Joining a U-Pad and a LowPass filter in a Balanced Audio Signal

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

Colortone

Member
Joined
Mar 15, 2022
Messages
22
Location
Sydney
Hi,
Would there be a problem with joining an in series U Pad and LowPass Filter in the Design Below?
The Outputs are coming from the classic non balanced to Balanced Opamp Circuit ( I have adjusted to a gain of 1 as I want it to feed a microphone level preamp).

Screen Shot 2022-05-31 at 11.13.50 am.png
It is for Guitar (instrument level) In and Mic level Out. As I still need to attenuate the output Signal to Mic Level I have added a @ 25 DB pad working to an assumption of feeding a mic Preamp with an impedance load of 1500 ohms.

Below is my literal attempt to place the LowPass after the PAD. It appears to work but im wondering if im approaching it the wrong way.
Screen Shot 2022-05-31 at 11.19.49 am.png
Many thanks,
AD
 

Attachments

  • Screen Shot 2022-05-31 at 11.03.13 am.png
    Screen Shot 2022-05-31 at 11.03.13 am.png
    392.8 KB · Views: 40
That could work. But I haven't checked the actual values and 560R in series is a little high for a mic input. And the 560R is not good for driving a long cable. But it should work. Breadboard it and check the frequency response before committing.

An alternative would be to add a series 1K / 100n to the input after the 1M.
 
That could work. But I haven't checked the actual values and 560R in series is a little high for a mic input. And the 560R is not good for driving a long cable. But it should work. Breadboard it and check the frequency response before committing.

An alternative would be to add a series 1K / 100n to the input after the 1M.
Thanks Bo Deadly,
Ill give that a go in terms of adding the Low pass on the input. I believe one of my problems is that I read adding a pad on the output reduces the signal but not necessarily the SN ratio. So I was attempting to see if I could lower some of the slight 'hiss' Im hearing on the output.

I only have an oscilloscope and mic level is really hard to read on it given its soo weak. Do you know of any good mac or PC free tools to check freq response, Ive been doing this by ear so far :/

I hadnt thought of cables, damn. What do I have to consider in that regard?
AD
 
Another alternative for a 5kHz LPF would be a 3n3 cap in parallel to the 10k R4.
Thanks For the reply Harpo, I read that a cap in the feedback loop of an Opamp has little effect when the gain is around 1. The above schematic does have a gain of 10 but ive reduced that gain to 1. I did try it on my board and could not hear the difference in this configuration. Maybe im wrong?
AD
 
Your U pad represents too much load for the TL072 opamps, so they will work with greater distortion, especially when larger voltage transients occur. I would recommend replacing the TL072 with a NE5532 or increasing resistors in the U pad and LPF.
 
HI Abbey Road D Enfer. The highest caps I have are 220nF. Could I use two of them in parallel to get close to the 500nF, or three caps 220n, 220n and a 47n?
Of course! Any combination of caps that gives the approximate value is suitable.
The 500nF figure is just to make the response similar to your original, but I think you'll need to experiment by trial and error.
BTW, If it's for taming the highs of an electric guitar, it won't be enough IMO.
After studying several electric guitar amp simulators, I've found that the most important single parameter is the HF cut-off, and a minimum 4-th order low-pass is desired.
This is because most of the electric guitar sounds we're used to listen to come through 10, 12 or 15 inch loudspeakers that rolls off sharply at about 3 to 5kHz.
Because we're creatures of habit, this is the reference sound we try to emulate. That's the nature of acquired taste. Lo-Fi has become a reference.
 
Thanks Again Abbey Road D Enfer.
For learning sake (instead of just getting an answer and running with it) would you be able to explain how you came to those values?
I had been using an online calculator For the UPad and LowPass Filtering that assumed the output impedance of the TL072 to be around 200 ohms and the load impedance (Preamp) to be around 1500 Ohms. I had not thought to account for cable length ...

Calculator Here Uneeda Audio - Standard Value Selector Program

The Low pass filter was just to tame some soft background noise I was hearing in the output of the circuit.

