Forums :: Resources :: Features :: Photo Gallery :: Vintage Radio Shows :: Archives
Support This Site: Contributors :: Advertise


It is currently Jul Wed 17, 2019 4:06 am


All times are UTC [ DST ]





Post New Topic Post Reply  [ 40 posts ]  Go to page 1, 2  Next
Author Message
 Post subject: Low pass filter addition to amplifier
PostPosted: Jan Fri 25, 2019 12:06 am 
Member
User avatar

Joined: Jan Thu 01, 1970 1:00 am
Posts: 20040
Location: Warner Robins, GA
I'm working on the following amplifier which is the subwoofer portion of a 2.1 channel Advent computer speaker setup.

I got the amp several years ago and want to use it for driving a 15" Jensen speaker from an Emerson console.

Attachment:
Subwoofer amp.png
Subwoofer amp.png [ 65.1 KiB | Viewed 1065 times ]


The following parts were added
1. LM741 buffer
2. 20K pot
3. 4.7K resistor
4. 10uF cap
5. 1 meg pot

The .001uF cap was changed to .006uF as is what determines the low pass frequency where a larger cap value reduces the frequency, but it isn't good enough given the filter is much too broad.

Any idea what changes I can make so the low pass function is better?

I want a low pass frequency of somewhere between 100-200Hz.

Now I'm thinking perhaps there may have been a filter before this portion of the amp, but am not sure as I don't have an actual schematic and had to draw this one from looking at the board.

When I finished drawing the schematic I realized that without the .006uF cap I may well have a HI-FI 20 watt amp that is very simple to build.


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Fri 25, 2019 4:24 am 
Member
User avatar

Joined: Jan Fri 02, 2009 11:32 pm
Posts: 618
Location: Equinunk PA 18417
That 6nF cap is not in a good place to do what you want. If you're game to add another opamp, look up Sallen-Key active filters. A 2nd order lowpass would require one opamp, better would be to use a dual opamp (TL082 for instance) to build two of those in series to make a 4th order lowpass. Typical subwoofer crossover -3db point is 120hz, but that is up to you.

You can build the circuit below with R1=R2 and C1=C2. I did a quick calc using standard values R1=R2=12k, C1=C2=0.1uF, and come up with -3db at 130Hz. Other values can bring similar results, and much depends on the impedence of the source. It is assumed the source impedence (your preamp?) is something like 1/10th or less of R1. Higher source impedence will work but will change the filter's curve a bit. Assuming low impedence source, and a Butterworth curve is wanted (which is smooth and well accepted for audio) make Ra=3800 and Rb=5600. With these values, which make the curve Butterworth characteristic, there will also be 4db gain per opamp. If you cannot tolerate the extra gain, you can make each stage unity gain by making Ra a jumper, and delete Rb. That will affect the roll-off curve making it less steep. Might sound okay, you'll have plenty of listening to do.


Attachments:
2nd order filters.jpg
2nd order filters.jpg [ 32.09 KiB | Viewed 1046 times ]

_________________
-Richard
Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Fri 25, 2019 4:35 am 
Member
User avatar

Joined: Jan Fri 02, 2009 11:32 pm
Posts: 618
Location: Equinunk PA 18417
Sorry, I should have noticed before, put new filter between your added buffer and the 20k pot, that will eliminate any fuss over input impedence. Doubt a little extra gain will be any problem because of your first 1M pot. Good luck.

_________________
-Richard


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Fri 25, 2019 5:32 am 
Member
User avatar

Joined: Jan Thu 01, 1970 1:00 am
Posts: 20040
Location: Warner Robins, GA
I saw that circuit.

That said I do have the dual 741 equivalent I could use which will allow both the buffer and low pass with one chip.

I saw one circuit where C1 was double that of C2.

Any benefit to that?



The source is a radio where I am replacing a 100K plate resistor with a 91K and 10K resistor combination with the 10K resistor fed from the filter cap.

