Call Search
     

New to Ham Radio?
My Profile

Community
Articles
Forums
News
Reviews
Friends Remembered
Strays
Survey Question

Operating
Contesting
DX Cluster Spots
Propagation

Resources
Calendar
Classifieds
Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement
About eHam.net

   Home   Help Search  
Pages: [1] 2 Next   Go Down
  Print  
Author Topic: Narrow roofing filters and SSB...  (Read 8905 times)
KT0DD
Member

Posts: 277




Ignore
« on: August 31, 2011, 08:22:03 AM »

Hi All, I admit my knowledge isn't on the engineering level, so I hope this question doesn't sound too stupid. I'm a 100% SSB operator / DX'er. I'm just wondering how much benefit a narrow 3k roofing filter is to SSB ops. I understand the benefit is mainly for CW. I'm not a contester, but I do try for weak signal DX. Will I notice much difference with a narrow roofing filter?  Thank You.

Todd - KT0DD
Logged
KA5N
Member

Posts: 4380




Ignore
« Reply #1 on: August 31, 2011, 11:10:35 AM »

A "3K" roofing filter is not a "narrow" filter.  2.7 K is a "standard" roofing filter.
Narrow would be around 2.1 to 1.8.

Best bet is to visit someone with a good selection of SSB roofing filters in their
rigs and listen and judge for yourself.  Be sure to tune where there is lots of QRM
and QRN.  A lot of listening is worth a 1000 words.

Allen
Logged
N4KZ
Member

Posts: 592




Ignore
« Reply #2 on: August 31, 2011, 11:24:59 AM »

For a few months, I've been doing A/B comparisons between my Icom Pro 3 and new Kenwood TS-590. The Icom has a 15 khz wide roofing filter - as a result of its up conversion high first IF design. On 160, 80, 40, 20 and 15 meters, the Kenwood uses down conversion with a low first IF and has a 2.7 khz roofing filter for SSB. Sounds like a big difference, huh? That's what I thought until I started my comparisons -- same signal, same antenna - just flipping quickly between the rigs with a coax switch. Most of the time, I can tell little or no difference when on SSB. That's under casual operating conditions. But when I have done my A/B testing during a major contest, the Kenwood's narrow filter does come into play but only if I'm listening to a signal with a huge, rock crushing signal very close in frequency.

Here's one such example. During the spring I was doing the A/B receive test during a big DX contest. I was on 40 meter SSB. I heard a weak LZ station calling CQ contest over and over and getting no answers. Then a strong U.S. station opened up just a couple khz away. On my Icom, the LZ could still be heard but most of the time he was not readable. He was covered up with SSB splatter. But on the TS-590, he was perfectly copyable at all times. It took a lot of knob turning on the Kenwood to get there but I could make him 100 percent copyable.

I've encountered that situation several times but only during big contests. During my normal casual ragchewing or sometimes DX chasing, I've yet to encounter the exact same scenario. So, for me, the Icom and its wider roofing filter is adequate at least 95 percent, or more, of the time. And as an overall package -- bandscope, Dual Watch receiver, size -- I still like the Icom a bunch. But its receiver isn't quite as good as the Kenwood but most of the time the Icom is still plenty good enough.

On CW the situation is much the same with the Kenwood being better but only under circumstances that are fairly unique to contests -- tons of strong, close signals often only a few hertz apart. Of course, keep in mind that on CW the TS-590 has a 500 Hz roofing filter. That's much better than a 3 KHz roofing filter -- particularly when the strong signals are within that 3 KHz range, which they will often be.

So, if you like to contest to any degree, narrow roofing filters can be beneficial from time to time but if you turn off your rig and run and hide when contests invade the bands, save your money because narrow roofing filters probably won't help often enough to justify the expense.

Not a very scientific reply to your question but just some real-life observations. Frankly, I expected to see a bigger difference between the Icom and Kenwood but how much difference one sees is very much dependent on your preference of operating -- casual ragchewing vs. competitive DXing or hardcore contesting.

