Creating quadrature LO for direct conversion receiver

[vectorotter]

I want to build a zero-if receiver for around 2 MHz of BW @ 1GHz.

I have a single ended feed from antenna, BPF, LNA although my plan is to eventually build the LNA out of discrete components so it could be differential out too.

The mixers are ADL5801 having differential inputs and outputs.  I've attached a JCad schematic.

I have a differential source for the LO marked A+,A-

My question is, how do I generate the 90 degree shifted LO having a good level match so as to achieve good I/Q balance?

I would think to place a 90 degree hybrid coupler from minicircuits in line but now the feeds are asymmetric.  Can I change the architecture somehow? Do I not know how to use the 90 degree hybrid correctly?  Where can I find an example similar to my set-up?

Best regards the otter with vectors.



 

[szoftveres]

You can use polyphase filters. Depending on how much LO power you have and how much is needed for the mixers some amplification might be necessary.
Or, you can use the 90 degree hybrid in the RF path (in place of the wilkinson on your schematic) - that way you can drive both mixers with the same LO.

[vectorotter]

Maybe buffer one or both LO feeds and trim the gain in other words,  irrespective of how the shift is implemented?

[mag_therm]

MAX2837 has the LO PLL, splitter and dual mixers integrated, along with onboard LPF.
To use it, you would need to upconvert your RF into MAX2837 band (~ 2500MHz)
That might be easier and more accurate than externally splitting LO into two separate mixers.

I just purchased a HackRF One board (open source)  for ham radio use. Here is a link to its block diagram showing the MAX2837,
and the upstream mixer RFFC5072.
https://hackrf.readthedocs.io/en/latest/hardware_components.html

[vectorotter]

Thanks. I'll give it some thought.  I really wanted to do it with the discrete mixers and LO I already have on hand.

[vectorotter]

Hrre's a JCad with Minicircits QCN12+ quadrature hybrid in the rf path.  Do I have the right idea here and what should the turns ratio of the baluns be for 50 ohm differential from 50 single ended be?

[MartinL]

It's not that common to see direct conversion receivers at 1GHz built from separate parts. Usually at GHz frequencies the quadrature LO synthesizer and the mixers are integrated into a single IC. Which is not to say you shouldn't do this, but rather to explain why you're not finding much information or options out there.

The MAX2837 already mentioned is an example of this approach for the 2.3-2.7GHz band, but for 1GHz, a better example would be the numerous direct conversion zero-IF receiver chips intended for satellite TV, which cover a range of around 900 - 2100 MHz. MAX2120 is one example.

But given that you want to do it this way, what you have there looks reasonable.

The transformers in your schematic should be 1:1. 50 ohms is 50 ohms.

Note though that the phase shift provided by the QCN-12+ is not exactly 90 degrees, more like 81-82 or so. You add the additional phase shift via transmission line lengths in the PCB layout. 8 degrees at 1GHz is about 4mm of microstrip on a typical stackup.

Which gives you another option: 90 degrees at 1GHz is only about 40mm of microstrip. If you have a specific frequency you want to operate at, then rather than using the QCN-12+ to phase-split the RF signal, you could just use 40mm of extra PCB track to delay the LO to one mixer by 90 degrees relative to the other. That would avoid any losses in the RF path.

[radiolistener]

you can use si5351 module to get IQ quadrature from 4 MHz to 220 MHz and above using it's phase shift register, but it's quadrature quality and limits are not good due to low phase shift resolution and range. And low frequency limit don't allows you to implement receiver for MW and LW band.

The better way is to use si5351 to generate twice higher frequency and then using external low jittter divider by 2 to generate IQ quadrature with 90° phase shift:


But note that using external inverter add some phase shift error for the second trigger. You can solve it with using second channel of si5351 with the same frequency and setup inverse flag for this channel, it allows to generate inverse signal with minimal phase difference.

This way allows you to get nice IQ quadrature generator from 5 kHz to 110 MHz.

PS: Also be careful with PCB layout, try to keep the same electrical length for both phases of IQ signal, it minimizes phase error. This is important because increase phase error leads to degrade your receiver performance and keeping this phase error for both mixers as minimum as possible is the main issue of analog quadrature receiver.

[ch_scr]

With the divide by 2 / 4 quadrature circuits, would a pair of D-FF clocked by Fin, one on each quadrature output, basically re-timing them, eliminate the quadrature error from gate delays?

[MartinL]

The OP was asking for a quadrature LO for a 1GHz direct conversion receiver. The Si5351 approach is not applicable as it can't go that high.

[radiolistener]

The OP was asking for a quadrature LO for a 1GHz direct conversion receiver. The Si5351 approach is not applicable as it can't go that high.

This circuit is still applicable for 1 GHz, but you're needs to get D-FF working at 2 GHz.
You can use ADF4351 to generate double frequency which will be 2 GHz.


But I think direct conversion from 1 GHz is a bad idea, because it will be very hard to get two exactly the same mixers and low phase error between LO IQ lines due to very high frequency, even 1 mm wire difference or a small solder drop will lead to a huge phase unbalance. Such receiver will have bad performance and will be very hard to design PCB and tune it.

