I would make the point that it would be straightforward to pinpoint where the differences are in systems, by doing real, technical measurements. My ears are telling me what's going on, and all that needs to happen is to create a measurement regime to mimic that detection process - which obviously the "usual suspects" don't do. Get the measurement approach right, and meaningful results will flow forth, in quantity - won't be trivially easy, requires a change in focus, but certainly achievable ...
I've taken to buffering my opamp outputs (AD797) with a simple single ended class A buffer running at about 25 mA. This allows me to drive low impedance feedback networks for low noise. The opamp output is also bootstrapped into class A. With everything in class A, the rails are also very clean - I heavily damp each stage with a 22 ohm and 100 uF cap to ground on each rail.
I am struggling to see how FM (PIMD?) distortion comes into it. With these very high loop gain opamps, it's a non issue - see Cordell for example.
You can read about it in my 'Symphony' preamp on my website. It's the smoothest sounding pre I've built to date, and I'd be quite happy to put it up against anything out there.
I am struggling to see how FM (PIMD?) distortion comes into it. With these very high loop gain opamps, it's a non issue - see Cordell for example.
You can read about it in my 'Symphony' preamp on my website. It's the smoothest sounding pre I've built to date, and I'd be quite happy to put it up against anything out there.
It doesn't until the uA741 plots get trotted out again along with the shaggy dog stories about how someone at some serious company designed it into an audio circuit 40 yr. ago therefore proving not everyone gets/got it.
from the same people who don't "get" Cordell's demolishing Otala's analysis, prescription for flat open loop gain
in theory and HARDWARE
they also ignore Walt Jung, Marshall Leach eventual "defection" from the Otala camp after some consideration, even though they both did build, publish circuits initially following Otala's prescription - they went on to publish that there was no essential link to FM/PIM and "sloping/low bandwidth loop gain"
in theory and HARDWARE
they also ignore Walt Jung, Marshall Leach eventual "defection" from the Otala camp after some consideration, even though they both did build, publish circuits initially following Otala's prescription - they went on to publish that there was no essential link to FM/PIM and "sloping/low bandwidth loop gain"
Last edited:
Ron Quan's latest measurements show that you almost HAVE TO add an output buffer to an IC, in order to remove the crossover distortion that is relatively easily measurable in many POPULAR IC's still in use here as well as in the vast majority of mid fi products. For example, is a TLO82 IC op amp, out of bounds, for measurement?
Still holding everyone hostage to what someone else chooses to do. Frankly who cares what mid fi products do.
There are piles of $2-$3 high performance op-amps, they wouldn't dream of budgeting that much per socket. PMA or jcx could easily find an IC to fit any of your needs for output drive, just what is the point of dragging old IC's that were not designed for the application through the mud?
BTW crossover distortion has little to do with PIM, moving target?
It's important because a chimp fashion audio reviewer will peek inside, see the op-amps, collapse on his fainting couch, then write a scorching review. You can't fix that with mere performance.
David, over at test equipment forum is trying to reduce distortion of his OPS opamp 's via an add-on buffer. His variable freq test oscillator has vanishingly low distortion products but cant drive low enough Z's and keep the distortion down. His goal is for 50 Ohm loads. Does anyone have a SOTA buffer to enable <-130dB distortion levels into 50 Ohms at several volts???
THx-RNMarsh
THx-RNMarsh
Last edited:
CFA DSL drivers can be better than buffers - can add gain to the loop
TPA6120 inside a loop will be loafing at only a few V into 50 Ohms - read the datasheet already, get a little more info from the THS6012 datasheet - same chip with DSL driver specs instead of the very thorough audio specs in the TPA version datasheet
old datasheet - I will raise a query on the new datasheet - looks like they slipped a decimal place in converting to dB in the 2014 rev
https://web.archive.org/web/20120424185528/http://www.ti.com/lit/ds/symlink/tpa6120a2.pdf
since they are duals its convenient to parallel the 2 in package with 1 Ohm output current sharing R, add a offset to the input and you can get Watts into 50 Ohms in Class A push-pull
not as good looing in the new 2014 rev - still fine in a multiloop loop with 30-40 dB gain from the op amp in front of it
http://www.ti.com/lit/ds/symlink/tpa6120a2.pdf
TPA6120 inside a loop will be loafing at only a few V into 50 Ohms - read the datasheet already, get a little more info from the THS6012 datasheet - same chip with DSL driver specs instead of the very thorough audio specs in the TPA version datasheet
old datasheet - I will raise a query on the new datasheet - looks like they slipped a decimal place in converting to dB in the 2014 rev
https://web.archive.org/web/20120424185528/http://www.ti.com/lit/ds/symlink/tpa6120a2.pdf
since they are duals its convenient to parallel the 2 in package with 1 Ohm output current sharing R, add a offset to the input and you can get Watts into 50 Ohms in Class A push-pull
not as good looing in the new 2014 rev - still fine in a multiloop loop with 30-40 dB gain from the op amp in front of it
http://www.ti.com/lit/ds/symlink/tpa6120a2.pdf
Last edited:
I don't know in general, but I smoke Danesmann cigarillos from Brazil. They suite my build.