73, Dave, N4KZ
Logged
W8JI
Member

Posts: 9304


WWW

Ignore
« Reply #3 on: August 31, 2011, 09:57:44 PM »

Hi All, I admit my knowledge isn't on the engineering level, so I hope this question doesn't sound too stupid. I'm a 100% SSB operator / DX'er. I'm just wondering how much benefit a narrow 3k roofing filter is to SSB ops. I understand the benefit is mainly for CW. I'm not a contester, but I do try for weak signal DX. Will I notice much difference with a narrow roofing filter?  Thank You.

Todd - KT0DD

Hi Todd,

Of course "narrow" is relative to what was original. Many roofing filters are pretty wide, up over 10 kHz wide.

The new narrower roofing filter CAN help things between the old filter's width and the new filter's width. Say you have a 6 kHz roofing and change to a 3kHz. Third order intercept and blocking performance in the overlap window of the new 3 and old 6 kHz skirts will get better.

This can help when the band is filled up with strong signals.

It usually doesn't do much inside the passband of the new narrower filter.

I would not expect night and day, or even anything noticeable, unless the original receiver is pretty bad or the band is just loaded with signals.   

73 Tom

Logged
KG6YV
Member

Posts: 504




Ignore
« Reply #4 on: September 01, 2011, 10:04:52 AM »

True "roofing filters" are those that go ahead of the first mixer.  They are sometimes effective on SSB depending on
the filter's BW and shape factor.  I have an FT-2000D with the Network Sciences after-market roofing filter which is a 69 mhz implementation with a 2.6 khz -6db bandwidth.  The filter isn't cheap ($300) but it seems effective.  The filter
has been lab qualified to improve IMD in the FT-2000 to put it in the performance category of an Icom 7800. 

OK, here is what I see/hear.  In crowded band conditions I can remove all grunge  IMD and splatter from any signal > 3 khz from the desired frequency.  Observational testing during last November's Sweepstakes made my casual contest
participation much more satisfying and less frustrating than it was before using the FT-2000's 6khz roofing filter.

My impression is that it was worth the investment,  IMHO

Greg
Logged
HB9PJT
Member

Posts: 262


WWW

Ignore
« Reply #5 on: September 05, 2011, 05:57:10 AM »

Ahead of the first mixer is the bandpass filter located. But at that place it is not possible to set a narrow roofing filter.

73, Peter - HB9PJT
Logged
KA5N
Member

Posts: 4380




Ignore
« Reply #6 on: September 05, 2011, 06:03:47 AM »

True "roofing filters" are those that go ahead of the first mixer.  They are sometimes effective on SSB depending on the filter's BW and shape factor.  I have an FT-2000D with the Network Sciences after-market roofing filter which is a 69 mhz implementation with a 2.6 khz -6db bandwidth.  The filter isn't cheap ($300) but it seems effective.  The filter has been lab qualified to improve IMD in the FT-2000 to put it in the performance category of an Icom 7800.  
IMHO
Greg

If a filter is "implemented" at 69 MHZ and is 2.6 kHz wide then it must be after the up converter (first
mixer) and before the second mixer.  Ahead of the first mixer is the full spectrum that gets into the
receiver.  Not so???

Allen
« Last Edit: September 05, 2011, 06:11:05 AM by KA5N » Logged
VA2AM
Member

Posts: 7




Ignore
« Reply #7 on: September 05, 2011, 06:06:09 PM »

Hi Todd

To be effective a roofing filters will be from the down conversion radio type (Orion II, TT 599, FTdx5000, TS590, K3)  some old Drake. This is very difficult to reproduce a shape factor at 65-70 mhz. You will see the main difference during contest time so important for cw and RTTY if the roofing filter matched the DSP bandwidth. Also at 100-500 hz there is no ringing effect that will please your ears for hours. The only thing we cannot eliminate are spatters from poor ssb operators that cannot believe on 50% ALC is enough.
I tried in the past many up conversion radios, the minimum is 6 khz and DSP cannot do miracle when the AGC is pumping. I worked with K3 and own FTdx5000 and as narrow 100 hz cw and 1.65 khz phone when needed with 100% copy.
73, Rejean va2am
Logged
KG6YV
Member

Posts: 504




Ignore
« Reply #8 on: September 07, 2011, 08:05:47 AM »

If a filter is "implemented" at 69 MHZ and is 2.6 kHz wide then it must be after the up converter (first
mixer) and before the second mixer.  Ahead of the first mixer is the full spectrum that gets into the
receiver.  Not so???

