The AM Forum

I ran some preliminary receiver tests using the HPSDR Mercury receiver against the FT-1000D. We all know the filter selectivity, noise blanker functions and dynamic range should be better on the SDR.

But what surprised me is * in the clear * the HPSDR could hear signals that were weaker than the 1000D could copy due to what I think is 1000D phase noise.  This is more pronounced on 20M and higher where atmospheric noise is lower, so the RX noise floor becomes more important.  On 40M and lower, the atmospheric noise usually masked this difference in performance.

I tried both on-air signals and using a signal generator. On 20M and above, the HPSDR front end preamp made a huge positive difference. Using a signal generator, a 14 mhz signal that was barely heard in the noise on the FT-1000D, I could hear clearly on the Mercury because its noise floor was much lower.  I reduced this signal lower, could hear it on the Mercury, but totally disappeared on the 1000D.  The 1000D seems to have cyclic areas throughout the spectrum where the phase noise is higher than other areas. To be fair, I was using a spot with the lowest 1000D phase noise.

Today I tuned in some Russian DX stations on 20M with the stacked 20M Yagis (using a real noise floor) and the weaker ones were easily better copy on the Mercury.

Considering SDR is still in its infancy, the next years of device improvement should leave the older receivers in the dust.


Rich/ETP was telling me about tests he and Chuck ran on 15M with similar results. Now I understand what they were seeing.

T

Interesting. Thanks for posting. No knock on the SDR, but I bet you would see the same or similar differences between the FT1000 and any higher-end new receiver, SDR or otherwise.

Yeah, sounds logical. The newer FTDX5000MP? that goes for $8K ranks way up the list in the Sherwood tests.  It appears as good as an SDR.

Keeping within a budget, I was always biased to think the 1000D was the best I could do here. It probably was optimum on 160-40M with the higher band noise, but now that low noise 20M-6M are opening up, this difference becomes significant if ya operate there looking for the real weak ones.


T

In my short time with the 1000, I've noticed that it is a bit nosier than some newer receivers I've used. It also exhibits some spurs here and there. To be fair, I should continue the comparison after a full alignment of the 1000.

The $8K is List; you can buy one for a mere $6.3K

Pete, if I don't eat for 6 months I can afford one at $6.3K.  It might be worth weighing 70 pounds to hear the weak Pascals of the world above 7 mhz.


HUZ, when stick your diddle wand in that 1000D just remember me - I can use another  backup for this transplanted 1000D amplifier... ;D

T

Yup....

I can feed my family of 4 on less than $5k for a whole year. And that's eating pretty well.

My priorities lie elswhere than dumping that wad on one little radio.

 ;D

Time to do more testing with the Racal and Cubic. Maybe I should drag out the old Heathkit SB303 with the full race cam. I wonder if you checked it against an FT102 that does not have a synthesizer.
The Cubic R303 has an Rf amplifier and input filter bans. It is very sensitive but has more close in phase noise than the Racal.
Mercury should be even cleaner since Flex uses a DDS source which makes close in spurs. Mercury has a crystal oscillator. Nothing is cleaner than a good crystal oscillator. The only hole is the FPGA code doing digital down conversion.
I have done a lot of testing on 80M and the Racal holds its own against Mercury. Mercury has ever so slight agvantage when there is high static.
Racal beat my copy of the Flex front end  hanging off the IF.
Receiver drag racing, I love the smell of two warm HP8640Bs driving receivers to their red line.

Question:

If you put a preamplifier (perhaps with a roofing filter to block strong, off-frequency signals) in front of a receiver, such as an FT-1000D's receiver... or any other decent receiver... and if the preamplifier uses the latest technology... why wouldn't the minimum "detectable" signal (on frequencies where atmospheric noise is lower than the thermal noise of the receiver's front end) be completely determined by the equivalent input noise of the preamplfier?

