It's sort of amusing that folks think this is a new idea. Lock-in amplifiers were using CMOS muxes as mixers in the 1970s, though discrete MOSFET bridges were more common. For instance, here's the schematic of a PARC 128A lock-in from 1983:
(See P. 54 of the service manual.)
Cheers
Phil Hobbs
Cheers
Phil Hobbs
I don't think it's the current steering (upper) stage of the Gilbert cell that's the problem--those devices are almost always ON or OFF.
The main issue is that the RF port drives a BJT diff pair that's in its linear range, so that its output current goes as tanh(e dV_BE/(2kT)), whereas in a diode mixer the RF port's load is the input resistance of the first IF, plus the switch resistances. Diode mixers work a lot better into a constant 50 ohms for this reason.
(Using a diplexer instead of a reflective filter after the mixer is a big win for linearity, and this is why.)
A strong interfering signal will drive the RF diff pair of the Gilbert cell nonlinear just the way it does an ordinary BJT amplifier. Running the diff pair as a current mirror, as in the LM13700 OTA, nominally turns the tanh characteristic back into a linear one, but that trick isn't often used at RF, I don't think.
Cheers
Phil Hobbs
Plus, unlike Findchips, the previous leader, you can do search on keywords as well as part numbers.
The SEO and redirect parasites have made Google dramatically less useful in the last couple of years.
Cheers
Phil Hobbs
AFAIK the business end of a CMOS analog switch is a pair of transistors. So in a sense it's a FET-ring mixer that someone found inside of a little bit of plastic.
embarrassed, I did NOT send the thank you, just did.
For modulation, I have the required 90 degree phase shift covered often with accuracies better than tenths of a degree, over the band of 10Hz to
100kHz. Supplying the required modulation will NOT be the problem. My problem is still with the carrier suppression and upper harmonic generation that comes out of a mixer, instead of out of a multiplier circuit. Perhaps what I really need is a 'perfect' analog multiplier that can operate up to 100MHz, similar architecture to the MC1496, but better specs??'Interfering higher tones' is a major concern, need them down more than
90dBfs [would like more than 100dBfs]. It is surprising how difficult it is to make physically realizable filters that can knock those 'specced' tones down enough.Two 2N3904's make a great 5 pole slightly ripply filter that operates up to just beyond 10MHz [not 100MHz], but the limit of rejection in the rolloff is ALWAYS the number of poles, no matter how you place them. ...Suppose could put two in series. It's too difficult to get all the poles placed properly at around 7 poles and upwards. Need that broadband capability, else would play games with notching out the tones. With active circuitry, might be possible to make a 'tracking' filter that follows the carrier. Not going to go there, yet.
If only. The real limit is usually due to strays.
Cheers
Phil Hobbs
Forgot all about SAW filtering. I know it's fixed filtering, but might have some advantages.
The required 90 degree phase shift in the modulation is covered. And, it's pretty easy to supply LO with quadrature, too. It's the 'balance' in the mixers I need to get that pesky carrier tone and 3rd harmonics down.
100MHz has 10ns period. 0.1nS is almost 4 degrees phase shift! Somehow will have to make the system accomodate that kind of sloppy action, too
True. but I've designed systems that can 'see' signals below Tempest levels. And they thought they were 'silencing' their electronics! NOT!
great article, thank you.
I tried to search for an 'original' SD8901 data sheet and all I get are wannabees.
Calogic is referenced as the originator, but couldn't find them, either.
Any EXACT URL to get me a 'clean' copy of that data sheet?
Would it be possible to wrap a feedback loop around the mixer balance, something like a carrier psd to servo carrier residual to minimum?
Just an idle thought.
piglet
I originally misread that you intended to use the carrier as some kind of pilot tone (based on the WITH in upper case :-), but apparently you want just ordinary SSB (J3E), upper or lower side band ?
There are a lot of "zero-IF" chips using I/Q (de)modulators, e.g. in various WLAN, DVB-T and in some equipment in various license-free bands. These might be usable.
Use an image canceling I/Q mixer on a fixed carrier frequency, say 300 MHz to generate a LSB SSB signal and put a simple band pass filter after it. A simple LC filter with Q=100 would have a 3 MHz bandwidth and quite a lot of attenuation at 600 MHz (wrap around carrier 3rd overtone lower side band) and 900 MHz (upper side band of carrier harmonics).
