Even More KISS Mixer Musings

Well,

The last time I fiddled with the KISS mixer from the Minima was back here:

http://g0mgx.blogspot.co.uk/2014/02/kiss-mixer-musings.html

Now, lots of QRPers have fiddled with this since and the musings of all of us have gone into the brain of Ashhar Farhan, VU2ESE and this has popped out as an improved design:

So, here's a first bash at giving it a try.

To test this properly I need to tap into the Si570 LO inside my minima so that the LO levels are as they will be, but for now I am trying something basic using the bench signal generators I have here.

So, for this test my LO will be at 27Mhz and generated by my EMRFD Sig Gen. Here's the signal into a 50R terminated 'scope:

For the RF signal I will use my DDS Sig Gen at 7MHz. Here's that signal into a 50R terminated 'scope:

So the LO signal is connected to the LO port in the schematic above (that's the port on the far left), the RF signal is connected to the RF port (that's the one at the bottom right) and the Spectrum Analyser is connected to the IF port (that's the only one we haven't used yet). Here's a picture of the mixer:

So, this is what we see on the Spectrum Analyser sweeping 0 to 100MHz:

The 10K pot is sat in a central position and I haven't made any attempt to fiddle with it. If I do have a fiddle then I can reduce the signals at marker 1 which is the RF port input at 7Mhz quite significantly (I can also make it much bigger too):

So, in the above image we have the following:

I now need to take my minima to bits and use the Si570 as the LO signal..... i also need to go back and review the results from the last tests on the previous incarnation of this mixer....

A quick check with the IF and RF port connections swapped round suggests a good isolation between the LO and RF ports of something like 45dBm, however the harmonic at 54 MHz is stronger and needs investigation:

If I now increase the LO drive level, thus, by adding a general RF amp to the output of the sig gen:

We now see the following with the initial configuration (Spectrum Analyser connected to IF port), just to keep you on your toes the SA is now sweeping from 0 to 80MHz.:

And with the SA connected to the RF port:

In both of the last two tests the pot in the circuit has allowed me to reduce the 7MHz signal drastically.

Finally, if we swap things about a bit and put the LO at 27MHz amplitude 14dBm and a 20MHz signal amplitude -5dBm into IF Port and SA on RF port we can go from this (just to keep you on your toes the SA is now sweeping from 0 to 80MHz):

to this:

by adjusting the pot in the schematic. It looks like a multi turn pot would be a most excellent idea as the adjustment is very critical but the 27MHz LO signal can be almost nulled out completely as we could with the 7 MHz LO signal in the earlier example.

We need to test this at the LO frequency range of the Minima and with the harmonic and other content of the Si570, but it does rather look like this mixer with a single lpf at 30MHz and just one trap at 20MHz will do just nicely.

Cat's not been helping much:

He did raise an eyebrow due to the excitement when the final SA image with the nice null was saved but soon settled again, he does like an RSGB book as a pillow

Fun though, egh?

More Mixer Musings

Well,

Following my total confusion over the KISS mixer, I decided to go back to basics and make certain my very little brain:

http://en.wikipedia.org/wiki/Winnie-the-Pooh

was up to speed and understanding the diode ring mixer as a starting point.

So, I made one of these:

Please accept my apologies for the simply appalling quality of the image above.

Now, the LO looks like this into 50R:

I make that 1.25V peak to peak which is 5.85dBm.

The RF input looks like this into 50R:

thats a whopping 2.36V peak to peak which is 11.44dBm (that's way too big!).

The output of the mixer looks like this (after a lot of fiddling with the 'scope):

So hopefully it's clear that ever time the LO (we could call it a switching signal) is negative the RF input is inverted on the IF output. If you then apply a filter to the IF output we can find a nice clean signal at the wanted frequency.

Now, if I keep the amplitudes as they are above, this is the output of the diode mixer on the SA:

I've put some markers on some notable signals; the 5.85dBm LO signal is visible at the output port at -32dBm. The RF input signal at 11.44dBm is at the output port at -20dBm and my wanted signals are at 1.7dBm. I also placed a marker on the harmonic at 40MHz - that's at -38.5dBm.

