Discrete sounds better than integrated? a possible reason

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
abbey road d enfer said:
I understand all that, but your remark, IIUC, was addressed only to the posted schemo.
Regarding the Helios and its 1:10 xfmr, the preamps in my mixers used a similar one (Melodium, if you remember them) with a BC109C operating at 50uA.
I measured -128dBu input noise with a 200r dummy.
About the Neve, the operating point must be chosen for the highest gain, since S/N ratio can only improve, unless someone does crazy things with attenuators.
I have designed a preamp where the input device (BC560) runs at 200uA and normally sees a 2k secondary. I dedicate it to drums and such brutal sources. It is not the quietest in the world but nobody complained so far.
The input pad is based on an alternate primary (20k nominal), so the transistor base sees about 20 ohms+ some DCR, so talk about mismatch! Anyway, the resulting S/N ratio is perfectly manageable and inobtrusive.
And yes, the output stage is class A.  :)

I knew you knew that,  sorry ;)

Well, I may be mistaken about the Helios then, but I seem to remember an input current higher than your 50uA so, mea culpa. 

On Neve, I remember that noise is optimized not for the highest gain which sees an impedance of 600 - 800 ohms (transformer secondary attenuator full up) but somewhere between that and the 1K4 when it's halfway down.

Maybe I'm just full of crap and remembering wrong!  :D

I don't remember Melodium no, I'd be interested in any info if you know of any online somewhere. 
 
Winston O'Boogie said:
I'm guilty of advising the testers at UA to test distortion on the LA-2A with the meter in gain reduction mode. 
They'd been doing it on the AP after checking line up levels etc.  and the meter was usually left in output monitoring mode.
I took a unit over to my FFT setup and showed them the effect of the meter on the distortion spectra.
It's quite clearly there even with a 600 om load.


Now in my defense, most folks will be using an LA-2A monitoring in gain reduction but, for my sins, flame away!

:D
Fair enough, but I think the user should have been warned. I must be nitpicking...
 
abbey road d enfer said:
Fair enough, but I think the user should have been warned. I must be nitpicking...

On Neve, or the LA-2A?

On Neve, yes because the meters are always there.
On the LA-2A, I did advise they put something in the manual regarding keeping the meter out of monitoring the output signal except to initially check things.

If it made it into the manual I can't say. 


Also, remember this is supposed to be a copy of an old unit so a meter buffer is not an option. 

Anyway, I wonder if any of the plug-in models of vintage devices have an option for "with meter" and "without" because, otherwise it ain't real man! 
 
Winston O'Boogie said:
I knew you knew that,  sorry ;)
Well, we must talk of things we take for granted, because others are watching us! They may not know...

I don't remember Melodium no, I'd be interested in any info if you know of any online somewhere.
Well, I was going to write that it is long gone, went bust in the 80's, but I checked and someone has revived the brand. https://www.melodium.fr/fr/
They were known for their microphones, not only the 42B (44BX rip-off), but the 75A, a compression chamber mic (!) that was found in almost every town and village that had some sound equipment. I made a lot of gigs where I sang into the 75A suppled by the local sparky.
But their xfmrs were quite good, with a more than decent mumetal shield.
 
Winston O'Boogie said:
On Neve, or the LA-2A?
On the Neve, it should have been discussed at the time of ordering. But I suppose the guy doing the sale would not attract the customer's attention on a sensitive issue. In french, we say "un sujet qui fâche" (a subject that causes anger).  :)

On the LA-2A, I did advise they put something in the manual regarding keeping the meter out of monitoring the output signal except to initially check things.
That's very considerate.

Also, remember this is supposed to be a copy of an old unit so a meter buffer is not an option.
Do you mean a counterfeit or a recreation?

Anyway, I wonder if any of the plug-in models of vintage devices have an option for "with meter" and "without" because, otherwise it ain't real man!
Maybe the emulation is different according to the meter being in level or in GR...?
 
abbey road d enfer said:
Do you mean a counterfeit or a recreation?

The 2000 Universal Audio reissue.  Recreation?
I suppose even the original Universal Audio was a recreation since the company who owned it and Teletronix was bought by Bill Putnam in the '60's.  Jim Lawrence was the original designer.
I wish someone at original UA had fixed that White Follower though, 1/gm for the top resistor instead of the ridiculously high value that's now, forever enshrined and set in stone. 
I tried to sneak the correct value into the build, but no go.