The better approach is to do down conversion to some IF, for example 10.7 MHz and then feed to quadrature mixers.

[MartinL]

But I think direct conversion from 1 GHz is a bad idea, because it will be very hard to get two exactly the same mixers and low phase error between LO IQ lines due to very high frequency, even 1 mm wire difference or a small solder drop will lead to a huge phase unbalance. Such receiver will have bad performance and will be very hard to design PCB and tune it.

The need to match the mixers and get the correct phase relationship between the LOs is of course a big part of why this is often done on-chip in the GHz range. But I see no reason to consider it intractable for a board level design.

We're not talking about putting something together with bits of wire. A PCB layout for the sorts of small surface mount components being discussed here will have tolerances much tighter than 1mm. And even a 1mm length on a typical PCB microstrip is only about 2 degrees of phase shift at 1GHz. If the LO signal paths are laid out to be identical except for a specific difference in length between them, it should be straightforward to get the phase error within a fraction of a degree. Solder joints etc should have little effect, as they will be very similar between the two paths.

Of course, the constraint of this approach is that the phase relationship would only be correct at a specific LO frequency. So this is only viable if the receiver is able to use a fixed LO, with any further tuning being handled in baseband processing.

There will also be differences in conversion gain between the two mixers, which will affect I/Q balance, but that can be corrected for at the baseband stage.

The better approach is to do down conversion to some IF, for example 10.7 MHz and then feed to quadrature mixers.

I don't think it's necessarily a better approach.

If you downconvert a signal to a 10.7MHz IF, then that conversion will be equally sensitive to interference at an image frequency that is 21.4MHz away from the signal of interest (either above or below it depending which side you put the LO). At HF or low VHF, that image frequency will be well out-of-band, and far enough away from the desired signal that it can be easily rejected by practical filter designs at the RF stage. At 1GHz though, an image 21.4MHz away is much closer to the signal in relative terms, and much harder to reject adequately. For some standard frequencies, there are SAW filters available that may do the job, but otherwise it may be impractical to achieve the necessary rejection.

In this regard a direct conversion architecture is a major advantage, since you get built-in image rejection without difficult filtering requirements at the RF front end.

[mag_therm]

A few years ago I did a Gilbert demodulator for an old HF receiver to receive the ham digi modes.
Even at 455kHz, I had trouble with fixed offset and varying duty cycle of the square wave out of ordinary D flip-flops on 3.3V supply. That reduced the performance of the MC1496 Gilbert.

I re-did the board with (2) of D FF SN74aup1G00 feeding an LVDS NBA3N011 to get a more accurate level shifted carrier into the Gilbert.
But looking at the specs for total prop delay, I would not expect good accuracy of those parts at 10MHz ( although they don't spec duty cycle)

[mawyatt]

There's a circuit we called the "Polyphase Mixer" that may be of interest.

https://www.eevblog.com/forum/rf-microwave/polyphase-or-n-path-mixer/

Has some very unique and amazing performance metrics, and naturally produces the I and Q baseband signals.

Best,

[radiolistener]

In this regard a direct conversion architecture is a major advantage, since you get built-in image rejection without difficult filtering requirements at the RF front end.

Yes, if you have ability to implement your circuit on the chip with good phase balance it will be really major advantage. But if you're planning to solder it as an amateur circuit, I'm afraid it will be complicated to do it at GHz band. This is why using some IF is better, because it is much easier to do it for 10 MHz than for 1 GHz...

[vectorotter]

There's a circuit we called the "Polyphase Mixer" that may be of interest.

https://www.eevblog.com/forum/rf-microwave/polyphase-or-n-path-mixer/

Has some very unique and amazing performance metrics, and naturally produces the I and Q baseband signals.

Best,

I'll look into this.  Thanks for the info.  I haven't got time right now but will report back when I get there.

[vectorotter]

But I think direct conversion from 1 GHz is a bad idea, because it will be very hard to get two exactly the same mixers and low phase error between LO IQ lines due to very high frequency, even 1 mm wire difference or a small solder drop will lead to a huge phase unbalance. Such receiver will have bad performance and will be very hard to design PCB and tune it.

The need to match the mixers and get the correct phase relationship between the LOs is of course a big part of why this is often done on-chip in the GHz range. But I see no reason to consider it intractable for a board level design.

We're not talking about putting something together with bits of wire. A PCB layout for the sorts of small surface mount components being discussed here will have tolerances much tighter than 1mm. And even a 1mm length on a typical PCB microstrip is only about 2 degrees of phase shift at 1GHz. If the LO signal paths are laid out to be identical except for a specific difference in length between them, it should be straightforward to get the phase error within a fraction of a degree. Solder joints etc should have little effect, as they will be very similar between the two paths.

Of course, the constraint of this approach is that the phase relationship would only be correct at a specific LO frequency. So this is only viable if the receiver is able to use a fixed LO, with any further tuning being handled in baseband processing.

There will also be differences in conversion gain between the two mixers, which will affect I/Q balance, but that can be corrected for at the baseband stage.

