The woofer is high eff too, so ideally I'd like to keep it's amp at the stock +20 dB (10x).
The tweet is very high eff, so ideally I'd like to keep it's amp as quiet as possible and really low (5x, 3x) or even no gain (1x) ... oh and keep the sound quality! Possible?
I just have a few DMMs so I can't do any fancy testing but I do have TINA and your files. Could I add the key parts of the fb system that I need to re compensate and get values 'close enough' from the simulation.
Thanks Tom,
Cheers,
Jeff
Nice attempt at deflection AndrewT but you started this by calling my statement wrong and then when I tried to clarify what was wrong about it by asking you to respond...well the readers can surmise the rest. I won't make any demands on you to admit your fallacies again but you should really google "difficulty with fault admission".AndrewT said:
Good point... I hope the Modulus-86 has been EMC tested! 🙄
FWIW, in EMC testing a piece of audio equipment would be allowed to make weird noises and distortion when subjected to RF. It only fails the test if it doesn't recover normal operation after the interfering signal is gone.
So actually there is no guarantee that the sound of your hi-fi equipment (or the accuracy of an Audio Precision test set for that matter) won't be affected by RF interference.
TI has appnotes about that, they specify EMIRR for some opamps, and have some EMI-hardened models. Not for audio through, they're mostly very slow "DC" opamps for sensors and instrumentation, where you don't want bogus data because of RFI.
Since it doesn't have any digital/RF stuff in it, there would be no emissions.
Since it has a balanced receiver at the input, it can be setup for proper shielding and immunity.
Well, IME, this IS the point ... my audio doesn't "squawk" but its quality is audibly degraded by by RF - it sounds "worse"! Therefore, distortion has increased, my ears can hear it - just because you haven't an easy, well recognised way of measuring it doesn't mean you have an excuse for ignoring it, sweeping it under the carpet, pretending it doesn't happen ... if an effect is there, jumping up and down and claiming that your ears are fooling you is a pretty lame excuse for an engineer to use ...
Jeff, even if you waste 20dB of headroom with resistor pad, you're still quieter than redbook CD noise floor! Your room noise will bury that! 😀
As one specific example, mobile or cell phones. I have done subjective experiments where a phone is a particular distance from the audio system, and I power it down, and then back up again; the quality degrades, specifically in the treble area while the phone is active - I'm not talking "noises" or obvious artifacts, I talking about constantly present extra distortion, the same sort of thing that one tries to engineer out of circuit topologies.
I haven't the equipment to do the sort of objective measurements that could identify the behaviour, but others do ...
You may want to chat up your AP rep. It's pretty remarkable the thinking they put into getting better correlation between measurement and perceived sound quality. One example is their 32-tone test signal. I've also seen a test where they step the input RMS voltage and measure how the amplifier responds to that on a long time scale. It was pretty interesting. I hope to reproduce those tests eventually, but right now I'm focusing on serving the immediate needs of my customers and on new product development.
I actually met the AP rep serving the Pacific Northwest before I bought my APx525. He was up here for other sales calls so we met over a few microbrews and dinner. Nice guy. He had lots of stories to tell, including quite a bit about APs approach to audio testing. They don't sit still, that's for sure.
~Tom
This is exactly why the Modulus-86 is equipped with EMI/RFI filtering on the input.
Broadband energy sources, such as spark gaps and relays, are causes of trouble as well. Without proper RF filtering, you'll likely hear a click or pop in the speakers when a high-current load (fridge, furnace, etc.) turns on.
The issue with RF injection into amplifiers is that the RF gets demodulated by the various diode junctions in the amplifier, resulting in a DC offset shift. This wouldn't be so bad if the RF was continuous, but in particular GSM wreaks havoc because of the 200 Hz frame rate. What this means is that the DC offset of the amplifier will change at a 200 Hz rate - well within the audio band and highly annoying.
Anyone interested, can check out the National Semiconductor/TI LMV851 EMI hardened op-amp. AN-1698 would be a good place to start. Note Figure 11 at the end of the paper. EMI/RFI filtering is also found in the LMP2021.
The bottom line is that in this day and age with WiFi, Bluetooth, cell phones, etc. all around us, an audio amplifier needs to include an RFI/EMI filter. That's why I included one on the Modulus-86.
