The AM Forum

Hola,

I wonder if anyone with access to active filter design software might be interested in designing up a brick wall 5KC low pass audio filter for AM use using op amps?    

The problem is I've found with my EQ rolled off after 5kc, the compressor after it still levels off the higher frequencies above 5KC. It just boosts and levels everything off.  I need a brick wall 5kc filter after the 6-band compressor. I was hoping to use say, a quad OP470G low noise quad op amp and use all four op amps in cascade to produce a very sharp roll off. I have a few 14 pin package 470 chips on hand here.

In the link below see the last circuit that shows three op amps in cascade for a 1KC low pass. Can someone model this for 5kc or design an even better quad circuit for 5kc low pass?  I will build it ASAP if you do and post the results. I'll bet others might be interested in building one too.


Check out the last circuit on the page:
http://www.discovercircuits.com/PDF-FILES/LowPassFilt1.pdf

Here's the datasheet for the OP470G low noise quad op amp:
http://chipdig.com/datasheets/parts/datasheet/041/OP470G-pdf.php


Thanks for any help!

Tom, K1JJ

I would be interested as well, although I have a dual 31 band eq, one is before compression, the other after.
It only cuts 12 DB though....

Brett


 

That's a neat idea.  I have a Nady 31 band but it is mono.  Something isn't right however, I think it's picking up RF.  Even when i have everything flat at 0db, I get a muddy sound with it in line.  I think it's time to put up on eBay and try a Dual band and configure it like you Brett.  My Behringer VX-2000 works very well and doesn't pick up any rf.  Maybe a matching dual band Behringer 31-band EQ would be nice.  I'm also considering the Virtualizer Pro for just a tinge of reverb. 

I too would be interested in seeing a good active brick wall filter even better if it could be adjustable somehow for say 4 to 6 khz. :)

FabMaster:

For rolling your own op-amp filters, look at TI FilterPro design software

http://focus.ti.com/docs/toolsw/folders/print/filterpro.html

or LT FilterCAD
http://www.linear.com/designtools/software/filtercad.jsp

Steep elliptical filters can be made with ladder filters with coils and caps. They are today emulated with series-ing up a few FDNR (frequency dependent negative resistance) notch filters. However, I feel there is an easier and better solution. With the suggestion of Stu, AB2EZ who used a different chip in Maxim's switched capacitor filter chip line for receive filtering (ala the commercial SCAF unit), I built a handy brick wall filter circuit in my class E PDM generator.

I used the MAX294 chip, which is an 8th order elliptic low pass filter chip that provides a stopband rejection of over 60db after cutoff and a rolloff of something like 36db/octave. Using a single oscillator frequency set capacitor, I have currently set my cutoff at 6kc, but wanna try 5kc sometime. After this filter should be a phase correcting network to un-tilt any now tilted square waves, but I just used a negative clipper the shave off the occasional peaks.

There is a little noise and distortion that limits the chip's dynamic range to about 50-60db, but as my filter is after my audio processing, this is not an issue at all. No one can tell I am using this chip. I am clean and not too wide.

The datasheet is here:
http://datasheets.maxim-ic.com/en/ds/MAX293-MAX297.pdf
http://www.maxim-ic.com/quick_view2.cfm/qv_pk/1443


Let me know what your thoughts are.

Just think of all the girls you'll get with this circuit! ;-)

73,
Dan
W1DAN

That looks like an interesting chip, Dan.

I like that it has an internal oscillator to use (with ext capacitor) and the design is very simple.
The data sheet talks about using the extra internal op amp as an anti-aliasing stage as an additional all-time low pass filter.  Do you have an actual design for the 6kc and additional stage? 

I wonder how hard it wud be to make it adjustable from 4-6 kc as suggested?

T

I have two (2) of these that I built a few years ago. They are now surplus to me needs. Basically they are simple modifications of a switched capacitor filter sold by this company:

http://www.idiompress.com/scaf-1.php

I changed a few resistors and capacitors to convert it from an adjustable CW (300 - 1000Hz) brick wall low pass filter to an adjustable AM (2000-6000 Hz) brick wall low pass filter. The rolloff is 96dB per octive. The box uses two(2) switched capacitor filter chips from MAXX, in series, each of which has a 48dB/octave brick wall rolloff.

The rolloff frequency of the MAXX chips is determined by a supplied clock signal. The rolloff frequency is 1/50th of the clock frequency... so, if you provide a 250kHz clock, the rolloff frequency is 5kHz.