Using a VCO/NCO tunable between 300.1 and 400 MHz and using a simple (non-I/Q) mixer followed by a simple LPF at 200 MHz, this will produce the 0.1 to 100 MHz signal you wanted.
Please note that due to the high side LO, the sidebands are swapped, so in order to generate the USB signal, the original I/Q mixer needs to generate LSB.
great thought! thanks. once thought of, becomes so obvious.
Might be easy to implement, too.
great thought! thanks. might be easy to implement, too.
Dropping another 20 dB lowers that noise floor just that much more!
Grin! Just slide your appreciation under the door.
piglet
What exactly is your problem, the suppressed carrier attenuation, the opposite (LSB) attenuation or some multiples of the carrier frequency ?
With analog components, the opposite side band attenuation in I/Q modulators can be at least 40 dB for the audio band. Doing something in digital domain will have better performance, e.g. by generating the SSB signal at a lowish frequency (2 MHz) and then translate to the final frequency.
Are there any gaps in the frequency band that could be used to handle the intermediate frequency (e.g. between 2 and 3 MHz) ?
What is the modulating signal, is it analog or already digital ? This might affect how to best optimize the whole _system_.
The original Si8901 mixer app note (AN85-29) was by Ed Oxner, back in
1988. I found the reference in this patent:The whole thing is in the 1989 databook, so I scanned it,
along with the datasheet:
Enjoy!
Cheers
Phil Hobbs
Siliconix made the Si8901, which is in my 1989 databook here on my shelf. The similar-vintage SD5400 series quad SPST switches were also commonly used as mixers.
After discontinuing it, the morons reused the part number for something completely different.
Cheers
Phil Hobbs
Years ago I got a few samples, still in my parts bin. As I recall this is an old article and there are updates with newer parts. I'll look for the updated articles. Here's a background article on mixers, modulators and Demodulators. The example here uses the SD5000 DMOD FET array.
Oh, just found this as an update for the array, FST3125.
On 04/04/14 20.43, amdx wrote: ...
Supplementary:
Found this some time ago (FST3125 test:
H-Mode mixer (Conclusion):
Both the FSA3157 and the NC7SZ384 mixers did not suffer at all from the fact that the mixer is constructed from two to four separate devices rather then just one, with regard to IMD. ... The 74AC04 fundamental squarer turned out to be a winning factor when it comes to getting rid of the spurs of a rather "spurry" unfiltered AD9951 LO-DDS. ... The FSA3157 in combination with the fundamental 74AC04 squarer and the ADTT1-1 transformer is very good when it comes to the reduction of spurs. ... There is no need anymore to try to make the RF-port "see" a 50 ohms resistive termination at all times to reduce the spur level. No more half or full diplexers anymore in front of the mixer. This greatly simplifies things! And consequently no extra (although small) losses introduced by those diplexers worsening the NF of the front-end. ... The category 6 mini-circuits transformers (ADTT1-6, ADT1-6T, TT1-6, T1-6T) turned out to be well worth investigating too. Although no winners in spur reduction, these transformers are the best if spurs are not the first concern. What these transformers have in common is that for an unexplained reason they almost completely eliminate the mixers bias point sensitivity. ... It turns out to be quite possible to construct a frontend with a dynamic range in excess of 120dB in SSB bandwidth on 40M. ..."
NC7SB3157, FSA3157 Low-Voltage SPDT Analog Switch or 2:1Multiplexer / De-multiplexer Bus Switch:
( main page:
SM 5 BSZ - I/Q mixer for direct conversion radio (Sept 28 2001):
Optimization of Quadrature Modulator Performance:
Down-converter to zero-IF: LT5517 - 40MHz to 900MHz Quadrature Demodulator:
Up-converter from zero-IF: LT5572 - 1.5GHz to 2.5GHz High Linearity Direct Quadrature Modulator:
Down-converter to IF: LT5560 - 0.01MHz to 4GHz Low Power Active Mixer:
/Glenn
Have something to add? Share your thoughts — no account required.
Ask the community — no account required