~Tom
Posted in the original MyRef thread. I have also done some simulation on that and the current LM3886 chip amp I am working with, seems like the MyRef was quite a bit better, so I am wondering about this amp. Audibly, the MyRef concept was quite cleaner when I last compared, but that has been quite a while.
And you can get fancy and add an extra output zobel at the speaker cable entrance:
like Cordell does to his chipamp in his Amp book (pg 540, Fig 27.3 The Super Gain Clone Employing a DC Servo)
On pg 541, he writes:
------------------------------------------------
...
"Output Network
The output network is a pi-section arrangement with Zobel networks on both sides of the L-R network. This provides enhanced resistance to RFI ingress via the speaker cables. The inductor is optionally constructed as an air-core toroid to keep the magnetic field circulating largely inside the toroid, reducing possible magnetic coupling to other circuits. The first Zobel, comprising C and R, is located close to the output of the LM3886. The L-R network and second Zobel are located close to the speaker output terminals, optionally on a small separate board. The output signal is carried from the main board to the output network board by about 6 in. of high-quality speaker cable."
---------------------------------------------------
Cheers,
Jeff
PS
Then use those EMI/RFI cable/panel connectors 454Casull recommended a few pages back:
http://www.diyaudio.com/forums/vend...er-achieving-0-0004-thd-n-53.html#post4162165
Then work on the power entry: any good RFI/EMI, w fuse, etc power connectors?
Then finally a proper case and proper grounding circuitry and layout to minimize RFI/EMI (any advice?)
Whew!
.... Damn then smartphones!
Last edited:
Please explain 'better'?
If you have a link to where I posted this originally, I would be grateful.
I note no one seems interested in the results of a properly conducted DBLT bla bla that actually comes up with a distinct preference 🙁
IMveryHO, these are the most important test results of all .. even more than sighted tests by deaf Golden Pinnae or even AP measurements. 🙂
Last edited:
For those who have not visited Toms website, his dayjob is designing stuff that has to live in a real RF environment, where there are potentially KW of EIRP floating about. It's been 12 years since I was last in that world, so I am a tad rusty, but safe to say if you live in that world then you take EMC/RFI very seriously. Lots of nice machined from solid bits to keep the high power bits from swamping the low noise bits. And it's a world where a dB matters a lot.
Aside: when I were a youth i got called on over the weekend to analyse some telemetry from a new satellite just launched that my employers had been prime on. Some odd temperature readings. As a grad did my work, got my overtime and thought no more. 6 months later discovered that, due to some incorrect testing a PTFE loaded cavity filter had developed a coronal discharge and turned into a glommy mess. Somewhat embarrassing when its up in orbit and not somewhere you can fix it.
I miss the machined from solid. I don't miss test chambers where you have to remove contact lenses before going in just in case...
One of the interesting issues I have discovered is that the same equipment will sound different in different setups, some I have found to be problems in internal design, some are related with either the speakers or interface between any two equipment. This is why I like to look at measurements that also show now loading interaction will effect performance. General measurements using constant resistance loading is good for validating internal design, but far from realistic. In the case of power amplifiers, it is more realistic to see how it will interact with an active load like a real driver which also acts like a microphone is reverse.
Listening is a good way to identify problems by experienced people, but to be sure where the problem really is, measurements under various situations are necessary. I remember in the early days when Japanese equipment measured well but sounded terrible. This is one reason I generally would recommend hooking the same equipment into different peoples systems to help identify trends listening with the auditor to that you can relate what they explain with what you hear even if you may not initially feel the same. In time, I generally can fine a relation with each specific persons description of sound.
The tricky thing about listening is that if it sounds good the first time, listen longer and you may find out something you have missed. If there is an obvious problem, you will identify it in the first minute, and you may find it only in an unfamiliar environment, this is one reason why I move around to different sites for listening tests.
Let me fix that for you. Any 2 collections of equipment, even if they share 1 item in common may sound different. If the different item is the speakers of course it is guaranteed to sound different. If the room is changed and the equipment is the same it will sound different.
I first learned about damping testing here.
http://www.diyaudio.com/forums/chip-amps/54571-my-audiophile-lm3886-approach-9.html#post683346
Some measurements I did with a real driver.
http://www.diyaudio.com/forums/chip-amps/54571-my-audiophile-lm3886-approach-333.html#post2820013
Pretty hard to find my Halfler measurements through all that.
You find out what is what by both listening and going back to the test bench. I think you are missing the whole picture.😉