplifiers with higher output impedances yield richer subjective perception of imaging, presumably due to a longer setting time in the impulse response as the drivers are less tightly controlled. This correlates positively with increasing distortion from <0.01% to around 0.1%, not negatively as you're suggesting.
It's probably not a monotonic thing, though. If an amp has 1% THD+N there's enough else going on an imaging improvement, if any, would likely be masked by other issues.
I'm not sure anyone really knows. However, reshaping the frequency response changes both the reverberant imaging cues in the recording as well as how the amp+speaker interacts with the imaging provided by room reflections. I would be surprised if flatness was the only factor but, in my experience with changing EQ whilst leaving everything else constant, it does have a small effect. Room treatment or switching between speakers with different directivity is rather more significant, however.
Last edited:
No, unfortunately - this is "the devil's in the details" type of thing, and very little to do with the easily measured, static, non-linear performance of the amplifier. Factors that matter are the dynamic behaviours of the power supplies, quality of all non-soldered connections anywhere in the system, amount of interference generated by particular components, susceptibility of each component to various interference mechanisms, electrostatic anomalies. These all impact in subtle ways, and add up usually in causing unsatisfying, lack-lustre sound. Only by doing experiments where all these factors are mitigated does one get convincing sound, I've found.
The items you mention, and can get measurements for, I have never worried about - the ear/brain can compensate for those anomalies, IME, but it has a much harder time with electronic "dirt" added to the mix by various mechanisms ...
In my case, by focusing on one of those "problem" areas mentioned above, doing everything that is reasonable, or even "unreasonable" to improve that aspect, and paying close attention to what the impact is, subjectively, on the sound. As an example, deliberately improving the performance of the power supply, say, tenfold in some aspects; e.g. resistance to sagging under load; conventional engineering wisdom may say that this is completely unnecessary, but perhaps that is not the whole "truth" - your ears may tell you otherwise.
Personally, I found that if I did 'enough' to improve performance in all the areas mentioned that a much higher level of SQ emerged; my ongoing experiments are to try and understand the cause and effect linkages, and how much is 'enough'.
Personally, I found that if I did 'enough' to improve performance in all the areas mentioned that a much higher level of SQ emerged; my ongoing experiments are to try and understand the cause and effect linkages, and how much is 'enough'.