The clock (in the Idiom press product) is supplied by a 555 timer chip... whose frequency is controlled by a front panel pot. 250kHz is a little high for a 555 timer, but it works.

At the bottom of the web page... you will see "Other Uses". Click on that, and you will see the article I sent to them... describing the modifications I made.

If you want to try one of these (Tom or Brett) I'll make you my usual deal that you can't refuse. [My regular E-mail is sdp2@verizon.net]

Stu

     Why not put the EQ AFTER the compressor instead of before? This is what I do. Also my voice does not have a lot of energy above 3 Khz, so I have no need for a brick wall filter. YMMV.

Jim
WD5JKO

Hi Jim.  Yes, this is what Brett was describing... using a dual band eq, one side ahead and the other side behind the compressor.  I think that's a grand idea.  But, I think Tom's point is that most EQs only have a 12db +/- range, which isn't exactly adequate to quell all of those unwanted frequencies, thus the desire for a brick wall filter.

Becareful using the generic 'stereo' eq's. There is some interaction between channels. YMMV

John,

   One approach to this EQ limitation is to put the sliders below 5 Khz at the upper end, and have the frequency most accentuated at the top of the slider. Then after 5 Khz, put the next slider at the bottom. This will give you a 24db drop, but NOT brick wall. I suppose the dB/Octave drop varies on the EQ design.

   I wonder if Tom uses a compressor that breaks the audio range into bands and then has multiple compressors?

Jim,
WD5JKO
   

Hi Stu,

Thanks for all the FB info.  YES!  I'd like to try one of those Maxx boxes. It would be perfect for the job.

I have a switched capacitor chip here, but yours has TWO of these type chips in series and the clock is already built in. Perfect.

I notice your article says 4500 hz max range and the post above says 6K.  6K max wud be FB, if so.

I'll send you an email and arrange things for payment, etc.  TNX MUCH, OM.


Jim:  I'm using a 6-band Behringer 9024 processor with the top band set at 4kc - 10kc.     I tried putting a second EQ at the output as suggested- the cut was OK, but not enuff - and it also added noise to the chain that was excessive for some reason.

T

Tom

ok...

Yes... it will go to 6kHz. I had to select a 555 timer chip that would work at a high enough frequency (300kHz) to do the job.

Stu

Excellent, Stu!

I received the schematic and modifications you did to the filter. This is a more sophisticated filter that I first thought.  It should do the trick in spades.

I appreciate your generousity, as you did before, and will put it to good use, OM.  You da man!


Thanks again!


T

It would be OK to place EQ after the compressor, but it should always precede the peak limiter.  EQ can alter the amplitude of peaks non-uniformly over the audio range, which would defeat the purpose of a peak limiter.

One thing I have noticed is that boosting the bass with EQ tends to reverse the polarity of asymmetrical voice peaks.  Also, EQ can generate a tilt in near-square waveforms.  These are both things to be avoided, if possible, before peak limiting, and must be avoided, after.

You might consider the LP filter in the upper right hand corner of this schematic from Steve's website:

http://www.classeradio.com/pwm_generator_rev_d.pdf

or make modifications to the component values to roll-off at say 4.5 kHz.

Phil - AC0OB

Since I have a dual eq, I cut off above 5kc (or whatever) with both eq channels.
That would be 24 db.
The after eq is only used to cut frequencies, not to boost them.

I think some sort of sharp acting filter would be better though...

Brett

You could always do it with free software. Works like a champ.

SCF has a lot going on under the hood with all those little buckets

I could have sworn this was covered in a two part article in ER Last year.  They had a big chart to make your own filters complete with values.  Not a Chip but a real filter. Does someone have those issues? I can Dig them up if nobody has them. No idea what I did with them.

I am interested in this since my New Transmitter will be done and it has NO audio limitation on the 1800 watt modulator. I am going to have to limit it for sure.  I was planning on using the EQ route but then realized this probably wont be enough after reading this post.

C

You are probably thinking of the article in the October 2009 ER issue on p. 14, Audio for Heavy Metal Rigs by K0HEH.  There is a schematic in there for a passive LPF on p. 16 for 600 ohms using caps and inductors and an audio transformer.  It was designed by K0RZ and is a Butterworth filter.  Different component values are given in a table for cutoffs of 10, 6, 5, 3, 2, 1.5 KHz and 770 Hz.  I glanced over the article to see if there was any data on attenuation above the cutoff but did not see anything. 

