Passive low pass circuit on output transformer secondaries

[Potato Cakes]

Hello, Everyone,

I am working on a unique project where a 1:1 (600R:600R) is in the middle of the circuit then uses buffer afterwards to unbalance the signal for the rest of the chain. Because of the level coming off of the transformer, I have to utilize a U-pad so as to no clip the input of the buffer. The client requested adding a high frequency roll off so quickest way I could think of is to use the Hi Cut circuit found in Vox amps, as shown in the top drawing. I ran into the same issue I have when I implement this on various push/pull amps which is the useful range is only last 1/4 of the turn. So I lowered the value of the potentiometer to 2k but then encountered a new issue where the signal was high passed until the full amount of high pass is engaged. I've messed around with a couple more capacitor changes but ended back to 250k and 1uF for this circuit. I used reverse log, log, and linear pots with similar results.

So the next idea is to use a dual gang pot and have separate circuits both referencing ground as shown in the bottom drawing (the pots labeled 10k should be 20k). The goal is to get the rate of audible change to be closely matched to the range of movement of the potentiometer. Of course, I could always add a low pass filter after the buffer and then add another buffer for make up gain for the insertion loss, but it would be simpler if this could accomplished passively.

I don't have a dual gang 10k pot to try this to I'll have to get one and report back.

Thanks!

Paul

 

[ruffrecords]

What buffer are you using? The problem with the passive methods at the transformer input is that they will be sensitive your driving source impedance. A way to avoid this is to achieve the high cut in the feedback loop of the buffer.

Cheers

Ian

 

[Potato Cakes]

The buffer is an NE5532. The filter is on the transformer output and not the input but I follow what you are saying.

The high cut in the loop makes WAY more sense. I was stuck thinking about tube guitar amps and how it was done on old designs, but what I've found is that I have the same issue with the high frequencies only start to be audibly cut in the last quarter of the turn as well.

I'll have to look at some schematics I have to see how to easily implement this.

Thanks!

Paul

 

[Brian Roth]

Paul....how steep of a rolloff is required? 6dB per octave is super easy. It gets more complex if you need a higher slope and/or a frequency "sweep" adjust.

Bri

 

[Potato Cakes]

6dB is perfectly fine. The trouble is having fairly even incremental changes throughout the turn of the knob.

Thanks!

Paul

 

[Potato Cakes]

What buffer are you using? The problem with the passive methods at the transformer input is that they will be sensitive your driving source impedance. A way to avoid this is to achieve the high cut in the feedback loop of the buffer.

Cheers

Ian

Oh, I surprisingly already know how to do this in the feedback. It's what is normally set as the feedback cap for stability in op amps. I had this issue one time when I put in the wrong caps and couldn't figure out why the whole circuit was low passed.

Once I determine the value of the capacitor to put in the feedback loop, will adding a potentiometer in series with that cap give me a variable frequency as or in this instance does a different method have to be used?

Thanks!

Paul

 

[ruffrecords]

Oh, I surprisingly already know how to do this in the feedback. It's what is normally set as the feedback cap for stability in op amps. I had this issue one time when I put in the wrong caps and couldn't figure out why the whole circuit was low passed.

Once I determine the value of the capacitor to put in the feedback loop, will adding a potentiometer in series with that cap give me a variable frequency as or in this instance does a different method have to be used?

Thanks!

Paul

As with all simple RC filter circuits there are two time constants and hence two frequencies. The high frequency one occurs when the capacitor has effectively become a short circuit compared to the other resistances in the network - this is the point the curve flattens out. The lower frequency is where the capacitance cause a 3dB change in response and the boost or cut begins. The trouble with a simple series resistor control is that it resistance affects both the total cut/boost and one of the frequencies. If you want just to vary the frequency then the best way is to use a switch and several capacitor values.

Cheers

Ian

 

[ruffrecords]

Just thinking about this some more, I see no reason to drive the 5532 with a balanced signal. So ground one side of the 600:600 transformer secondary and load it with 600 ohms so the source impecance is reasonably we defined (not more than 600 ohms). You could then combine your pad with the HF roll of control. For example a 91K and 10K resistor network will drop the lleve by 20dB. You could then add an RC across the 10K to achieve the HF roll off.

