john curl said:
I like it. I hope my customers like it. It may or may not sound like the real thing. My ad copy writer will use his creative writing skills to convince them to like it.scottjoplin said:
My understanding of neutralization is that it is used to reduce the effects of feedback capacitance of a device by adding feedback of the opposite polarity. In most cases neutralisation is negative feedback to counter the intrinsic positive feedback.traderbam said:
Reminds me of the old modules from Telefunken, Neuman, Siemens ect.
http://audio.kubarth.com/rundfunk/index.cgi
(single ended topology, Al electrolytic coupling caps, input-output transformers)
For some Neve schematics
Schematic Vault (Neve) - Dan Rudin Recording & Production
George
Attachments
Just updated my Bode-100 software and there is an improved test suite for capacitors: YouTube
I don't have the test fixture though, which probably is another $ 400. It never stops...
Jan
Usually a diy approach to a fixture is a money saver.
It is no mystery but more a matter of careful construction and execution; most measurement devices offer sort of a calibration routine to account for external fixture differences.
Might be though that you need a calibration capacitor for a complete procedure, then it it depends on your need wrt measurement precision and uncertainty.
I totally agree.
I don't know something that satisfy my ears and measure bad.
Remember, I wrote: "of course, distortion, bandwidth and linearity, slew rate, damping, every technical aspect has to be good enough."
Good enough depend of your degree of exigence ;-)
If you have the choice between two solutions, the one that measure the best seems the best. But when we are between two solutions with different results, one better in harmonic distortion, the other in IM, as an example, i will listen ... (and most of the time, I will prefer the best IM). Just an example.
Well, when I said good enough, we can achieve nowadays very low distortion numbers.
After that, the differences lie in the details, and distortion don't tell everything.
My point of view was about DBT. (That I use only when I am in uncertainty)
A dream ? In real world, I don't know two brands of amplifiers with the exact same "sound". And if I can find a couple of them, the solution is simple (buying the cheapest one, or the better looking, or the one you think it will last the most, ... what is important for you ;-)
But I agree on one point: the differences can be very thin and subtle, so, better concentrate before on speakers ...
I'm not really sure he would agree at all with that assessment, the Neve consoles were actually known for good, solid, conventional engineering, and lots and lots of op-amps (5532 and 5534 if memory serves) in the early ones. (Nothing boutique in the parts used or the design IMVLE) Most mainstream studio gear was not deliberately interpretative, there were and are lots of effects devices that can be introduced into the recording chain as needed, but the basics were, well basic..
I'll second that, we had a 48-channel 1980's Neve and it ran 18 V on the 5532/5534s. It was taken on the road for shows (in the largest road case Anvil had ever made to that point) and it was a full time job replacing those ICs, it seemed like every time the board was powered up there would be a dead bug somewhere.
And relevant to recent discussions it was full of electrolytic coupling caps...and it was a great sounding board with boatloads of headroom. I cannot imagine how many well-regarded recordings audiophiles love were mixed on Neve consoles...
Cheers,
Howie
Well the AD797 isn't complying with either of our definitions so I don't think the term neutralization applies to it.
Hi Bill,
Dunno exactly when he stopped consulting for his namesake. I have talked to label recording engineers who thought Rupert's discrete SS consoles sounded worse than either the earlier tube or later IC based consoles, but you know what they say about opinions...
Cheers!
Howie
More apt was an analogy to a gyrator creating a -C. I was simply trying to make the point that there is no sensing of the error and amplifying it a lot in an additional conventional feedback loop. The operation is easy to demonstrate with an ideal current mirror and a few sources. If the output buffer is ideal the correction capacitance is bootstrapped perfectly and the main loop GBW is still determined solely by gm and C. The error in a real buffer is small and only enough -C is seen to cancel it. If I can find the AC plot of the output impedance the gyrator action is more clearly seen.
I was simply trying to differentiate the operation from what I have seen in other EC schemes where the delta Vbe in the output is measured, amplified, and used to create a compensating voltage. Here the effect of the voltage error is replicated as a current and injected in the opposite sense.
Last edited:
That would be helpful.
Yes
That's a little unclear. I get the current through Cn. Where is the other current coming from?
