AMS Neve 1073 Sweeps

[Aaronrash]

Also figured I would post this, I just found this on my computer and forgot about it. I got bored one day and hooked up over 30 different preamps and did a talking test shootout. I only included my super high end ones in this test but figured I would post it...

The order is as follows:

API 512
GML8304
NEVE 1073
PACIFICA
UA-610
ADL 600
MILINEIA
MANLEY SLAM

http://www.mediafire.com/listen/nufwqbdr7iuq077/Ultimate_Pre_Shootout.wav

 

[TheJames]

Hey Aaron,

Thanks very much for your insight.  It's been very helpful if for no other reason than making me dig in and test out what's going on with what I've got.

Apparently I botched a file name when I was making initial measurements because when I put my NV73's back together as "stock," they did not measure the same.  Upon further investigation, the unit I labeled as "stock" in my graphs actually had a shorted 0.01uf cap in the zobel.  I tracked that down by starting with one unit and swapping transformers to see if the response followed the transformers (which it did) and lead me to the issue.

I think there is something different about the AMS transformers (specifically the output).

Initially I was thinking there was something up with the input transformer...So I played with the zobel there and found that 300-400pf made things look a little better.  But...

I don't think the VTB1148 iron is quite right at all.  At least not as a drop-in replacement.  If you just completely remove the zobel, the top end extends a bit and lines up much more accurately with the AMS sweep posted here:

http://groupdiy.com/index.php?topic=54748.msg705360#msg705360

Unfortunately....There's a bit of a hump in the sub-sonic low-end, and the curve between 100hz and 2k isn't quite as exaggerated.  It has the same general shape from about 80hz on, with the relative levels at 10k and 20k in-line with the AMS unit.  This is without messing with the input transformer at all.

One more thing to reference....

CJ made this post:

http://groupdiy.com/index.php?topic=54748.msg699345#msg699345

If you look at his graphs about the various output transformers, you'll see that the Carnhill is rolling off whereas the 1166 is rising.  I'm not sure which Carnhill he was using, but it indicates that there's something different between the 1166 and that particular Carnhill.

I'd be interested to hear someone else do the same and relay their experiences.  In the meantime I'm going to setup an experiment where I can split and level match a sound sample for everyone.

 

[TheJames]

Aaron posted a series of sweeps on this page starting here:

http://groupdiy.com/index.php?topic=54748.msg711374#msg711374

All four of them are different...But notice the roll-off on the EZ1073.  I experience that same roll-off with the zobel on the NV73.  If someone has a EZ1073 and wants to attempt to brighten it up to make it sound more like the AMS, try lifting the zobel and report back.

 

[Aaronrash]

I also believe the AMS transformers are different as well... Both the input and output.

There are several reasons I believe this...

For one, I measured different characteristics between the VTB9045 and the TF10003 (AMS Green can) even simple continuity test are different. If they in fact go down the same production line and are wound on the same run. They are using a different internal chassis for the transformer. Does it sound different? I don't know but they are without a doubt different.

Also, don't ever put your DMM on the transformer. Geoff has said this will magnetize the transformer.

The LO1166 (AMS TF12000) measures differently than all the other peoples stock Carnhill sweeps I have seen. Now of course, this could be due to other factors, but I suspect that if the input transformers have differences than the output must too.

Also, AMS took their transformers off their website witch means nobody can buy them anymore. If their transformers were the same as stock Carnhills, why would they take them off their site? Also why would they talk so much about them being different?

You have to remember when I bought my stock of AMS transformers they costed ALLOT more than the regular Carnhills. AMS could make a ton off selling their transformers.

I believe they took them down because they are part of their sound and don't want cloners to sound as good as their products they charge premium for.

Especially when they get online and see threads like this where a guy is etching orignal cards and building modules with their transformers.

Even then... After building my own modules with original layouts and original components with AMS transformers. I don't have the original specs that AMS have. I didn't remove every resistor and capacitor from my AMS modules and measure them. Components and transformers don't mean anything if the module doesn't perform to the same spec. I'm not talking about simple frequency sweeps and distortion and phase graphs like we are doing here. There are much more specific test AMS performs on their modules at certain levels to make sure it performs exactly true to the original.

Long story short, that's part of the reason I just bought another AMS1073.

 

[ruffrecords]

There is one thing that puzzles me about this thread. Initially it was about the audible differences between three implementations of the same Neve circuit. Subsequently it seems to be have shown that the measured response of these three is so similar that there is not way it can account for the the audible differences. The discussion has now moved on to the detailed physical and constructional differences between the three in an attempt to divine the 'magic' in the original Neve circuit.