More important is an addendum article in the December 2009 ER on p. 23, Audio for Heavy Metal Rigs, Revisited that has more information on constructing the filter and sources for the parts.  The filter looks extremely simple.  Four coils, six caps in a ladder arrangement driven by the transformer secondary, another cap in series on one side of the primary.  That's it.

BTW the problem with using an eq after a compressor is that the eq is not where it is supposed to be in the audio chain.  The eq should be used to literally equalize everything across a (for voice limited) range so the compressor can level the audio, then the peak limiter can keep everything to a set limit to prevent over modulation.  If the eq is last it undos all the processing ahead of it.  You can add a second eq at the end set up to just be a filter but my experience is that it adds distortion and is a step back.  Some other kind of filter is needed. 

Besides the simple filter in ER (which I would try first because it is simple) take a look at the LPF section on the schematic of the Inovonics 222 limiter.  Your LPF is IC 8, 9, 10 and 11 at the bottom of page 33 of the manual.  This one is for 10 KHz LPF but the general design is one you can compare to others you have seen.  There is a general circuit description of the filter on p. 27 of the manual.

http://www.inovon.com/download/222%20Manual.pdf 

Rob

Stu:

Does the SCAF and your circuit not use an elliptic filter IC?

I thought that was the main difference between your circuit and mine. I believe the Maxim chip I used was a different model.

Thanks,
Dan

 wow! what a great idea! narrow up your signal so the sperm burping sidebanders can move in even closer to us and still complain about our "AM splatter".

LOL..   Thats exactly what will happen!

How can I have one antenna produce a 1500 watt carrier.  One seperate transmitter produce 1500 watts of side band energy out another antenna.  Then I can have 3000 watts and still be under the legal limit.

C

While I know the "switched capacitor filters" work, and Stu made them and put them on the air, I could always hear the effect of the ones that Stu breadboarded when he had them on the air. I did not like the overall sound at all. I don't think it was the effect of the filter that i was hearing, rather the effect on the pass band audio.

I vote for an eliptical filter myself... the CAD programs make that somewhat easier, and it's all resistors and caps, opamps.

Just my opinion... ymmv.

                    _-_-bear

Sure, we can defend the frequency like it’s the Alamo and use electronic warfare to keep the sidebanders away. But, the idea to run wideband audio or "turn up the audio to keep the ssbers away" doesn't work. It just causes wars, bad feelings and more interference to fellow AMers.  

This weekend, like it or not, the AM QSOs from 3865 to 3895 operated the whole AM Rally event with 5kc spacing between QSOs. Sure, we could have spread out with wider spacing, but we didn't, as usual. We could have even spread out down the band (other than 3725 area) or gone anywhere for that matter – but we chose not to… no surprise.  

With only 5kc channel spacing, by right we should be running 2.5kc per side to prevent interference to one another.. (2.5 kc audio) This is ridiculous for AM, of course. However, many were running much more than 5kc audio and it was at times challenging to copy some weaker AM QSOs cuz of adjacent channel sidechatter.  That’s why it sounded like chaos in the ghetto as mentioned by a few here. Again, no surprise, just physics.

The point is it's a good idea to be able to switch in an audio filter to limit our AM bandwidth WHEN we crowd up…. and we will sometimes.  When the band is clear, we can switch it out and run “full” bandwidth, whatever that means.

The idea of operating down in the 3700 area is hi hi FB, but most of the AM stations will not do this. So, when crowded in, we need a filter to keep from beating each other up.  


The point is, no matter where we operate, we still need to have the ability to limit our bandwidth when the band becomes crowded.  The alternative will only create ssb jammers, AM haters -  and very important - the FEAR of most AMers to operate anywhere out of the cozy 3885 or 3725 areas.  With the ability to run a narrower bandwidth when needed, we can fit in most anywhere and slowly start migrating to other parts of the band without  “shocking” the locals all at once.  

As an EXTREME example, if someone were to operate a stock ricebox on AM with 2.4kc audio, he could go anywhere on the band and rarely get flack.  We need to have the ability to do something in-between this narrow  and  wideband audio WHEN needed.

Bottom line: When the band gets crowded, a low pass audio filter adjustable from ~ 4kc to 5kc is probably a good tool to have.  (even that's too wide for 5 kc spacing, however)


T

Further, most people's receiver aren't going to hear audio past about 5 kHz. To run frequency response much greater than this is just plain stupid.

