Pad network to feed mic transformer with line levels

[user 65064]

Hi guys, this is my very first thread, please be kind!  ;D

I'm asking your advices since I have an old mixing console with mic transformers per channels.

Since I don't use mics inputs at all I decided to try modding their newtork to let it handle hotter signals (in my case line level unbalanced signal).

I decided to go with a padding network and my first attempt was a 10 kohm resistor in series with signal path + a 470 ohm shunt after it, in parallel with the primary winding.

This solution is good for the input impedance but then the signal feeding the transformer was almost killed needing to be followed by an absurd gain stage that would add a lot of noise and nonlinearity..

So i tried experimenting with resistors and a network that let me keep a good secondary signal level is a 1kohm resistor in series with the signal + 220 ohm shunt, in parallel with the primary winding.

The transformer was used to show a 200 ohm input impedance and has a ratio of 1:10 and the secondary is loaded with 47 kohm (but since I use it with an unbalanced connection the ratio should be halved if I'm not wrong).

Anyway I just wanted to know if this route may have sense since in this case my input impedance is quite low respect to normal line level input impedance (in my case is around 1,2 kohm).

I assumed that this wouldn't be a critical load for a modern output stage, isn't it?

I know this padding thing is not good for S/N ratio but it's just for fun and experiment a bit.

By the way, do you thing it could be a better alternative swapping the secondary load, originally 47 kohm, with a 1 Mohm?

If I'm not wrong in this case the primary winding should exibit an impedance of 10 kohm without needing any network, may this be a better choice?

In the latter case, I'm ok to adjust the following preamp stage to avoid clipping since the voltage would be boosted of 5 times (since I use it unbalanced).

I thank you guys and I hope to learn as much as I can, this forum is handsome!

Cheers.

 

[abbey road d enfer]

Hi guys, this is my very first thread, please be kind!  ;D

I'm asking your advices since I have an old mixing console with mic transformers per channels.

Having a schemo would help understanding the problem.

  my first attempt was a 10 kohm resistor in series with signal path + a 470 ohm shunt after it, in parallel with the primary winding.

This solution is good for the input impedance but then the signal feeding the transformer was almost killed needing to be followed by an absurd gain stage that would add a lot of noise and nonlinearity..

This would indicate that the actual input impedance of the mixer is absurdly low...which I don't really believe. I would rather think you made a mistake with your connections.

So i tried experimenting with resistors and a network that let me keep a good secondary signal level is a 1kohm resistor in series with the signal + 220 ohm shunt, in parallel with the primary winding.

The transformer was used to show a 200 ohm input impedance and has a ratio of 1:10 and the secondary is loaded with 47 kohm

That would result in an input impedance of about 500 ohms, which is quite low according to current standards, but may have been considered normal at the time.

(but since I use it with an unbalanced connection the ratio should be halved if I'm not wrong).

No. This 6dB thing is an issue with some sub-par electronically balanced stages. Not with transformers. Have you connected the cold leg of the input to the cold leg of the source (which may be the "ground" in an unbalanced output)?

Anyway I just wanted to know if this route may have sense since in this case my input impedance is quite low respect to normal line level input impedance (in my case is around 1,2 kohm).

As long as you know it and use it only with low-Z sources, who cares?

By the way, do you thing it could be a better alternative swapping the secondary load, originally 47 kohm, with a 1 Mohm?

Depends on what's there, a tube, a transistor, an opamp...?

If I'm not wrong in this case the primary winding should exibit an impedance of 10 kohm without needing any network, may this be a better choice?

That would be true only if the inductance of the xfmr's primary is high enough; if not, there may be a lack of bass.

In the latter case, I'm ok to adjust the following preamp stage to avoid clipping since the voltage would be boosted of 5 times

No; the step-up ratio of the xfmr is fixed. The input impedance may change, but the step-up gain will not.

(since I use it unbalanced).

I don't understand that comment; the ratio of a xfmr does not change with how it's connected.

 

[user 65064]

Having a schemo would help understanding the problem.

I know and I'll write it down asap, I haven't found any schematic on the net currently I'm brutally working directly on the boards

This would indicate that the actual input impedance of the mixer is absurdly low...which I don't really believe. I would rather think you made a mistake with your connections.

The mic Input Impedance is super low since was planned to give 200 ohm impedance, quite old design.

That would result in an input impedance of about 500 ohms, which is quite low according to current standards, but may have been considered normal at the time.

I modified the secondary loading it with 47kohm but originally the secondary load was around 22 kohm
letting the input show more or less 200 ohm to the source.

No. This 6dB thing is an issue with some sub-par electronically balanced stages. Not with transformers. Have you connected the cold leg of the input to the cold leg of the source (which may be the "ground" in an unbalanced output)?

Yes, I'm quite sure to have wired it correctly.

Anyway in case I had swapped the hot and the cold wires this wouldn't affect anything but the phase.

As long as you know it and use it only with low-Z sources, who cares?

Nice to hear that! I just wanted to be safe in case I face some gear with a bigger output impedance (I mean something around 1k).

Depends on what's there, a tube, a transistor, an opamp...?