It occurs to me that there is one chunky flaw in the whole thread. There is an assumption that the differences between the three cannot be measured and must be due to something incredibly subtle.  In fact, there is one huge difference between the way the sweeps were were obtained and the way the audio samples were obtained and that is that for the audio samples a microphone was connected to each amp. It is quite possible that the differences in perceived sound is entirely due to the the way the input circuit of each amp reacts to the microphones as a source. That factor is absent when making the sweeps.

It seems to me the way to resolve this is is to use the microphone connected to each  pre in turn to record and audible sweep from a loudspeaker in a dead acoustic environment. It does not matter what the actual frequency response turns out to be; all we would be looking for would be differences in response.

Cheers

Ian

 

[TheJames]

I experimented a little more with the VTB1148 output transformer termination/zobel to try and match up the sweep response of the AMS 1073.

If you look back at the EZ1073 sweeps (and my "modded" NV73 ones), you'll see there's a rise in the LF from about 40hz down, peaking at around 10hz.

I strapped a pot across the output of the transformer and started sweeping until I got the response the flattest and then disconnected/measured the resistance of the pot.  LF response was flattest with 120r.  That seems pretty low...But it's what it took to get rid of the LF hump.  The not so good part of that is the HF response rolled off more drastically.

The most AMS-like HF response was achieved by using a 681r terminating resistor, with a 604r/1000pf zobel.

All of this was tested with the outputs connected to my converters.  I have not tested with the device plugged into another piece of gear with 600r input impedance to see how that reacts.

As always, I'd be curious to read others thoughts and experiences.

 

[TheJames]

It occurs to me that there is one chunky flaw in the whole thread. There is an assumption that the differences between the three cannot be measured and must be due to something incredibly subtle.

I don't think they're immeasurable.  You can clearly see the response differences in the graphs that we've all captured.  They are very small differences....But as I got my response curve closer to that of the AMS curve, my real-world captured sound got closer as well.  I'm not comfortable saying that I've found "it" yet because I can't nail that spectrum.

I'm also at a disadvantage in that I don't have Aarons AMS 1073 or EZ1073 to compare with.

I am however thinking about renting in an AMS or vintage unit just to see what the experience is.

In fact, there is one huge difference between the way the sweeps were were obtained and the way the audio samples were obtained and that is that for the audio samples a microphone was connected to each amp. It is quite possible that the differences in perceived sound is entirely due to the the way the input circuit of each amp reacts to the microphones as a source. That factor is absent when making the sweeps.

This is a good point...And because of a lack of detail about how the sweeps were initially captured, it resulted in some confusion in subsequent data collection.  I've probably made over 100 sweeps at this point and a dozen or so audio comparisons with a microphone split into stock and modded NV73 preamps.  So far I haven't hit on anything that is "night and day" different, however tweaking the output termination, I have changed the response of the NV73 and gotten some of that "air" that you hear in Aarons samples.

I have a couple of "EZ1290's" with the same Carnhill iron in them.  I'm going to test those as well.

It seems to me the way to resolve this is is to use the microphone connected to each  pre in turn to record and audible sweep from a loudspeaker in a dead acoustic environment. It does not matter what the actual frequency response turns out to be; all we would be looking for would be differences in response.

Cheers

Ian

I don't have a dead enough space for that....But it's not a bad idea.  But really, you could just inject a signal into the head amp of a microphone and get the same result no?

 

[Aaronrash]

This thread is really a learning in progress type thing. You have to remember, I started with AMS modules in the first place and have been my go to tracking chain for the past 10 years. That is why I noticed clones don't sound the same!

Ian your points are well taken and I have already done Y cable test long before I even started this thread. You have to remember that I wouldn't be going through all this trouble and spending this much money if something didn't sound drastically different. I also believe that it doesn't take Y cable test to hear huge differences between units. Even units that measure the same. I also believe there's much more going on that is measurable that we simply cannot measure with REW and a interface, it's just a guild-line.

We started with PCB suspicions and it paid off, I was able to get my EZ unit to measure and sound much better. Still nothing as good as my AMS1073. it really is the benchmark in my situation and the differences are most obvious in the high end. Not EQ but harmonic

 

[TheJames]

I think one thing that was interesting about hearing Aarons samples was the "punchiness" and "articulation" of the AMS compared to the more "laid-back" sound of the EZ1073.

