Video není dostupné.
Omlouváme se.
44 - More Mixer Testing
Vložit
- čas přidán 4. 10. 2023
- Nick M0NTV tests the accuracy of the Constant Current Source he built to help match the diodes in a double balanced diode ring mixer. Then he tackles the thorny question of whether matching diodes actually makes a difference. The results are interesting!
For details of the construction of the W7EL Audio Diplexer:
• 34 - Improvements to t...
For the RF Diplexer:
• 21 - Diving Deeper Int...
Pete Juliano (N6QW) has a great 3-part series on constructing diode ring mixers. Part 1 is here:
• Homebrew Double Balanc...
Thanks very much for watching. 73, Nick M0NTV
I guess that explains why using a diode tester on a multimeter is 'good enough' in most cases, and my rigs built over the years have worked. Well done for taking the time to do those experiments Nick, and thanks for sharing.
Thanks Steve! It was your question about whether it makes a difference or not that motivated me to try the experiment. I hope at least that it will encourage budding homebrewers across the world to have a go at building a mixer - safe in the knowledge that there is a pretty good chance of it actually working! Thanks again. 73, Nick
Thanks Nick! Well done. The diode thing was very useful, but what really got me was your explanation of why we homebrew -- we want to know how it works! 73 Bill N2CQR
Thanks Bill! I think like Steve (G0FUW) your instincts about the diode matching were probably right! Sometimes good enough is enough. As for the spirit of homebrew, I think it is something that people either get or not. For me it is an insatiable curiosity about how things work. I will often read about a circuit, then model it in LTSpice and then build it and try and get it to work. Somewhere in this process I begin to understand more about what is actually going on. This understanding is as satisfying as the building itself - in fact it adds another dimension to it. Thanks for all your inspiration and encouragement. 73, Nick M0NTV
Nice work Nick,
The matching as you show makes a big difference on the various spurs, (I don't call reducing the spurs by 6-10db insignificant ! )which can hover well above the noise floor, and they will cause "birdies" if the rx is sensitive enough.
I agree that it is the construction that makes all the difference.
Andy
Cheers Andy. I know you've done a lot of work in this field. Thanks for the other resources you sent me too. Much appreciated. 73, Nick
Keep up the good work, very interesting. However schottky diodes have a lot of leakage compared to silicon diodes. I found this out when designing a charge discharge circuit into a mosfet. Maybe it would be better if the leakage was matched.
Thank you very much and that's an interesting idea about the diode leakage. I must confess that's another area I don't know much about ... yet! There's always something new to learn isn't there? Thanks again. 73, Nick
Fascinating - thank you Nick! Like you, I expected diode matching to make a big difference. 73 Ian G4COL
Thanks Ian. Yes, it was a bit of a surprise to me. Good to know that sometimes RF can be a bit more forgiving though! 73, Nick
Well that was both interesting and surprising! It does, however, prove that good standards of construction are (or should be) top of the list. Nice one matey!
Thanks Alan. It was a bit surprising but every day is a school day! Hope you're both OK. See you soon :)
👍great work,
Thanks very much. 73 Nick
Well!, who'd have thought it?!! Thanks for the video and experiment Nick. What a surprise. Given that there is a small advantage, will you be matching in the future, not bother or only in critical applications?
Hi Ace! I think I will keep on matching as I still want to build as carefully and efficiently as I can. Mind you, I won't have to do it for a while as I've got another 12 sets of matched diodes to keep me going now! Thanks very much. 73, Nick
@@M0NTVHomebrewing Thanks for your reply Nick, I had a feeling you will be using matched sets long after your stock is exhausted. Using 5W, every milliwatt counts!
Good video Nick. You've touched on the what I think is most important point-- symmetry & balance is everything in double balanced mixers.
Diodes, the transformers/ baluns + wires + layout -- the lot. Stray coupling (L and C) may also affect balance and I have unwittingly changed these during test runs of various mixer components. The goals of double balanced versus unbalanced or single balanced -- boost inter-port isolation, reduce spurious responses and hopefully improve IP3.
I tend to match diodes for port isolation as this is likely the best indicator of mixer balance: e.g. to cancel input signals from outputs at the 3 mixer ports
LO is the strongest tone -- so it's easy to measure the LO at the IF ( or RF ) port with the spectrum analyzer.
Isolation = LO input available power / LO power at the IF ( or RF ) port.
