AM Receiver IF filter

[N4LTA]

I am working on a homebrew tube receiver or "high fidelity AM". I have completed the power supply and audio section and am very happy with it (two 6V6s in PP with a 6SL7 phase inverter.


I have heard it said that a Collins mechanical filter has poor sound fidelity becuase of phase distortion. Collins makes a fairly inexpensive 7000 hz wide filter. The widest crystal filter is 6000 hz and quite a bit more expensive.


Any info would be appreciated.

Pat
N4LTA

[WA1GFZ]

Hi Pat,
Very good you started your design at the speaker end of the RX. I found this is the best way to test stages as you go. The most high fi IF filter would be LC to avoid the phase problems of a mechanical filter. I'm told crystal filters are cleaner but that depends on the filter design. Look at the IF strip in a R390 or R392 for ideas. You might fire up a simulator and try your hand at a design. I played with one once and think it took around 5 sections of parallel tuned circuits to get a nice response but I really didn't look at the phase. Most phase changes are in the filter skirts so a sharp skirt like a mechanical filter will have phase changes. Good luck!

[N2DTS]

I used kiwa filters in my homebrew receivers, and am really happy with them.
I used their cheap ($50.00) filter modules, they seem to work like mechanical filters (very sharp) and I dont notice any distortion in them.

They are little cubes, they have an op amp input, an op amp output, and three ceramic? filters of slightly different center frequencies I guess.
What makes them work so well is the op amps I suppose, the filters look into the impiedance they want to see and everything is very close together.

Not sure what bandwidths they are making, they used to make every 500 hz, 3.5, 4.0, 4.5, etc....

In my receivers, the kiwa filter makes for performance, off frequency stuff does not pass.
Digital readout makes it accurite, simple design and the hi fi detector makes for fidelity and low noise.

At the moment, I don't own any commercial receivers...I never had anything that worked better on AM than the homebrews. The only limitation is they don't go above 40 meters.


Brett
N2DTS

[N4LTA]

Yep - I have been homebrewing for a long time and I always work from back to front. Usually the PS is the easiest stage and the audio next.

I may try to put three tube IF stages together and see what king of bandwidth I get. I may do it as a module  - possibly on a PC board. The ceramic filters seem to be cheap but I think they may have phase delays also. Going with a low IF like 50 or 100 khz would make the filter easier but betting the inductors may be a problem - but then ferrite toriods might make some fairly high Q inductors at those frequencies. This is not going to be a short project - but it should be fun.

Where did you get the KIWA filters?

Pat
N4LTA

[N2DTS]

From kiwa of course.

http://www.kiwa.com/kiwa455.html

[k4kyv]

Sometimes at hamfests a  parts-rig R-392 can be had for $50 or so.  Check to see that the i.f. module is still there and looks reasonably intact.  If you want to go solid state, it shouldn't be difficult to replace the tubes, which run at starvation-level voltages, with solid state devices.  Ask Steve, WA1QIX.  The receiver has a switchable 2-4-8 kHz i.f. strip using L-C tuned circuit filters.  The selectivity is almost as good as that of a R-390A with mechanical  filters, and the audio quality is much better.

Remember, the R-390A was a cost cutting measure over the old R-390, not an attempt to improve performance.

[KM1H]

With 3 stages of 455 kHz IF you can expect around 10 kHz on the nose and 15-18 kHz at 60 dB down. Vary the coil spacing as desired. I like the performance of the 3 stage HQ-129X/140X for an example

If images arent a bother then use old auto radio 262 kHz IF's. Its hard to beat the sound of those early 50's Buick radios and duplicating the audio and IF is simple. Autos used 262kHz from the 30's to the end of the tube era so just about anything will give good impedance matches to tubes.

I modified a rather useless SX-140 with double conversion and 262 IF's and its a real nice performer at our summer place. Cant get much smaller and still have tubes in it.

Kiwa filters are excellent. Ive used them and Murata's in several boatanchors to get great skirts. The HRO-50, SX-28, and others are much improved during QRM.