Same with the wrong value resistor in the 1176LN

abbey road d enfer said:
Maybe the emulation is different according to the meter being in level or in GR...?

That would be correct and fair. 

That Melodium mic seems interesting, were the originals considered as good as the RCA 44 or ? 





 
Winston O'Boogie said:
The 2000 Universal Audio reissue. 
Well, we all know reissues are somewhat different than the original, some being re-creations, others having no more differences than a grounded cord and no stinger cap..

Same with the wrong value resistor in the 1176LN
Which one?

That Melodium mic seems interesting, were the originals considered as good as the RCA 44 or ?
In their dreams! It was a decent mic, but just a substitute for unaffordable RCA's.
They came out later with a more modern mic, the RM6 IIRC, but it couldn't compete against the heavyweights that were available then.
These guys that are reviving the 42B seem quite knowledgeable and determined, so maybe an act to follow.
 
abbey road d enfer said:
Fair enough, but I think the user should have been warned. I must be nitpicking...
Perhaps another factor is the fact that in the early days, there were not always VU meters on every channel. Often they woujld only be on groups plus another one for PFL and gainstaging.

Cheers

Ian
 
Winston O'Boogie said:
It isn't the device, it's the fact it's optimized for lowest noise with the source impedance.

You mention the BC109 which is what Neve used early on.
They (Neve) couldn't optimize for lowest noise because the impedance seen by the first BC109 varies depending on attenuator setting. 
This because they chose to not vary the amp gain too much and opted to go for a divider network on the IP transformer secondary.  So the current in the input transistor is at a compromise setting.
I think you must be mis-remembering. On the 1073, for example, for gains from 80dB down to 55dB, the first amp gain is fixed and there is an attenuator at the amp output.  From 50dB gain down to 20dB gain, the first amp is switched out and an attenuator is placed on the secondary of the mic transformer but it is arranged to be fairly close to a 4K8 load at all positions.

Cheers

Ian
 
ruffrecords said:
I think you must be mis-remembering. On the 1073, for example, for gains from 80dB down to 55dB, the first amp gain is fixed and there is an attenuator at the amp output.  From 50dB gain down to 20dB gain, the first amp is switched out and an attenuator is placed on the secondary of the mic transformer but it is arranged to be fairly close to a 4K8 load at all positions.

Cheers

Ian


I often do mis-remember :D

I should have been more explicit but, of course, you are correct in your description of the complete 1073 et al.

The bit I left out of explaining my simple analysis was that, in my head, I was attempting to compare apples to apples and so I took only a single BA283 gain amp of the 1073 against the single gain amp of the Helios and the single gain amp circuit that was posted of Ricardo's.
So, gains up until only 55dB on the Neve.  Maybe not that relevant,  but it seemed a better way to compare the 3 circuits in my head. 

In my thinking, I did however forget that gains above 55dB had the attenuator post the BA284 amp, but this would make perfect sense, having an attenuator right at the beginning of 2 gain amps wouldn't be good. 

thanks Ian



 
abbey road d enfer said:
Which one?
 

The wrong resistor in the 1176LN isn't a wrong value in the original unit.  But was an incorrect value that had been implemented by mistake in the UA reissue.
When I came onboard, they'd already sold gazillions of them and, when I was looking at the circuit on another matter, I noticed the issue.

The reason the engineering head gave me for not wanting to correct it was that someone might buy a 2nd corrected unit and it wouldn't match one they'd bought the year or two before.

I don't remember the exact resistor now, but I did post about it here about 16 years ago.  It severely decreased headroom of a transistor in the output stage and only managed to limp through the finish line due to the negative feedback negating the effect somewhat.

Thanks for the info on the Melodium mics.  Yep I've heard of Kerwax, the studio folks reissuing it, someone to watch for sure.


 
ruffrecords said:
  it is arranged to be fairly close to a 4K8 load at all positions.

Forgot to address this bit. 
I was meaning the source impedance is changing to the first transistor, rather than the transformer load. 
Line inputs via the 31267 transformer at full up attenuator setting being much lower source z than a mic source when the post 10468 transformer attenuator is at the -6dB or so position. 

