Marinair 31267 headroom...

Hello,

I recently acquired a pultec clone, with a tube makeup gain amp after the passive eq section. Looking under the hood, I noticed that the output of the amp section was driving an Edcor 10K/600 transformer for the outputs. As an experiment, i thought I would try substituting a Marinair 31267, which can be wired for 10k/600 operation, to see if there would be any usefully interesting difference.

Long story short... it seems like the mariner can't really take much past +4 db, which isn't as much fun at this stage in the signal path.

So looking to replace it with a trafo that has a higher input rating, like a Peerless 15095 or UTC h-26.

Any thoughts? Have i missed something here, like demagnetizing, or is this a basic limitation of this particular transformer?

that transformer has a real small nickel core that will not take those levels,

so you are not missing anything like magnetization,

what kind of 'tube make-up'?
what is wrong with the edcor?
Pultec uses 3 transformers, output has tertiary winding NFB--clean make-up stage.

Thanks for the replies, confirms the hunch and ears analysis.

The Edcor seems a bit grainy on the very top and a bit loose on the low end, which I would like to be either more neutral or smoother, and all of the different 1:1 transformers in my little collection have different characters, so am looking for a choice of tastes here. ultimately if the edcor is the only thing that would work, then back it goes, but it just seems like a budget part and not the only or the 'best' option for the function.

Don't know too much about the make-up amp, it's based on 4 russian 6N2P tubes

That cannot be right. The 31267 is rated to at least +20dBu at the input when wired 10K/600. Is it in a circuit that is passing dc current through it??

Cheers

Ian

we can estimate flux for the 31267 at 20 hz,

cross section:.45 cm^2
turns:2700 (10k side)
f:20 hz

b=100,000,000/4.44KfAN, K=.9 stacking, so b=25 million/fAN
b=25 million/20hz * .45 cm^2 * 2700 turns
b=25 million/24,300
b=1030 V-ac,

so k-gauss at 20 hz and 10 volts  might be close to 10,000
nickel core 50/50 might take 12,000 gauss, so we max out at 11.5 volts

divide by turns ratio, 4:1, so 3 volts max on 600 side, = 12 dbu

LOL, I forgot he was using it as an output transformer. Neve consoles operated with a maximum output of +26dBu so via a 31267 wired 10K:600 that would becomes +14dBu.

Cheers

Ian

Fantastic insights! thanks CJ for the analysis, & Ian for the perspective.

Not wanting to offend any delicate sensibilities, but this is an IGS Audio prototype EQP-2, which I'm using as the master program EQ after a customized/hacked Neve mix amp that is rated to +26dbu.

And it does seem that the maximum level I can reach without distortion is about +3 to 4db, which lines up with CJ's analysis of the input headroom topping out (on the 10k side) at about +12db.

So the 31267 is not working here. I think the Edcor is rated to +20db on its input, and will be trying some other transformers in this spot, and open to recommendations. I'll be trying out a Peerless 15095 (15K/600) this week, which is rated slightly higher, to +18db (above 40hz)

I think something with ample headroom, wide frequency response and a quick transient response would be ideal.  I felt that the Edcors may be slightly spongy as well, but if they are the only transformer capable of working here, which I doubt, then so be it.

Regarding the 31267, what is it's optimal configuration? As a line input transformer, was it commonly used to provide initial gain (600 -> 10K step-up).  If used as a 2.4K=2.4K (1:1-ish), would the input headroom & max level be any different?  Thinking to use it on a group/bus insert, and as long as the levels are reasonably close to +4db, should be OK, right?

quantyk said:

Regarding the 31267, what is it's optimal configuration?

Click to expand...

10k:600 step-down.

quantyk said:

As a line input transformer, was it commonly used to provide initial gain (600 -> 10K step-up).

Click to expand...

Never, to the best of my knowledge.

Here are some original speces for the T1442 which is equal to the 31267.

Impedance:  10.000/2.500 : 2.400/600
Nominal ratio: 2+2 : 1+1
Rated maximum input +19,7dB (at 40Hz)

quantyk said:

Regarding the 31267, what is it's optimal configuration? As a line input transformer, was it commonly used to provide initial gain (600 -> 10K step-up).  If used as a 2.4K=2.4K (1:1-ish), would the input headroom & max level be any different?  Thinking to use it on a group/bus insert, and as long as the levels are reasonably close to +4db, should be OK, right?

Click to expand...

The 31267 is a line input transformer. The classic single ended class A Neve designs it is used on have a supply voltage of 24 volts. This means the maximum theoretical internal level is just over +20dBu. In practice, distortion starts to rise above +18dBu. Neve designs always aimed for a 26dB headroom which means with a +18dBu ceiling, the nominal  internal operating level needs to be 26dB below this i.e at about -8dBu. A +4dBu input to a 31267 wired 10K:600 comes out on the secondary at -8dBu. This is not a coincidence!

Cheers

Ian

So it seems this would be happiest handling line-level source signals, rather than summed mix signals.

And in the original configuration, as a step-down, it was followed by amplification, whereas in my setup, it is following amplification, so wrong horse for the course.

I think if I use it 1:1 in a context where I can control the input level to live around 0-4dbU, it shouldn't complain too much. Ultimately I may build a 1272 version, or find a circuit that would benefit from what this unit can add in terms of character.

thanks again for the insights and info.

BTW, the Neve amp i'm using is the 10852, with +25/-25v power supply rails. Balanced summing network of 4 stereo signals into differential transistor inputs , driving output transformers, then feeding the pultec clone before the A/D.

So this is last point in the signal path, and just not up to handling the high signal levels.

Success so far with the peerless, a bit more headroom, but still able to saturate it a bit too easily.

Would an output transformer work better here? Seems like a plate-to-line level conversion job.

Are there any recommendations for transformer types? The circuit seems to be single-ended, as the ground to the primary is signal/chassis ground.

quantyk said:

Success so far with the peerless, a bit more headroom, but still able to saturate it a bit too easily.

Would an output transformer work better here? Seems like a plate-to-line level conversion job.

Are there any recommendations for transformer types? The circuit seems to be single-ended, as the ground to the primary is signal/chassis ground.

Click to expand...

Try a VTB2291.

Cheers

Ian