Stu

Tom,
Just for the heck of it listen on 10 or 15 and pump your signal generator into the RX. Crank up the level until you hear noise change. Now slowly tune across the bands to see if you have any responses. An OK RX -30 to -20 dBM, a good RX -20, Kick butt 100 db above the noise floor of the RX.
I agree with Syu a good preselector with some gain will help. Harris, Cubic, McKay, Collins etc made some cool prepost seletors made for that putpose.
I need to put mine in service. I would love to be smart enough to have HPSDR control it. It tunes in 10 KHz steps.
Also another cool box that you might like is the Racal dual preselector with 13 bandpass filters .5 to 30 MHz. This would be perfect for diversity. I landed a new one but they come up on ebay once in a while for around $100. In a 3 inch rack. My tunable preselector is in a 1U rack MSR6300. Both have a little gain.
I have a collins unit that has a 4 gang cap as the tuner and full of vacuum relays to select inductors

Sure, I'd like to try a good quality preselector for the Mercury.  I don't mind manually tuning it either if it will show better results.  Let's say I don't have any images or spurs - just what will the preselector enhance otherwise?



"An OK RX -30 to -20 dBM, a good RX -20, Kick butt 100 db above the noise floor of the RX. "

Please explain what you mean here ??


T

This will tell you where the RX responds to out of band signals or overloads. If you had 2 generators you could do dynamic range testing. Another thing I try is use a hybrid combiner with the antenna on one port and the generator on the other input. Listen to a weak signal while you tune the generator around to see what it takes to trash the signal you want to hear.  A preselector on mercury would limit the input bandwidth and will filter out some of the broadband noise.

A preselector and a preamp will do nothing for internal spurs, birdies, oscillator phase noise and filter leakage.

By adding the preamp you will likely reduce the dynamic range of the receiver.

The FT-1000 has a preamp. It is on when the Front End control is set to the Normal position. It is out when this control is set to IPO position.

One advantage the HPSDR has is there's no LO, so no phase noise from the LO.  Most likely the clock from the A/D converter might contribute some to noise in the HPSDR however.  Nothing is noiseless  ;D

Phase noise from the LO will modulate the signal and can't be overcome by preselectors or preamps.  In some receivers the phase noise is so bad the hiss is apparent even on strong signals.  I'd bet that if you drove the Yaesu with a couple of HP-8640s it'd be roughly equivalent to the HPSDR.

Last, and this is purely conjecture on my part due to unfamiliarity with DSP algorithms, but perhaps another factor is the demodulator code and/or the I/Q nature of the analog input in a SDR might tend to minimize demodulated phase noise.

All: thank you for your comments.

Clarifying my earlier post:

The original post in this thread is focused on weak signal performance. The observation is that the HPSDR receiver can "hear" weaker signals than the Yaesu FT-1000D can hear.

If one uses a preamplifier that employs the same transistor technology as is used in the HPSDR, and if one incorporates a preselector to avoid dynamic range problems caused by strong off-frequency signals (if needed)... then one should be able to

a) amplify the incoming signal and the associated incoming noise by (for example) 10dB

b) achieve the same ratio of incoming signal-to-(incoming noise + preamplfier noise) at the output of the preamplifier as is achieved at the output of the HPSDR's first stage front end.

Once you have added the low noise preamplifier (presumably having somewhat lower preamplifier noise than any of the preamplifiers that are in the circa 1989 generation FT-1000D), you no longer have a "weak" signal coming into the FT-1000D.

It is reasonable to assume that all "birdies", mixer-induced phase noises, A/D conversion artifacts, etc. are introduced after the front end preamplifier... and would be present with both weak and strong incoming signals.

Therefore, if there is a difference in performance of the two receivers that is observable only with weak incoming signals... you should be able to overcome that with the use of a preamplfier employing the latest technology... as is employed in the front end stage of the HPSDR.

Stu

P.S. The reason that modern technology can enable the construction of a preamplifier with somewhat lower noise (maybe a fraction of the decibel or a decibel) vs. a 1989 design is that modern transistors have smaller feature sizes, and, in some cases, higher electron mobility... that yield higher gain-bandwidth products. This allows the use of low noise transimpedance amplifier designs whose noise figures (when driven by a 50 Ohm source signal having the equivalent noise of a 50 Ohm resistor at room temperature) are closer to 0dB. I don't know what the noise figure of the lowest noise FT-1000D preamplifer is at 30MHz ... but I would guess it is around 2dB. It is likely that a design incorporating the latest transistor technology would have a noise figure of less than 1dB.

