In an earlier posting I suggested that this new mixer might be a drop-in replacement for the much-discussed J-KISS. In the interest of moving the project along, I'd like to suggest that we now have two alternatives, one that functions (albeit difficult to reproduce and of unproven performance) and one that tests out very well on the bench. From a functional standpoint, they're pin-compatible at the block diagram level. I want to suggest once again that the board developers move forward as though the mixer issue is resolved, but put provisions on the board so that a 'daughter board' replacement mixer can be put in if/when it proves to be better. Actually, it wouldn't be bad to make provisions for doing this at every stage. The actual circuit implementations are likely to change regularly, but the block diagram is (I think) sound. If this is an acceptable bridge between present roadblocks and the long-term roadmap, it will allow us to get a set of boards out that more folks can build and improve on without fear of obsolescence. If we're going to do this, we should agree on the number/function of the pins required for each daughter-board. Joe ----- Original Message ----- From: Joe Rocci To: minima@xxxxxxxxxxxxx Sent: Wednesday, August 27, 2014 7:28 AM Subject: [minima] Re: New Mixer Design Farhan I believe there's no 1st order difference in the placement of the switches, primarily because it's the same circuit either way, and either way the FETs have to handle all the 'signal' currents (except the LO drive). It's all a matter of where you place the AC ground point in the schematic. In the KISS, the AC ground is placed at the switch common point and the output is taken from the transformer CT. You could just as well have AC grounded the CT and take the output from the switches' common point. In this case, it sure looks a lot like a diode single balanced mixer, doesn't it? This equivalency holds for other common circuits too. For instance, if you take a bipolar colpitts oscillator and move the common AC ground point from the collector to the emitter, then stare at it for a few seconds, you'll realize that it now looks like a Pierce oscillator. AC ground points are often moved around in the interest of easier ground-referenced drive or load coupling. In any case, the loop currents in the circuit remain essentially the same. In the KISS mixer, the output current loop is formed by whichever switch is on at the instant, the half of the secondary winding it connects to, and the load. This is the same current whether the switch is ground referenced or floating....just a difference in the common point reference. However, in the quad arrangement, you have true 'push-pull' operation which halves the current that each switch has to handle and improves balance and symmetry. There might be higher order effects from changing the reference point, such as IP3 (which is finicky about just about anything), but from a 1st order point of view, I think they're equivalent. Joe ----- Original Message ----- From: Ashhar Farhan To: minima@xxxxxxxxxxxxx Sent: Tuesday, August 26, 2014 9:49 PM Subject: [minima] Re: New Mixer Design joe, the earlier IMD setup couldn't be used beyond 20dbm. i am attaching a simpler and better design that i am currently using. this is excellent work, it points to something that we secretly suspected all along : the placement of the switches, either in the signal's path or on the grounding leg of the transformer, isn't going to make much of a difference. - f On Wed, Aug 27, 2014 at 2:44 AM, Joe Rocci <joe@xxxxxxxxxx> wrote: For reference on my previous, here's a scan of my return loss bridge terminated in a precision 50 ohm load. It's about 50 dB over most of the HF range, so no questions about the accuracy of my previous RL scans. Joe -------- Original Message -------- Subject: RE: [minima] Re: New Mixer Design From: "Joe Rocci" <joe@xxxxxxxxxx> Date: Tue, August 26, 2014 5:08 pm To: minima@xxxxxxxxxxxxx I quickly hooked the new mixer board up to the test set for some more scans. The conversion loss looks identical, maybe even a tenth or so better. All the mixing products up to 100 Mhz look identical at a cursory glance, except the LO leakage which looks about 10 dB worse at -39 dBm. Still very respectable but I'll need to find out what changed. I used the transformers from the breadboard, so it's not them. Possibly something to do with using a different pair of switches in the IC package. I attached scans of return loss at both ports. They look almost identical, with the LO popping up at 34 MHZ....to be expected. This is a good sign of performance symmetry and also a good sign of very little energy wasted due to mismatch. More to come... Joe -------- Original Message -------- Subject: [minima] Re: New Mixer Design From: "Joe Rocci" <joe@xxxxxxxxxx> Date: Tue, August 26, 2014 1:16 pm To: <minima@xxxxxxxxxxxxx> I got a chance today to design and etch the attached PCB, so hopefully I can get some more data tonight. Note that I changed the usage of the available switches in the package to make a single-layer PCB easier. Now we'll find out if the performance is reproducable. Farhan I saw on your HFSignals site that you were developing an IMD test set. Is that working? Joe