Neve-style Square Wave Riddle

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Anthropic

Active member
Joined
Jun 23, 2005
Messages
34
Location
New York
Hello fellow DIYers,

I have a bit of a riddle for you:  I've just completed a 4 channel 1290-style microphone preamp, using a PCB I designed/derived from the original BA283/284 cards and schematics.  The PCBs were also designed to accommodate the 2006 Group Buy Carnhill transformers (VTB2168), meaning that the PCBs (on which the input TX is mounted) incorporate the changes needed to use the differently-labeled input transformer properly, as CJ intended. 

Upon my first listening tests, all is well.  They sound darker than some of my other preamps, but that isn't necessarily a problem.  What I found unusual, however, was when I did some square-wave testing on them.  Using a BK Precision 4040 function generator, and feeding a 1kHz square wave to the preamps (with a 150 Ohm resistor across pins 2 and 3 of the input XLR to simulate a microphone), I get some strange results.  While feeding the same signal to my REDD47, Syteks, and Ampex 351, I got nearly perfect square waves on their outputs.  But, when going into the 1290-style preamps, I got this:

https://photos.app.goo.gl/LFzMq8RnzstgXCAT8

All channels exhibit an identical "bass-cut" square wave profile, so it doesn't appear to be a simple bad component.  Plus, using my scope probe to measure the output of my function generator, I get a perfect square wave, but as soon as I attach the function generator to the XLR of the preamps, I get the skewed square wave (even measured prior to where the signal hits the input transformer!). 

For further testing, I tried feeding the preamps a 10kHz square wave, then a 3kHz square wave, respectively.  Both exhibit a much flatter "horizontal" portion:

https://photos.app.goo.gl/EHJUDrfB4hwQmNdv7

https://photos.app.goo.gl/QR4kJxZg2SFc2Vzg7


Does anyone have any clue what is happening here?  Is this what a typical Neve-style preamp will produce?  Is it possible that the input transformer is loading the input signal in some unusual way, such that a square wave is caused to become skewed even when measured at Pin 2 on the back of the input XLR? 

Thanks! 
 
I don't know if it is relevant but if you want to make your (low output impedance) sig gen look like a microphone you need to wire 150 ohms is series with it rather than across the hot and cold of the XLR - 75 ohms in each leg will do it.

Cheers

Ian
 
Anthropic said:
Does anyone have any clue what is happening here?  Is this what a typical Neve-style preamp will produce?
That is utterly normal. This is what any transformer-based mic pre will do. The inductance of the primary, combined with the source impedance create a HighPassFilter, that's why you see it both at the primary and at the preamp's output..
Regarding the 3k/10k squarewave response, I would think it's an artefact of the oscilloscope's dual-channel switch (insufficient blanking).
 
Greetings,

Could it be the input TX requires a different compensation network across the secondary? Curious what the results of a sine sweep from 20-20K HZ would be?

Regards, Jim
 
Thank you, gentlemen.  All of your input was helpful.  I'll perform some additional testing, including a sweep, and report back.  Thanks, again.
 
abbey road d enfer said:
That is utterly normal. This is what any transformer-based mic pre will do. The inductance of the primary, combined with the source impedance create a HighPassFilter, that's why you see it both at the primary and at the preamp's output..
Regarding the 3k/10k squarewave response, I would think it's an artefact of the oscilloscope's dual-channel switch (insufficient blanking).

But the Redd47 and Ampex 351 are transformer input without the same anomaly?  Is there maybe a problem with these  Carnhill transformers (VTB2168) ?
 
hitchhiker said:
But the Redd47 and Ampex 351 are transformer input without the same anomaly?  Is there maybe a problem with these  Carnhill transformers (VTB2168) ?

This was my fear as well since, as you correctly point out, both the REDD47 and Ampex 351 are transformer in/out.  I suspect more testing will be necessary to get to the bottom of this. 
 
hitchhiker said:
But the Redd47 and Ampex 351 are transformer input without the same anomaly?  Is there maybe a problem with these  Carnhill transformers (VTB2168) ?

Both the Redd and Ampex are tube units with large step ups.  Tube grids tend to be quite hi-z.  Maybe not the best comparisons.
 
That function generator has a 50 Ohm output impedance, so you need to add a pair of 100 Ohm resistors to each leg to present 150 Ohms impedance to the mic input.

Bri

 
Brian Roth said:
That function generator has a 50 Ohm output impedance, so you need to add a pair of 100 Ohm resistors to each leg to present 150 Ohms impedance to the mic input.

Bri

I think you mean 50 ohms in each leg for 100 ohms total added.

Cheers

Ian
 
Brian Roth said:
That function generator has a 50 Ohm output impedance, so you need to add a pair of 100 Ohm resistors to each leg to present 150 Ohms impedance to the mic input.

Bri
This is moot, since the HP effect is visible although the source impedance is lower than expected.
Indeed, with a higher source Z the effect will be more pronounced.
It just means the primary inductance is significantly lower than that of other xfmrs he has tested this way.
It looks like designers in the late 60's/early 70's figured out they could use less inductance and get away with it.
After all, who really cares if the response at 20Hz is 0.3dB down or 0.1?
 
David Kulka said:
It would be helpful to know the frequency response.
Yes, I was going to suggest that, but the overall frequency response may be significantly different than the input xfmr's response, particularly if a deliberate LF cut-off is introduced.
 
Hello gents,

UPDATE

It appears that the HPF phenomenon with the 1 kHz (and below) square wave function was due to the source resistance in the feed from the signal generator to the preamps.  In my initial tests, I had voltage dropping resistors installed in the input XLR connector (along with a 150 Ohm resister across pins 2 and 3), as suggested by Ian in this thread: https://groupdiy.com/index.php?topic=55200.0

When I, instead, took a line straight from the function generator (50 Ohms output) to the mic pre input (1200 Ohms input impedance), the resulting 1 kHz square wave measured on the preamp's output was much more level (as good, or better, than the slope of the 3 kHz square wave shown above). 

So, it looks as though the lower ratio of the Carnhill input transformer interacted unfavorably with the dropping resisters in series with the input signal, unlike the higher-ratio transformers in my REDD47 and Ampex 351 (or the electronically-balanced inputs of my Sytek and console mic inputs). 

Thank you for sharing your expertise. 
 
Anthropic said:
Hello gents,

UPDATE

It appears that the HPF phenomenon with the 1 kHz (and below) square wave function was due to the source resistance in the feed from the signal generator to the preamps.  In my initial tests, I had voltage dropping resistors installed in the input XLR connector (along with a 150 Ohm resister across pins 2 and 3), as suggested by Ian in this thread: https://groupdiy.com/index.php?topic=55200.0

When I, instead, took a line straight from the function generator (50 Ohms output) to the mic pre input (1200 Ohms input impedance), the resulting 1 kHz square wave measured on the preamp's output was much more level (as good, or better, than the slope of the 3 kHz square wave shown above). 

So, it looks as though the lower ratio of the Carnhill input transformer interacted unfavorably with the dropping resisters in series with the input signal, unlike the higher-ratio transformers in my REDD47 and Ampex 351 (or the electronically-balanced inputs of my Sytek and console mic inputs). 

Thank you for sharing your expertise.
Assuming the resistors have been connected correctly and the 150r resistor being correctly across pins 2 & 3, which results in the correct 150r nominal source impedance, your first observation indicates the Carnhill xfmr has a lower inductance than the other xfmrs you have tested in an identical manner.
Indeed, presenting the xfmr with a lower source impedance reduces the HPF effect and tends to flatten the horizontal part of the square wave.
 
Back
Top