My 1975 SB303 could hear down to -140 dBM with a tunable preselector. But it only had 85 dB of dynamic range after a lot of modification. Stock it was 67 dB. It took me another 15 years to build stuff in the mid 90s. Jay and I go back about 35 years in that game.
HUZ very easy to buld a high dynamic range preselector. My cubic will handle 1/2 watt. There is a guy in UK selling high dynamic range tunable preselectors
The point of the preselector is to limit the crud getting into the RX chain.
Yes the DDC function of the FPGA will introduce some distortion but in the case of HPSDR check the test data. It is not an I/Q mixer but a straight A/D sampling at 122 MHz. The I/Q is generated by the DDC. My TCI/BR 8174 does the same thing at IF.
I wonder if a simple low pass filter would help the fT1000 to limit lower frequency crud from getting into the front end.

A/Ds have jitter, which is roughly analogous to phase noise. If the HPSDR designers were smart, the A/D rate was selected so that during down conversion much of the noise is out of band.

If a preamp is not being overloaded, placing a filter in front of it will make no improvement. The FT-1000 has a bunch of front-end bandpass filters that are switched in for each ham band.

Using a preamp with a 1 dB NF is overkill on 20 meters. Unless you are using a very directional antenna, the incoming noise (galactic, atmospheric and man made) is equivalent to a noise figure well above 1 dB.

Typical rice box noise figure at 30MHz is 12-15dB. My souped up 75A4 is 6dB without a preamp and the performance differential is dramatic.

For the TS-940 where I replaced the IMD prone general purpose band range diodes with PIN's, I switch in a 1.5dB NF @ 11dB gain, high IP3 preamp, which brings the overall NF down to a managable level. An identical one can also be cascaded and has helped on a few occassions where sky noise on 10/12M was exceptionally low.

For 10M AM these days I use a souped up HRO-60 which hears better than the stock TS-940 or 950SD. I had a 1000D for awhile in the early 90's and got rid of it as the receiver performance was much worse than the TS-940.

The Racal RA6830 increases the noise floor when I connect it to the T12 on 10 meters. It also has about 12 dB noise floor. I would think it could use a few dB of gain on 10. Racal had a step attenuator on the front end piggy backed on the input board of the DF receivers so it would be easy mount a nice high dynamic range push pull Norton anp with a pair of 2N5109s in that spot. 

"There is a guy in UK selling high dynamic range tunable preselectors"

Frank,

Do you have a link to this bloke? Is it auto or manual?   I'm getting to the point of wanting to try a preselector since there is really nothing on the front end for selectivity in stock form. With the 40M stack, it's like having a freight train of BC stns rammed down the front end. I'm surprised the Mercury works as well as it does.

I should search the HPSDR archives, but has anyone implemented an available preselector with Mercury that is auto-controlled?  I'll bet some of the UK or German guys have already, but some of their creations can be homebrew nightmares to duplicate.



T

You need a Harris RF-551A pre/post selector.  These were used in full-duplex HF installations; the transmitter and receiver were spearated by 3 or 4 MHz... the exciter drove a RF-551A which drove a 1.5KW leenyar.  Another RF-551A was used ahead of the RF-590 receiver.

Then there's this little gem which I found when photogoogling for a RF-551A http://www.radiocronache.com/2009/11/rf-preselector-or-not-tiny-scr-digisel-vrf-rf-551a/

Tom,
Clifton Labs makes a nice BC filter Warren K2ORS has one. Check back through HPSDR for the link to the guy who makes the preselectors. They are not cheap and computer control. Not sure if anyone has interfaced it to HPSDR.
You should see the Flex 5000 at w1aw get trashed by BCI

My collins unit is like the 551. My cubic is all relay switched in 10 KHz steps.
they usually have a 10% bandwidth with skirts down about 45 dB or better.

I did some reading on the  Harris RF-551A and it looks like a winner. It says remote controlled via BCD.  What does this mean? Can it be easily interfaced to the computer/Mercury/PSDR software or will it need Mr. Goldberg's assistance?

How much do they go for and are they ever available? I see none on the web right now.



T

Go nuts and get a K&L - 5% BW. These can be controlled by RS-232, GPIB and BCD. BCD is binary coded decimal.

They're available from time to time, Tom, keep an eye peeled on eBay.