Carl
KM1H

[N4LTA]

262 Khz would be a good IF choice.

It probably would be a good idea to make it a double conversion scheme , maybe with a single stage of 1700 Khz IF or so to get rid of the images. I see that "Playthings of the Past" has many 262 Khz IF transformers listed. I may try to build a 262 Khz IF Strip and see what I can do with it.

Did you use 2 or 3 stages?

Has anyone ever tried to use high performance broadband Op/Amps in a low frequency IF scheme much like an active filter?



Thanks,

Pat
N4LTA

[WA1GFZ]

I've used high performance op amps to over 30 mhz. Check Analog Devices. some have low Z outputs that can drive coax.  Look at AD810 AD811

[N4LTA]

I think you need far less performance than those Op/Amps. I'm thinking at 50 Khz or so you can almost mimic an RC active filter - with lots of bandwidth control.

The RC type may be hard to get enough Q though.  Not sure for a multiple pole wideband 6 -8 Khz wide filter. that 12-16% of the IF frequency.  Anyone seen it tried?

Pat
N4LTA

[KM1H]

The SX-140 has a 1600 kc IF and a single useless 6BA6 regen IF stage. That became a 6BE6 mixer driving 2 stages of 262 kc with 6BA6's. I used 2 IF cans between stages as there was sufficient gain available.

Thats about the lowest frequency I would go for good AM with both sidebands. A SX-100 at 50 kc for instance is too sharp.

Ive no experience with SS in a ham rig except keeping my Kenwoods running. My modern working years were all at microwaves.

Carl
KM1H

[N4LTA]

I think you are right. I am going to order some 262 Khz IF transformers and build a 6BA6 IF strip with possibly a switchable ceramic  or mechanical filter to narrow it down a bit if I need to.

Did you have a serious image problem or is the front end sharp enough to strip it off. I have looked at Ray Osterwalds Front end in ER Nov 2008 and it looks good on paper. I am only interested in 40 meters as the highest frequency and will use it principally on 80 meters.

Pat
N4LTA

[KM1H]

Finding suitable filters at 262 kc can be a problem, you can always add a 3rd IF stage and more cans to get better skirts without narrowing the nose by careful alignment. Or you can get real creative by adding 3rd windings to the forms and switching several bandwidths.

Ive had no image issues with the SX-140 but Ive only used it a few times on 10M. On 6M there isnt anything to get images from. It also has an antenna trimmer as built and a pretty sensitive front end. I use it mostly on 40, 20, 15, and 6M with simple antennas....40/15 dipole with 20M at right angles and a 4 el on 6M, all at about 25' high on the Maine coast.

Carl
KM1H

[N4LTA]

Duh - I will have a difficult time with a 262 Khz filter. I didn't engage the brain.

Ever seen a variable resistance used to de-Q an IF transformer to broaden it?


Thanks for the ideas,

Pat
N4LTA

[WBear2GCR]

Collins did/does make mechanical filters that are/were wider than 7kc.

Dunno about the phase shift being audible or not...

Fwiw, I like what i have heard out of the Murata ceramic filters in my 440.
If anyone is interested there are some 500kc I think 16khz wide Collins made ceramic filters for not too much bux at Surplus Sales of Nebraska. I got a pair and tried them in series and found that they are pretty good with a decent stop band like that. May not be "flat enough" in the passband for your application, I don't know.

One way to go with passive is with eliptical filters... there is a nice write up on a LPF version in October 1958 QST, page 21 - will give the general idea, a BP filter that is similar can be created.

These days one might go with a solid state implementation that is all active filters too... but that might be considered "cheating" and not part of the design paradigm for your project?

            _-_-bear

[KM1H]

I had originally typed a few comments about mini relay switching several swamping resistors but gain equilization can get complicated so I deleted it..

My HB 160-10M project has taken a different route. It starts off with a Heath LMO and filters, then down to 500 kHz and into a 4 filter bank from a Racal followed up by the xtal filter assembly from a 51J4. With a switching matrix controlling mini relays I can have a wide range of selectivities from 10 kHz to 250 Hz. The Heath parts and 500 kHz were chosen for very cheap availability compared to more common frequencies such as 8.8 and 455 where prices are always high. Besides the mixer spur program looked excellent for all 10 bands.

