5903	2011-03-10 07:24:42	Harold Smith	another Mixer level question
Message	Date	From	Subject
As a variation on the RF level question, I'm wondering if that -10 dBm limit also applies to the IF port, when it (SRA-1 or equivalent) is used as a modulator with audio applied. de KE6TI, Harold
5904	2011-03-10 11:13:14	KK7B	Re: another Mixer level question
This topic is discussed in detail on page 9.47 of EMRFD beginning with the sub-secti
5905	2011-03-10 13:34:13	Wes	Re: another Mixer level question
Hi Rick, and group, The topic of mixer drive level is also discussed on page 5.14 and 5.15. Most of that discussion is with regard to mixer IIP3 performance with respect to drive. Your suggestion of using -10 dBm drive per mixer is in line with my own experinece when low IMD is required. There should probably be a sentence in Chapter 5 that makes reference to the detailed discussion within Chapter 9. I'll take care of this in the book errata on line. Thanks Rick. 73, Wes w7zoi
5906	2011-03-11 08:26:42	kb1gmx	Re: another Mixer level question
Hi All, This is not new information. Going back to the 6H6(diode tube) based mixers(modulators) its was common to read RF must exceed audio by a factor of 6 or more for good signals. DBMs are fairly predictable building blocks and popular for that reason. The "however" is always whats better? Is there one building block that allows for more output for more input with less RF needed? The context in this case is as a modulator where the inputs are audio and carrier and the output is high quality DSB for use as SSB by filter or phasing methods. I'll leave those questions for possible answers. This is likely more a subject of opinion but something based on applicati
5907	2011-03-11 11:04:53	Tim	non-diode-bridge mixers
I have been tinkering around with gilbert cell mixers (MC1496, SA602, discrete transistors ala EMRFD Fig 5.28) for a while now. In HF radio we commonly use them in two ways: 1: Both inputs are scaled nicely for nice nondistorted mathematical multiplication. With clean sine wave inputs at reasonable levels (e.g. very low hundreds of mV) and just a little effort put into balance, the output is super duper clean and you have to look hard to find anything other than sum and difference frequencies. BUT any variation in input levels shows up exactly as a variation in output level. e.g. if carrier drops by 20%, then output levels drop by 20%. 2: Carrier input is driven into saturati
5908	2011-03-11 11:28:48	ehydra	Re: non-diode-bridge mixers
If you want to see the scientific view of the different possible mixer modes then look for Rubiola Mixer with Google. You can then expand to sampler theory. All done, nothing new. - Henry -- ehydra.dyndns.info Tim schrieb: > I have been tinkering around with gilbert cell mixers (MC1496, SA602, discrete transistors ala EMRFD Fig 5.28) for a while now. > > In HF radio we commonly use them in two ways: > > 1: Both inputs are scaled nicely for nice nondistorted mathematical multiplication. With clean sine wave inputs at reasonable levels (e.g. very low hundreds of mV) and just a little effort put into balance, the output is super duper clean and you have to look hard to find anything other than sum and difference frequencies. > > BUT any variation in input levels shows up exactly as a variation in output level. e.g. if carrier drops by 20%, then output levels drop by 20%. > > 2: Carrier input is driven into saturati
5909	2011-03-11 11:31:22	Chris Trask	Re: non-diode-bridge mixers
> > I have been tinkering around with gilbert cell mixers (MC1496, SA602, > discrete transistors ala EMRFD Fig 5.28) for a while now. > > In HF radio we commonly use them in two ways: > > 1: Both inputs are scaled nicely for nice nondistorted mathematical > multiplication. With clean sine wave inputs at reasonable levels (e.g. > very low hundreds of mV) and just a little effort put into balance, > the output is super duper clean and you have to look hard to find > anything other than sum and difference frequencies. > > BUT any variation in input levels shows up exactly as a variation in > output level. e.g. if carrier drops by 20%, then output levels drop > by 20%. > > 2: Carrier input is driven into saturation but audio is kept below > saturation. The output is not super clean... lots of odd harmonics of > the carrier going out to infinity. A lot like a DBM. After getting > rid of the harmonics of the carrier ends up quite clean. > > BUT the advantage is that variations in carrier level produce little > to no net change in output level. > > My takeaway: if using a Gilbert cell with a saturated input there's > not a lot of advantages (other than inherent gain) over a diode > bridge DBM. > > Using a diode DBM or a Gilbert cell in saturation, works out quite > well (odd harmonics easily cleaned up via simple filters and you are > insensitive to even major variations in carrier level) until you want > to broadband the output. > All of which is correct. An additional option is to use the Gilbert Cell mixer with feedback added: http://www.home.earthlink.net/~christrask/fbmxrs.html http://www.home.earthlink.net/~christrask/ISCAS02A.pdf Chris Trask N7ZWY / WDX3HLB Senior Member IEEE http://www.home.earthlink.net/~christrask/
5910	2011-03-11 11:52:38	Tim	Re: non-diode-bridge mixers

5911	2011-03-11 12:23:17	Chris Trask	Re: non-diode-bridge mixers
> > > All of which is correct. An additional option is to use the Gilbert > > Cell mixer with feedback added: > > > > http://www.home.earthlink.net/~christrask/fbmxrs.html > > > > http://www.home.earthlink.net/~christrask/ISCAS02A.pdf > > All nice but it seems to me that rather than increase linearity (or sensitivity > to carrier level) in a low spur mixer, what we might desire is the clean output > of a multiplier being operated in linear-linear region but with some magical > insensitivity to carrier level. > > Off the top of my head I'm thinking, use really mushy but also well-matched > active devices. Like CD4007 style. I've tried to implement this before but > with poor results. Probably just a very bad idea. > I have done the same thing, using a CA3046 array which provides five matched transistors. Two were used as a translinear predistorter for the LO, two others (the differential pair) were used as the multiplier proper, and the fifth was used as the RF voltage-to-current converter which drove the differential pair emitters. The results were far less that what I had anticipated. I tried other variations of the basic idea, but all of them produced equally unsatisfactory results. One thing I've learned after many years of tinkering is that in the commutating mixers (aka Gilbert Cell), the switching delay times of the transistors is primarily at fault. None of them turn on and off at the same speed, and you end up with a "dead zone" where all four transistors are on (or off) at the same time. I've tried using Schottky switching diodes in place of the switching transistors, but the circuit complexity is not worth the effort and the results are no better than for a simple diode DBM. The only thing I've managed to come up with in all these years after the low-noise feedback mixer is a diode DBM that has just 3.5dB of conversion loss, very close to the theoretical minimum of 3dB. IMD performance remained the same as for a conventional diode DBM. > > Some sort of simple ALC for carrier level is probably more appropriate than > trying to do ALC in the mixer. e.g. Ulrich Rohde's ALC crystal oscillator, fig > 10.26 in the 2006 ARRL handbook. That's exactly the BFO oscillator I used > with a MC1496 product detctor BTW. I was very happy with the results. > Yes, that is necessary if using the analogue multiplier approach. Chris
5912	2011-03-12 09:45:55	kb1gmx	Re: non-diode-bridge mixers