Take a look at the simple little mixer above. I think I first saw it in SPRAT. Thinking that it was really just a simplified version of the two diode Doug DeMaw mixer that I had been using for years, I couple of years ago I built it into a little Direct Conversion receiver. It worked great. But later, I began to have doubts about it. In the words of young James Clerk Maxwell, I started to wonder about "the particular go of it."
You see, the way the DeMaw mixer is set up, both of the diodes are simultaneously on and off. This has the effect of "chopping up" the incoming RF at a rate set by the VFO frequency. Boom. Fournier. Mixing. Great.
But look at the mixer at the top of this post. Here the VFO signal is coming in on the wiper of the 1k pot. The same signal is hitting both diodes at the same time. The diodes are not being fed differentially. So D1 and D2 are NOT both simultaneously tuning on and off. Instead, when the wiper goes positive, D2 turns on while D1 is off. On negative swings of the voltage at the wiper, D1 turns on while D2 is off. For me, this made it a "mystery mixer."
This reminded me of the sub-harmonic DC receiver I built earlier in the year: The VFO runs at half the operating frequency, but the diodes are set up to switch on and sample the RF TWICE each VFO cycle. This is the equivalent of having the VFO at the operating frequency.
Could it be that this was just a sub-harmonic mixer with the VFO at the operating frequency? (I should note that Doug DeMaw published a design that actually made this mistake. See: https://soldersmoke.blogspot.com/2011/07/doug-demay-and-polyakov.html ) I knew that this would sort of work, but it would not work very well. And the mystery mixer seemed to work very well. Hmmm.
I was loaning the DC receiver with the mystery mixer in it to a local high school. I worried that I was loaning them something that I didn't really understand. I remembered that I'd been trying to figure out this mixer since early 2021: https://soldersmoke.blogspot.com/2021/02/some-thoughts-on-singly-balanced-mixers.html
Our beloved book, Solid State Design for the Radio Amateur (SSDRA) has an explanation of this circuit on page 74. But this explanation didn't seen to work for me. Check it out. YMMV.
Bottom line: I still couldn't figure this circuit out, so left it alone for while.
The other day I woke up and looked at it with fresh eyes. Suddenly it hit me. Although the VFO was hitting the diodes in the same non-differential way as is done in the sub-harmonic mixer, the RF (signal) is entering the mixer in a differential way. This means that the two diodes are taking turns sampling the upper side of L2, then bottom side of L2, via L1 and L2. This results in a complex repeating waveform that is similar to that of diode ring mixer. Within that complex repeating waveform, there are sum and difference frequencies. I did some noodling on this:
The key difference between this mixer and the sub-harmonic mixer is the way L2 is positioned: In the sub-harmonic mixer, there is no differential feed of the RF. Both diodes get the same polarity of RF. The VFO switches on D1, then D2. The RF is sampled at twice the VFO frequency. But in the mystery mixer that had me scratching my head, the RF is fed to the diodes in differential form. So while the diodes here are -- as in the sub-harmonic mixer -- being switched on and off sequentially, they are taking turns sampling the top and the bottom of L2. That provides the complex repeating waveform that we need to get the sum and difference frequencies. In a DC receiver the difference frequency is audio.
What do you guys think? Do I have this right? How would you characterize this mixer: Is it multiplying by 1 and 0? Or is it multiplying by 1 and -1?
This would be good mixer for a school project. It is simpler than a mixer with a tri-filar toroid.
Posts
Right on the nail, Bill! *1 &*-1, just like the 'ring'! Audio level will be less, with the RF 'singly-fed' vs the trifilar toroid's transformer ratio, but that can be overcome by simple turns ratio: your L2 is not restricted to simple multiple of toroid winding. L1 needs to be resonant at RF if tuned by variable capacitor, to keep L2 balanced, but permeability tuning of L2 might be workable. Set your school class to work on that?
ReplyDeletePART 1
ReplyDeleteI see Doug’s and the other mixers as fundamental mixers.
While most mixers can technically work as harmonic mixers, we apply that particular anti-parallel diode pair in subharmonic mixers (or let’s just say for harmonic mixers) since they offer the advantage of suppressing fundamental and other odd harmonic mixing products as well as even harmonics of the local oscillator. The diodes are switching ON and OFF from harmonic energy of the LO /mixer system.
I've never seen that SPRAT mixer, however, it seems a reduced version of a simple popular product detector used in the 1970s.
First off – let’s also say that diodes are sampling switches that get turned ON or OFF by the polarity of the LO signal. E.g. on each ½ of the LO AC signal the diodes go either forward or reverse biased so they function as switches whether in parallel, series ,or anti-parallel assuming there is enough LO to fully bias them ON or OFF.
We may place our diode(s) in multiplier circuits that offer balance and symmetry to achieve certain effects including suppression of even order mixer products and to provide port isolation. We enjoy many ways to do this, and some are simple, novelty grade versions for low-complexity receivers.
A center tapped secondary transformer is the common way to provide mixer balance – 2 are needed to get double balance in most passive mixers. Let’s take the double balanced diode ring mixer and focus on the secondary winding of each transformer. Essentially, we have two bifilar transformers: The LO transformer T1 has its center tap connected to ground, while the other transformer T2 does not --- the center tap is the IF PORT for T2.