I have no idea what could cause that in a "linear" amplifier.  Inadequate capacitance on the power?  High impedance ground?

 

[ruffrecords]

The really big problem here is trying to find a cause to a corresponding effect. The measured differences between the samples is so small as to be insignificant. This means, we are either measuring the wrong thing or there is something wrong with the test method. My ears are 63 years old and even they can easily detect audible differences in the three samples. Clearly, what is happening is not subtle. That is why I suggested a sweep with the microphone used in the audio tests because that is the principal difference between the two types of test that have been done.

I have been in electronics engineering for over 40 years and one thing I have learned is that there is no such thing as a subtle fault. There are many very puzzling faults that seems to defy reason and be very subtle, but when you finally find the cause it turns out not to be subtle at all.

Cheers

Ian

 

[Aaronrash]

Awesome advice Ian, trust me I'm a novice compared to you. I don't claim to know it all. I only can speak on what I'm hearing.

My AMS1073 (second unit) should be here next week and I will set up the sweeps you're talking about. If anything it should be informative and fun.


Aaron

 

[TheJames]

Hey Aaron,

What units do you have now and what transformers do you have?  I'd gladly ship you VTB9045 and VTB1148 input and output transformers for testing in your units to verify that the "sound is in the iron."  (unless you want to send me a 1073/1084 to play with)

 

[Aaronrash]

Here's the process so far...

Owned a pair of 1073's and 1084's for about 10 years. Studio closed in 2012 they were sold

Built a pair of EZ1073's in July last year. Started out stock and ended up with AMS transformers and expensive components. Sold them to a buyer in Hong Kong, they were badly damaged during shipping so buyer sent them back to be repaired. During repairing them I discovered the wiring tricks and got them to match my 1084...

When I sold my EZ1073 units I purchased a AMS1073 from VK this was early Feb this year

I owned the AMS1073 for a little over a month and sold it immediately when I discovered a used AMS1084 for sale with the same build date as my old pair. Owned it for a little over a month and decided it didn't sound like my old 1084's or like the AMS1073  It just had a blanket phasey effect in comparison, however it sounded the same as my home etch builds and  the EZ1073 with my mods

I just recently sold my 1084 and bought yet another AMS1073 that's on its way and it's brand new. So when it gets here I will run more test.

As far as transformers go, my home etch units have AMS transformers throughout. The EZ1073's I built have stock inputs but AMS outputs.

As far as the AMS1084 not sounding like the AMS1073 could be a million reasons and I wasn't able to figure it out. I worked with a close friend of mine who is a brilliant tech as well as Avedis. Sent them both sound files and they both agreed that the 1084 sounded harsh and muffled in comparison to my AMS1073. The high end just wasn't coming through. Oddly enough my home etch builds sounded identical and so do my EZ1073's so whatever is missing is literally the magic or "the sound" it's hard to explain... You would have to really be used to tracking your own voice on the same setup for over 10 years with the same mic... When something sounds off it's apparent really fast.

The AMS 1073 had quite a bit more THD in the HF range around 20kHz where all my other units measured the same. My tech as well as Avedis didn't think that was the problem though. Not enough to notice anyway... Strangely enough my AMS1073 broke up differently as well when running it hot. It really just sounded pretty damn good no matter what. This AMS1073 I just bought I will never let it go. I'm sticking with it... I'm tired of switching modules, I'm just ready to get back to making music here at home. The home etch build's I did as a favor for a friend who is starting up a studio so those won't be here, also the EZ1073 will have to be sold again to get back all the crazy amounts of money I've spent on this Neve obsession.

The fact that I know how a tip top module is supposed to sound doesn't help either. If I would have never owned my AMS units for so long I wouldn't have been so in sync with that sound and would have been happy with the EZ1073 initially. Ideally I would have rather kept my 1084 since it's a better unit but it didn't sound right unfortunately. That's the thing when you're buying used gear, you don't know where it came from, who owned it, how it was treated and if and components were ever replaced etc... So there could have been a million reasons as to why it sounded bad in comparison to my new AMS1073 any changes made to it could have brought it out of factory spec.

That's why I'm starting to agree with all the people that say "the 1073 is often cloned but rarely recreated". There's something about that circuit that is so damn picky, it's hard to get just right obviously.

That said.. I would also have to agree that pretty much every clone still sounds good because of the circuit. But whatever it is with the 1073, whenever it's dialed in and cloned properly it really is unbeatable and I wouldn't want to use any other pre, ever.

