Better than mine, the SMA is a much better connector than the SO239 I used.

I suppose if we had the base of the enclosures made from a piece of copper then the heat would spread much more easily and so not get so hot.

Yours is neater than mine anyway :)

This is my attempt.? Different design due to different scrap material in the garage and SMA scavenged from some unidentified thing from the club junk box.? 100W element.? Dimensions are 35x35x15mm.??







By itself 10W is about the limit as it gets HOT.? Attached to a scrap CPU heatsink the hottest part gets to 70C at a continuous 100W, so well within the 100C limit from the datasheet.



The cover was left off for the tests so the resistor flange temperature could be measured.

Thanks for the prompt into action!? Nice to do some mechanical stuff.

73

Dave M0WID

Information my load R50 -?#1561
------------------------------------------------------------------------------------------------

I like that Igor, that's nice !

You'll have to show us a smith plot.

Having a lathe must be extremely handy/useful for all sorts of work.

I've used my 50W load element (with the heat sink arrangement I showed) with nearly 200W PEP SSB without problem, though I think the whole heat sink gets quite hot on 100W continuous. So these little 50R elements can take quite a bit of power.

After calibrating the VNA with the usual 50R SMA dummy load in place of the big one, this is what I get. The dummy load SO239 definitely needs to be an N-Type socket.

Even better without the short coax.

Need to replace the S0239 socket with an N-Type and use a straight SMA to N-Type to the dummy load to improve it I think.

This is from 10kHz to 900MHz, considering it's going through an SO239 socket, PL259 plug, BNC socket, BNC plug, SMA socket, SMA plug, another SMA plug and finally another SMA socket, I don't think it's doing too bad really ..

On Fri, Sep 4, 2020 at 12:20 PM, Dragan Milivojevic wrote:

That is a very nice construction. Do you have a RL measurement for it?

The SO239 socket really needs to be an N-Type socket. The PCB track from socket to 50R element should be 50R (3mm track on 1.6mm FG) so that ought to be OK but the measurement just done (as below) it's not as good as was hoped for, but is most likely down to the liberal use of S0239 and PL259 connectors in this test.

The sweep is 50kHz to 1.5GHz - the green smith chart line.

Very neat.? I think I have some scrap aluminium in the garage that has been waiting around 10 years to become useful, and there is a similar 50R in the component box.? A nice project for the weekend, in between g-qrp club convention sessions.? Thanks for sharing it.

Dave

Here's one we did earlier (50W 50R load element) ..

Dave,

The commercial tinySA has a correction table for the RSSI of each of the 57 possible RBW
--
HBTE Files section:?/g/HBTE/files
Erik, PD0EK

Observing a few odd things and following on from the post in the tinySA group where the 2.6kHz filter was removed I added a test mode to my ESP32 version to test the FIR filters, I called it RX Sweep, and added a trace store facility to the web page.

The test mode uses a fixed LO set and fixed input - I used the 15MHz -30dBm reference, and varies the RX frequency.? In this case the IF is set to 434.12 which is at the upper end of the SAW filter passband and minimises the effect of the IF breakthrough at low frequencies.? The results are essentially the same at different IF in the SAW filter passband and anyway the signal to the receiver is fixed frequency and strength for this test.

Results for the 2.6kHz(light blue - store 1), 2.8kHz(orange - store 2), 3.1kHz(purple - store 3), 3.2kHz(green - store 4) and 3.7kHz (blue) filters are shown below:




Some interesting observations.? Most significant is that the passband is offset from the nominal IF on these narrow filters, and some of the filters have significant ripple (2dB on the green 3.2kHz filter).? It looks like there is some benefit from using the 2.6kHz filter


Here is another capture, this time? 112.1kHz(light blue - store 1), 181.1kHz(orange - store 2), 225.1kHz(purple - store 3), 248.8kHz(green - store 4) and 335.5kHz (blue) filters?.



By now the filters are centred on the IF but there are significant differences in the RSSI values for different filters, some filters give a nice flat passband response, others have significant dip mid band.

I am using cheap modules ex China - I wonder if this is repeatable from chip to chip or if these devices are cast offs that have failed a test in some way and found their way onto the usual auction sites?

Either way looks like some form of self calibration to correct the RSSI value for different filters is needed, and an offset in IF or LO to shift into the passband at narrow filters.? I expect this is already included in the tinySA product?

73

Dave M0WID

开云体育

Thats a winner...

On Sun, Aug 30, 2020 at 07:22 PM, Glenn wrote:

Here's a plot a 20M BPF I had in my junque box.?? Comparing "simpleSA" with a Rigole SA-TG. Same span (15MHz.)

The simple SA is not quite as good, but its pretty close for a cheap instrument.

Glen,
? ?Thanks for the reply.? Considering the cost difference between the Rigol and the simpleSA, I think your hardware build is well worth the effort.? I am especially impressed by how smooth and noise free the tracking waveform is.

? ?I own Erik's tinySA and was looking into a similar tracking set-up with it.? Erik has suggested investing in an external amplifier and attenuator at the LO output to increase the dynamic range for tracking generator applications.? I wanted to see what you were achieving before investing in the additional components.

- Herb

Here's a plot a 20M BPF I had in my junque box.?? Comparing "simpleSA" with a Rigole SA-TG. Same span (15MHz.)

The simple SA is not quite as good, but its pretty close for a cheap instrument.

EDIT:- the simpleSA top of screen is actual "0dB" point in this case for tracking.?? So you can compare BPF losses.

glenn
vk3pe

What kind of dynamic range are you seeing with the tracking generator add-on module?? One of the advantages of using the HP856x series spectrum analyzers with a HP8444 tracking generator was the LO tracking allowed a much greater dynamic range than a similar stepped sweep using a CW generator.
? A plot of a bandpass filter with 70dB or better skirts would be appreciated.
- Herb?

What kind of dynamic range are you seeing with the tracking generator add-on module?? One of the advantages of using the HP856x series spectrum analyzers with a HP8444 tracking generator was the LO tracking allowed a much greater dynamic range than a similar stepped sweep using a CW generator.

? A plot of a bandpass filter with 70dB or better skirts would be appreciated.

- Herb??