Jacco, that's just me Look at it this way: I'll pay about the same for a regular say BC 546/556B pair and a pair of BC 639/640. I am buying them to act as current boosters, so what's a more logical choice, 546 or 639? I am not likely to ever need its peak current, but why not? Give me one good reason why not.
I am (a little) obsessed with current capability, make no mistake. I know it, but I figure having more in resreve us better than a very close shave. So, in my headphone amps, I used 50W devices at the output (MJE 15030/31 or IRF 540/9540 in the MOSFET version). Again, it's hardly likely I will ever need what they can deliver, but having a healthy reserve is something dear to my heart.
It's really like sizing one's output stage on a power amp. Maths and practice say that for nominally 100W/8, with a solid 4 Ohm capabiliy, requires 2 pairs of 200W devices (MJL 3281/1302 in my case). It's mostly done with 2 pairs of 150W devices, typicallay Toshiba 2SC5200/2SA1943 in the comercial sector. The extra power reserve I have using more powerful devices means I have that little edge in improved thermal behavior and the extra 0.4A capability per transistor.
So for an "economy" design, I'll stop at 3 pairs because I want more current capability, better damping factor and better heat source distibution. For myself, I'll throw in yet another pair, for the idea that you can never have too much current capability, fully realizing that John for example, in his 1205 power amp, nominally 100W/8 did use 3 pairs of 150W devices, if memory serves. It didn't come across as lacking by any means.
More or less the same reasoning for their PSUs. I tend to oversize power transformers for the sake of "just in case" logic. I do not "SAVE" on the electrolytics, either. I am fully aware that in view of my speakers' characteristics, I will never need all that, ever, but there's that nagging feeling of satisafction when one knows it's a job properly done. It eliminates a lot of possible variables from the equation and allows me to concentrate on getting it to sound right, in full knowledge it's not bogged down wherever I could help it.
You were probably drinking full fat milk the day you designed it.
I agree with Scott, these days there's an op amp built for just about anything you want. And even some op amps not made specifically for audio, such as say AD 818, can be made to sound truly impressive. As can some old favorites, like the OP 37.
I find it odd, to say the least, that those reviewers who insist on short signal paths should bash op amps for giving us the chance of really making them short. These days, with a good selection, you can reduce the size of a preamp down to the size of a cigarette pack (without the PSU).
That's where the true nature of it lies, I think. Open it up and you see a relatively small PCB containing the audio electrocs coupled to a hefty price tag, and people start wondering what are you charging them for? To the layman, log ago poisoned by popualr concepts, that's a crime.
And it is, in many ways, such as that you are not paying for the BOM, you are paying for the brains which made it work well. Let's face it, these days, knolwedge is not respected.
That is the opposite of engineering. "Engineering" means doing numerical analysis to understand the actual requirements, and adding in a healthy safety margin. When you spec 50W devices for a headphone amp you are going far beyond that. You are picking devices that are inappropriate then making up a justification.
The TL082 is the 90-ies uA741. Get serious John.
Why don't you talk about, say, the ADA4870? 1A output current, current feedback topology, 2500V/uS?
Maybe you stopped reading about opamps 50 years ago? That would explain a lot of your diatribes that have no basis in what's going on these days.
Jan
Last edited:
I can get to that level with my class A headphone amp - but, I think the quickest route is to follow jcx's advice and use the TI part.
I've done a wide loop gain bandwidth amp (15W class A CFA sx-Amplifier - its up on my website) that sports a 60 kHz loop gain bandwidth. Its a fantastic sounding little amp (runs hot as hell . . . ) - but I would not say its because the LG BW is wide. (I did refer to Bob Cordell's in the TIM papers and conclusion in the write up).
For a conventional VFA, you can always employ TPC - 40 kHz loop gain BW also perfectly achievable.
- Status
- Not open for further replies.