Being an ass because someone else is an ass is great fun.
Get those really wide signals on the band and take care of the guys running ssb, that will show them!

I was in a qso yesterday on 3885, and one of the usual very wide guys was on 3875 with his 25 to 30 KHz wide strapping signal, so I joined in and opened the EQ up to 20,000 Hz and leaned into the audio.
The guy I was talking to thought it sounded very good, his passband might go out to 3100 Hz on the pro 3.

It was great fun being 6 years old again!


The way we need to work it is have one qso on the lower part of the band, say 3810, and the other qso can be on 3900, if you want to start another qso, go to 40 meters.

If we can get the signals 60 or more KHz wide, there will not be anyplace for the ssb guys to go until they get the 3cx15000 amplifier working, which will give us time to get our 3cx30,000 finals warmed up.

If anyone listens to ssb on a good modern rig, they would find out that even a very strapping AM signal does not bother ssb much at all. They can get VERY close and have a fine qso if they want, and going out to 30 KHz is not going to bother them one bit unless they live in town....

During the 'contest' there were 4 or 5 qso's with about 5 KHz spacing and it seemed to work great at my qth.
Most people had quite narrow signals, even if the passband was wide, because they don't mulitband compress without limiting the highs, so the extreme high stuff is way down and does not seem to cause trouble.

Brett

Most wont implement the filter.  The big audio sound is going to be killed.  I would make or buy one.  I think its needed.  But I really cant see others using it.

C

Dan

I don't remember which MAXIM switched capacitor filter is standard in the Idiom Press product. I'm pretty sure it wasn't an elliptical filter. I think it was a Butterworth filter, but I'm not sure.

MAXIM has a large variety of switched capacitor filters (I think much a larger variety than what they offered at the time I bought the Idiom Press SCAF-1 kit... several years ago).

Since I already have an audio chain that can insert a low-pass brick-wall filter with a cutoff of 4.5 kHz or greater...  I only used the switched capacitor filter, on-the-air, when I wanted to reduce my single-sided bandwidth below 4.5 kHz. So I don't think anyone ever head me on-the-air with it set to 5 or 6 kHz.

Like anything involving audio... you need to try it to see if you like the way it sounds with your voice and your microphone,..., and your ears.

It should be easy to substitute a different switched capacitor filter (e.g. elliptic) ... and they only cost about $5.00 in quantities of 1.

In my shack, I am now using digital filters... which were not available in the past... in the form I wanted, and at the price I was willing to pay. I didn't want to build a box from individual components to implement digital filtering... and I couldn't find anything that met my requirements commercially.

As pointed out, these days you should be able to implement any kind of audio filter you want, using a pc with any plain vanilla sound card... if you can find the application software that meets your needs (free is good).

Each passing month there are more free applications available to download that can be used... and the user interfaces keep getting simpler and more versatile. I haven't yet found a freeware application that can implement brick wall filters... but there may be one out there.

Keep in mind that the delay through any filter (LC or digital or whatever) is lower-bounded by (roughly) 1/w; where w is the change in frequency that corresponds to the transition from no attenuation to high attenuation. So if a brick wall filter is very steep (as opposed to very deep), the delay through it will be relatively large. For example, if you design the filter to go from no attenuation to high attenuation when the frequency changes from 5000Hz to 5050 Hz... then the delay through the filter will be at least 1/50th of a second = 20 milliseconds. This follows from mathematics... and is independent of how the filter is constructed.

Stu

Stu,

I use http://www.reaper.fm/  It's capable of using just about any plugin available, and it comes with some that do brick wall.  I use it as my audio chain, with a few commercial plugins as well as freeware / free-to-use ones.

IIRC, Reaper is free for non commercial use....  Or it used to be, the one I have IS.  If you want it for commercial use, pony up the cash.  It's EVERY bit as good as Adobe Audition, but much less in the wallet area.


--Shane

Dan

I checked the schematic of the Idiom Press SCAF-1 when I got home yesterday. The filter component is a MAX 295 (Butterworth, 8 pole)

http://www.maxim-ic.com/quick_view2.cfm/qv_pk/1370
http://datasheets.maxim-ic.com/en/ds/MAX291-MAX296.pdf

Also, the schematic of the unit, as I modified it, is attached below

Stu

Hi Stu:

Thanks for the schematic and information. I knew that as the filter gets sharper, the delay (and phase shift) gets longer.

