'T' attenuator response interactions at higher attenuations

[emrr]

Let's hear some theories.  I certainly hope there's no 'mistake' in the measurements, and can't think of a thing that would surface during attenuation that wouldn't also surface without it, or at moderate levels. 

I'll look at it again, and some other examples when I can. 

It's a typical setup, MOTU 16A through 25 feet of balanced wire to a patchbay, out to the DOT, back into patchbay through another 25 feet of wire to the 16A.  I've seen it on other amps, in different eras with different converters, shorter more direct wire runs, etc. 

I can also look on my AP ATS-1. 

Jeff, curious what you see below -31dB (8 o'clock).  What's it look like fully counterclockwise  (7 o'clock)?

 

[Ilya]

That's rather interesting. I'm getting the same frequency responce from my 1176 input stage (which is t-pad into Cinemag O-12 replica). Although this is valid for the very first few degrees of the T-pad travel. Approximately the last 4-6 dB of attenuation. Otherwize it's ruler flat.

 

[ruffrecords]

Are you loading the attenuator output with 600 ohms?

Cheers

Ian

 

[Ilya]

Are you loading the attenuator output with 600 ohms?

If this question is directed to me, than yes, the t-pad is loaded properly since it's inside the original circuit.

 

[ruffrecords]

If this question is directed to me, than yes, the t-pad is loaded properly since it's inside the original circuit.

Actually it was directed at EmRR but..

Yours may not be loaded with 600 ohms. The load is sees is whatever is connected across the secondary of the Cinemag reflected to the primary i parallel with any resistor that may be across the prmary. Can you post a schematic of the input circuit?

Cheers

Ian

 

[emrr]

Ian, I am pretty sure I have seen it loaded and unloaded, I have seen it on both inputs AND outputs of differing equipment. 

 

[emrr]

That's rather interesting. I'm getting the same frequency responce from my 1176 input stage (which is t-pad into Cinemag O-12 replica). Although this is valid for the very first few degrees of the T-pad travel. Approximately the last 4-6 dB of attenuation. Otherwize it's ruler flat.

Yes, the extremes of the range.  Interesting to see other confirmation.  Looks like you've got a 30dB treble  boost starting around 1kHz?

 

[emrr]

Yours may not be loaded with 600 ohms. The load is sees is whatever is connected across the secondary of the Cinemag reflected to the primary i parallel with any resistor that may be across the prmary. Can you post a schematic of the input circuit

1176, Ian.  Those are a T pad into the 500/600 winding of a UTC O-12 that's loaded with 270Ω on the 200/250 secondary.  It's proper, but influenced directly by the source, which from what I've seen is predominately amount of attenuation for a given knob rotation.    Offhand I don't see specs on a Cinemag clone, but that's what it should be. 

 

[Ilya]

Here's the input circuit (I'm using the 3-deck T-pad, while the circuit shows 2-deck equivalent):

And yes, the Cinemag is spot on. I have vintage O-12 somewhere, but I'm pretty sure they'll behave the same. Actually, I have another 1176 unit with NOS T-pads and original O-12 and I'll measure it as soon as I can get my hands on it.

Now, EmRR is correct. I apologize for the abscense of frequency numbers on my graph. The frequency responce starts to increase at around 1k and has approximately 30dB boost.
Attenuation maximum of the T-pad is 74dB. The frequency respons flattens around 48dB of attenuation (which is at the very beginning of the pot travel). After that (if you turn the pot CCW) the attenuation increases rapidly, and the frequency responce looses bass.

 

[Ilya]

Here's the frequency responce of the Mallory NOS T-pad into the vintage O-12 (fully CCW):

I'm pretty confident that original UREI will measure the same.
Please note that these NOS T-pads have less max attenuation than the new Bourns ones - 56dB vs 74dB.

Here's the frequency responce of the new Bourns T-pad that is set to the 56dB of attenuation (that is the maximum value for the NOS T-pad):

 

[ruffrecords]

1176, Ian.  Those are a T pad into the 500/600 winding of a UTC O-12 that's loaded with 270Ω on the 200/250 secondary.  It's proper, but influenced directly by the source, which from what I've seen is predominately amount of attenuation for a given knob rotation.    Offhand I don't see specs on a Cinemag clone, but that's what it should be.

I am not so sure. That T circuit should maintain a 600 ohm source pretty much over the entire range provided you drive it with a 600 ohm source.