The BCD input interfaces with the RF-590 receiver or RF-1310 exciter.  It's fully usable on its own with the decade knobs though.  Selectivity is 10% at -60dB.

551 is a cool box. Also there is a tunable preselector module that goes inside the 590. Some 551s come with decade switches on the front panel.
I have powered up my cubic and it works quite well.

Keep in mind, even a 5% BW is still 700 kHz at 14 MHz. So, the preselector will do little or nothing for strong signals within the 20 meter band.

right but it will beat the crap out of a signal in the BC band or 10 MHz away.
Tom, La Pointe made a high Q preselector in a 3 foot rack. it was very sharp. I worked on the prototype  and it was very cool. It went on the deck of a ship. It had a wip antenna but we built adapters for coax input for testing and tuning. You could tune it from a remote control box. Imagine a couple of them into a pair of mercury receivers.
It had 1/4 inch copper tubing silver plated inductors.

If those take out your RX, it's junk.

well yes we are talking about the ft1000 but again you should see the flex 5000 freak out at w1aw and local bci

Maybe W1AW is getting beat up with WPOP, 5KW about a mile? away. And WTIC, 50KW maybe 5 miles?


Steve, about the 5% BW.  Lets say it's down 60db 700 khz away.  What would that roughly equate to at 50khz, 100khz  or 200khz away, 5-10db or so?  Maybe there is some small value on the same band if the offender is far enuff away.

Frank, a 3' high rack preselector? Sounds like a beast even Vortex Joe wouldn't haul into his cellar... ;D

Maybe I'll look into a simple triple 365 variable section cap and two coils for a multi pole filter, just like a tube RX front end..

T

Ive been using 2N5109 Nortons for about 30 years here, its one tuff transistor especially when you clip on a heat sink and push the current up.

Im using remote tuned PP U-310's in the 2 turn loops made from 1" CATV hardline and they have been fine, however the Q is very high and the tuning is sharp so overload has been impossible.

Yup, bias them up to about 60 ma each and use the large 3 fin radiator

Anyone have a workable circuit to build up?

This one was passed to me by a friend who has had very good luck with the 5109.

http://w7iuv.com/preamp50/preamp_r50.pdf

I'd like to put one at the feedpoint of each of my receive loops. There is a simple BCB reject filter circuit near the end of the pdf unless you are inclined to pursue the preselector path.

There are several nice ones on that CD I gave you.

Rob, Those amps will work ok but a norton will have higher dynamic range and less gain. I really like the push pull norton  set at about 10 dB of gain. Biased at 60 ma and you have to be careful because it will also make some power

For the Mercury, I'm curious why I need gain and an active device which gives away dynamic range and reduce IMD cleanliness?  Isn't the exisiting built-in preamp enough gain?   I was thinking of just three (inactive) tunable ganged L/C circuits in cascade. This shouldn't have much insertion loss I wud think.

I see Carl used two turns of 1" alum hardline as the coil. A super high Q with large caps seems like a winner for selectivity.

If I'm wrong then kick me in the GDB next time you're in Jersey.

T

BTW, on the subject of HPSDR, Mercury and receiving diversity, here's a post from a NE guy (Joe/K1RQG, a HossTraders founder) who uses Mercurys in diversity. He has some results that may interest guys with smaller lots or looking to do a big vertical array with null steering.


T


----------
David and all
 
I have been using K5SO diversity software for quite some time now, months, and
during this time I have transitioned from using 2 horizontal antennas (80 meters),
to using 3 selectable RF probes. I am trying to document the hardware aspect of
my setup and let the software speak for itself.
 
I am still experimenting with different antennas at this time, but bad weather has
slowed my progress. My initial tests with the 2 horizontal antennas on 80 meters
proved to be quite effective bringing DX signals "out of the noise" certainly by
minimizing louder signals from a different direction. Also found it quite effective
to eliminate "birdies" and "UFOs" caused by switching power supplies etc.
 
I have since gone to using 3 RF probes set up in an equal lateral triangle about
50 feet apart. These probes are nothing more than 3 inch square pieces of
double sided PC board connected directly to a high impedance unity gain
amplifier feeding RG-6 back to a software selectable (ALEX control) switching
system that not only provides any two of six selectable inputs, but also provides
RF protection for the Mercury boards. (The RF probes are about 20 feet in the air)
 
The beauty of the RF probes is that they are NOT resonant, and believe it or not,
they perform much better than any of my wire antennas in the Broadcast band
and work well from about 300 Khz to 20 meters.
 
