13159	2016-09-18 12:49:01	Chris Trask	Mesmerising About Mixers
Message	Date	From	Subject
Ever since I devised the KISS mixer, I've been contemplating other ways of making better mixers, but there are distinct obstacles to b dealt with. The LO power required to drive diode ring mixers results in LO-RF and LO-IF leakage, plus they need to be matched over a wide bandwidth in order to get decent IMD performance. Substituting switching FETs and buss switches for the diodes gives better performance, but the high frequency range is limited. Active mixers such as the Gilbert Cell can give higher frequency performance, but the manner in which they work results in switching the bias current of the switching quad on and off. This increases the problem of LO to output isolation, and the transition zone between ON and OFF is where most of the distortion takes place. Analogue multipliers once seemed to be a good choice, but their IMD performance is not much better than that of a diode ring mixer. Using translinear topologies only makes the matter worse. There is no ideal solution to this. The use of buss switches such as with the KISS mixer have given highly desireable IMD and NF performance, but they are pretty much limited to HF frequencies. It would be nice to have monolithic dual small signal switching MOSFETs available, but the ones I've looked at are terribly slow and are oriented towards usage as precision current mirrors for instrumentation applications. Grumble! Chris When the going gets weird, the weird turn pro - Hunter S. Thompson
13160	2016-09-18 15:16:03	kb1gmx	Re: Mesmerising About Mixers
Chris, Two parts to look at: Fairchild COOLMOS dual MOSfets. designed for class e and other switching uses. Very low Rds-on lower G-D capacitance. Another is the GAN mosfets, Rds-on in the really low range and very fast switching. Optimized for class D and E power at HF. EPC 80xx and 90xx parts are intersting. Allison
13161	2016-09-18 21:56:17	Ashhar Farhan	Re: Mesmerising About Mixers
Folks, I just realized the other day that most of us have been fooling around with JFETs as switches. The output had a spikey waveform of the local oscillator. it was quite puzzling. it wasn't present in the bus switch versions. It was just the other day, I realized that this was due to driving the gate into conduction. I tried the same circuit with 2N7000 and the spikes were gone. I must try a full KISS mixer with 2N7000s driven by a 74HC74 as a square wave push-pull driver. - f
13163	2016-09-19 09:17:01	Chris Trask	Re: Mesmerising About Mixers
Thanks for the info, but these are all power devices with high capacitances and slow switching speeds. A MOSFET for a mixer application at VHF would need to have switching speeds in the order of a few nSec and very low parasitic capacitances. > > Fairchild COOLMOS dual MOSfets. designed for class e and other switching uses. > Very low Rds-on lower G-D capacitance. > > Another is the GAN mosfets, Rds-on in the really low range and very fast switching. > Optimized for class D and E power at HF. EPC 80xx and 90xx parts are intersting. > Chris When the going gets weird, the weird turn pro - Hunter S. Thompson
13168	2016-09-20 22:46:40	ehydra	Re: Mesmerising About Mixers
You could try the BF862 lowest-noise highest-gain, or the dual J310 in a single package PMBFJ620 for symmetry. Or 4xarray SD5000/SD5400. Or one with separated substrate connection 2N4351. I would like to see the results. - Henry Chris Trask christrask@earthlink.net [emrfd] schrieb: > Thanks for the info, but these are all power devices with high capacitances and slow switching speeds. A MOSFET for a mixer application at VHF would need to have switching speeds in the order of a few nSec and very low parasitic capacitances. > >> Fairchild COOLMOS dual MOSfets. designed for class e and other switching uses. >> Very low Rds-on lower G-D capacitance. >> >> Another is the GAN mosfets, Rds-on in the really low range and very fast switching. >> Optimized for class D and E power at HF. EPC 80xx and 90xx parts are intersting.
13169	2016-09-21 09:16:55	Chris Trask	Re: Mesmerising About Mixers
After looking through the Farnell website, the best device I can find so far is the Fairchild FDC6301N. > > Two parts to look at: > > > Fairchild COOLMOS dual MOSfets. designed for class e and other switching uses. > Very low Rds-on lower G-D capacitance. > > > Another is the GAN mosfets, Rds-on in the really low range and very fast switching. > Optimized for class D and E power at HF. EPC 80xx and 90xx parts are intersting. > Chris Trask N7ZWY / WDX3HLB Senior Member IEEE http://www.home.earthlink.net/~christrask/
13173	2016-09-22 06:00:11	ehydra	Re: Mesmerising About Mixers
This FDC6301N is a power DMOS-FET. The rds_on is way too low! And so is the gate capacitance to high and would need big driving power (Because it is almost entirely capacitive the driver has to waste the power as heat). I would check for devices with rds_on approx. equal to system impedance (i.e. 50 ohms). And for low current rating (= small die) and of course low voltage (= small cells). And because a power MOSFET die is large and consists of thousands of small cells, if ones drives it to fast some cells won't be activated or delayed (The internal gate interconnections are equivalent to different RC-delay lines. The gate interconnection is doped Si whereas the power-driven S and D are metall with much lower resitivity). If the D-S current is high the devices will break because of hot-spots, if the current is low you just waste driving power and have sluggly transition edges. Harris makes MOSFET drivers for diagnostic ultrasonic piezos. Maybe there is something for you. They work 100MHz+. The problem is how to route an analog signal with it to construct the mixer archicture. - H. Chris Trask christrask@earthlink.net [emrfd] schrieb: > After looking through the Farnell website, the best device I can find so far is the Fairchild FDC6301N. > >> Two parts to look at: >> >> >> Fairchild COOLMOS dual MOSfets. designed for class e and other switching uses. >> Very low Rds-on lower G-D capacitance. >> >> >> Another is the GAN mosfets, Rds-on in the really low range and very fast switching. >> Optimized for class D and E power at HF. EPC 80xx and 90xx parts are intersting.
13174	2016-09-22 09:29:50	Dana Myers	Re: Mesmerising About Mixers

13175	2016-09-22 10:06:46	Chris Trask	Re: Mesmerising About Mixers
> >> >> After looking through the Farnell website, the best device I can find so far is the Fairchild FDC6301N. >> > >Are there any suitable purpose-built RF switches still around? HP once >had some GaAs switches with a ~2nS switching time and response >up to 6GHz (and decent response down at DC, err, HF). Looks >like all Broadcom/Avago has now is a chip product spec'ed for DC-40GHz. > Most, if not all of these GaAs RF switches have awful switching times. I could not find any information at all on HP switches, outside of an application note for the MGS series. > >Using any common MOSFETs seems to run into the problem of high >inter-electrode capacitance right away. > There are compromises, and sifting through what's available is time consuming. I did manage to find a dual MOSFET switch that appears to be promising: http://www.nxp.com/documents/data_sheet/PMGD8000LN.pdf The switching times, on resistance, and capacitances are not much better than for the 74LVC1G3157 (NXP equivalent of the FSA3157), but the linearity of the on resistance over voltage is tremendously better. I would expect this device to have much better IMD performance, though the OIP3 of the KISS mixer is tough to beat. A dual 2-input XOR gate such as the NC7WZ86 would be needed to provide the needed logic switching. Chris When the going gets weird, the weird turn pro - Hunter S. Thompson
13176	2016-09-22 11:44:30	Dana Myers	Re: Mesmerising About Mixers