That way I have a low impedance output from the radio and the voltage level is already reduced to a level the amp can easily handle.

I chose the buffer as the amp section has a low input impedance which as I now know is defined by the 4.7K resistor so it is effectively 4.7K.

The 1 meg pot was chosen so as to present as little load as possible to the radio.

Now the .001uF cap should I keep it or remove it? That cap does roll off the highs, but it is at 6db/octave.

Extra gain is fine as I can adjust the 20K pot to compensate.

Another thing is I cannot for the life of me figure out what the 220K resistor is for.


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Fri 25, 2019 6:12 am 
Member

Joined: Nov Wed 17, 2010 11:14 pm
Posts: 167
Another thing is I cannot for the life of me figure out what the 220K resistor is for.



The 220 K resistor I suspect is to help equalize the input offset voltages. There is a specified input offset current in the 741. To force the input offset voltage to zero, I would like to have the inverting and non inverting DC impedance matched in a manner that forces that offset V to zero.

Alan


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Fri 25, 2019 6:37 am 
Member

Joined: Jan Thu 01, 1970 1:00 am
Posts: 2284
Location: Saskatoon
Also, that 0.006µF cap in parallel with the 470k resistor is a phase shift network, not a low pass filter. It's there to prevent the amplifier from going into oscillation by correcting the inherent phase shift of the amp. It does behave somewhat like a low pass filter, but that's not really it's intended purpose. It's there to stabilize the amplifier.


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Fri 25, 2019 12:33 pm 
Member
User avatar

Joined: Jan Thu 01, 1970 1:00 am
Posts: 20040
Location: Warner Robins, GA
Oh ok so I should leave the .001uF cap there then.

Now I doubt that I can easily find a 3.8K resistor in my stash. Will a 3.9K resistor work?

With the 3.8K resistor I get

Voltage gain ~1.678571
Amplification ~4.499 dB

With the 3.9K resistor I get

Voltage gain ~1.696429
Amplification ~4.591 dB

Here's the circuit

Attachment:
Subwoofer amp 3.png
Subwoofer amp 3.png [ 57.61 KiB | Viewed 1015 times ]


Here's the measurement

Attachment:
Measurement.png
Measurement.png [ 2.4 KiB | Viewed 1015 times ]


Now does that seem right?

Also if I were to build an identical filter would that be better?


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Fri 25, 2019 3:30 pm 
Member
User avatar

Joined: Jan Fri 02, 2009 11:32 pm
Posts: 618
Location: Equinunk PA 18417
edit: Somehow I missed the bottom part of your post with measurement. 200hz vs. 40 hz, your numbers show a 10db attenuation. It should be near 12db attentuaion at 240Hz, so I suppose your filter is operating right. Another opamp stage will make a steeper roll-off curve, if that is what you're after.

One 2nd-order filter stage is certainly better than what you have now. Two stages of Butterworth roll-off characteristic is 12db per octave just like a series pair of passive RC filters. 12 db attenuation at 240Hz leaves plenty that will be audible, and perhaps audible up to around 400Hz where hopefully the main speakers mask it entirely. Could be that amount of roll-off is a benefit or an annoyance, just depends on the other speakers, their placements, your tastes, etc. Try it.

3.9k is fine. It will change the curve a little bit away from Butterworth, making the rolloff slightly steeper and in theory allowing a tiny bit more of passband ripple in its curve, which on a response graph will look like a slight bump just before the -3db point, but I doubt the change will be audible so tary on! In fact I'll bet you already have by the time I have replied.