Allen

Allen is 100% correct, I meant to say that the roofing filter goes behind the first mixer. 

Greg
Logged
M0HCN
Member

Posts: 473




Ignore
« Reply #9 on: September 10, 2011, 09:18:14 AM »

Yep, and it is worth noting that even if you have narrow selectivity filters immediately behind the roofing filter, these often sacrifice IMD3 performance for shape factor so a good roofer helps even then.

All of this does of course only show under contest conditions,  and even then everything else needs to be spot on to make the IF filters the weak point.

Regards, Dan.
Logged
ZENKI
Member

Posts: 906




Ignore
« Reply #10 on: September 18, 2011, 06:15:00 AM »

Well first ask the question, what is the IMD dynamic range of your transceiver. If you dont have a receiver with a decent IMD dynamic range adding a roofing filter is not going to buy you much. All the tests of the after market roofing filters have revealed that they dont improve things much at close signal spacings.

You far better off buying one of the top 5 radios than trying to improve a radio with a roofing filter when the receiver has a marginal IMD dynamic range. The Elecraft K3;s narrow filters are pretty impressive, however you soon learn the valuable lesson when using a good receiver like the K3. The main limitation that prevents you from exploiting the K3;s excellent receiver performance is the poor state of other stations  signals especially on SSB. It does not matter how tight your receiver or roofing filter is, it wont prevent another station from bleeding IMD and other spurious products in your excellent receiver even with a very good roofing filter and receiver. Roofing filters are best left to the CW end of the band where most stations have good to excellent keying sideband suppression. On CW its easy to use the advantages of a receiver with excellent dynamic range  and excellent roofing filters, that is until some op with a bad FT1000MP that causes keylicks comes along.

Bottom line is that ham manufacturers really need to start improve transmitter design. When I read the RSGB review of radios like the Ten Ten Eagle and see that it has 3rd order IMD figures as low as 20db below pep I just shudder and think hey these designers just dont get it! Dont be too obsessed about receiver performance on SSB, your money is soon squandered by the crud on the band from other stations that no receiver can block.
« Last Edit: September 18, 2011, 02:20:59 PM by ZENKI » Logged
M0HCN
Member

Posts: 473




Ignore
« Reply #11 on: September 18, 2011, 11:37:04 AM »

IMD3 twenty below PEP, OUCH!

30dB below ONE TONE is about the minimum anything around here does, any worse then that and it gets fixed (-30dB ref one tone is not that hard if you design everything conservative like, as long as big and heavy does not bother you).
I have a rig on the drawing board that should manage better then -60dB ref one tone at competitive power levels and efficiency.

Note that the use of PEP rather then ONE TONE level for the test gives the manufacturer a 6dB free pass!

I should design a digital mode with a constellation diagram that only a really clean PA can reproduce, and try to get it popular, maybe get my full license and go and do some DX with it from a **REALLY** rare location.

Regards, Dan.
Logged
ZENKI
Member

Posts: 906




Ignore
« Reply #12 on: September 18, 2011, 02:31:23 PM »

Excellent work Dan, maybe you can licence your design to Ten Tec, Elecraft, Yaesu, Kenwood and Icom. They certainly dont know how to design a clean solid PA's that have acceptable IMD performance. All the  excuses  and lame arguments from seemingly intelligent hams against mandated IMD standards for commercial  ham radio equipment serves only the interests of the ham radio manufacturers profit margins. There is really no technical reason why all ham transceivers should not have a PA design  with 3rd order IMD  figures of least 36db below PEP. -60db below 1 tone is certainly an extraordinary figure, how are you achieving this kind of performance and what RF devices are you using?

IMD3 twenty below PEP, OUCH!

30dB below ONE TONE is about the minimum anything around here does, any worse then that and it gets fixed (-30dB ref one tone is not that hard if you design everything conservative like, as long as big and heavy does not bother you).
I have a rig on the drawing board that should manage better then -60dB ref one tone at competitive power levels and efficiency.