The T1 bifilar wound transformer’s centre tap provides a low impedance path to ground and each arm carries the local oscillator signal differentially (180 degrees out of phase from 1 another) for each polarity change of the LO signal.
Now we add in the primary windings so each transformer is now trifilar. The left-sided primary of T1 goes to the LO and the right-sided primary winding of T2 is the RF port.
In a diode ring mixer, when the LO signal goes positive, 2 diodes turn ON and 2 diodes switch OFF. This unbalances the right-sided secondary transformer and provides a ground return for the IF port through the center tap of T1’s secondary.
When the LO signal goes negative, the opposite 2 diode pairs turn ON and OFF which once again unbalances the T2 secondary winding and once again gives the IF Port ground return through the T1 secondary center tap. All the while, the RF port signal passes through to the IF port alternating 180 degrees out of phase from T2’s winding resulting in a NULL for RF at the IF port. Balance within the switching diode ring also suppresses the IF from reaching the RF port
Since T1’s secondary tap is at ground or 0 volts, the potential for the LO to appear at the IF port is also 0.
Thus from a SYMMETRY STANDPOINT, some authors (like Wes in SSD) talk about the secondary transformers as having no voltage difference between various nodes where the T1 and T2 differential secondary windings connect to the diode quad ring ---- so theoretically these voltage are all the same and = 0. They refer to this in the context of port isolation via balance of the diodes in the ring + the transformer symmetry/balance.
Again, this is nuanced and theoretical, however, pretty decent port-port isolation ensues in a well-designed mixer, My personal measurements show commercial diode ring mixers enjoy a 20 dB or more boosted isolation between LO to RF and LO to IF than my home
brew diode rings mixers.
Part 2
ReplyDeleteBack to the SPRAT Mixer
My version of this product detector comes from reading done long ago – Rather than use a trifilar transformer, you may just use a conventional transformer with the primary as the RF port and the secondary 2 wires forming 2 arms that do no connect to ground. Just like the SPRAT mixer shown.
Each arm gets a diode that is placed anti-parallel to the other just like the SPRAT mixer.
To the right of the diodes , two 470pF caps get connected to each arm at 1 end and then each other at the other end (caps in series). The BFO is connected between them so that the BFO signal will go to each diode and switch them ON and OFF oppositely. Finally, just like the SPRAT figure -- on the left side of the diodes, near the secondary winding, a 1K pot goes across each arm and the wiper goes to ground. Thus you can adjust the balance of this product detector using a pot. The pot also provides the important DC return path for the diodes.
It seems, they have removed the 470pF capacitors and lifted the wiper and just used this to feed the BFO signal to the diodes with some potential to balance the 2 sides via pot adjustment. Where is the DC return path? I’m sure it works, but how well? Where are the measures to show balance (port isolation)?
Again, this might product detector work fine for low-complexity receivers. Switching is normal like a single balanced diode pair --- but balance is likely reduced compared to a conventional, transformer single-balanced 2 diode mixer with a proper DC return path.
You can always make a bifilar transformer for differential feeding of a mixer or anything really -- and put a few links around it for a primary winding. This is a wonderful compromise to trifilar transformers ----- and may give good 0 to 180 degree balance from my experiments.
Best to All!
Thanks Vasily.
ReplyDeleteOn Doug's mixer (figure 5-3 page111 "W1FB's Design Notebook" I have never seen this kind of subharmonic mixer used with the VFO at the operating frequency. DeMaw said he was presenting a simple DC RX to familiarize people with how they operate. This use (misuse?) of this kind of exotic mixer might not have been the best choice.
Regarding the SSDRA and SPRAT mixers, yes the SPRAT version seems to omit a needed DC return. I see that in the SSDRA version there is a resistor to ground. Maybe the SPRAT version worked because the author had a large cap to ground. The author of the SPRAT article was quite innovative in building the transformer: he just glued two chokes together! One was the primary, the other the secondary! Nice.
But what do you think of my analysis of how the diodes switch (at the VFO rate) between the top and the bottom of L2? Will this result in a complex repeating waveform with the sum and difference products in there? Is that how this circuit really works? 73 Bill
Hi Bill.
DeleteRE: W1FB’s D Notebook --- Page 111. I didn’t remember this receiver. About as bare bones as it gets! Who really knows, but I prefer to think Doug was trying to go uber minimalist here and this might be unintentional error --- he wrote in 7-7.3 MHz and designed the VBFO for that instead of ½ that value as he might have originally intended. I too make many such errors when drafting my schematics and writing up stuff. It’s odd to operate a LO at ½ or ¼ of the RF value when you are used to fundamental mixing.
It seems cool that he included the APDP product detector and cleverly drives it with a tap in his input tank secondary -- and this also provides a DC path to keep the AP diode pair biased properly.
Good work on your diagram Bill – the operation and output looks accurate to me.
73 OM
Bill, look at Sprat 35, 1983, your mixer is there
ReplyDeleteW1FB's Design Notebook, pg 51, fig 3-1, has your mixer
ReplyDeleteYea, I think he has it correctly there. But in that same book look at the receiver he has on page 111 figure 5-3. That is a subharmonic mixer, but he is running the VFO at the operating frequency. I think that is wrong, a mistake. 73 Bill
Delete