The combo of Ra and Rb determine the opamp gain which determines all filter characteristics except the -3db calculation. You asked about different values for C1 and C2. Each resistor-capacitor pair operate as a single RC filter, and RC values can be worked separately to tune a filter design for a specific outcome. Response ripples and group delays will change for different RC values (because different RC values may produce different timing characteristics), so they can be matched to either add or cancel. Filter design is an art in itself, one that I am not well versed in. You're only working with a 2 order filter, and since a Butterworth characteristic has no appreciable response ripple, I don't see much benefit to playing with different RC values. It is much easier to plug in equal values to get the filter in the ballpark. You can breadboard and mess around with different values after that if you are inclined.

About that .001/.006 cap, you can leave it as .006 if you want it won't hurt anything or switch it to the original value. It only affects uppwe audio frequencies. As Bob says, it exists to quench circuit oscillation. There are two reasons it is not very effective for low passing. First, as you've noticed, it is only 6db per octave but the bigger issue is it is in the feedback path and as such the circuit gain will not fall below unity so the roll-off will "flatten out" at higher frequencies as the stage gain becomes unity. That's worse performance than a passive RC filter.

_________________
-Richard


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Fri 25, 2019 3:39 pm 
Member
User avatar

Joined: Jan Fri 02, 2009 11:32 pm
Posts: 618
Location: Equinunk PA 18417
One point I forgot to make before, the amp appears to invert audio polarity from input to output. You might have to switch wires going to the woofer to match the other speakers.

_________________
-Richard


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Fri 25, 2019 3:53 pm 
Member
User avatar

Joined: Jan Thu 01, 1970 1:00 am
Posts: 20040
Location: Warner Robins, GA
I meant to clarify about the additional filter. I would build it just like the first and put it between the buffer and first filter.

The main issue is this.

The point in the radio where I'm taking the signal from is the first audio tube which uses a 100K plate resistor and feeds a driver tube grid through a coupling cap.

I plan on making the 100K resistor 91K and 10K with the 10K connected to the filter cap so that I have a low impedance output at a lower voltage level.

I plan on reducing the coupling cap in the radio some to raise its -3dB point some to hopefully match the bass amplifier's 3dB point.

Should I go with a third pole in the bass amp so the phase is the same between the bass amp output and the radio's output?

That said based on what I looked up on OP-AMPs the .001uF 470K combo is a 1 pole low pass filter only the transistors are in the feedback loop.

So perhaps I can try a larger cap across the .001uF cap.

EDIT:

Looking at the output I can see where the .001uF 470K combination is affecting the phase given if there were only a two pole filter I should see the input and output 180 degrees out of phase, but they're not exactly 180 out.

Now the .001uF cap and 470K resistor combination gives a frequency cutoff of 338.8Hz. I would need a ,22uF cap in parallel with it to get a cutoff of 102.67Hz.


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Fri 25, 2019 4:22 pm 
Member
User avatar

Joined: Jan Fri 02, 2009 11:32 pm
Posts: 618
Location: Equinunk PA 18417
You could try a larger capacitor, it may give some additional upper frequency roll-off but another Sallen-Key stage will make a much bigger change.

Sallen-Key filters do not invert audio polarity. Neither does your buffer stage since you are applying audio to its + input. The inversion happens at the opamp driver just ahead of the transistors. It's not a big deal since you can easily swap wires at the woofer. edit: this is a separate issue from group delay of the filter itself.

Cross-over from subwoofer to full range speakers is a whole other topic. First, electrically, there is group delay (a.k.a. phase shift) of frequencies near the -3db points of the filters. A 1st order filter and a 4th order filter will have very different amounts of phase shift, possible opposite directions, at the same frequency. What's worse, speaker placement dramatically changes the phase relationship between speakers simply due to speed of sound through air, enclosure acoustics, placement within a room. So, it comes down to trial and error in your listening space and swapping speaker polarity to see which way is better. That said, it is often better to use steep roll-off curves so that the overlap between sub to full range is more narrow, less range to be affected.

edit: It is not uncommon to find switchable phase shift networks on subwoofer speakers, to give more options for best match.

_________________
-Richard


Last edited by richfair on Jan Fri 25, 2019 4:29 pm, edited 1 time in total.

Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Fri 25, 2019 4:29 pm 
Member
User avatar

Joined: Jan Thu 01, 1970 1:00 am
Posts: 20040
Location: Warner Robins, GA
Went ahead and tried the .0033uF cap and here's what I get.

Attachment:
Measurement 2.png
Measurement 2.png [ 2.38 KiB | Viewed 1005 times ]


I'll try that and see what I get.

If I don't like the results I'll build a second filter.

That said if I do wind up building a second filter I'll remove the .0033uF cap so that it doesn't affect the response any.


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Fri 25, 2019 4:33 pm 
Member
User avatar

Joined: Jan Fri 02, 2009 11:32 pm
Posts: 618
Location: Equinunk PA 18417
That's 12db attentuation at 200Hz. Not too bad, in terms of attenuation. Not sure how the group delays will mess you up.

_________________
-Richard


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Fri 25, 2019 4:37 pm 
Member
User avatar

Joined: Jan Thu 01, 1970 1:00 am
Posts: 20040
Location: Warner Robins, GA
Will have to listen to it.

haven't decided if I want the Jensen 15" bass speaker front firing or down firing.

Here's the updated schematic where things are at now.

Attachment:
Subwoofer amp 5.png
Subwoofer amp 5.png [ 58.31 KiB | Viewed 990 times ]



Will 20 watts into a 15" Jensen speaker with alnico magnet be enough when the radio it is going to be used with puts out 10 watts through an efficient field coil speaker?

Now if I wanted a filter in the radio in addition to the filers in the amp I could use a 100K and a 10K in series and place a cap across the 10K which would bypass the frequencies above a certain point around the 10K resistor and would thusfore not be sent to the amp.

That would affect the operating point of the first audio tube though, but I think that at lower frequencies the impedance would be higher and it may actually create a bass boost in the radio's amplifier.


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Sat 26, 2019 11:05 pm 
Member
User avatar

Joined: Jan Thu 01, 1970 1:00 am
Posts: 20040
Location: Warner Robins, GA
I tried the amp last night with a HI-FI radio and am external speaker that doesn't have much bass response.

The amp worked well driving the Jensen.

For being in an open back cardboard box the Jensen sounds quite good far as bass response goes. I can only imagine it will be much better in a proper open back wood box.

Now if I take the original circuit and remove the .006uF cap and replace the 470K resistor with a 500K pot will I then have a variable gain control?

If so I could eliminate the 20K pot.

That would be done only if I decided to build a HI-FI version of the amp.

Thanks so much for your help.


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Sun 27, 2019 5:29 am 
Member
User avatar

Joined: Jan Fri 02, 2009 11:32 pm
Posts: 618
Location: Equinunk PA 18417
Quote:
Now if I take the original circuit and remove the .006uF cap and replace the 470K resistor with a 500K pot will I then have a variable gain control?
Keep that .001uF cap in place (or .0033 or whatever you've got now), and be sure to keep the 4.7k resistor currently on the 20k's wiper. I had wondered why you added the 20k pot since you've got a 1Meg pot on the input. The 20k pot has an unexpected downside being connected to the inverting input. In that position its operation presents a variable impedance to the inverting input which affects the loop gain of the opamp+transistors output. That stage's gain will be MORE when the 20k pot is midway, and LESS when the 20k pot is at either extreme. You probably never noticed, though.

A 500k pot will do a good job controlling volume, but as it changes stage gain some DC may appear on the output, which will go through the speaker. Probably millivolts, not volts, so probably not a huge problem. That 220k resistor is there to cancel out current imbalance (as pointed out in an earlier post), but I wouldn't be surprised to see the output's DC offset change as you spin the control. I will guess the output stage will otherwise remain stable through the pot's range, but that 's just a seat of pants guess. I'd do a thorough check before buttoning up. Also, as the pot in a feedback loop ages and gets "scratchy", the noise will be amplified compared to one used like that 1 meg pot.

