"Doctors say that Nordberg has a 50/50 chance of living, though there's only a 10 percent chance of that. "
Interesting history -- second hand info was that Dean had the idea to make the transformers be flat for group-delay. he took credit for that. I never used transformers for anything... except once in my first and only tube power amp build. So I wasnt aware of who actually constructed them. no matter now. Dean killed himself so we'll never know. But he was an early thinker on why things sound poorly and narrowed it down to group delay for transformers. When thought about, improving group-delay (especially with speaker systems) always rewards. Thx for the history lesson. RNMarsh
Last edited:
I used two AD811s in ~parallel and an AD797, with some fairly complex compensation and an overall closed loop gain of 6dB, for the never-quite-finished buffer amps in the Harman R&D computer-controlled attenuator. There's a 1997 AES preprint that can probably be downloaded. I needed the current as the nominal impedance of the switched resistor network that followed was fairly low, and I wanted to offload the dissipation changes in the 797 as well. I did need a ferrite in the input to the whole thing, which raises eyebrows a bit although the signal current was very small there.
The output buffer was very different. The overall balanced maximum output signal to rms noise ratio was nearly 144dB, although the paper misstates that my voltmeter was true-rms, making the published result a little optimistic.
The output buffer was very different. The overall balanced maximum output signal to rms noise ratio was nearly 144dB, although the paper misstates that my voltmeter was true-rms, making the published result a little optimistic.
PMA dabbles in all things, not sure why you're trying to 'nail' him to one technology? He was just playing with a JLH 10W amp (about as simple and discrete as you can get!) and before that did lots of revisions of a great discrete preamp. But before that he was playing with a OPA627/BUF634 linedriver/headamp ...
Hey PMA, can you be more specific about your more appropriate layout this time around and what this "buffer" is (discrete or chip)?
Cheers,
Jeff
I would never use a 50k pot. And there are other possibilities than to put the pot directly at the input. This does not depend on age, though.
Let me explain, everyone, a few points leading to successful audio line amp design.
There are, of course, a number of routes to making a line amp with a volume control.
The most simple, is just a resistive or tapped transformer as the entire line amp. In certain limited circumstances, this is a pretty good solution. It is 'clean', simple, and has almost no possible overload threshold. I have used it in the past, with good success.
Before the Blowtorch was put in my system, it is what I used.
Now, what if you want 'more'? You know, like a little forward gain, output buffering and even conversion from single ended to balanced?
Under these conditions, a number of approaches have been tried over the decades, and new approaches are becoming available.
Many decades ago, it was decided that line inputs should have the gain control at the very input in most cases. This is because it acts as a buffer to the input of the active circuits included with the line amp, and tends to keep the distortion lowest, overall.
Another, EXOTIC approach, as used by Ayre, is to add local feedback to a quality input stage. This has the advantage of being somewhat quieter, allowing a very high input Z, and being a pretty good solution, especially for lower level inputs, such as microphones.
So a microphone preamp (the type takes a number of types of microphones and further amplifies them, usually balanced in and out) can take advantage of this sort of level control. We usually do not really call it a volume control, because it is almost impossible to reduce the gain without ultimately adding a series switch in series with the attenuator.
This is not so with a potentiometer input, but it has its own problems and tradeoffs. First, it is ALWAYS a balance between higher input resistance, effective bandwidth with setting, added resistive noise, and even sound quality. For a high input resistance like 100K-500K, popular with tube circuits, the worst case added noise is usually 1/4 the nominal pot value, even with a jfet or tube input. Bipolar input is almost impossible.
For 10K-50K pots, either bipolar or jfet input is possible, BUT any tube source, such as a Fisher or Marantz tube FM tuner (I own 1 each of both) will be sincerely compromised with, let's say a 10K input, that would be set by the pot.
Yet, 10K pots often sound 'better' than 50Kpots. Oh, Oh! (more to come)_
There are, of course, a number of routes to making a line amp with a volume control.
The most simple, is just a resistive or tapped transformer as the entire line amp. In certain limited circumstances, this is a pretty good solution. It is 'clean', simple, and has almost no possible overload threshold. I have used it in the past, with good success.
Before the Blowtorch was put in my system, it is what I used.
Now, what if you want 'more'? You know, like a little forward gain, output buffering and even conversion from single ended to balanced?
Under these conditions, a number of approaches have been tried over the decades, and new approaches are becoming available.
Many decades ago, it was decided that line inputs should have the gain control at the very input in most cases. This is because it acts as a buffer to the input of the active circuits included with the line amp, and tends to keep the distortion lowest, overall.
Another, EXOTIC approach, as used by Ayre, is to add local feedback to a quality input stage. This has the advantage of being somewhat quieter, allowing a very high input Z, and being a pretty good solution, especially for lower level inputs, such as microphones.