The better approach is to do down conversion to some IF, for example 10.7 MHz and then feed to quadrature mixers.

I don't think it's necessarily a better approach.

If you downconvert a signal to a 10.7MHz IF, then that conversion will be equally sensitive to interference at an image frequency that is 21.4MHz away from the signal of interest (either above or below it depending which side you put the LO). At HF or low VHF, that image frequency will be well out-of-band, and far enough away from the desired signal that it can be easily rejected by practical filter designs at the RF stage. At 1GHz though, an image 21.4MHz away is much closer to the signal in relative terms, and much harder to reject adequately. For some standard frequencies, there are SAW filters available that may do the job, but otherwise it may be impractical to achieve the necessary rejection.

In this regard a direct conversion architecture is a major advantage, since you get built-in image rejection without difficult filtering requirements at the RF front end.

I would have to agree with a lot of this.  I have actually worked on a board level design of a direct conversion receiver designed for operation around 400 MHz, so about factor 2.5 lower than for what I intend for this one.  I didn't even use proper power dividers and came out with something like -30dB of I/Q imbalance in the end. 

I't just been a few years since I worked on it and I forgot a lot of the fine detail.

I remember it maybe having centre-tapped transformers as baluns hence why I wonder about the turns ratios and also wonder about possibly using two quadrature hybrids, presumably using the 0 degree port in one arm and the ~90 degree port in the other arm to achieve approximate level matching.

People build PCBs designed to operate far above 1GHz with discrete components quite frequenctly, and yes, it is challenging, but that's part of the reason for doing it.

[IconicPCB]

Just looked at polyphase filter mixer.
Reminded me of  a little project I did close to 20 year ago.
A poly phase filter to synthesise a variale phase shifter to minimise interfering signal in an active antenna.

The structure is a lattice RC network driven by a sample of interfering signal which would saturate the input FET of the active antenna.
The phase shifting was achieved by combining the quadrature outout of the filter through PIN diode attenuators.
This would then produce variable phase/ gain signal producing interference of 30dB at up to 30degrees phase shift from maximum attenuation notch center.
I was able to inject an interfering signal into an off air PAL TV signal ( back in the good old analog days ) causing the TV to loose both audio and video ( through the active antenna and then by processing the intrference ( using an Atmel AT mega32 micro ) recover reception.
SO You could try the quadrature phase shift trick, process the LO into quadrature signals and achieve the down conversion.

[cncjerry]

I think what is called a twisted wire quadrature circuit will work at that frequency.  Can't remember but I think Breed was the person that developed it and then it showed up a lot in EMRFD in the R2 receiver and T2 transmitter.  I use it all the time, its clean, passive, no extra jitter or phase noise impact and you can put it together in a few minutes with some beads at 1Ghz.  A couple people used it in DC SSB radios a lot higher than 1Ghz.

[ch_scr]

I think what is called a twisted wire quadrature circuit will work at that frequency.  Can't remember but I think Breed was the person that developed it and then it showed up a lot in EMRFD in the R2 receiver and T2 transmitter.  I use it all the time, its clean, passive, no extra jitter or phase noise impact and you can put it together in a few minutes with some beads at 1Ghz.  A couple people used it in DC SSB radios a lot higher than 1Ghz.

This implements the calculations for the "Lumped element 3rd-order polynomial-based quadrature Hybrid coupler", as described in "David Andrews - Lumped Element Quadrature Hybrids", with selectable bandwidth ratio from 1.2 to 3 as per the tables in the book (trade-off is mid-band amplitude imbalance). It seems to me like the slightly-more-fancy "3 core version" of the "original" twisted wire quadrature circuit you mentioned. It's response looks a lot like an "coupled microstrip" hybrid coupler, which is nice I guess. The table also estimates the turns needed based on the calculated "needed L" and the Al value of the cores, I'd guess if you put an "Al of one" in there you could get an idea for the turns needed to what's described as an "air core bifilar wire selonoid" in some of the older sources. Even at 10MHz you only need a few turns already on cores with an AL of 20 and 80; and at 1GHz it's about 1.1 and 2.5 turns with an Al of 1 (air core), suggesting hitting the limit for what frequency is possible with this particular design. I guess at 1GHz a coupled microstrip implementation is more repeatable / (already small enough to be) practical / etc.

[mawyatt]

There's a circuit we called the "Polyphase Mixer" that may be of interest.

https://www.eevblog.com/forum/rf-microwave/polyphase-or-n-path-mixer/

Has some very unique and amazing performance metrics, and naturally produces the I and Q baseband signals.

Best,

I'll look into this.  Thanks for the info.  I haven't got time right now but will report back when I get there.

Be careful this is a very deep dive into what appears as a simple circuit but in fact has very involved behavior as well as some incredible metrics (better than theoretical Mixer NF for example) and why it quickly spread throughout the academic world. Dates back to mid 2000, became general public with IEEE articles in 2009, and has found a way into some advanced SDRs and Apple products.

Spend some time with the IEEE article & references, if you are an intellectual curious type you'll find this deep dive fascinating.

Best,