High side injection and low side injection

[stanley1]

For 540 kHz commercial AM radio broadcast station, considering 455 kHz as the intermediate frequency, we get two values of local oscillator frequency
1.) 85 kHz (low side injection)
2.) 995 kHz (high side injection)
I noticed that 540 kHz/85 kHz = 6.353 while 995 kHz/540 kHz = 1.843.
What can I conclude/learn from this? What decides the choice of side injection?

[ms660]

Calculate the tuning ratio (max/min) for the oscillator and all be be revealed when you consider the min and max frequency that the oscillator needs to tune to using only a variable capacitor to tune the whole of the BC band (scroll down to Local Oscillator Frequency):

https://en.wikibooks.org/wiki/Commun...s/AM_Receivers

Lawrence.

[Radio Wrangler]

One other factor is check where the image frequency goes... They can cross loud stations and stress the RF selectivity.

David

[G0HZU_JMR]

85kHz would be a poor choice for the LO in this case because a typical mixer will respond to signals where the absolute difference of n*RF +/- m*LO = 455kHz. These responses won't be as strong as the main mixer response but they might only be 10-30dB down. A lot depends on the mixer choice.

So in this case, there would be fairly significant and unwanted mixer response terms at:

455kHz+(85*2) = 625kHz
455kHz+(85*3) = 710kHz
455kHz+(85*4) = 795kHz
etc

These responses will be difficult to reject with a basic (tracking) image reject filter if the receiver is tuned to 540kHz.

Work out the equivalent terms for high side LO injection and these terms will be way outside the MW band and so they can be filtered easily.

[G0HZU_JMR]

One big unwanted mixer response with the 85kHz low side injection would be 455kHz - 85kHz = 370kHz. This is outside the MW band but it would need to be rejected by any front end preselection. There would also be strong birdies at tuned frequencies where n*LO = 455kHz.

It really isn't a good idea to use a low side LO for this particular frequency plan...

[Argus25]

I have considered this issue myself over the years and noted that for most radios, shortwave, medium wave and for TV's (VHF types at least) the designers always seem to select high side injection. Its fairly obvious for the reasons for LW & MW radios due to the required tuning ranges of the L/O and birdies & images. But at much higher frequencies there are more choices.

I only have one piece of superhet based apparatus in my entire collection of radios and TV's that uses low side injection and that is the HMV904 television set that runs the L/O below the received vision and sound carriers. In the Bush TV22 for example and most superhet TV sets, the L/O is high side injection.

So one other reason why one system is chosen over the other, in TV work at least, when you are receiving two carriers, is that the choice of high side injection acts like a Mirror, so if your transmitted video carrier is above the sound carrier frequency, it comes out below it from the mixer, but if say you use low side injection, the relationship of one carrier above the other remains the same. And this has implications in the ease of design of the sound and vision IF stages. Though they were pushing the upper limits of the oscillator frequency in the HMV904 with the X41 valve, so that might have been another reason they went for low side injection.

[turretslug]

With some HF receivers using "typical" US pattern multigrid mixers (oscillator injection at grid 1, signal injection at grid 3), the lower bands are aligned osc-high but the highest frequency band is aligned osc-low. The AR88 receiver uses this type of mixer (6SA7) but the RCA manual stipulates osc-high alignment on all bands. I found that aligning the highest frequency (22-32MHz) band osc-high resulted in scale discrepancy approaching almost 100kHz mid-band however exacting I was about setting the trimmer and slug adjustments. Re-setting to osc-low (quick and easy to do, so trivial to try!) resulted in the tuning scale discrepancy never being more than 30kHz, so I assumed that the manual was not necessarily correct. Though it's always possible that changes in value of tuned circuit componentry in ageing radios could produce this outcome.

[Jon_G4MDC]

I once tried to swap LO injection on a Pye PF8 because it suited the crystals I already had to hand.

I didn't bargain on the fact the RX was PLL based and had automatic frequency control (AFC). It seemed to work fine except the AFC tuned backwards and it ran away from any signals!

I ended up spending the money on the right crystals. The AFC line was a buried track on layer 3 of the PCB that could not be broken and inverted.

[ms660]

So far as I can remember low side osc advantages on HF of SW is covered in RDH4 (Langford Smith)

A Marconi CR300 I once had used low side on the highest range, also seem to remember that some VHF/FM receivers did this to avoid interference to Band III TV.

Lawrence.

[G6Tanuki]

There's also the consideration that if you're using low-side injection the receiver is likely at some point to tune through the 2nd and 3rd harmonics of its own LO.

High-side injection ensures this won't ever happen.

[G0HZU_JMR]

Decades ago I can remember doing this type of analysis at work using paper based mixer spurious charts. These printed charts were the 'smith chart' of the frequency planner in those days and they looked like a spiders web and looked even worse once the analysis was completed. The potential for mistakes was immense.

To escape this I wrote my own mixer spurious analysis software in Quick Basic about 30 years ago. This was able to predict all mixer terms that fell in band or close to band and also any in band birdies caused by the mixing of LOs and their harmonics. This was mainly for a wideband triple conversion system that tuned from HF up into the GHz region.

MiniCircuits and Synergy both produced mixer software in the 1990s that made the mixer charts obsolete and then Agilent produced some software for Genesys called WhatIF that took it to the next level.

I'm not sure if MiniCircuits still distribute a version of their Minispur software but this old DOS SW had mixer spurious charts built in for all of their mixers and allowed the user to choose the LO, RF and IF and it predicted what the in band problems would be and at what level they might occur vs RF drive level. It wasn't very good at predicting the levels but it could at least show an interactive display of all the mixer terms and the display resembled a spectrum analyser.