During the daytime, I can null a local broadcast station (13.5 miles away) by more
than 50 dBm and this holds pretty steady. However as time progresses into
darkness and ground wave starts to fade and sky wave increases, the null goes
down somewhat to around 30 dBm.
 
Using diversity at night on 160 and 80 meters varies with propagation, and using the
same probes, I see anywhere from 10 dBm to 25 dBm null, but this is very dynamic
given the variations of the received sky wave.
 
My plan is to try some larger (20 foot) probes and evaluate them when the weather
allows. Then I will try 3 resonant verticals as well.
 
I find I use the RF probes on 80, 40 and 20 meters more than I use the full sized
antennas. They are very useful on 20 meters when I am talking to Europe and
NA stations at the same time, as I can rotate the diversity control to "peak" up
the station in one direction and just click 180 shift for the other. I do run full legal
limit on these bands, thus the need for protecting the Mercury boards even though
the RF probes are 100 yards from my transmit antenna.
 
In conclusion, I find the diversity software extremely useful, and intriguing to experiment
with. I also would like to hear from others about their experiences and how they
use it. If anyone has any other questions regarding my set up, I will be happy to
respond.
 
I want to thank K5SO & W5WC for their continuing improvements of the PowerSDR
software as applied to HPSDR, and of course thanks to FLEX radio and KD5TFD as
well.
 
73 de Joe
K1RQG

Cool stuff. I wonder how long it will be until they add this option for transmit?

Tom you don't need any extra gain unless a preselector ahead of mercury has a lot of loss. This will depend on the configuration and Q of the preselector you use. Many guys have been playing with series tuned circuits rather than parallel tuned tanks. I don't know which one is better. I have a QEX article somewhere by the guy in UK who is selling them. I would like to see what Steve sent you.
I've had a couple emails with Dallas Lankford the past few weeks and he is now using small loaded loops and prefers them over probes. You can't rotate a loop pattern though.
That 3 foot preselector had a vertical on top of it. three would be very cool

Tom, Im using 3 365pf TOKO varicaps in parallel and running the DC down the feedline. The coil is 1" Trilogy which uses poly spacers instead of continuous foam, it has the least C per foot. Its probably overkill and regular foam hardline would be OK for most users. Solid dielectric cables such as RG-8/58 have too much C and really ruin the Q; those are better for the casual or dimbulb SWL/BCL ;D

Why is coax used at all? You can easily make coils with a Q over 200 from wire. Some guys are making loading coils appraoching Qs of 600.

You have examples of these wire claims for loops?

Wire loops? Coils. I'm talking about an LC preselector. Talk to W6ANR, W7ZV, W6KW about hi-Q coils. All these guys are using them as loading coils in their transmit antennas. W6ANR claims a Q of 650. If he was FOS, it would have been exposed by now. Measurements on numerous coil types at the link below.

http://www.w8ji.com/loading_inductors.htm

http://www.herostechnology.co.uk/pages/tiny_SCR_preselector.html

Check out this preselector from the UK that will work with HPSDR remotely.  Costs about 120 British Pounds. 160-10M, no 6M though.  Tiny, not like a 3' high rack... :-)

Frank found it yesterday.   It might do the trick.  Notice the other offerings at higher prices.

T

Tom,
It looks like they have another model coming out soon. They did a QEX article on the design and would be easy to duplicate if you wanted to tune manually. Heck you could gear a variable inductor to a variable cap and do the same thing. The trick is they transform down to a low Z and go through a series tuned circuit then a second transformer to get back to 50 ohms. The specs look pretty good until you get to the highest band which seems a bit broad. Two filter sections with a push pull low gain Norton in the middle would kick butt. A kit would be easy to build.
I bet my MSR6300 is better.
The cool thing is it can be set up to track Power SDR

There is nothing new there about solenoid or toroid coils. Ive been using Q Meters at work or home since the early 80's and others found similar long before that.

What I was referring to is a balanced receiving loop and you will never get high Q with the over the counter RG stuff.