I agree- even in commercial circuits, you may find imbalances that sometimes disappoints for all the $$$$
Different diodes may incur variances in ON resistance and capacitance. Port matching (especially the IF port) and temperature may affect mixer performance measures and that port termination itself is worthy of experimentation. A great mixer (well balanced) with poor port termination may function as a "good mixer".
I've found that balanced mixer function tests sometimes vary with frequency -- probably not at HF though. Most mixer data sheets will show port isolation across their intended bandwidth -- we tend to just measure at 1 LO + 1 RF frequency.
Strive for balance -- enjoy the results whatever they may be. This the point I get from you. Great thoughts. Thank you.
Best! -QRPHB-Todd-VE7BPO- qrp-popcorn.blogspot.com/
sorry. it's minus not " / " for isolation - We measure feed through
Thank you very much for watching and for taking the time to share your thoughts. I think your comments are very helpful. 73, Nick
Thanks for SA pics, very educational!
I saw datasheet for HP matched set of mixer diodes for their SA mixer and they not just match diodes by curve stats but also by capacitance, which is crazy, sadly cant remember datasheet name
Thanks Igor. Yes some folk take this matching business very seriously indeed. I just do my best! 73, Nick
👍Thanks for sharing this Nick. Certainly indicates that your matching device works very well. In terms of matching, I appreciate that you aimed for the best vs. worst case. However, a random batch of x4 is likely to fall into the "reasonably matched" category rather than have one or more "outliers". So, I suspect that x4 random diodes from a batch may give results even closer to the results you got with the matched ones.🤔
Cheers Steve! Yes, it certainly seems that RF (at HF frequencies at least) is a bit more forgiving of forward voltage differences than I imagined. 73, Nick
The diodes are just one part of the equation. One that is looked at too often with too much precision for the use case of many amateur radio designs. Perfectly matched diodes are not going to make much of a difference with typically quite asymmetric, hand-wound toroid transformers. Here's an Experiment that's worth the effort: Rewind the transformer on the left side in picture shown at 0:11 , do not solder the midpoint-windings together and do not connect them to ground. Instead take each end of a winding that you would ordinarily solder together, and connect them each, individually to the outer two pins of a low value (10-100 Ohms) potentiometer. The wiper of the pot then should be soldered to ground. You now have a very cheap and easy way to adjust the balance of the balanced mixer. Repeat your modulator test and adjust the pot while viewing the output spectrum. You'll find that you can achieve a significantly higher carrier supression now.
Thanks for your suggestion. I actually refer to this very idea in the concluding section of the video and suggested folks check out the work of Pete Juliano (N6QW) in this respect. I used to include 100Ω pots in my mixers until I saw a video of Charlie Morris (ZL2CTM) where he found it actually reduced carrier suppression - so he left it out! I've tended to do the same. But never say never! Thanks again for watching and commenting. 73, Nick
@@M0NTVHomebrewing Interesting, I always have good results with that. I watch Charlie's videos regularly and noticed that he is no longer doing that. Another thing I do different is to actually shield the toroids from each other. Too commonly in hobbyist designs, inductors (especially in filter circuits) are aligned in a way to allow magnetic coupling between them. You see multi-stage pi-filters assembled as they're drawn with all inductors lining up. I'd suspect that ensuring symmetry and including (magnetic) shielding between transformers can do magic. Might do a test myself and post it. Don't have much time right now though. Anyhow, just my 2 cents. Anyhow, great video, I've been watching for a while and am always happy to see new content from you!
Thanks very much. Yes, you are right, shielding is a fascinating subject. I really had to get a grip of it with my last transceiver build which had the steerable IF section. So I had weak input signals very close to much stronger output signals and hey presto my amplifiers had turned into oscillators! A bit of shielding and careful routing of coaxial connections fixed it. It took a while though! Thanks again. 73, Nick
Thanks for the video, I found it really interesting and I think I'll have a try at building a mixer this homebrew season.
