Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Two audio filters. Please help me modernize them

Status
Not open for further replies.

neazoi

Advanced Member level 6
Joined
Jan 5, 2008
Messages
4,119
Helped
13
Reputation
26
Reaction score
15
Trophy points
1,318
Location
Greece
Activity points
36,918
Here are two audio filters for receivers.
I would like to test them but I want to make them using NPN transistors instead. Is it enough to connect replace the PNP transistors with NPN and put a positive VCC where a negative exists now?

Also any guess of their operation principles?
 

Attachments

  • fl1.PNG
    fl1.PNG
    142 KB · Views: 144
  • fl2.PNG
    fl2.PNG
    245.8 KB · Views: 142

Is it enough to connect replace the PNP transistors with NPN and put a positive VCC where a negative exists now?

Yes, that is the general rule. Also reverse diode directions, but there are none in your schematics.]

- - - Updated - - -

Also check capacitor orientations.
 
  • Like
Reactions: neazoi

    neazoi

    Points: 2
    Helpful Answer Positive Rating
The first schematic amplifies two bands: 70-500 and everything above 1500 Hz.

It has the series LC which diverts a band to ground. So there is reduced gain at 500-1500 Hz.

This may obey an equalization curve, in order to restore audio fidelity.
 
  • Like
Reactions: neazoi

    neazoi

    Points: 2
    Helpful Answer Positive Rating
They are not going to give the sharp peak you were looking for in your other thread.

I thought about your requirements for a peaking circuit that avoids using feedback to increase 'Q' and an idea came to mind. I have not tried this and I can foresee some impedance matching problems but bear with me and maybe try it for yourself:

My preferred selective network is a twin-tee. It uses easily available components, is simple to construct, has in theory an infinite 'Q' (although not in practice!) and most importantly, it conducts DC between it's input and output. The problem is they work exactly in reverse to what you want, they have high impedance between in and out at the resonant frequency, making them a notch rather than peak filter.

The obvioius way to invert the notch is to place it in the negative feedback path of an amplifier instead of in the signal path. If the negative feedback is notched, the gain is highest at that frequency so a selective peaking effect is achieved.

My next thought was you don't want to use ICs and you do want to keep it simple, so what is the least complex amplifier that you can add a notch filter to. I think the simplest is a single transistor common emitter stage with the 'notch' between the base and emitter pins. The bonus that the filter conducts DC can be used to simplify the bias circuit.

What I would propose is you try this out: use a single transistor amplifier stage, input fed to the base and output from the collector, you can use NPN or PNP as long as you provide the appropriate supply polarity. It needs a reasonably high emitter resistor to work, I would suggest 4.7K as a good starting value and it must not be bypassed with a capacitor as you need signal on the emitter for this to work. Then wire a twin-tee network between the base and emitter using two 10K series resistors with 44nF to ground and two 22nF capacitors with 5K to ground. Those values will give a peak at about 750Hz which is nice for CW. Note the values need to be exact but you can make 5K from two 10Ks in parallel and 44nF from two 22nF in parallel.

See what happens and let me know. If it shows promise, there is a simple modification that allows true bandwidth adjustment.

Brian.
 

    V

    Points: 2
    Helpful Answer Positive Rating
I remember that I built variable band-pass/band-stop audio filters as a student, to be used for audio effects and as a measurement tool. They always used operational amplifiers in a "state-variable" respectively "universal active filter" configuration. https://en.wikipedia.org/wiki/State_variable_filter
 
  • Like
Reactions: neazoi

    neazoi

    Points: 2
    Helpful Answer Positive Rating
Agree with Brian, the twin tee in the feedback will give you a huge gain spike at one particular frequency.
Its extremely effective.
 
  • Like
Reactions: neazoi

    neazoi

    Points: 2
    Helpful Answer Positive Rating
What I would propose is you try this out: use a single transistor amplifier stage, input fed to the base and output from the collector, you can use NPN or PNP as long as you provide the appropriate supply polarity. It needs a reasonably high emitter resistor to work, I would suggest 4.7K as a good starting value and it must not be bypassed with a capacitor as you need signal on the emitter for this to work. Then wire a twin-tee network between the base and emitter using two 10K series resistors with 44nF to ground and two 22nF capacitors with 5K to ground. Those values will give a peak at about 750Hz which is nice for CW. Note the values need to be exact but you can make 5K from two 10Ks in parallel and 44nF from two 22nF in parallel.

See what happens and let me know. If it shows promise, there is a simple modification that allows true bandwidth adjustment.

Brian.
Something like this?
An estimation for the bias resistors R1 and R2? Maybe R2 can be omitted.
 