_________________
-Richard


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Sun 27, 2019 8:42 am 
Member
User avatar

Joined: Jan Thu 01, 1970 1:00 am
Posts: 20040
Location: Warner Robins, GA
Didn't think the 20K pot would affect gain since the inverting input is a virtual ground.

The idea for the 20K pot is because the gain of the amp without the buffer is on the order of less than 100mV produces full amplifier output. So the pot is a way for the overall amplifiers gain to be reduced. I chose a variable instead of fixed resistor as I don't know what the radio's output will be and I don't want too much or not enough gain.

The 4.7K resistor affects gain somewhat which means the impedance to the inverting input is low.

Didn't realize that changing the 470K resistor would affect the dc offset.

Figured given the + and - supplies the dc offset would stay at 0 or real close to it.

Now if the .001uF cap was removed would it make the amp have a full frequency response?

If it will without affecting stability I may build another and have a HI-FI amplifier.

Might just build another anyways and experiment with it. See just what I can do with the circuit.

EDIT: so if I take the power supplies which are approximately 20 volts positive and negative to the op-amps I can handle about 14Vrms on the input before distortion. Factoring in the voltage gain of the filter I can get an input of around 8.33Vrms before distortion. Those are rough calculated numbers though.

So if for instance the radio outputs a max of 10Vrms across the 90K 10K series resistor combination I just need to set the 20K pot so that the input sensitivity to the amp is 1Vrms.

At the B+ point where the driver tube transformer connects the audio voltage on the B+ line increases as the audio frequency goes lower due to the low value of filter cap used.

So perhaps I can make a hack using that to feed the amp. Just thinking though.

I did notice however that as the heatsink heats up the maximum output voltage possible from the amp goes up from about 9.2Vrms to 9.6Vrms. So perhaps the circuit needed some sort of compensation which was deemed not necessary given it is a subwoofer amp.


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Sun 27, 2019 5:00 pm 
Member
User avatar

Joined: Jan Fri 02, 2009 11:32 pm
Posts: 618
Location: Equinunk PA 18417
Your post got longer while I was replying, I suppose you edited it. I may not be addressing anything in your bottom half, but hope this is helpful.

"Virtual ground" does not equal "ground". It is the opamp's output that is "virtual ground" since it is a very low impedence point even if it is not at 0 volts. At the opamp's inverting input there is some current flow although it may not be obvious at first. Current flows through the feedback resistor into the inverting input, which the opamp reacts to and immediately alters its output to reduce that current flow, and that's what allows the opamp to function as an opamp. Anything else connected to the inverting input may also draw some current, away from the opamp, which changes the current through the opamp input and thus the stage gain will change. The original circuit's gain would be 1+(470k/4.7k)=101. NOTE: the following image is generic and does not represent your amp exactly but the gain calculations are the same.
Attachment:
opamp gain.jpg
opamp gain.jpg [ 24.37 KiB | Viewed 942 times ]


Things are not as simple as you want them to be. An opamp stage's AC performance is often different than its DC performance. Capacitors block DC current and form RC time constants, which require us to calculate DC gains separately from AC gain. Audio is AC.I've redrawn the generic circuit below. Pretend that audio is being amplified and is already present at the opamp stage's output. In terms of AC, your 20k pot is connected to signal ground on one end and to a virtual ground on the other. That virtual ground happens to be another opamp's output which is a very capable current sink similar to signal ground, and our AC signal sees each in the same way. The wiper floats between ground and virtual ground. When at either extreme the pot's resistance to ground or virtual ground is effectively zero as far as AC is concerned, and some current is split away from the inverting input and sunk to ground or virtual ground through the 4.7k resistor. (edit: the 10k resistance of the pot does nothing since zero ohms in parallel with 10k ohms is still zero.) Voltage gain is 1+(470k/4.7k)=101, just like the original circuit. What happens with the pot is set midway? AC signals now see the pot as two 10k resistors in parallel to ground and virtual ground, or 5k total resistance. This is in series with the fixed 4.7k resistor, so now opamp gain is 1+(470k/(4.7k+5k))=21.9. Your opamp gain dropped from about 40db to about 27db as you twist the pot from end to midpoint. See how that works?
Attachment:
opamp gain pot midway.jpg
opamp gain pot midway.jpg [ 29.18 KiB | Viewed 942 times ]