Cheers

Ian

 

[Tubetec]

I found a link to documents lately ,
It was a series of publications from one of the major US manufactures of sound equipment and transformers ,
Some of it was restricted ,
It was concerned with all aspects of impedence matching and equalisation using old fashioned iron cored components
it appeared to consist mainly of charts and slide rules , so various matching conditions , driving/recieving impedences and component values could be read off for filters of various orders .

I think it was RCA or WE , I forgot to bookmark the page
Maybe EMRR Doug might know ,

 

[Tubetec]

I found it again ,
https://www.bunkerofdoom.com/lit/altec/altec_and_we_transformers_amps_eq_1948.pdfYou can see the items in red are restricted ,
There are chapters on RC eq design and attenuators which might be of interest .

Would be nice to find the missing parts of this document .

 

[drtechno]

Hello, Everyone,

I am working on a unique project where a 1:1 (600R:600R) is in the middle of the circuit then uses buffer afterwards to unbalance the signal for the rest of the chain. Because of the level coming off of the transformer, I have to utilize a U-pad so as to no clip the input of the buffer. The client requested adding a high frequency roll off so quickest way I could think of is to use the Hi Cut circuit found in Vox amps, as shown in the top drawing. I ran into the same issue I have when I implement this on various push/pull amps which is the useful range is only last 1/4 of the turn. So I lowered the value of the potentiometer to 2k but then encountered a new issue where the signal was high passed until the full amount of high pass is engaged. I've messed around with a couple more capacitor changes but ended back to 250k and 1uF for this circuit. I used reverse log, log, and linear pots with similar results.

So the next idea is to use a dual gang pot and have separate circuits both referencing ground as shown in the bottom drawing (the pots labeled 10k should be 20k). The goal is to get the rate of audible change to be closely matched to the range of movement of the potentiometer. Of course, I could always add a low pass filter after the buffer and then add another buffer for make up gain for the insertion loss, but it would be simpler if this could accomplished passively.

I don't have a dual gang 10k pot to try this to I'll have to get one and report back.

Thanks!

Paul

RC filters are unbalanced and is more complicated to run in a symmetrical balanced circuit because the caps and controls have to match or suffer from common mode noise, just like matching series resistors. So the common practice is to make the secondary unbalanced then install your filters then go into the non inverting or inverting op amp. Its strange that its only low pass, because I would want a high pass then a low pass But I guess this is application specific. For recording I would want an adjustable high pass from 50hz to 400 hz and a adjustable low pass from 10KHz-25Khz but every engineer is different when it comes to this.

 

[drtechno]

There are chapters on RC eq design and attenuators which might be of interest .
Would be nice to find the missing parts of this document .

I have that, and yes its restricted. Because they are RF/radar items in the army signal core publications. I have them in print, because I went through a course in the Air Force, But they are declassified because aviation has moved on. So I imagine you could find them if someone scanned them in. But its only relevant to FCC class B techs and RF/ham radio people that would have most, if not all of that in their AARL data.

The Western Electric repeater coil supplement should be out there, because I have downloaded that before just to see what it was. So its out there with the exception of the example circuits for radio and radar jamming that caused it to be restricted.

 

[Potato Cakes]

Just thinking about this some more, I see no reason to drive the 5532 with a balanced signal. So ground one side of the 600:600 transformer secondary and load it with 600 ohms so the source impecance is reasonably we defined (not more than 600 ohms). You could then combine your pad with the HF roll of control. For example a 91K and 10K resistor network will drop the lleve by 20dB. You could then add an RC across the 10K to achieve the HF roll off.

Cheers

Ian

The problem is the primary side of the transformer has one of it's legs going to ground. So grounding a leg of the secondary will connect the primary to the secondary. I know I didn't draw it on the schematic so there was no way for anyone to know. For this project, the secondaries need to stay balanced, the u-pad needs to be in place to keep the signal from the transformer from distorting the NE5532.

I think your switchable feedback cap might be the easiest solution. It'll just take some testing to get the right combination.

Thanks!

Paul

 

[ccaudle]

because the caps and controls have to match or suffer from common mode noise, just like matching series resistors

In the second drawing of the original post, if you changed the two 22nF caps to gnd for a single 11nF from leg to leg, no connection to gnd so no reduction in CMRR, like an instrumentation amp front end.