Attachments

  • fl.PNG
    fl.PNG
    8.6 KB · Views: 134

hmm.... I would add a collector load resistor or the output will be Vvveeerrrryyyy quiet!
Yes, miss R2 out and calculate R1 according to the transistor Hfe.

The problem with such a simple implementation is that for best results, the input impedance should be very low and the output load very high impedance, both of which this isn't able to satisfy. Try it though and see how it performs. Remember the R/2R and C/2C rule and make the values by parallel resistors/capacitors rather than using nearest values.

Brian.
 

hmm.... I would add a collector load resistor or the output will be Vvveeerrrryyyy quiet!
Yes, miss R2 out and calculate R1 according to the transistor Hfe.

The problem with such a simple implementation is that for best results, the input impedance should be very low and the output load very high impedance, both of which this isn't able to satisfy. Try it though and see how it performs. Remember the R/2R and C/2C rule and make the values by parallel resistors/capacitors rather than using nearest values.

Brian.

I made it. I put a 4.7k on the collector to vcc and a 470k on the collector to base. I removed R2 (base to gnd) totally. I have used ceramic 22nf capacitors and two of these parallel for the 44nf.

The circuit had no audible effect on the sound, maybe it narrowed the bandwidth a bit, because I could not hear very high pitch tones and hiss, but not the desired effect, not at all.

Any suggestions are appreciated.
 

I feel there is just insufficient voltage gain in your single transistor amplifier to get the gain spike you are hoping for.

It would work much better with an op amp if you have one to try.

What might work would be a Wein bridge oscillator circuit, held just below the threshold of oscillation.
 
Last edited:

That is what I feared. In fact the feedback path from the emitter has no voltage gain at all. When I've used it in the past it has been in the feedback of op-amp circuits where the gain is high and the impedances better controlled. It might be possible to add gain by using a second transistor stage but as the function of the twin-tee is a notch you have to be careful to not to invert the signal while amplifying the voltage, that may not be so easy.

Just as an experiment while you have it built, try injecting the signal at the emitter instead of the base. I have no idea if it will make any difference but it makes it a common base amplifier which might exhibit slight gain.

Brian.
 

You mean like my second circuit in post #1?
Similar idea, but I would do it differently.
Wein filter.jpeg
Tweak the pot until it oscillates at 1Khz, then back it off just below the threshold of oscillation.
It should have plenty of gain there at 1Khz.
 

Just as an experiment while you have it built, try injecting the signal at the emitter instead of the base. I have no idea if it will make any difference but it makes it a common base amplifier which might exhibit slight gain.

Brian.

No, it did not make an audible difference.

- - - Updated - - -

Similar idea, but I would do it differently.
View attachment 135135
Tweak the pot until it oscillates at 1Khz, then back it off just below the threshold of oscillation.
It should have plenty of gain there at 1Khz.

I have done that with the second circuit in post #1 in this page https://www.edaboard.com/threads/362949/#post1554876 It worked to some extent but read that thread for the problems I encountered.

I am curious about the Wein bridge method mentioned in post #10 here. Is that the second circuit in post #1?
I have not tried it yet.

- - - Updated - - -

I have also found this circuit. but it is much more complex than an opamp. It seems to be a combination of twin-t and phasing filter. promising?
 

Attachments

  • fl3.jpg
    fl3.jpg
    73.7 KB · Views: 126
  • flt3desc.jpg
    flt3desc.jpg
    102.8 KB · Views: 122
  • fl3curve.PNG
    fl3curve.PNG
    410.6 KB · Views: 131
Last edited:

To be really effective, I think it would need it to be within a hair of oscillating.
That is about the best you can probably do with only a single stage.
 

To be really effective, I think it would need it to be within a hair of oscillating.
That is about the best you can probably do with only a single stage.

Thanks Tony, we cross posted. See the circuit and the description in the previous post. It seems he uses Q3 to cancel out, out-of-phase signals. Not a one transistor circuit, but it uses the twin-T not in the negative feedback way (as the circuit Brian suggested), but at a notch filter. When the filtered output is applied to Q3 along with a out of phase full-signal, all the frequencies are phased out, except from the ones that existed in the notch. They do not affect the imput signal.
I do not know how much better will it work than the single transistor regenerative filter described in this thread https://www.edaboard.com/threads/362949/#post1554876 but at least it won't oscillate.
I think one might be able to do it without Q1, by just feeding the input signal at the base of Q3, if the input signal is high enough?

- - - Updated - - -

Now that I am thinking of it, this filter should be called regenerative and phasing filters combined into one. Regenerative, because there is positive feedback and phasing because it works by phase cancellation. Please correct me if my thought is wrong.