Here's a really good example of "the sound" instantly recognizable even on my laptop speakers https://www.youtube.com/watch?v=Jy-K5u-hMz8

 

[TheJames]

Ugh...I heard it in that video too.  It just sounds "right."  Not that the other pre sounded "bad" or "wrong" really....But in comparison, the Neve just sounded better.

:-\  :-\  :-\  :-\

 

[Aaronrash]

It would have been allot more obvious with a condenser. I'm surprised it was that obvious with a dynamic!

 

[tommypiper]

It was obvious to me too, right away listening to the first singer, on the first word of comparison.  However, it sounded maybe a bit louder too, which would favor the neve and make it sounder better no matter what.  Definitely the richness and harmonic character is there.

Seems like you are chasing your own tail.  I like the microphone recording standard sweeps idea from Ian, in a controlled setting.

I've also worked with original Neve modules that sounded like a damp blanket, and others that had the magic.  This means to me it's got to be a setting.  It's not the actual Neve manufacture or traces or iron, or the layout, wires, or any part, etc.  (Well iron maybe -- I should correct myself -- there are differences from different companies which may make some diff like St Ives vs Marinair... vs Carnhill... but it's not going to be big difference like you are experiencing.)

Beyond that, I believe it's a setting, not a construction element or part that is causing the differences.  If two original Neves can sound so different, one right and one wrong, then it means it's not the parts or the construction.  There are obvious variations like zobel settings and hookup settings, input Z setting, output amp bias setting, and power supply variables.  Power supplies DO effect sound.  (Have you made controlled tests with same PS and diff modules, and same module and diff PS?!  You may be surprised.) 

I think you need to find the common variable that is consistent with the tests, as Ian said.  Any of the variables I mentioned above in preceding paragraph can cause the blanket and other effects you have complained about. 

It may be one of those, rather than a more esoteric and elusive one.  I know from my own troubleshooting that I often overlook the actual fault even when I collect the data that could identify it, because I am looking elsewhere or convinced that I have to find a certain type of fault, etc.  It often takes an outsider with no agenda to steer me back to the obvious fault. 

 

[Aaronrash]

Hey guys,

I've actually been so incredibly happy with my new unit that I forgot to post back here. This unit has a special sound like my first one did. They are both issue 7 so maybe that has something to do with it. Each issue is different. But the sheen/harmonic/mojo factor on this particular unit is unreal. It also does that "bell like" warm pronounced and aggressive upper mid range thing that is impossible to get any other way.. I'm never getting rid of it. (sorry my sound describing terms suck) but that's how I hear it.

Vintage King are really great peeps, If you're looking for a AMS unit they really are the best people to buy from

I'll make it clear now though that this unit sounds nothing like my 1084... Waaaaaay more mid forwardness and silk with this unit.

More test and measurements soon... from the first sweep though I did just for a quick reff looked pretty much the same as the 84, also the same distortion specs.

 

[Tekay]

I also believe the AMS transformers are different as well... Both the input and output.
For one, I measured different characteristics between the VTB9045 and the TF10003 (AMS Green can) even simple continuity test are different. If they in fact go down the same production line and are wound on the same run. They are using a different internal chassis for the transformer. Does it sound different? I don't know but they are without a doubt different.

The Carnhill L10468 and Marinair T1444 was not winded the same so it might be that he TF10003 is a copy of the T1444.
Even if they didn't match both were used in vintage Neve modules.
Also heard that the AMS output has a smaller gap to match the LO1166 inductance, but a smaller gap gives you a lower headroom

 

[ruffrecords]

It would have been allot more obvious with a condenser. I'm surprised it was that obvious with a dynamic!

The opposite might be true. Condensers have nice, low, almost resistive output impedances, across the entire audio band and act much the same as a test generator when driving a mic pre transformer. A dynamic mic, on the other hand, has a significant inductive element and its impedance varies widely over the audio band. I would not, therefore, be surprised to find that it reacts quite differently with different input transformers.

Cheers

Ian

 

[Aaronrash]

Thanks Ian... That's good info to know. Learn something new everyday on here

That's something I didn't know.

But also I hear allot more high end detail with condensers and that's particularly what I love so much about Neve modules is what they do in the super high end range.

I have a SM7B that I like to use sometimes though when I want a darker tone that works great too. Trying different dynamic mics out though with the Neve would be allot of fun.