I used the elliptical filter as it is a brick wall version (actually two in series with a negative clipper), and is sonically a little brighter than other filters with the same -3db point. I also used it as the analog Orban units use elliptical filters.

What digital filters do you use? I do understand the digital Orban units have somewhat adjustable digital filters.

Shane:

How is Reaper different in features than Adobe Audition (which I use)? I may consider using the Breakaway Broadcast PC based radio processor in the future.

73,
Dan

I've had Stu's modified filter in use for about a week now on the transmit audio chain. At -96db/octave roll off, it's a keeper.

For example, I can set the TX roll off to start at 5,000 hz and put a tone thru the transmitter to set it at 100% modulation.  (The filter is on the output of the audio chain after the 6-band processor.)  Starting at 5K audio, 100% modulation,  the 5500hz point is down to around 5% modulation. At 6kc there is nothing getting out at all.

In practice I find starting the roll off at 6kc is a good all-around audio bandwidth.  Though 4500hz is tolerable when band cornditions get crowded.  There is definately a difference!

BTW, after a year's hiatus, I just re-installed my SP-600 receiver and tapped off the detector into an audio 12 band EQ. The EQ makes the SP-600 flat into the stereo audio system. The performance is superb - flat from 30hz to about 7kc.   Recently I was running my FT-1000D off its detector, etc. I thought the receive audio was decent until I swept it with my modulated hi-fi transmitter. It actually sucked. Now with the SP-600 system, I can hear who is truly HI-FI on the air and who is more communications audio. Beforehand I was fooling myself. Also, I think I have the recording system ironed out and ready to give out some on-the-air playbacks. The recording gets re-transmitted using the FT-1000D's transmitter balanced modulator which is quite transparent.

Back to the receiver - Bottom line is it's vitally important to sweep our whole receiver system from antenna to audio output to see how flat it really is. IF strips can alter things. Even the SP-600 off the detector needed an EQ to flatten things out.  And remember, the 16 khz receiver filter position passes ~8kc audio, not 16k audio, ( 16kc divided by 2)  unless we tune off the center frequency somewhat.


T

Where is information on STu's filter?  Got a picture?

I run the Sp600 JX14.  Its fantastic. I tapped off the back also in a Marantz amp and then a nice speaker. Its great. This makes sense for the 16K filter position. I can roll around slightly and alter the tone.

I have been doing all kinds of tests here.  I got a Real nice Eico Audio Gen off Fleabay. It works great and I can sweep my system and recievers.

C

Clark,

Go back a few posts for Stu /AB2EZ's schematics and information. His modified schematic is there too. You wud need to buy the unit and modify it. Works as well as I imagined the best filter I cud possibly build. Two chips in cascade are really something else for a brick wall.


BTW, if your SP-600 is like mine, you may need some slight EQ boost betwwen 20-70hz and again between 2kc-7kc to flatten it out. It didn't take much to get a perfectly flat line from 30hz - 7kc.  Without it the extreme highs and extreme lows lacked slightly. The EQ goes between the detector and audio amp, or wherever is easiest in the post receiver audio chain.

T

T

Mine can hear out to 10K.  I have not measured it though..   I got the Mute working on mine tonight. I am going to make a post. It was a simple Relay and socket.  I used an old wall wart to power the relay coil with teh T368s contacts out the back.  I tacked on two wires on the send recieve and ran it out the back of the 600.  My 600 had no relay socket.  Its a real pleasure to have this mute when I flip the remote switch at the desk. 

C

Clark,

To have 10kc audio, did you modify the RX to have a wider filter position than the stock 13kc?   Even tuning off freq a little, I found 7kc is all I could squeeze out in the 13kc position. It needed some EQing to pull it out.  Though, I find 13kc is about all I can tolerate on the bands most of the time anyway.   I mean, I can still hear 8-9kc audio, but it starts rolling off up there.

T

Tom,

I'm curious about the 1000D prod. det. audio.  Any idea why it was so poor coming out of the tap?  I don't know much at all about the D.  Perhaps Yaesu used a 455 KHz IF filter for AM that's too narrow?   They used a 6 KHz murata ceramic AM filter in the 455 IF of my 1000MP Mk V.  I had to pull that sucker out.  3 KHz audio--much too narrow.

Rob