However, the regeneration here, is not used to narrow the bandwidth and increase the gain of a phasing oscillator, but to narrow the bandwidth of the notch in the notch filter.

In that sense I believe that the notch amplifier will not easily fall into oscillation, as it amplifies all frequencies except from the notch, in comparison to the phasing oscillator method, that amplifies just one frequency (hence easily oscillate into that frequency).

Please correct me if I am wrong in this aspect.
 
Last edited:

Yes you are quite right. The twin tee notched feedback just allows the amplifier to operate at its natural maximum gain.
For a single transistor amplifier that may not be a lot. An op amp offers far more potential.

The Wein bridge has positive feedback, and potentially a much taller and narrower peak. But its tricky and potentially unstable.

Only other consideration might be that the Wein is easier to tune requiring only two equal value resistors, which is dead easy. If that is a required feature.
The bridged tee, requires three tunable resistors, which becomes a bit more difficult.

I can see advantages and disadvantages either way.
 

The Wein bridge has positive feedback, and potentially a much taller and narrower peak. But its tricky and potentially unstable.

Only other consideration might be that the Wein is easier to tune requiring only two equal value resistors, which is dead easy. If that is a required feature.
The bridged tee, requires three tunable resistors, which becomes a bit more difficult.

Please tell me which circuits you refer to? The second circuit in my post #1 is the Wein bridge you refer to?

- - - Updated - - -

That is what I feared. In fact the feedback path from the emitter has no voltage gain at all. When I've used it in the past it has been in the feedback of op-amp circuits where the gain is high and the impedances better controlled. It might be possible to add gain by using a second transistor stage but as the function of the twin-tee is a notch you have to be careful to not to invert the signal while amplifying the voltage, that may not be so easy.

Just as an experiment while you have it built, try injecting the signal at the emitter instead of the base. I have no idea if it will make any difference but it makes it a common base amplifier which might exhibit slight gain.

Brian.

Maybe I judged your idea earlier than needed.
Look what I have found about twin-t networks
"The use of poorly matched capacitors in the twin-T net work will result in a low-Q filter. Capacitors should be 1% silver mica for CI, C2, C3, or they should be matched on a bridge. Do not attempt to substitute ceramic capacitors."

I used ceramic capacitors unmatched. Maybe the Q was too low because of this. If I could use lower value of capacitors (up to 470pF-2.2nf max) I could substitute them with silver mica 1% types.
Then maybe this single transistor circuit could work.
But I find the idea of post #14 interesting.
 

I suspect ANY regenerative or 'Q multiplier' filter will suffer the same problems of instability as the audio levels change and a degree of hysteresis in the regeneration control. Neazoi has already noted the effect of the filter starting to oscillate and how much louder it is than the signal being filtered.

The schematic in post #14 is what I was describing in post #11, adding a second stage to increase the feedback in a standard twin-tee arrangement. I am unhappy about the way the tuning works in that schematic though as R1 & R2 must always be the same value and R3 must always be half that value or the Q rapidly drops to almost nothing. As noted with capacitor values, the resistors are just as critical.

If a second stage has to be added, I see no problem with using #14 schematic but using the fixed component values already tried but omitting Q6 and all the associated metering circuits (who wants a meter to tell how loud it is anyway?) and omitting everything after the volume control. I would leave the 47K resistor in series with the volume control there though so reduces loading effects of subsequent stages. The purpose of the Q1 stage is to present a constant impedance at the summing point (base of Q3) but depending on what you feed it from, you may be able to leave that out too.

Brian.
 

The purpose of the Q1 stage is to present a constant impedance at the summing point (base of Q3) but depending on what you feed it from, you may be able to leave that out too.

Brian.

The filter will be placed at the output of the second transistor (regenerative detector) in this circuit **broken link removed** after the 10k resistor (used for isolation).
Do you believe Q1 stage in the filter, would be needed?

I have also found this one https://www.qsl.net/va3iul/Homebrew_RF_Circuit_Design_Ideas/Low_Distortion_CW_Filter.gif This is an Elektor project, straight forward. 3 transistors, but no tuning or weird filter effects. I thought that the 100mH chokes (the complete article, which I have on my phone SD and can't upload now, states that any 100mH choke can be used) were expensive, but on ebay, 5 of them cost about 3 euros, so it shouldn't be a great deal.
The filter has something like 400Hz bandwidth, at -6db but the responce at lower dB is better than the regenerative filters, which seem to have a low bandwidth nose but "open" quite rapidly.
If these components at the twin-t must be kept so precise (hard), an lc filter could be an easier to implement option. hm... difficult decision.
 
Last edited:

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top