Thanks Again
AD


Of course! Any combination of caps that gives the approximate value is suitable.
The 500nF figure is just to make the response similar to your original, but I think you'll need to experiment by trial and error.
BTW, If it's for taming the highs of an electric guitar, it won't be enough IMO.
After studying several electric guitar amp simulators, I've found that the most important single parameter is the HF cut-off, and a minimum 4-th order low-pass is desired.
This is because most of the electric guitar sounds we're used to listen to come through 10, 12 or 15 inch loudspeakers that rolls off sharply at about 3 to 5kHz.
Because we're creatures of habit, this is the reference sound we try to emulate. That's the nature of acquired taste. Lo-Fi has become a reference.
 
For learning sake (instead of just getting an answer and running with it) would you be able to explain how you came to those values?
I had been using an online calculator For the UPad and LowPass Filtering that assumed the output impedance of the TL072 to be around 200 ohms and the load impedance (Preamp) to be around 1500 Ohms. I had not thought to account for cable length ...
The screenshot is of LTSpice so presumably it was simulated which is the best way to figure out the values and frequency because the impedance is not trival when combined with the filter. With a balanced attenuator the resistors look like half the impedance at AC because the voltage on the other side is changing in an equal and opposite direction. Meaning 1V across a 1K resistor is 1mA but if the other side of the resistor is -1V and not 0V, the current is 2mA which makes the resistor look like it's 500 ohms and not 1K. Factor in the low-pass cap and it deserves simulation to be sure.

The output impedance of a TL072 might be 200 ohms (not sure about that) but with negative feedback it's AC output impedance is near zero like just about all op amps when negative feedback is used. What makes op amps different however is how much current they can deliver. As the output voltage swing increases, there is a point where the op amp fails to deliver current sufficient to maintain the correct voltage which equates to distortion. The TL0X series does have fairly limited current capability so you cannot load it with (560 + 39 + 560)/2 = 580 ohms. You'll get distortion at even medium signal levels.

The 2K + 130 + 2K will look like a load of ~2K which is easy going for a TL072.

Note that the cable length is not a factor for the pad / filter circuit posted by Abbey because the output of the pad is simply defined by the 130 ohms resistor. So the output impedance is actually 130 ohms which is fine for driving a moderately long cable.
 
For learning sake (instead of just getting an answer and running with it) would you be able to explain how you came to those values?
That's the most difficult part for me, explaining what was almost instinctive. :)
I started with the constraint of load applied to the TL0, 2k. Then I determined the value of the shunt resistor for 30dB attenuation. I didn't use a calc or anything. I know that 10dB is 3.16 (square root of 10) and that 20dB is factor 10. So 30dB is 31.6. So the shunt resistor for a single arm is 2000/31.6, which turns out to be 2 times 31.6 (by virtue of the powers of 10), or 63.2. So for the two arms gives 126.4, for which I chose the next standardized E24 value of 130 ohms.
The opamps see 2065 ohms and the mic input sees a little less than 130 ohms, which is perfectly adequate since it expects between 150 and 200 ohms.
Now for the LPF, I used a bit of mental calculation for estimating a rough figure.
I start with 1nF@1kHz => roughly 150 kiloohms (actually 159). So 130 ohms is about 1/100th of 150k (I told you it's an approximation!) and 5kHz is, guess,...5 times 1k.
So the resulting guesstimate is 1nF divided by 5 (cause frequency) and multiplied by 100, for about 250nF.

And whizz bam, that's how I figure I posted the picture with the wrong value! Ah Ah :eek:
Doh!
500nF puts the -3dB point at 2.5kHz, which may or may not be a problem.
 
That's the most difficult part for me, explaining what was almost instinctive. :)
I started with the constraint of load applied to the TL0, 2k. Then I determined the value of the shunt resistor for 30dB attenuation. I didn't use a calc or anything. I know that 10dB is 3.16 (square root of 10) and that 20dB is factor 10. So 30dB is 31.6. So the shunt resistor for a single arm is 2000/31.6, which turns out to be 2 times 31.6 (by virtue of the powers of 10), or 63.2. So for the two arms gives 126.4, for which I chose the next standardized E24 value of 130 ohms.
The opamps see 2065 ohms and the mic input sees a little less than 130 ohms, which is perfectly adequate since it expects between 150 and 200 ohms.
Now for the LPF, I used a bit of mental calculation for estimating a rough figure.
I start with 1nF@1kHz => roughly 150 kiloohms (actually 159). So 130 ohms is about 1/100th of 150k (I told you it's an approximation!) and 5kHz is, guess,...5 times 1k.
So the resulting guesstimate is 1nF divided by 5 (cause frequency) and multiplied by 100, for about 250nF.

