Mic Input Transformer Sweeps - Vintage and Modern

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
ruffrecords said:
So they can and often are asymmetrical? Maybe this explains the poor HF response I have been having with some OEP 1:10 input transformers?

Yes they "can" be.  The only current company I've seen publish anything within its specifications when it might be an issue with a specific transformer is Lundahl.  So it's worth checking for sure.
 
most of your dual coil transformers should have balanced C.  this might make a dual coil xfmr better to flip around if you were going to use an output as an input, like they do with UTC A-26  on the 670,  Peerless also states that the S-217-D can be used backwards, it too is a dual coil xfmr.

most of your Lundahl transformers are dual coil. Jensen, Cinemag seem to prefer the single coil stuff.  UTC is about 70% dual coil the rest single.

you might want to give that UTC A-10 a few minutes running at hi level lo freq to "wake it up'" and then repeat the test to see if the low end distortion does not improve.

that UTC LS-51 that we just tested was really dragging primary current when we first plugged it in, probably had not been hooked up in a while. the difference was quit drastic, it was drawing about 4 times as much current when fist plugged in, so we had to repeat the pri ind. test.
 
Ok, new test of the ADC A5508 with primary windings in series:

Source Z: 150 ohms
Level: -40dBm
Frequency: 1kHz

Balanced cable to clip leads, terminals 3+4 tied with soldered jumper.  Positive signal input to 1, Negative to 6, Shield to pin S.

Out of the transformer secondary, balanced cable with pin 8 +, pin 7 - and Shield to pin S.

Input Z of the Amber: 100k ohms
Signal level: -21.1dBm

Referenced to the level at 1k I found:

20Hz is -0.07dBm
100Hz is -0.01dBm
2kHz is +0.17dBm
5kHz is +1.4dBm
10kHz is -1.3dBm
15kHz is -10.6dBm
20kHz is -19.3dBm

So an interesting little bump around 5k, and much better at 10k but still drops like a rock past that.  Note that at 20kHz it is essentially 1:1.  I am thinking of building a preamp with this since I have the matching output transformer A5509 which I have already posted about here.  Any thoughts on that are welcome.  I am thinking in BA-2A/SA-70 sort of directions.
 
CJ said:
try it without the shields connected...

CJ said:
try it without the shields connected...

Everything exactly the same until...
10kHz is -0.85dBm
15kHz is -9.78dBm
20kHz is -17.75dBm

So we gain 1dB of top end.  Again, this is sitting on my bench with ~1 meter of Canare star quad cable going to alligator clips each way. Pic: https://www.dropbox.com/s/2q8109229h0kaz5/ADC%20test.jpg?dl=0
 
mjrippe said:
Everything exactly the same until...
10kHz is -0.85dBm
15kHz is -9.78dBm
20kHz is -17.75dBm

So we gain 1dB of top end.  Again, this is sitting on my bench with ~1 meter of Canare star quad cable going to alligator clips each way.

Thanks for running these tests. Makes me feel better that I was testing it correctly as we have similar results.

I have a whole pile of old transformers to test this evening. I'll report back.
 
BluegrassDan said:
Thanks for running these tests. Makes me feel better that I was testing it correctly as we have similar results.

I have a whole pile of old transformers to test this evening. I'll report back.

I have a handful as well, but gotta get some paying work done first.
 
I have a bunch of response curves for various transformers in a Google Drive folder I shared on GDIY a couple of years ago. I haven't added to it in a while. Few of them are iconic or desirable units - most were pulls and oddballs I tested to see if they were usable in commissioned projects for people that wanted a preamp or compressor or somesuch but didn't want to spend for higher end stuff. But there are some classics in there. One of these days, I'll get around to documenting all the good stuff.

This stuff was done using a Gen Rad white noise generator instead of signal generator sweeps, but the result is the same. You're still putting the same amount of voltage in across the frequencies of interest. There's a slight dip between about 13kHz and 17kHz on some of the plots due to the sound card I was using at the time.  So you can ignore that.

And don't get hung up on the levels - I was using a Pete Millet home built buffer/attenuator on some of them, and not on others. Levels were appropriate for the transformer type tested, usually set by testing with a single 20, 30, 40 or 50Hz distortion test beforehand. Also, some were log plots, some linear. If they aren't labelled one or the other, I think they're linear plots. Again, it's been a long time. It's a motley bunch as it was all done over a period of a couple of years or so.

Maybe some of it will be of interest. Maybe not.

https://drive.google.com/drive/folders/0B__Vr0ik2GcgLVlidXRjd0c2b1U?usp=sharing
 
rackmonkey said:
I have a bunch of response curves for various transformers in a Google Drive folder I shared on GDIY a couple of years ago.

I remember that and I had downloaded some of the files, but there were a LOT of them.  Just took a big google dump of the folder so I can peruse it at my leisure.  Thank you again for sharing this!
 
BluegrassDan said:
Here are two sweeps of the Peerless with different situations. The red line is with a -30dB PAD line-level attenuator (150 ohm) between the interface output and the transformer input. The blue line is straight from the interface output (5 ohm) with the levels attenuated -30dB digitally.

[The upload folder is full, so here is a dropbox link.]
https://www.dropbox.com/s/qtmm5n4bx73iygp/Screen%20Shot%202020-06-13%20at%204.08.46%20PM.png?dl=0

Probably a silly suggestion but have you tried increasing the sweep time?

I remember in the past I used to get similar results at the low end by having the sweep time too short.

Just thinking out loud really ☺
 
I used white noise, not sweeps. It was my setup at the time. Since then, i’ve gotten a much improved interface and the hump is gone. I assume the hump was from stray capacitance, or the size of the cap on the input of the old sound card. (The test setup in the shop is too far from my recording setup to use the good stuff, so it was fairly janky by comparison until a couple years ago).

You can see from the plot of just the test setup in that folder that the hump only appeared with the DUT in the path, hence the resonance theory.

Also keep in mind that many of those are linear plots rather than log, and low end levels relative to high end are thus higher than a log plot. On the log plots, they look pretty close to what one would expect in most cases. But the hump was there, no doubt.
 
I always use pink noise and move settling time to longest once it’s mostly settled.  White certainly fine too.
 
if 20-20K curves get boring you can step it up a notch and start checking BH curves on antique transformers,

application article>

https://www.ee.iitb.ac.in/course/~emlab/assets/bh_curve22.pdf

 
BluegrassDan said:
Well, here are a few more interesting examples from days of yore.

An RCA N-6567 and some little rectangular transformer with "TE 5266 F" on the side. The top lettering is worn away.

https://www.dropbox.com/s/7624lpd1i63ll5z/RCA%20N-6567.jpg?dl=0

https://www.dropbox.com/s/5v5pqs5vt0e1r26/TE5566F.jpg?dl=0

Stumped on 6567, that's hard to do.    From memory Audio Cyclopedia says loading typically results in a 3dB loss, so that looks like the 100Kish line. 

TE goes to RCA MI-9268 / MI-9285 / MI-10289-A mag film preamps.  Probably Hollytran.  30Ω primary; that's why it looks bandpassed. 


 
Back
Top