I have simulated the feedback loop (in Fig 33 of the AD797 data sheet) formed by Cn around the unity gain output buffer A (which I have given an arbitrary pole at 10MHz), in the absence of any input current from the input stage. This loop is unstable when Cn=Cc (as predicted by my maths).
So I'm curious as to what the values of Cn and Cc are in the real op-amp. If they are the same you must have had to make the loop stable by some other means...?
Sure. This is a feedback loop around the output buffer. It doesn't affect the overall gain of the buffer except at higher frequencies where it extends the buffer's bandwidth.
When you look at HEC implementations you'll see the same effect. A near unity gain output stage has a loop put around it which has high loop gain; the gain cannot be made infinite and stable at the same time so total distortion elimination is impossible. There is no "cancellation" in an arithmetic sense going on at all; it isn't a feed-forward system.
I think its pretty obvious from interviews with Rupert Neve that he designed the best he could do at the time. Case in point, he improved his earlier boards by designing an IC based (what became the 5534, still prototype at the time) module to replace the discrete modules he had originally designed.
There is a small difference in sound, unless you use it a certain way, ie overdrive everything. That is the way Dave Grohl uses the Neve console, for its smoother overload margin and character, not as it was originally intended to be used within its dynamic range.
I was sent to decommission an 80 series Neve, (that's a style number, not a time reference) the studio owner was complaining that he never was able to get the fabled Neve sound out of it, it was too neutral he said. I asked him about his levels and working process. He was using it as intended and was delivering a neutral sound, as designed. Let the boys in Seattle run the same board and it will sound the "grunge" way, because it can have several flavors of sound.
There are many ways to produce a recording, pop music in all its varieties is usually a creation process, don't expect "real", there is no real before its made.
Regarding the later Neve boards Howard was referring to, I have recapped them before, 150 electrolytic caps per channel! And heat!! Well known as a board to make you sweat, because its too damn hot! I commissioned one for a hip hop studio in Seattle, they ended up running a separate AC system directly into the frame of the board. Only one I know of where you didn't drip onto the board.
Cheers
Alan
There is a small difference in sound, unless you use it a certain way, ie overdrive everything. That is the way Dave Grohl uses the Neve console, for its smoother overload margin and character, not as it was originally intended to be used within its dynamic range.
I was sent to decommission an 80 series Neve, (that's a style number, not a time reference) the studio owner was complaining that he never was able to get the fabled Neve sound out of it, it was too neutral he said. I asked him about his levels and working process. He was using it as intended and was delivering a neutral sound, as designed. Let the boys in Seattle run the same board and it will sound the "grunge" way, because it can have several flavors of sound.
There are many ways to produce a recording, pop music in all its varieties is usually a creation process, don't expect "real", there is no real before its made.
Regarding the later Neve boards Howard was referring to, I have recapped them before, 150 electrolytic caps per channel! And heat!! Well known as a board to make you sweat, because its too damn hot! I commissioned one for a hip hop studio in Seattle, they ended up running a separate AC system directly into the frame of the board. Only one I know of where you didn't drip onto the board.
Cheers
Alan
Last edited:
Cc and Cn are exactly the same and the triple-darlington has more BW so at 50-70MHz there is a small pole zero network but that is not fundamental to the basic operation. The phase margin was pushed a little hard because applications for the noise performance are predominantly at substantial gain and input snubbers can almost always be made to work.
For sake of argument let the output buffer be ideal then Cn can't do anything including making any loop unstable. The plots are on the data sheet they are with an external 50pF mica IIRC and the internal Cc is 50pF +-10% or so. I used a 10pF plus 47pF ceramic trimmer for optimizing the plots in my AES presentation.
EDIT - jcx did the same as PMA also, and I think I posted an SK170 input version with generic 4401/4403 bi-polars for the rest of the circuit.
EDIT EDIT - See figure 21 on DS for output impedance (with Cn falls at 40dB/dec rather than 20) same holds for distortion components. You might also be confused in fig. 33 the operation of the current mirror and bootstrap forces A and B to be at the same voltage which is also the output voltage.
Last edited:
Talk about deja-vu! I just stumbled on this. I seem to be repeating myself every 9 years: John Curl's Blowtorch preamplifier
Sigh. Sorry if I'm being boring by going over this again. I had better wade through those old posts.