And whizz bam, that's how I figure I posted the picture with the wrong value! Ah Ah :eek:
Doh!
500nF puts the -3dB point at 2.5kHz, which may or may not be a problem.
Nice rules of thumb. I am afraid I am much more reliant on my HP calculator with RPN...
 
Nice rules of thumb. I am afraid I am much more reliant on my HP calculator with RPN...
My physics teachers insisted (rightly IMO) that we master mental calculation, particularly for avoiding gross mistakes (powers of 10).
In my whole professional life, I never needed more than a basic pocket calculator that does logs and square roots.
Today I'm perfectly content with the calc in Window accessories.
I'm not criticizing anyone, it's just an observation.
 
My physics teachers insisted (rightly IMO) that we master mental calculation, particularly for avoiding gross mistakes (powers of 10).
In my whole professional life, I never needed more than a basic pocket calculator that does logs and square roots.
Today I'm perfectly content with the calc in Window accessories.
I'm not criticizing anyone, it's just an observation.
Back in the day using slide rules to calculate filter values we had to do the ballpark calculations in our head, and just used the slide rule to provide the significant digits.

JR
 
My physics teachers insisted (rightly IMO) that we master mental calculation, particularly for avoiding gross mistakes (powers of 10).
In my whole professional life, I never needed more than a basic pocket calculator that does logs and square roots.
Today I'm perfectly content with the calc in Window accessories.
I'm not criticizing anyone, it's just an observation.
Yes, that is a good practice. In my case, more than being great to do mental calculations, I am able to remember tons of constants and formulae, I think that some things must always be remembered. That is what I tell my gen Z engineering students who don't want to commit anything to long term memory and rely heavily on their phones. I tell them that it just looks bad when you don't know some stuff, and I tell them this story:

Suppose you go to the doctor's office, you explain to him what is wrong with you, and afterwards, you see the doctor reach for his phone in front of you and say "Hey Siri, symptoms are: headache, loss of apetite, fever, etc....". I then ask my students: What would you think?

I know it is an extreme exaggeration, but serves to illustrate the lesson I am trying to teach them.
 
Last edited:
My brain just slightly exploded with that simple calculation :/ :)
I was following up until the LPFilter bit.
You mentioned the original picture you posted was incorrect, are you saying the difference is just the Cap value which should now be 250nF for a cut over 5k?
Thanks, and appreciate your knowledge and time.
AD




That's the most difficult part for me, explaining what was almost instinctive. :)
I started with the constraint of load applied to the TL0, 2k. Then I determined the value of the shunt resistor for 30dB attenuation. I didn't use a calc or anything. I know that 10dB is 3.16 (square root of 10) and that 20dB is factor 10. So 30dB is 31.6. So the shunt resistor for a single arm is 2000/31.6, which turns out to be 2 times 31.6 (by virtue of the powers of 10), or 63.2. So for the two arms gives 126.4, for which I chose the next standardized E24 value of 130 ohms.
The opamps see 2065 ohms and the mic input sees a little less than 130 ohms, which is perfectly adequate since it expects between 150 and 200 ohms.
Now for the LPF, I used a bit of mental calculation for estimating a rough figure.
I start with 1nF@1kHz => roughly 150 kiloohms (actually 159). So 130 ohms is about 1/100th of 150k (I told you it's an approximation!) and 5kHz is, guess,...5 times 1k.
So the resulting guesstimate is 1nF divided by 5 (cause frequency) and multiplied by 100, for about 250nF.

And whizz bam, that's how I figure I posted the picture with the wrong value! Ah Ah :eek:
Doh!
500nF puts the -3dB point at 2.5kHz, which may or may not be a problem.
 

Latest posts

Back
Top