Maybe this is one reason that the EMI Neve's supposedly didn't use the 31267 for line inputs but used the 10468 with a U pad in front.  ?

 
Winston O'Boogie said:
Forgot to address this bit. 
I was meaning the source impedance is changing to the first transistor, rather than the transformer load. 
I am not sure what you are saying here. The source impedance seen by the first transistor does not change as the gain is altered from 80dB to 55dB. The source it sees is effectively the mic output impedance times 4. Often this is not optimum for the transistor - usually 4K8 or so is the minimum noise source impedance for the BC109. But all preamps have this problem simply because you do not know what the source impedance is going to be but you can only optimise noise for one impedance value. In addition. you want to make the load seen by the mic to be about 1K2 so as not to unduly load it
Line inputs via the 31267 transformer at full up attenuator setting being much lower source z than a mic source when the post 10468 transformer attenuator is at the -6dB or so position. 
True but first transistor noise is only really an issue for the first amplifier stage which is not in circuit for line input. And in any case the question is moot because the line input source is transformed down by the 31267. A typical 600 ohm line input source is transformed down to about 90 ohms. The 600 ohm pad network connected across here make little difference to impedance presented to the amplifier.
Maybe this is one reason that the EMI Neve's supposedly didn't use the 31267 for line inputs but used the 10468 with a U pad in front.  ?
I have never been sure why this was done. It might simply have been a shortage of 31267s. The problem with the pad before the 10468 is to get the input impedance up to 10K and match the 10468 input the attenuator needs to be more than the 12dB loss you get from the 31267. This means the output from the 10468 is lower than from the 31267 so its noise must be worse not better.

Cheers

Ian
 
ruffrecords said:
I am not sure what you are saying here. The source impedance seen by the first transistor does not change as the gain is altered from 80dB to 55dB.

Forget 55 - 80dB gain for a minute.  Let's look at the 20 - 50dB gain range.  This is where it'd be set for 90% of cases anyway.

ruffrecords said:
The source it sees is effectively the mic output impedance times 4. Often this is not optimum for the transistor -

Yes, it  4 X the mic impedance when the attenuator is full up or the attenuator is somewhere around -18dB down.

But at the -6dB point it is (transformed mic impedance + attenuator impedance)/ 4

So in the Neve, it would be: (600 + 4800) / 4
= 1K35


I know the Neve doesn't have a -6dB position, but the attenuator impedance is roughly 4K8/5K from 35dB down and source z gets higher than the 600 ohm straight transformed mic z close to/either side of the -6dB point.

ruffrecords said:
usually 4K8 or so is the minimum noise source impedance for the BC109.


No.  Assuming rbb is decently low, then optimum noise source impedance for the BC109, like any bipolar,  depends on the bias current.   

If we have a low source impedance, we want more current in the transistor. 

In Abbey's circuit with the 1:10 transformer, he biased it at 50uA

With a straight through 1:2 transformer such as the Neve 10468, giving a 600 ohms source z, we'd normally be seeing bias currents around where the Jensen 990 is biased, at over a mA.

For a low source, such as the 60- 90 ohms from the Neve 31267, we'd have 2 - 3mA and be settling for a lower Hfe but low rbb device like the 2N4401 or an extinct Rohm 2SB device.  Or parallels of whatever gets rbb down.   

Most low noise transistor mic amps have a fixed source impedance and therefore the bias current in the 1st transistor can be optimized.

But due to the nature of the Neve circuit , they settled on a compromise since it's not possible to satisfy minimum noise conditions for all settings of between 60 ohms and 1K35 ohms

I'm not saying the Neve is terrible.  Just intended to make a comment that, in Ricardo's circuit posted above, he'd set the first transistor bias to be noise optimized at, whatever it was -  90uA or so? - for the 1:5 transformer.  And that source impedance doesn't shift around.




 



 
On Neve gains of 55dB to 80dB, anyone who's used the gains there will tell you that the noise jumps up.

Since we're dealing with a fixed 600 ohm source though, noise could be lowered by simply re-jigging the 284 stage for more current in the 1st transistor.