In your actual circuit, AC signal is applied to the inverting input, which means it goes through resistors, voltage dividers, and its current is split away at the inverting input through that feedback resistor toward the opamp's output (another virtual ground). With the voltage divisions reducing signal voltage available to the inverting input, overall stage gain APPEARS LOWER than the opamp's actual gain. You can figure that with ohms law if you want. No matter whether the signal is applied to the inverting input or non-inverting input, the gain of the opamp itself is calculated with the same formula.

You'll want a cap of some value where the .001 cap has been, to ensure the circuit will not break into high frequency oscillation at high gain settings. Common to find values in the 20-50pf range around single opamps. This stage includes other components so you may need to experiment for optimum value. With no cap, assuming circuit is not already oscillating, apply square waves through the amp and 'scope out what they look like, you can probably find some ringing that can be damped with a cap small enough to not affect audible frequency response. EDIT: Different loads on the output stage, particularly reactive, will affect circuit stability so check all possible loads you may apply. Some people like to connect capacitors across the output to simulate worst case. There is usually some capacitor value that will cause the output to become unhinged, and possibly damage output devices, so be aware!

_________________
-Richard


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Mon 28, 2019 12:34 am 
Member
User avatar

Joined: Jan Thu 01, 1970 1:00 am
Posts: 20040
Location: Warner Robins, GA
Oh ok now I understand. Thanks.

Now if I do away with the 20K pot and adjust the 470K resistor will that provide a variable gain or should I just duplicate the original circuit first posted and just use a small enough cap value to prevent oscillations.

Had I known then what I do now i would have looked at the original resistor values used to sum the stereo signal to mono then halved that and used that value where the 4.7K resistor is.

The virtual ground does explain why the bass control pot was just in series with the summing resistors with no ground connection.

Now the amp does have a cap to ground on the OP-AMP output. Am assuming that serves to provide some sort of stability to the amp without affecting frequency response.

Think I'll order the parts to build another amp so I can experiment with it.

Given the amp only uses a metal plate as a heatsink building the amp should be relatively easy.


Top
 Profile  
 
 Post subject: Re: Low pass filter addition to amplifier
PostPosted: Jan Mon 28, 2019 3:40 am 
Member
User avatar

Joined: Jan Fri 02, 2009 11:32 pm
Posts: 618
Location: Equinunk PA 18417
Quote:
Now if I do away with the 20K pot and adjust the 470K resistor will that provide a variable gain or should I just duplicate the original circuit first posted and just use a small enough cap value to prevent oscillations.
Are you planning to keep the 1M pot before the buffer? If so, I would do away with the 20k pot. Since you've got a lot more gain than you want, experiment with a smaller value fixed feedback resistor instead of that 470k, to reduce gain. Maybe 100k as a trial? Keep an eye out for the troubles i've already mentioned. If no instability and DC offset of the output is acceptably low, then you're done. You already know what to do for oscillating instability, and lower gain will mean less chance of trouble. If DC offset needs attention, try adjusting the 220k resistor on the opamp's non-inv input.

_________________
-Richard


Top
 Profile  
 
Post New Topic Post Reply  [ 40 posts ]  Go to page 1, 2  Next

All times are UTC [ DST ]


Who is online

Users browsing this forum: Jthorusen and 3 guests



Search for:
Jump to:  




























Privacy Policy :: Powered by phpBB