If I now add 30dB of attenuation to the RF signal, the output looks like this:

So now, the 5.85dBm LO signal is visible at the output port at -32dBm (no change), but my RF input signal now at -18.5dBm is visible at the output at -42dBm. My wanted signals are at -23dBm and my random harmonic is at -50dBm.

So, back to the KISS mixer confusion:

Let's look at exactly the same input conditions with the KISS mixer configured as per the schematic:

So, here we have the LO signal at 5.85dBm, the RF at 11.44dBm, with these input signals the bias pot has no effect at all that I can see:

and here, the LO signal remains at 5.85dBm but the RF signal is at -18.5dBm, with these signal levels I can make a slight change to the amplitude of the LO signal in the output with the bias pot - I have set it for minimum LO:

What does all this tell me? Primarily that I need to go out more!

I'm going to let those far cleverer than I draw the conclusions.

Interesting, egh?

KISS Mixer Musings

Well,

Part of the Minima transceiver I made back here:

http://g0mgx.blogspot.co.uk/2014/01/my-minima-is-alive-and-well.html

was the so called KISS mixer:

Above I have stolen the mixer part from the Minima schematic and labelled the ports. In my test setup I have the LO connected to a test crystal oscillator I had lying about, it's running at 20MHz (ish), the RF port is connected to my bench signal generator and the IF port is connected to the Spectrum Analyser.

Here's what the test setup looks like:

Now, the author of the Minima Farhan gave us some information on the KISS mixer here:

http://www.hfsignals.org/index.php/Minima#The_KISS_Mixer

and also a reference here:

http://www.phonestack.com/farhan/mixermusings.pdf

Now, here is what I see when I have a play. I have 20MHz LO, 15MHz set into the RF port from my bench sig gen and the spectrum analyser connected to the IF port.

Here's the output spanning 0 through 100MHz with just the LO (20MHz) signal:

This is just my oscillator and it's harmonics - so far so good. Now lets add in the 15MHz signal to the RF port and keep the 100MHz span:

So now we see the 20MHz signal plus harmonics, the 15MHz signal plus harmonics plus all the differences and sums which land at every 5MHz interval. So far so good.

Now, lets zoom in on one of the mixed signals at 35MHz:

Looks good to me.

Now, if I connect my x10 'scope probes to the gates of the FETs, here is what we see:

The 'scope struggles to trigger on anything here, because there is all sorts of frequencies in the signal, however, I think we can safely say that the two signals are the same but out of phase with each other as we would expect.

Now, if I look at the drains (my sources are connected together and the bias supply), we see this:

and if I add the LPF in the scope to bandwidth restrict the inputs:

If we replace the SA on the IF port with a 50R load, then look at the signals on the drains with the SA, here's drain one:

and here is drain two:

Now, in all of the tests above I have tried altering the bias pot and I can see no difference at all under any circumstances. If I look at the voltage at the wiper of the pot I can see 0-5V as expected. However the voltage at the sources read from -0.12 to -0.15V. That isn't what I expected either! If I remove the +5V supply the output doesn't change nor does it if I connect the sources to ground.

The FETs in the KISS mixer effectively replace the diodes in a more traditional diode ring mixer; this chops the RF signal thus providing the IF signal at the centre tap of the RF transformer.

So I replaced the single bias pot with one on each FET, under these circumstances I can alter the amplitude of the unwanted signals in the output but not make any difference to the wanted signals.

In my case the wanted signal is the 20 + 15MHz, here is the mixer output with the two bias pots adjusted to minimise the unwanted signals:

However, if I just remove the 5V line, the bias pots and just ground the two sources on the FETs, here's the output:

And these are the signals on the drains with no bias (the sources grounded):

Any volunteers to explain?