The hardest part will be getting the Collins crystal to the correct frequency for the filters and then getting the passband tuning to track.

Carl
KM1H

[w3jn]

Passband tuning tracking is easy.  One osc driving 2 mixers.  Filter is in between the 2 mixers.   With one osc the PBT will automagically track.

input -->  mixer --> filter --> mixer -->output

^-------  osc    ----^

[N2DTS]

The Gonset G76 used a 262 Khz IF, and also used resistors across the IF cans to broaden the response, and also a small cap like 2pf from primary to secondary, you could play with the values and get different results, but nothing was razor sharp filtering.

I used 455 Khz in my receivers because I wanted to fit a filter, and did not want to roll off before the filter.
Let the if amplify, let the filter do the filtering...

I have a bunch of new 262 Khz IF cans for car radios.
 
Brett
N2DTS

[k4kyv]

To help the image problem, Hallicrafters tried a scheme in the late 30's that was some sort of variable notch filter in the front end of the  receiver that allowed you to null out the image.  It was supposed to be more effective than simply adding an extra rf stage or an additional tuned circuit between antenna and 1st stage.  Might be worth some research.  There may have even been an article or two in QST with homebrew construction details, but I am not sure.  I would look through 1938-41 issues since I recall it being pre WW2.  Look for articles and Hallicrafters ads, which can be found by the advertisers index.

[WQ9E]

Don,

That was the SX-18 Sky Challenger with the "image rejector" built in.

Apparently it didn't work too well.  There was a review where the image rejection tuning reduced the desired signal by about as much as the image and the net gain was around 3 db of rejection at the higher frequencies.  However, it would be interesting to see how well the idea worked if built with a higher Q rejection circuit where cost wasn't the primary consideration.

I am intrigued by a receiver design presented in one of the older RSGB handbooks.  It was built by Rhodes and Schwarz and achieved its selectivity through the use of low pass filters in the audio range.  The IF is mixed down to the audio range and passes through one low pass filter and is then inverted and passed through another to set both sides of the selectivity curve.  It was then mixed back up and conventionally detected.  The mixing scheme allowed variable selectivity and passband tuning. 

Rodger WQ9E

[KM1H]

I have a SX-18 and all that magic circuit is a series tuned trap that is as broad as a barn door and adds to the front end noise figure which is already sky high.

Halli only advertised it a few times and it was rapidly gone and replaced along with the SX-16 by the SX-24. The SX-17 isnt half bad and the SX-28 is in a class of its own.

Ive collected many 30-40's era radios mainly as curiosities that get restored and become shelf queens. A small number do get used regularly.

Carl
KM1H

[WBear2GCR]

Whoa! Mixed down to audio then back up?? Wild!

One would think that going that low in freq would be not as good as staying higher because of the electrical size of the passive components, the electrical size being large too... but if R&S did it then there is some reason/trick that makes it work. Would like to know what that is... and how did they "flip" the signal, making the LF high and the HF low??

Inquiring minds want to know! 

           _-_-bear

[N4LTA]

Brett,

You don't want to part with a few of those transformers , do you? If so , let me know.

I have also pondered putting a fet across the IF transformer and using it as a remote controlled variable resistor to broaden the responce.


Pat
N4LTA

[N2DTS]

I have a crap load of them, and dont plan on using them, so you can have some for free.

I dont have any image problems at all, single conversion, 455 Khz IF, but I use a very high Q tuned circuit on the antenna input.
I can TUNE it to the image and get stuff, but when its tuned correctly, I get nothing.
 
B+W coil stock, link to the antenna, variable cap tunes it to 80 or 40 meters, I marked the knob with the positions for 80 and 40 meters. Simple, silent and effective...

Brett
N2DTS

[w3jn]

I put a Collins 3KC mechanical filter on my HP 3577A network analyzer and there were 5 or 6 180 degree phase shifts throughout its bandpass.  Crystal filters have largely the same issue.  I haven't swept a L/C IF stage with it yet to see how it compares with the xtal/mech filters.