Theres one significant detail that's coming to mind here. Those little pads the diodes are on are above the ground. The two tests may have had variations in capacitance to ground, by chance. If you look at the schematics for balanced modulators in older HF transceivers, (FT7B springs to mind) they use a trimmer across the diode bridge for the balance modulator, theres also a preset potentiometer. I hope this gives you fodder for exploration
Interesting thought about the pad capacitance. I have tried using the Pete Juliano 100Ω trimmer trick to compensate for unevenness in the windings of the transformers. Sometimes it has helped and sometimes not. I have to say the best addition has been the W7EL audio diplexer on the audio (IF) port. Thanks very much for watching and for sharing your thoughts. 73, Nick
@@M0NTVHomebrewing Yes I saw that, went on a binge of your videos last night. But I think some adjustability would be a step up again. The "diplexer" and some further nulling adjustments, you may get the carrier much lower again in different operating situations like a different excitement frequency, power level or impedance. The yaesu FT7B balanced modulator circuit I actually played with about 20 years ago. It uses normal germanium schottky diodes... I think very similar to a 1N34, but has a 20pf ceramic trimmer and a 100 ohm preset pot for nulling- you go back and forth between the two to get the carrier as low as you can. I've also seen very similar arrangements in older 23 channel crystal synthesis SSB radios like the kraco SSB Deluxe. Just giving you some direction for a bit of experimental research since unlike me, you have good bench equipment to really put things through their paces, and seem to really like tinkering 😁
@@M0NTVHomebrewing it might allow you also to do away with the RF "diplexer" (one port in and out, it's actually a bandpass filter) and just use the audio diplexer, keeping things a lot more rudimentary
Hi there. Although the RF diplexer I built might look at first glance like just a regular 2 port LC band pass filter actually there's a bit more to it than that. Technically it is a three port device - it's just that the third port here is shunted to ground - but it needn't necessarily be! It takes an input on one port and sends a signal on 13.3 MHz to the opposite port and everything else to the bottom port (ground). Best of all is that all signals see a constant 50Ω impedance regardless of frequency. They can be fiddly to build - mainly because of the low inductance value in the parallel tuned (tank) circuit at the bottom - but very effective if you can get them right. 73, Nick
@@M0NTVHomebrewing I wasn't going to say anything, but it's still not a diplexer. It's a bandpass filter. ONE frequency is involved, especially since both tuned circuits are on the same frequency. Couldn't help myself..
Cheaper is good if the results are so accuraste and repeatable,
I think series connected diodes, anode to anode or cathode to cathode would enable capacitance measurements but from the results here that looks like it's not going to yield much improvement at HF
Thanks for watching and commenting. 73, Nick
I wonder how hard it would be to match BAT54S Diode Array Schottky sets with each other. And wonder if it would make any difference to the uBITX ? At about 19:00 I do see spurs way down but definitely there that are not there on the left. Aha, you did talk about them.
I did briefly mention those spurs - yes. Basically I am of the opinion that the more you can do to build your mixer carefully (e.g. diode matching, symmetrical layout etc) the better will be your results. All the best with your own building. 73, Nick
Your testing was done with a single frequency of a constant amplitude. But vary that audio frequency, vary the carrier frequency and the amplitude, and what do you get? Perhaps more harmonic content, more distortion.
The observation about the quality of construction is the key point. I think building a radio is a lot like building a race car. Build each and every single thing as near perfect as you can manage, and when you're done, you'll have a winner. Do a little sloppy work in multiple steps, and you're left with a very sloppy result. In the mixer, everything from lead length to your ground-plane beneath those pads affects the inductance and capacitance of the circuit, inducing resonances, harmonics, and imbalances. If I go to the trouble of building my own transceiver I'm going to match, balance, work and rework, optimize the filters to get the best skirts possible, and do everything I can to build the very best radio I can, whether I can easily notice the difference or not, because that's what it's all about.
I agree, race car or radio a small plus at each stage adds up in the final result...every little bit helps.
Thanks very much John. Yes, I'm with you on this - my favourite homebrew transceiver is No. 4 which I call 'The Optimiser'. It is basically a glorified BitX-like design but every stage is optimised to give the cleanest signal. There are more filters and attenuators in it than I care to remember and it took an age to complete but the result is well worth it. I had it on the air on 40m just two nights ago and people still comment on the good modulation and great audio. Fair point about my experiment being a single frequency sign wave - from a SigGen as well and not even an Si5351 VFO. I've often found that when testing mixers things start to look different once you introduce a microphone and a real VFO! Thanks again for watching and commenting. 73, Nick
I totally agree with @johnwest7993. You have at least an input (LO or IF) pretty constant power, but the other is not. Try both with a power ramp or sweep, and you will get different results from the two mixers, I suspect more evidently on the low power side, where you can different diodes turned on and off. Also consider that you calculated the voltages into a 50ohm system, but you are using transformers. Are they 1:1? If not, measure the voltages on the diodes AFTER transformation.
I always enjoy your content.
IZ4HDM Chris
Nice science, Nick! de Phil M1GWZ
Thanks very much Phil! 73, Nick