My Minima Progresses

Well,

Following a whistle-stop tour of the Middle East, I am back and have done some more with my Minima. After much deliberation and head scratching, I have decided to box the linear amplifier and Low Pass Filters separately. This way I can re-use the amp later for other projects should I decide to do so.

So here's the finished Minima:

You will see that I have used a DIN socket to take the BCD band selection information from the processor to the rear panel, this will allow me to automatically switch the LPFs in the linear as the radio tunes. I have also added a simple PTT line to the rear panel to switch the linear into TX.

Here's the linear and filters in a state of incompleteness:

So, the BCD output is configured in accordance with this table:

I've included 6 low pass filters with the frequencies listed in the column TX LPF above. So as the radio tunes when the frequency passes one of the thresholds the output binary value is set on the three lines I have attached to the DIN socket. The only thing to note is that the 19MHZ TX filters is engaged with BOTH the 15MHz and 30MHz TX filter. This has meant I have to use a simple diode matrix on the output of the 4028 CMOS chip to enable the same filter on two different outputs. So in the linear box we have the 4028 CMOS BCD converter plus a DIL package with darlington pairs to physically switch the relays.

If you need a schematic of the linear control wiring that takes the BCD signals from the processor and switches the LPFs email me and I will create one.

I've made a few modifications to the software, including the logic needed for my LPF switching in the linear. Rather than explain all the details I have simply uploaded my code here:

http://www.qsl.net/g/g0mgx//files/Minima_MJ_Final.ino

If you were to use the most excellent utility Beyond Compare to create a difference listing between my version and the original, you would be able to see the changes very clearly.

So, on with the linear....

Good, egh?

JBoT - A linear for Minima

Well,

In an attempt to finish up my Minima tranciever, I have today been making the linear amplifier to follow the gubbins we have so far.

The Author has used this:

http://www.hfsignals.org/index.php/File:Qrp-linear.png

and I saw no reason not to follow suit.

Here's what the linear looks like right now:

Mine is built with 2N3053s (because I had some). I had a spot of bother getting the correct output from T2, this should produce 2 equal amplitude but out of phase signals. I eventually managed to make something that seemed to work, the output of T2 looks like this:

and the output of the final transformer after the 4 finals looks like this into a dummy load:

So, I make 56V peak to peak about 7.8 watts, so I suspect I am driving the amplifier a bit hard with my sig gen and attenuator set up. I also think I have a bit of a tendency for the finals to oscillate.

So, we will have to see. The next bit I need to do is understand the band selection outputs from the Minima software - I am sure these are designed to allow me to switch LPFs after the linear, which I need to do as the linear will have to be followed by suitable filters to remove any harmonic content.

Good progress though, egh?  

Minima - Remaking the 15MHz LPF

Well,

Now that I have some life in my Minima, I need to remake the 15MHz Low Pass Filer. You may remember the sweep I did originally:

There is a bit too much attenuation at the 14-15MHz frequency range - this will result in a very deaf radio on the 20M band!

So I used the QUCS tool to design an 11th Order filter with a cut off at 15MHz, I then changed the capacitor values to "real world" numbers and ended up with this:

I then made the filter and after winding each toroid I measured the value and carefully spaced the turns to get as close to the required value as I could.

The physical construction looks like this - all the toroids are T50-6s:

and when I sweep the filter on the Spectrum Analyser I see this:

So this looks much more like what we need - I will use this and replace the one in the radio right now.

Good, egh?

And at that point - it ground to a halt!

Well,

Here's how my Minima is looking on the bench right now:

I've left just enough room between the front mounted processor and associated parts and the boards on the base to fit a vertical RF screen if I find I need it:

And here is the back of the case with a reset switch, DC inlet, antenna socket and connector for the USB serial adaptor to connect.

Unfortunately I have to return to working for a living tomorrow and that means Stockholm this week, so progress will now have to halt for a while.

I hope to have enough room on the base of the case to the rear for the PA and also the right hand wall will be used for the output filters. I have made the RS232 interface and that's on the back wall of the case.

I will finish this when I return....

Watch this space.