Injection Locked Peltz Oscillator with Bode Analysis

[mawyatt]

Having worked with Injection Locked Oscillators quite a bit in the past we decided to see if the Peltz Oscillator would be a good candidate for an investigation for using a DSO Bode Function (SDS2000X+) while under the influence of external signal injection.

For the theory behind injection locking see Adler's famous 1973 IEEE paper, also Razavi 2004 IEEE paper. Note the gain/phase graphs Fig 2 is Razavi and similar graph from Adler. 

https://www.eevblog.com/forum/projects/simple-sinusoidal-oscillators/

https://www.eevblog.com/forum/projects/things-coming-together-bode-plot-diy-isolation-transformer-peltz-oscillator/msg4288363/#msg4288363

R. Adler, “A study of locking phenomena in oscillators,” Proc. IEEE, vol. 61, pp. 1380–1385, Oct. 1973.

http://www.seas.ucla.edu/brweb/papers/Journals/RSep04.pdf

Attached is a schematic of the Peltz Oscillator, and the Bode setup. The Injection signal is the stimulus from the AWG which is monitored by DSO CH1, this feeds the Peltz Oscillator thru Ri and the Output is monitored by DSO CH2. The injection current is ~ Vi/Ri and the bias current is ~ (Vbias -Vbe)/Re.

Q1 and Q2 are 2N3904
Ri is 100K
Re is 10K
Vbias is 1VDC
Vi is 10VPP
L is 470uH
C is 10nF

The oscillator free running frequency is ~75.5KHz.

Anyway, here's the Bode Plot which shows the Injection Oscillator Response in Amplitude and Phase!!!

Best,

[moffy]

Nice plots. Correct me if wrong, but it looks like at those levels the injection locking works/holds from about 70kHz to 80kHz, where those discontinuities on the phase plot are.

[mawyatt]

If you study the references you'll find the locking range is defined as the range where the oscillator will frequency lock onto an incoming signal, and this is defined as:

delta F = +-[Fo/(2Q)](Ii/Ib),

where Fo is the free running oscillator frequency, Q is the oscillator loaded Q, Ii is the injected signal current and Ib is the oscillator bias current.

This delta frequency range is where the phase between the injected signal Ii and oscillator signal is between +-90 degrees. The oscillator can hold lock slightly beyond delta F and this is shown with the Phase peaks above and below +-90 degrees.

 In the plot shown below you can see where the DSO has some trouble and the amplitude and phase are jittering around the lock range edges. The markers indicate the ~ +-90 phase and show a delta frequency of 2.56KHz which agrees with the 2*(1.347KHz) = 2.69KHz predicted from the notes.

This measurement is particularly difficult for the DSO to perform since the oscillator signal is large and near the Input Signal frequency, so the DSO must dig out the reference signal, and frankly we're surprised this DSO can perform this measurement, and speaks highly of the Frequency Selective Sampling Technique utilized. Honestly, we did not expect these results!!

Anyway, this is an involved subject as you can tell, Razavi does a nice job and recommended as a reference.

Best,

[moffy]

thank you for the detailed explanation. It looks like your second plot is significantly narrower than your first, guess it has more resolution. With a PLL and multiplier mixer, the pull in range (going from non lock to lock) is usually significantly smaller than the hold in range (maintaining lock), is it the same for injection locking?

[mawyatt]

You are welcome, hope this helps some.

The lock-in delta F equation predicts a range somewhat proportional to the injection signal level to bias current ratio. The first plot showed a much larger ratio and thus a wider lock range as well as a cleaner plot since the DSO has a bigger signal to detect and process (that's why we showed it).

Once in lock, the hold in range is slightly larger than the lock-in range.

Here's another plot with a higher ratio Ii/Ib which produces a wider lock-in range than the second plot, but less ratio than the 1st plot. You can see the markers show a difference of 3.6KHz between +-90 degree and the delta F equation predicts 3.7KHz, it's a nicer looking plot with less DSO uncertainty due to the larger injection signal level.

Impressive performance from this DSO IMO.

Best,

[moffy]

If I may answer my own question,  , a PLL has a low pass filter after the mixer, which determines the pull in range, whereas the hold in range is how far the VCO deviates when subjected to the maximum phase error voltage. With injection locking there is no low pass filter after the mixer, so they are probably the same.

Sorry it appears our answers collided in time.

[mawyatt]

The injection locking mechanism does have a somewhat memory like effect, so one could think of it as a low pass function. This is in the Q of the network, it has a circulating current which acts like inertia.

We utilized this concept as a means for demodulation, it was the key enabler in a new type single chip receiver which utilized the injection locking oscillator as a local oscillator, direct down conversion demodulator, RF filter and IF filter, see patent 5603111.

Best,

[moffy]

The injection locking mechanism does have a somewhat memory like effect, so one could think of it as a low pass function. This is in the Q of the network, it has a circulating current which acts like inertia.

We utilized this concept as a means for demodulation, it was the key enabler in a new type single chip receiver which utilized the injection locking oscillator as a local oscillator, direct down conversion demodulator, RF filter and IF filter, see patent 5603111.

Best,

noted and thanks.

[mawyatt]

Just for fun we built up a Peltz Oscillator based upon a couple 2N3904s, 2.2mH, 10nF and 10K Re emitter bias with 3 Volt supply. Output was sensed with a 100nF and 1K to a SA (-32dB attenuation). Injection Locking was provided with a 10nF and 10K series R to the emitters.

The setup produced a 32.7KHz frequency. Injection signal was from a AWG and 50mvPP thru the 10K and 10nF series cap.

PNG2 is the free running plot from the SA with 100Hz Span at RBW of 1Hz.

PNG1 is Injection Locked at 1X or 32.7KHz from AWG, note the improved close in Phase Noise (PN) vs. Free Running (PNG2).

PNG4 is with AWG at 2X or 65.4KHz, the oscillator is acting as a %2. Note improved PN vs. Free Running (PNG2).

PNG6 is AWG at 3X or 98.1KHz, also note PN relative to free running.

In the past we've used this Injection Locking concept in many applications, one was above mentioned Synchronous Homodyne Microwave Receiver (Direct Down Conversion) where the Local Oscillator Injection Locked to the input signal, another was in selective frequency dividers ~100GHz (beyond where conventional dividers could operate).

Anyway, an interesting technique you don't see mentioned much today.

Edit: Don't know why these plots get mixed up, also note PNG1 has a bandwidth of 200Hz all the others are 100Hz.

Best,