Cheers

ian

 

[emrr]

I am not so sure. That T circuit should maintain a 600 ohm source pretty much over the entire range provided you drive it with a 600 ohm source.

I'm not sure what you're not sure of. 

These modern T's (last 20 years?) are an approximation, a pair of series 500Ω linear decks and a 1KΩ shunt deck, no open position.  This translates to a load range from 500Ω to 1KΩ without an external load; an approximation that works ‘ok’ with many SS sources.  They are, as far as I've seen, always used in transformer coupled situations where they are on an input that's driven with a lower Z that may or may not have 'proper' loading, or an output where they are driven with a lower Z and fed to a higher Z.    And that translates directly to vintage equipment used in modern environments.

Here's the frequency responce of the Mallory NOS T-pad into the vintage O-12 (fully CCW):

I'm pretty confident that original UREI will measure the same.
Please note that these NOS T-pads have less max attenuation than the new Bourns ones - 56dB vs 74dB.

Here's the frequency responce of the new Bourns T-pad that is set to the 56dB of attenuation (that is the maximum value for the NOS T-pad):

Interesting to see the difference.  Very little change with the NOS Mallory, and at very deep attenuation.

 

[Ilya]

Interesting to see the difference.  Very little change with the NOS Mallory, and at very deep attenuation.

Indeed. I'm curious about the new Bi-Technologies t-pad that is sold by Hairball. How does that compares to the Bourns one... If anybody can do any measurements it would be very interesting to see them.

 

[Ilya]

For what it's worth, here's the true resistor T-pad cobbled together with 3 resistors I had at hand (2x604R and 1.2R) fed right into the 1176 input with it's T-pad fully CW:

 

[ruffrecords]

For what it's worth, here's the true resistor T-pad cobbled together with 3 resistors I had at hand (2x604R and 1.2R) fed right into the 1176 input with it's T-pad fully CW:

Whih seems to imply it is not the pad that is causing the HF rise.

Also I do not understand why the levels are so low - or is this just the noise spectrum we are seeing?

Cheers

Ian

 

[ruffrecords]

I'm not sure what you're not sure of. 

If" it is proper but influenced by the source"

Cheers

Ian

 

[emrr]

Here's a Daven T-321-G into 10K.  Source Z 200Ω. 
0, -10, -20, -30, and off.
‘off’ is -26 into 10K, -31 into 600.

full size pic HERE

The above attenuator tested with an Audio Precision ATS-1, 600 source/load, -10 and -20 dead on for attenuation value and totally flat.
1kHz 'off' is -56.7dB into 600Ω, or -50.68dB into 100KΩ.  Looks about the same shape 600Ω or 100KΩ.

10Hz -83dB 100Hz -63.8dB 1kHz -56.7dB 10kHz -46.9dB  20kHz -41dB 30kHz -38dB.

-30dB looks like:

10Hz -30.1dB 100Hz -30dB 1kHz -29.72dB 10kHz 29.64dB  20kHz -29.4dB 30kHz -29.1dB

Here's a Daven decade T attenuator, also 200Ω source Z.  This shows the relative attenuation for both 636Ω load and 10KΩ load with respect to labeled setting. 

load-> 636 10K -10 -11.6 -6.9 -20 -19.8 -14.9 -30 -25.8 -20.8 -40 -29.1 -24.1

full size pic HERE

 

[Ilya]

Whih seems to imply it is not the pad that is causing the HF rise.

You can see from the graph that the simple resistor pad doesn't have a huge rise in the HF as the pots. 2dB@20kHz vs 3dB (Mallory) vs 7dB (Bourns)

Also I do not understand why the levels are so low - or is this just the noise spectrum we are seeing?

The level is low because there's a 56dB attenuator there. The noise spectrum of my system is around -140dBFS. I probably should've adjusted the output to get the signal up to -70dBFS, but I didn't want to change any settings on the units because they were matched to each other.

 

[moamps]

Ilya, could you do a frequency response at 56dB of attenuation but with only 1kHz input signal to see is the rise of the freq. response at HF only signal or combination with noise?
I believe the problem is in impedance-contact noise of the middle section of the attenuator  where the current thru the wiper is at maximum at high attenuation.

 

[Ilya]

Ilya, could you do a frequency response at 56dB of attenuation but with only 1kHz input signal to see is the rise of the freq. response at HF only signal or combination with noise?

Here you go