Not only is it possible, it's been done.

Another trick I've played with is on the 31102 module which uses a separate little buffer for the line stage.  Shove a bit more current in that and your line ins are a bit quieter.



 
Winston O'Boogie said:
No.  Assuming rbb is decently low, then optimum noise source impedance for the BC109, like any bipolar,  depends on the bias current.   
No argument there, but I don't know if you rememeber the old BC109 data sheets, there used to be a graphs of noise figure versus source impedance for various collector currents and I am pretty sure the minima was at about 4K8. For some reason you don't seem to get that on transistor data sheets any more.

Cheers

Ian
 
ruffrecords said:
No argument there, but I don't know if you rememeber the old BC109 data sheets, there used to be a graphs of noise figure versus source impedance for various collector currents and I am pretty sure the minima was at about 4K8. For some reason you don't seem to get that on transistor data sheets any more.

Most all the Neve's I played with had BC184C transistors, so a later iteration. 
It is possible to bump up current on those for sure but I honestly don't know about the BC109C's

These days we can cheat and put a low rbb high Hfe BJT in there, although it requires re-jigging the circuit as everything in those things is intertwined and there're only 3 transistors anyway.
You might change one resistor, and then the 2nd transistor gets starved and farts on ya.



Anyway, onwards and upwards.  :)
 
Winston O'Boogie said:
Most all the Neve's I played with had BC184C transistors, so a later iteration. 
It is possible to bump up current on those for sure but I honestly don't know about the BC109C's
I think only mentioned BC109s after you did but I may be wrong.
For some reason the BC184C does not appear in the Blue Mullard book from September 1974 (when I was at Neve) even though the later BC549 does. Wikipedia argues these were all part of a sequence BC109-->BC184-->BC549. From the 1974 Mullard data book the BC549 has identical noise curves to the BC109 so it is reasonable to assume the BC184 is the same.

If you look at the NF curves for BC109 and 549 you will see the lowest NF of 1.75dB can be obtained with a collector current between 0.1 and 1mA and a source impedance of around 800 ohm which I think I said is the typical impedance a 200 ohm mic presents to the circuit. It is difficult to see how simply altering the collector current could make much difference in NF. (pdf of NF curves attached)
These days we can cheat and put a low rbb high Hfe BJT in there, although it requires re-jigging the circuit as everything in those things is intertwined and there're only 3 transistors anyway.
Which low rbb high Hfe BJT do you have in mind?
You might change one resistor, and then the 2nd transistor gets starved and farts on ya.
Yeah but that is just tinkering not proper design.

Cheers

Ian


 

Attachments

  • BC549NF.pdf
    46 KB
ruffrecords said:
If you look at the NF curves for BC109 and 549 you will see the lowest NF of 1.75dB can be obtained with a collector current between 0.1 and 1mA and a source impedance of around 800 ohm which I think I said is the typical impedance a 200 ohm mic presents to the circuit.

I think we're talking at crossed purposes here
So we can bias for lowest NF at a source of 800 ohms then.  You said that 4K8 was where the minima of the BC109 was. 

ruffrecords said:
It is difficult to see how simply altering the collector current could make much difference in NF. (pdf of NF curves attached)

By bumping it up to where I said it was better at that source impedance -  at 1mA or a bit more, as it is in the Jensen 990 which was specifically designed for a similar transformer turns ratio. 




ruffrecords said:
Which low rbb high Hfe BJT do you have in mind?

Ah well.  Those devices are now moot for the most part since they were EOL about 20 years ago.  Rohm 2SB7... something, something was one.  Great for mic amps and phono pre-amps and used by many folks.
Now, people seem to use the 2N4401  which is really a switching transistor.  Again paralleling 2 devices will get the rbb contribution down though and a theoretical 3dB improvement which, in practicality usually ends up about a 2dB improvement.
Only probably worth it anyway for in the 31102 with its own line amp buffer. 

ruffrecords said:
Yeah but that is just tinkering not proper design.

For sure.  Which is why I added that bit in, to dissuade anyone thinking of just throwing in another transistor and expecting major improvements without doing a proper redesign. 

Edit:  a redesign also looking at the impedance of feedback network too of course.



 

Latest posts

Back
Top