Next stage is a buffering gain stage that uses a tl071 so I was thinking to rework it to avoid headroom issues.

That would be true only if the inductance of the xfmr's primary is high enough; if not, there may be a lack of bass.

Much thanks! I didn't thought about it.. ;D  I need to keep a good bass response.
Maybe a I just could use a secondary of perhaps 470 khom, it may be a good compromise, what you think?

No; the step-up ratio of the xfmr is fixed. The input impedance may change, but the step-up gain will not.
I don't understand that comment; the ratio of a xfmr does not change with how it's connected.

Since the primary is center tapped if I use it with an unbalanced connection the current I feed into it runs an halve of its windings and the ratio should step up in this case, am I wrong?

By the way much thanks man, you already showed some goos spot to me!

 

[abbey road d enfer]

The mic Input Impedance is super low since was planned to give 200 ohm impedance, quite old design.

I modified the secondary loading it with 47kohm

So, with your attenuator of 10k in series and 470r shunt, that would result in about 34 dB attenuation. That is not in accordance with your comment "the signal feeding the transformer was almost killed needing to be followed by an absurd gain "; a line level with 34 dB attenuation would result in about -30 dBu. The preamp gain should be about 30-35dB, which is not absurd. So one element of data is missing here. What are you using as line source?

Since the primary is center tapped

That is quite unexpected!  If you're using only one leg for signal, that may explain the extra loss.

  if I use it with an unbalanced connection the current I feed into it runs an halve of its windings and the ratio should step up in this case, am I wrong?

True, but the impedance would be much lower, 50r instead of 200. So the attenuation would be 12 dB more.
Really, a good schemo, with the description of your connections would help solving the mystery.

 

[user 65064]

So, with your attenuator of 10k in series and 470r shunt, that would result in about 34 dB attenuation. That is not in accordance with your comment "the signal feeding the transformer was almost killed needing to be followed by an absurd gain "; a line level with 34 dB attenuation would result in about -30 dBu. The preamp gain should be about 30-35dB, which is not absurd. So one element of data is missing here. What are you using as line source?

I mean "almost killed" since the attenuation was quite heavy and I was concerned that then an aggressive gaining stage would bring up a lot of unwanted noise.

That is quite unexpected!  If you're using only one leg for signal, that may explain the extra loss.

You totally get my situation, it's a bit tricky!

True, but the impedance would be much lower, 50r instead of 200. So the attenuation would be 12 dB more.
Really, a good schemo, with the description of your connections would help solving the mystery.

I've attached the schematic of my last trial, using a 1 kohm series resitor and 220 ohm shunt I was able to have, at the same gain sitting, a level comparable to the straight line input (with as input impedance the 47 kohm in parallel to secondary).

By the way much much thanks for your time an advices!

;D

 

[abbey road d enfer]

The problem is you are shorting one half-primary, that's what "kills" the signal.
I would suggest first you disconnect the center-tap; I don't think there's any good reason to keep this medieval arrangement.
Then, is your source really unbalanced?
I would also increase the 47k, significantly; something like 220k. The transformer may ring then, you would notice that as an excess of treble. A Zobel may be necessary for taming this ringing. Now you need some test equipment for tuning the Zobel, but it's not too difficult; you can use your DAW as test instrument.

 

[user 65064]

Nice tips, I'll try 'em!

About my padding network would you sudgest  to increase primary impedance just working on secondary loading or you think it may be good to keep the pad?

By the way, my sources are all unbalanced connections (from a close hardwired patchbay).

Will let you know asap.

 

[abbey road d enfer]

About my padding network would you sudgest  to increase primary impedance just working on secondary loading or you think it may be good to keep the pad?

This is not an option; you need a pad. Increasing the impedance will not magically decrease the level that hits the opamp.

 

[user 65064]

Yep I know it and I'm not concerned about the following amp stage.. I'll rework it when I'm happy with the trasformer issue.

Since it is a non inverting stage based on a tl071 (great input impedance) I'm just ignoring it for now..

Without the pad I think the main problem would be a strong nonlinearity, it's small transformer it can't handle much power.

By the way I found a quite good balance using a 2,2kohm input resistor + a 68ohm shunt on primary and 47kohm + a 220 nF as Zobel on the secondary.

Running a square wave through it the output isn't distorted as far as I can see (measure by oscilloscope), I'll stay around this values and make some better measurement later.

Thanks again!

 

[abbey road d enfer]

By the way I found a quite good balance using a 2,2kohm input resistor + a 68ohm shunt on primary and 47kohm + a 220 nF as Zobel on the secondary.

220 nanoFarad, really? Typically, a Zobel would use 100pF-1nF; here it's exactly like you were loading with a 47k.

 

[user 65064]

Lower cap values where uneffective and on my first squarewave test I had a really distorted wave (poor bass response).

I guess the reason is that the transformer were handling more power than expected so the low freqs were struggling, the 68 ohm shunt resistor on primary helped a lot.

Since the primary and secondary impedances are connected maybe this loading config. is compensating (primary imp. higher at low freq. -> less power transfomer) but I should investigate more to be sure about it.
By the way I'm still assuming TL071 input imp. infinite.