At work, we still use our own software for stuff like this. I'm a little rusty these days but I can usually do this stuff in my head. However, our SW would frequently be working with high multiples of the LOs. eg 5*LO1 +/- 11*LO2 would be typical multiples in a dual conversion receiver analysis.

[G0HZU_JMR]

I ran up this frequency plan using the old Synergy LOCUS software we bought at work 25 years ago. This old DOS program allows the user to enter the target IF frequency (0.45-0.46MHz) and then the total tuning range of the RF and the LO. It then works out what mixer spurious terms can walk through the IF as the RF and LO are swept over their full ranges.

The result is as below and it shows the terms I predicted earlier. However, it does require the user to reverse engineer where these terms appear and how hard they will be to filter out with a tracking preselector.

There is another feature called SPECT in the software that makes this process easier.

You can see how dirty this frequency plan is if the LO is chosen to tune across 0.085 - 1.100MHz. The main spurious terms are as below but this is for a fairly large RF signal of -20dBm and it models a generic level 7 diode ring mixer.

[Terry_VK5TM]

Originally Posted by ms660

So far as I can remember low side osc advantages on HF of SW is covered in RDH4 (Langford Smith)

A Marconi CR300 I once had used low side on the highest range, also seem to....

Lawrence.

The Realistic DX160 (and probably others from Tandy), do the same thing.

As far as any "advantages" in that method, don't know about that as there are images galore on the highest band in the DX160.

[Skywave]

An example of a numerical calculation which should help make things clear.

Let's say: MW = 530 → 1650 kHz & IF = 465 kHz.

Case 1.
Osc. on 'high side', i.e. Osc - RF = IF, so Osc. = RF + IF.
For RF = 530 kHz, Osc. = 530 + 465 = 995 kHz
For RF = 1650 kHz, Osc. = 1650 + 465 = 2115 kHz
Hence, Osc. freq. = 995 → 2115 kHz. (A freq. ratio of 2.13 :1)

► For a typical tuning capacitor, say 20 pF → 500 pF, a calculation reveals that the Osc. coil = 42 µH (to nearest µH).
The tuning capacitor will require an additional 116 pF in parallel. (Will consist of strays and coil self capacitance and a fixed capacitance)

Case 2.
Osc. on 'low side'. i.e. RF - Osc. = IF, so Osc. = RF - IF
For RF = 530 kHz, Osc. = 530 - 465 = 65 kHz
For RF = 1650 kHz, Osc. = 1650 - 465 = 1185 kHz
Hence, Osc. freq. = 65 → 1185 kHz. (A freq. ratio of 18.2:1)

► For a typical tuning capacitor, say 20 pF → 500 pF, a calculation reveals that the Osc. coil = 13 mH (to nearest mH).
However, the design will not be feasible, since the required additional capacitance is a negative value!

Note: the requirement for 'acceptable tracking' is not considered in the above calculations. This can also be done, but the arithmetic can get quite heavy. Nevertheless, the above set of calcs. is illustrative to this thread.

Al.

[Skywave]

Originally Posted by Terry_VK5TM

..
The Realistic DX160 (and probably others from Tandy), do the same thing.

It's been a very long time since I had to 'get heavy' with a DX160. The one I had was aimed at the European market: it had Long Wave coverage.

IIRC, the osc. was not only set to the 'wrong side' of the incoming RF, but was also running at the wrong freqs. - so the set would not tune to anything on LW. There were no signs at all that this set had been tampered with - which really puzzled me. However, I re-designed the cct. for correct osc. freqs. & with the osc. on the 'other side' of the incoming RF - and then all was well. I also have a vague notion that the other (higher) freq. bands were also on the 'low side': poor image rejection, yes, but the calibration & sensitivity (for a set of this class) seemed reasonable.

Al.

[G0HZU_JMR]

One practical test to try would be to get a mixer and feed it with an 85kHz LO. Feed the IF port to a (optional) 455kHz BPF and feed the output of this filter to a an SDR or a receiver or a spectrum analyser tuned to 455kHz.

Then feed an RF signal to the RF port and sweep it from 200kHz through to maybe 1600kHz. Make sure the mixer LO port can support a local oscillator frequency as low as 85kHz.

You should find that the mixer will not just respond to 540kHz, it will respond to a lot of signal frequencies including the main ones I listed earlier. So a MW receiver with this 85kHz LO would be effectively listening to lots of frequencies in the MW band at the same time. Not just 540kHz! The receiver would be listening to 370kHz and 710kHz quite effectively plus loads more frequencies in the MW band at the same time. Inject an RF port signal at 710kHz and a signal should appear at the IF at 455kHz for example.

[Terry_VK5TM]

Originally Posted by Skywave

It's been a very long time since I had to 'get heavy' with a DX160. The one I had was aimed at the European market: it had Long Wave coverage....

Al.

Mine has LW and all bands except the highest one are all high side injection.

[Skywave]

Originally Posted by Skywave

It's been a very long time since I had to 'get heavy' with a DX160. The one I had was aimed at the European market: it had Long Wave coverage....

Al.

Originally Posted by Terry_VK5TM

Mine has LW and all bands except the highest one are all high side injection.

Presumably, 'highest' & 'high side' refer to frequency, not wavelength.

As for the case I had to deal with, it might have been caused where it was set up wrong at the factory. I've never heard of the problem reported anywhere else, then or since: all very mysterious.

Al.