So a microphone preamp (the type takes a number of types of microphones and further amplifies them, usually balanced in and out) can take advantage of this sort of level control. We usually do not really call it a volume control, because it is almost impossible to reduce the gain without ultimately adding a series switch in series with the attenuator.
This is not so with a potentiometer input, but it has its own problems and tradeoffs. First, it is ALWAYS a balance between higher input resistance, effective bandwidth with setting, added resistive noise, and even sound quality. For a high input resistance like 100K-500K, popular with tube circuits, the worst case added noise is usually 1/4 the nominal pot value, even with a jfet or tube input. Bipolar input is almost impossible.
For 10K-50K pots, either bipolar or jfet input is possible, BUT any tube source, such as a Fisher or Marantz tube FM tuner (I own 1 each of both) will be sincerely compromised with, let's say a 10K input, that would be set by the pot.
Yet, 10K pots often sound 'better' than 50Kpots. Oh, Oh! (more to come)_
Richard, i don't understand the first schematic: the signal at the - input of the upper OPA will be very low and contain lot of distortion. So it is a matter to sum the two ( non symmetrical ) outputs ?
Last edited:
I have been browsing the thread a bit and would like to throw in my two cents on listening tests. From 1970 until 2011 I worked in radio broadcast engineering doing both studio and transmitter work.My experiments were circa 1980 with vinyl & mastertapes from the record companies. I thought I'd gone into all this earlier in this thread.
I'm interested in whether the State of the Art is still at the stage where eg 20kHz brick wall sounds better.
Of course there are those who are uninterested in what sounds better and prefer to have their Golden Pinnae myths unblemished by truth.
At most popular radio stations it is not the station engineer that dictates how the station audio sounds, it's usually programming. The main criteria is be louder than the other stations and that usually means clipping and compressing the audio.
On a few occasions I challenged the audio preference of programming, two are my most favorite. The first was around 1980 when I worked in St. Louis at KWK-AM. The program director set the audio up himself on an Orban processor with the 100HZ high pass filter in, this was a brick wall filter that lopped everything below 100HZ off. His claim was on an AM radio you can't hear anything below 100HZ, I said he was wrong. To prove my point we set an AM table radio on the bench in the shop for me to listen and he was in the other room with the Orban processor. He would switch the filter in and out and I would yell when the filter was in or out, I got it every time. He was not happy but still insisted on having the filter in.
The second time was in 1990 at WWMX-FM in Baltimore. I actually had no problem with the program director, it was the station manager that kept complaining about us not being as loud as our competition. I was at a disadvantage because we had a muzak sub-carrier eating up 10% of our modulation. I also reset the processing from the previous engineer for less clipping, the competition was clipping so much that their high end was mostly harmonic distortion. I asked my station manager if he listened to them and he said he could not stand listening to them for very long. I pointed out to him that he couldn't listen because of all the high frequency distortion and it would be a plus if we kept our audio clean. It took about three months but eventually we started beating the competition in the ratings and higher ratings means the station can charge more for commercial time.
I've also experienced an occasion where someone kept bragging about how much better his Sony CD player sounds since he replaced all the capacitors in it with high grade caps. He said it now sounded better than the Denon players at the radio station. I told him I doubt that very much, bring it in and we'll listen. We hooked his CD player up to an input in the production studio mixing console, found two copies of the same CD and started both players at the same time. Using the mixer we went back and forth between players and the Denon made the Sony sound like low fidelity. He admitted the capacitors apparently made no difference.
There was one other element in the audio in Baltimore. Although I had the radio station sounding very clean it seemed to lack realism. Thinking back to my early days of audio as a kid I remembered my brother-in-laws mono Hi-Fi having realism. The difference I concluded was vacuum tubes. I spent a couple months at home and built up a stereo all 12AX7/6AL5 vacuum tube broadband compressor that gently rode levels but did not sound compressed. When I was happy with the sound I went in one night and without telling anyone put my vacuum tube processor in front of the Orban processor. I then backed off on the broadband processing of the Orban and let my processor do the work. I went home and listened and the realism I was missing was now there.
Within a week I was getting compliments from other station engineers on how good we sounded. One day a music consultant was driving through town and came to see me. He said he was driving through town and had to find out why our station sounded so unique. I opened the back of the rack and his mouth dropped open when he saw all the tubes glowing.
Listen is subjective, I've had program directors request some change in the audio, more bass, more highs. I'd go away, make no changes then come back 15 minutes later and they would say it sounds much better.
For myself, I've slipped back to vacuum tubes, always did prefer the sound.
- Status
- Not open for further replies.