I don't have any datasheet unfortunately, I'm going just by a "trial-error" method and since the impedance of primary and secondary influences each other, it's tricky!

 

[abbey road d enfer]

Lower cap values where uneffective and on my first squarewave test I had a really distorted wave (poor bass response).

I guess the reason is that the transformer were handling more power than expected so the low freqs were struggling, the 68 ohm shunt resistor on primary helped a lot.

These are two different subjects; LF response indeed is influenced by level and source impedance, and the Zobel network acts at HF.
The 68 ohm resistor providdes the primary with a low source impedance, which helps with LF frequency response. The attenuation provided by the 68 ohms and whatever series resistance reduces the level, so it helps LF distortion.
HF response is tamed by the 47k resistor, and the 220nF is useless, you may as well have a jumper there. This is brute force, as it loads the secondary all over the audio spectrum; a Zobel is more subtle, as it loads only in the frequency band where it's necessary, in order to compensate the resonance of the secondary.
The 47k is probably ok, but you should decrease the capacitor, whilst observing the frequency response and optimizing the HF peak.

 

[ruffrecords]

Source impedance can also affect secondary resonance.  I have observed this in three 1:10 mic transformers by Sowter, Cinemag and Jensen. With a 150 ohm source their response is platform to 40KHz with a 150k secondary load.  With a 50 ohm source there is a 4dB peak around 25KHz.

Cheers

Ian

 

[abbey road d enfer]

Source impedance can also affect secondary resonance.  I have observed this in three 1:10 mic transformers by Sowter, Cinemag and Jensen. With a 150 ohm source their response is platform to 40KHz with a 150k secondary load.  With a 50 ohm source there is a 4dB peak around 25KHz.

Cheers

Ian

Not can, does. Optimization of a Zobel network implies providing the correct source Z to the device. A mistake commonly made by younger designers is to let their generator at its lowest impedance and optimize the Zobel there; I know, I've been there.
Indeed, with a non-constant-Z attenuator, compromises are to be made.

 

[user 65064]

These are two different subjects; LF response indeed is influenced by level and source impedance, and the Zobel network acts at HF.
The 68 ohm resistor providdes the primary with a low source impedance, which helps with LF frequency response. The attenuation provided by the 68 ohms and whatever series resistance reduces the level, so it helps LF distortion.
HF response is tamed by the 47k resistor, and the 220nF is useless, you may as well have a jumper there. This is brute force, as it loads the secondary all over the audio spectrum; a Zobel is more subtle, as it loads only in the frequency band where it's necessary, in order to compensate the resonance of the secondary.
The 47k is probably ok, but you should decrease the capacitor, whilst observing the frequency response and optimizing the HF peak.

I've used this value because starting with 470pf, increasing progressively routed me there.

Making the secondary loading variable with frequency and then the power handled by primary (helping it with the 68 ohm shunt) worked for me in a situation where no data where available.

I just know the trafo. was designed to work with 200 ohm source, mic signals nothing more.

My network just wanted to pad line signals offering a decent input impedance, considering a device/generator connected to the circuit it should see as input impedance: its output impedance + 2,2 kohm resistor + max 68 ohm (at low freq. the secondary loading is behaving as open circuit, then the primary shows an high input impedance compared to 68 ohm).

What would you suggest to do?

Could you suggest a better padding network that matches your design prospective?

Anyway thanks for your time, I'm having a lot of great tips!

 

[abbey road d enfer]

Well, my primary recommendation was to disconnect the center-tap from ground; have you done it? Any attempt at optimization is vain as long as it is not done, because it actually shorts the primary.

 

[user 65064]

Well, my primary recommendation was to disconnect the center-tap from ground; have you done it? Any attempt at optimization is vain as long as it is not done, because it actually shorts the primary.

I tried to disconnect it but the the trafo didn't worked, I was able to just hear some tiny high freqs but nothing more.

Anyway the original wiring configuration was:

Primary (CT):
- Black -> Signal (Hot)
- Withe -> Phase (Cold)
- Grey -> Ground

Secondary:
- Red -> Signal Transformed
- Green -> Ground

To have output on the secondary I needed to have the Grey wire grounded.

 

[abbey road d enfer]

I tried to disconnect it but the the trafo didn't worked, I was able to just hear some tiny high freqs but nothing more.

Anyway the original wiring configuration was:

Primary (CT):
- Black -> Signal (Hot)
- Withe -> Phase (Cold)
- Grey -> Ground

Secondary:
- Red -> Signal Transformed
- Green -> Ground

To have output on the secondary I needed to have the Grey wire grounded.

If you disconnect the center tap, you have to connect the cold leg (white) to the ground of the incoming signal; the circuit needs to be closed. then you'll have 200 ohms input impedance, instead of 50 (or even less).

 

[user 65064]

If you disconnect the center tap, you have to connect the cold leg (white) to the ground of the incoming signal; the circuit needs to be closed. then you'll have 200 ohms input impedance, instead of 50 (or even less).

The withe was grounded.

Maybe I had a failing contact, I'll try again this morning.

 

[abbey road d enfer]

The withe was grounded.

Maybe I had a failing contact, I'll try again this morning.

This is what you should end up with.