OK. I thought you were talking about a coil. You're making an coax loop with a split shield, right?  Wouldn't you get just as high a Q with a wire/copper tubing loop?

Yes, several have made their own coax loops using copper pipe, fittings and wire which is all sorta getting pricey these days. Whatever is used the idea is to get the C as low as possible if you want selectivity. The 2 turn balanced loop with a PP Jfet preamp gives better performance than a single unbalanced turn.

Forming pipe into a 2 turn loop could be a challenge ::)

Would a multi-turn loop remove the need for copper tubing?

Copper tubing or "pipe"  is quite easily formed into loop configuration.  Using basic tools this is easily done in the home shop.

Anything larger than 0.5 inch is overkill anyway, especially for RX.

Multi turn open loops dont have the noise reduction or selectivity, they are OK for general purpose use and go back to the 19teens.

Multi wire shielded loops get right back to the high C issue and are fine for direction finding as done in WW2.

At least with CATV hardline you dont need a pipe bender, and other tools to be a plumber...plus its free.

My preference was the need to minimize noise bandwidth as well as offer enough selectivity to reduce QRM and other crud on 160 during pileups and contests. This also works for BCB DXing and the noise reduction is good for the 2 experimenters bands.

Dig this test:  There is a unique reason for needing a preselector on the Mercury receiver.  At night, with the triple 40M Yagi stack  connected to Mercury listening Europe on 20M, the ADC goes into overload. The noise floor goes up at least 25db. There are SW BC images all over the 20M band from 40M. This is at night when the SW BC sigs are crushing, of course, and during the day it is quiet. This is with Mercury's preamp on.   The same thing happens on the higher bands, 10-15M too.
 
This is a VERY demanding test, of course, with BC signals 60+ over. But there is still some SW 40M BC stuff in there when using the 20M stack, though very weak.  The preselector would probably cure this completely.  I think the existing Merc front end (as little as there is) is right on the brink and could be improved, especially when the Merc's preamp needs to be on for 10-20M..

With the same test, my own FT-1000D does NOT have 40M BC images on 10-20M listening on the 40M stack at night. Huge difference.

I'm not complaining - just pointing out that a preselector is a very good investment for the Merc.

I got an email from a guy who said the UK preselctor we discussed will fit the Merc to a "T".

http://www.herostechnology.co.uk/pages/tiny_SCR_preselector.html

T

Tom,
The Alex filter may be enough in your case. Actually the FT1000 has a filter system is similar to Alex. Look at the heros schematic. It is dirt simple and could be easily duplicated with two BB transformers , a tapped inductor and a variable cap.
The cool thing is that SCR will track Power SDR. I see it has not been updated to handle the Metis interface but that will come. It would be cool to interface 1 controller board to two filters so diversity is possible. I bet it would just take a different ribbon cable.

Consider using a Taylor hybrid. It will improve things greatly. I've added one to the FT-1000. The difference is night and day.

Last night I checked the 10 meter sensitivity using an 80 meter dipole. When I connected the antenna the background noise came up. This tells me there is plenty of sensitivity since I run my Mercury with a 6 dB pad on the input.
Without the pad would be even better. You do need to select the preamplifier "on" though.
I didn't think to check 6 meters but the RX noise floor is sitting around -140 dBM. (-107 dBM is 1uV)

Try the Mercury on 20M with the 40M ant connected at night. Preamp on.

No preamp = quiet. With preamp on many images and noisy 40M BC crud.


I built a new QIX overload board for the 1000D and ready to test.   The LP filter controller board from TX got lost in the USPS mail. Just my luck.

Added digital readout to voltage and current on the 1000D amp.   Coming along.

T

I'll give that a try over the weekend.
Almost ready to wire my power supply for HPSDR Erb final. Just a few more holes. Decided to run aluminum strap across the 5 filter caps on the 45 volt supply. Man it will be a packed box but will deliver over 1KW of power with a 24 volt 3 amp supply for predriver and control. All in a 5 inch rack.

Copper tubing and/or pipe are easily worked into loops with common hand tools . Loops fabricated from tubing or pipe require minimal support from added brackets and braces.  Plywood forms are convenient fixtures for pulling the tubing into the required shape.

  The techniques are more related to Metal Model Making than the Plumbing trades.