Note that the use of PEP rather then ONE TONE level for the test gives the manufacturer a 6dB free pass!

I should design a digital mode with a constellation diagram that only a really clean PA can reproduce, and try to get it popular, maybe get my full license and go and do some DX with it from a **REALLY** rare location.

Regards, Dan.
Logged
M0HCN
Member

Posts: 473




Ignore
« Reply #13 on: September 18, 2011, 04:14:17 PM »

The trick is all in the feedback!

basically I am using cartesian envelope feedback with the group delay in the PA compensated by using two DDS chips (one for the forward IQ modulator carrier and one for the reverse), then by tweaking the phase control registers I can bring the returned signals into phase with the forward path, add a suitable band limited error amplifier and close the loop when the measured values come right and the distortion just vanishes.....
36dB of overall loop gain cures a lot of ills if you can make it stable.

As long as the group delay causes less then a few tens of degrees of phase shift change over the loop bandwidth (~70Khz in my design) no further loop compensation is needed.

This is all in 20+ year old literature, I found a paper about a design done for IIRC Racal that was doing this in the late 80's (Granted DDS makes it easier), so there is no excuse.

My biggest pain at the moment is (because the loop runs at baseband) dc offset drift, I might need some sort of chopper arrangement.

The PA  at the moment is nothing more then a pair of old BLW96 driven by BLW50F, nothing exciting (-30dBc or so open loop @250W), I am sure a modern SD whatever would do as well. Yes I know my intermediate license only allows 50W at the feed point, not using it on air yet, relax.

One trick I am working on is based on the realization that with all this feedback in place I can afford to vary the gain of the finals (slowly compared to the loop bandwidth) as long as the drive power stays within limits, this opens up the possibility of going to a sort of partial polar loop wrapped around the cartesian loop, in that the collector/drain voltage can follow (and lead on just a bit) the envelope, which with a DSP modulator is trivial (delay the audio a ms or so, and ramp the switchmode PSU as appropriate). The win is improved load matching at all power levels and less heat in the finals. The inner loop will compensate the gain changes automatically allowing the PSU modulation to be open loop.

Further for key down modes (or CW) I can start off with the PA bias in class AB, once full carrier power is achieved, bias can be ramped down to improve efficiency (the inner loop will automatically ramp the drive to compensate), possibly all the way to zero bias class C, with the reverse being done just before key up to allow the decay to happen.   

Getting both the bias control and the supply voltage control laws right is proving a bit nasty (they interact in horrible ways, especially on 10M, miller effect I suspect).

A key realization is that for SSB, it really does not matter what the efficiency is at PEP, what matters is the efficiency at 30% of PEP as the rig spends far more of its time down there then it does at full power.

Oh, also ALC between amps and the rig sucks, bad idea, don't do it, you know the amps rated power and the amps gain on each band, so just program the correct rig output power for each band..... AGC between gear which was not fairly carefully designed to work together will always suck as the composite transfer function will not have been designed in any sane way (Splatter city). 

Transmitters really are the low hanging fruit in ham radio at the moment, particularly as the HF receiver is fairly close to being at the point where QR{N,M} is the limiting factor in most situations (LO Phase noise is really the last part of that puzzle).

My problem is I am never satisfied, I really should just stop, design some nice boards and write this thing up for QEX/Radcom/DUBUS or whatever.

Regards, Dan.
Logged
M0HCN
Member

Posts: 473




Ignore
« Reply #14 on: September 18, 2011, 04:24:20 PM »

Oh all the manufacturers you mention know exactly how to do it, it is not rocket science.

But doing it right costs money, and space, and weight and potentially efficiency, and you really do NOT want to do it with 12V parts, so possibly you need a boost converter in there if the rig really must run off 12V.....

 Big heatsinks, 400W devices in 200W PA strips, power rails that stay put under full modulation, magnetics and wire that are man enough under 3:1 VSWR, audio stages rated at 0.1% THD instead of 10%, it is enough to bring a beancounter out in hives!

Also you do it this way and you get ONE receiver and one transmitter in a $5,000 radio, not popular these days.

Regards, Dan.
Logged
Pages: [1] 2 Next   Go Up
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!