 

[ruffrecords]

The problem is the primary side of the transformer has one of it's legs going to ground. So grounding a leg of the secondary will connect the primary to the secondary. I know I didn't draw it on the schematic so there was no way for anyone to know. For this project, the secondaries need to stay balanced, the u-pad needs to be in place to keep the signal from the transformer from distorting the NE5532.

Thanks!

Paul

I am sorry but this does not make sense to me. There is no reason why one side of the primary of the transformer cannot be connected to ground AND one side of the secondary also connected to ground. If all you are feeding is an NE5532 there is no reason for the secondary to stay balanced. I think you need to explain this project in a little more detail so we can understand the decisions you have made.

Cheers

Ian

 

[Potato Cakes]

I though electrically connecting both sides of a transformer to ground was ill advised, that is why I am using the NE5532 to go back to unbalanced signal. If there is no problem with an audio transformer having both sides then that makes things much easier and I can go back to a passive variable low pass filter and use the NE5532 as makeup gain. The transformer is part of the effect and is not used for isolation.

In the meantime, to get this first build working for the client, I've calculated several different corner frequencies based on a 2k7 feedback resistor and the closest value I have in parenthesis.

7.5kHz - 7.9nF (8.2)
4.8kHz - 12.3nF (12)
3.2kHz - 18.4nF (18nF)
1.6kHz - 36.8nF (33nF)
700Hz - 84.2nF (82nF)

I had some old mini 2x6 switches that will fit in the space already drilled in the enclosure.

Thanks!

Paul

 

[Brian Roth]

Hi, Paul....you mentioned "a 2k7 feedback resistor" and I'm confused about your newest design approach.

Bri

 

[Potato Cakes]

The drawing has a 2k7/130/2k7 U-Pad in front of NE5532 buffer. That same buffer is also providing make up gain, which is what the 2k7 is doing since there are two 2k7s on both the + and - input terminals which in other instances I've seen only half of the value of the input resistors are needed for the feedback for unity gain. I used the same value which is double the gain, but since the input resistors part of a U-pad this was just a guess which worked out to be good enough.

I think what has happened is that I was working off the premise that I could not tie one leg of the primary and secondary of the transformer to ground, which led me down all kinds of rabbit holes and compounding bad ideas on top of each other. I think what I need to do is start over since everything I've build is point to point on a prototyping PCB and all of the connections from soldering over and over are creating an unmanageable mess inside the enclosure which I am using. What is happening is that a client has a project where the transformer being used is to be driven from mild to extreme saturation. There are other circuits afterwards that use ICs which require a pad on the front end to keep from clipping the op amps, which sounds terrible. They are wanting to possibly turn it into a thing in the future so I'm avoiding showing all the puzzle pieces out of discretion. The only thing that I haven't sorted out is the high frequency roll off which was requested. I can go back to the original version that is based off of the Vox Hi-Cut circuit which only rolls off at the last quarter turn and then start on a fresh build since this one has been touched by a soldering iron one time too many.

The original build they brought me to use as a starting point did have a leg of both the primary and secondary tied to ground and it seemed to work as they wanted, but I thought this was bad circuitry design and thus all of the questions and confusion I created above. If I would have just to that I would have been done by now.

Also the cap values in the feedback loop did not work as intended. There is a permanent roll of it seems and when turning the switch to increase the capacitor value lower the gain/creates a high pass filter, probably because of the interaction with the U-pad.


Thanks!

Paul

 

[Brian Roth]

OK @Potato Cakes Paul <g> , you made me find a pencil and then fire up Ye Olde Win XP computer that works with my Even Older HP scanner. then sneaker net to my main computer! LOL!!!!! Too lazy to spin the attachment 90 degrees <g>

First of all, is the "A" section.....Any needed loading and/or Zobel to keep the transfo happy.

"B" is a bit convoluted, but offers ways to "skin the attenuation cat". Keep in mind, this diagram uses an inverting opamp, which presents a virtual ground node.

Why do I suggest an inverter? The only way to make the roll-off work. A non-inverter can only drop down to unity gain. In this case, the HF roll-off goes on forever.

My goofy idea is try to break up the requirements into chunks that should "play nice" with each other.

Bri

 

[gyraf]

..have you considered using a zero-field transformer circuit? That's probably the most invisible way to transformer isolate, if you have opamp psu at hand anyways..

Lowpass would be placed in feedback loop, like in Bri's above (which is quite close to a zero-field input, give or take the R(?)'s in series with inverting input)