Sigh. Sorry if I'm being boring by going over this again. I had better wade through those old posts.
Once again, I have to go back into the past. In 1974, while hired to construct a new recording studio in Montreux, Switzerland, I was going to make an all discrete board with Levinson quality designs that I had prototyped. At this time, I was approached by a Swiss couple (American wife) who wanted to invest into the recording studio. Their one insistence was using a Neve board for the heart of the recording system. I politely declined, and it was to my regret. Ultimately, they fought for control of the studio, and won! So after 1 year of prototype development, we were without a studio to put together.
Now, what if I had to do it over again? I was very idealistic at the time, and I knew that Neve made a GOOD but not SOTA electronics. It had a good sonic reputation, better than most others, except earlier tube consoles. I should have compromised, as I had with the Studer mastering recorders that we ordered. I could have 'fixed' any moderate oversights, and saved myself from a lot of work building an entire console myself. In fact, my vision was a console to record classical music almost exclusively and it was not versatile enough to do pop music at the time. It would have been better than the Neve in overall clarity, but limited to live performances almost exclusively.
Now, what was wrong with the Neve console, and what was really good about it? First, looking at the discrete schematic, you find the lack of PNP transistors. This is because silicon PNP transistors were not as popular with the British as they were more expensive at the time. In the USA, we started to use small signal silicon PNP transistors in audio since the mid-1960's because they became available from Motorola, only slightly more expensive than their NPN complements. Ampex made similar circuits as Neve, but with a few more PNP transistors, in their tape recorder electronics at the time, and they sounded OK as well. Usually we used DC Biased Tantalum caps for input and output coupling, as they were popular then. Aluminum was really large in those days, so was avoided if possible. Ceramic compensation caps were also used that could distort at the extremes.
Now what was GOOD about these designs?
They were class A, input to output, for one.
They had pretty good gain bandwidth, because they were not limited by IC processes.
They subsequently had a fairly high open loop bandwidth, because they had very limited open loop gain, as well. (gee, what do you know, could this be a good thing?)
They used selected transformers for input and output, keeping out a lot of RFI, etc.
There may be more, but I can't think of them at this time.
In any case, while I designed more sophisticated circuits at the time, these early discrete designs could be pretty darn good.
In hindsight, the discrete designs made by Studer, Ampex and Neve, are still classics, that can do better than most later stuff with cheap op amps, etc. "Old 'amps' for new!" '-)
Now, what if I had to do it over again? I was very idealistic at the time, and I knew that Neve made a GOOD but not SOTA electronics. It had a good sonic reputation, better than most others, except earlier tube consoles. I should have compromised, as I had with the Studer mastering recorders that we ordered. I could have 'fixed' any moderate oversights, and saved myself from a lot of work building an entire console myself. In fact, my vision was a console to record classical music almost exclusively and it was not versatile enough to do pop music at the time. It would have been better than the Neve in overall clarity, but limited to live performances almost exclusively.
Now, what was wrong with the Neve console, and what was really good about it? First, looking at the discrete schematic, you find the lack of PNP transistors. This is because silicon PNP transistors were not as popular with the British as they were more expensive at the time. In the USA, we started to use small signal silicon PNP transistors in audio since the mid-1960's because they became available from Motorola, only slightly more expensive than their NPN complements. Ampex made similar circuits as Neve, but with a few more PNP transistors, in their tape recorder electronics at the time, and they sounded OK as well. Usually we used DC Biased Tantalum caps for input and output coupling, as they were popular then. Aluminum was really large in those days, so was avoided if possible. Ceramic compensation caps were also used that could distort at the extremes.
Now what was GOOD about these designs?
They were class A, input to output, for one.
They had pretty good gain bandwidth, because they were not limited by IC processes.
They subsequently had a fairly high open loop bandwidth, because they had very limited open loop gain, as well. (gee, what do you know, could this be a good thing?)
They used selected transformers for input and output, keeping out a lot of RFI, etc.
There may be more, but I can't think of them at this time.
In any case, while I designed more sophisticated circuits at the time, these early discrete designs could be pretty darn good.
In hindsight, the discrete designs made by Studer, Ampex and Neve, are still classics, that can do better than most later stuff with cheap op amps, etc. "Old 'amps' for new!" '-)
Last edited:
- Status
- Not open for further replies.