Author Topic: Oscillator design  (Read 2379 times)

0 Members and 1 Guest are viewing this topic.

Offline JeanFTopic starter

  • Regular Contributor
  • *
  • Posts: 143
  • Country: fr
Oscillator design
« on: May 06, 2020, 08:54:11 pm »
Hi,

not too long ago I took apart an old OCXO to see what was inside, some of you may remember I posted a thread in Beginners section about the heater circuit.

Here I would like to talk about the oscillator circuit; I mostly understand what's going on but I have a few specific questions, maybe some of you guys can point me in the right direction!

I tried my best (and spent a lot of time) to avoid mistakes when tracing the circuit but obviously it's maybe not perfect, so if you find something suspect I can check once more.


So this seems to be a Colpitts quartz oscillator. I cropped the scanned image in order to leave the heater circuit out of the frame, it's just a basic comparator circuit with an opamp and a thermistor.

At first I was intrigued by L1 and C12 and then I read in "Design of crystal oscillator circuits" by B. Neubig (here, page 5) that it was a resonant circuit, which is capacitive at nominal frequency, to be used in place of the usual capacitor to prevent the oscillation at the third or higher overtone of the crystal, or other unwanted resonances. So that's sorted out.

My questions are:
- why in that case the bottom of L1 / C12 is not tied to ground but to the little R24/ C11 filter instead ; and do these filters have a more specific name than "parallel RC filter" so I can read up on them (I think R12 / C9 seems very similar, same question for them)
- what is Q1 doing ? I guess it is some kind of feedback. I think it is injecting current into R7 and into the filter mentioned previously but I don't trust my measurement enough. I would have been tempted to think it could be positive feedback to help sustain the oscillations, but when I desolder R7 (just to see what could happen) then the amplitude at the base of Q3 goes through the roof instead, so that's more like negative feedback? but why? I also tried to change the value of R7 a bit (between 1k and 2k) but I couldn't see any difference anywhere.
- in any case, what could be the point of the R14 / R15 voltage divider which does not divide by much? What would be the drawback of tying the left pin of C8 to the emitter of Q4 instead?
- (this last question is also related to the fact that I don't get what Q1 is doing) - would it have been possible for the designer to reduce BOM by 1 transistor by removing Q1 and tying R7 to the output of Q5 instead?

Thank you very much. This is not for work, not for school, I'm just curious :)
 

Offline ArthurDent

  • Super Contributor
  • ***
  • Posts: 1193
  • Country: us
Re: Oscillator design
« Reply #1 on: May 06, 2020, 09:46:48 pm »
The feedback (as you found out) is like automatic gain control. The oscillator will shift frequency somewhat if the level of oscillation doesn't remain constant. The voltage on C11/R24 comes from C7 (4u7) which is D.C. This is similar to keeping the frequency stable by keeping the temperature constant. There are a number of variables that affect the frequency and they try to compensate for these with additional circuitry.
« Last Edit: May 06, 2020, 09:52:25 pm by ArthurDent »
 

Offline moffy

  • Super Contributor
  • ***
  • Posts: 2218
  • Country: au
Re: Oscillator design
« Reply #2 on: May 07, 2020, 01:17:10 am »
Exactly as ArthurDent said. The gm/gain of Q2 is dependent on the amount of current flowing through it. Less current then lower gm. Q1 and R7/C7 provide that feedback. As the amplitude increases R7 steals current from Q2's emitter lowering its gm until a stable amplitude is found. It would also tend to make the output more sinusoidal with fewer distortion products.
 

Offline pardo-bsso

  • Regular Contributor
  • *
  • Posts: 235
  • Country: ar
Re: Oscillator design
« Reply #3 on: May 07, 2020, 03:19:48 pm »

My questions are:
- why in that case the bottom of L1 / C12 is not tied to ground but to the little R24/ C11 filter instead ; and do these filters have a more specific name than "parallel RC filter" so I can read up on them (I think R12 / C9 seems very similar, same question for them)


They are there to give more gain at AC. When you calculate the bias point the transistor 'sees' only the resistor. However, for the ac analysis the capacitors are considered as a short.
 

Offline JeanFTopic starter

  • Regular Contributor
  • *
  • Posts: 143
  • Country: fr
Re: Oscillator design
« Reply #4 on: May 08, 2020, 10:23:06 am »
Thank you very much everyone  :-+ I really appreciate it !
this circuit is a good opportunity for me to learn more about transistor theory, which I've been neglecting a bit until now :)
 

Offline sorin

  • Frequent Contributor
  • **
  • Posts: 279
  • Country: de
Re: Oscillator design
« Reply #5 on: May 10, 2020, 10:53:39 am »
In my opinion the schematic is not correct because Q1 is connected as short circuit (on AC). It cant inject anything into R7 because C7 act as a short circuit at 10MHz. Q1 don't work also because is incorrectly biased, the base of Qi should be connected to ground not to + 6.8V, it is a PNP transistor, the base should be negative compared to the emitter. Maybe you forget to draw some resistors or you put the C7 on the wrong place?

- "At first I was intrigued by L1 and C12 and then I read in "Design of crystal oscillator circuits" by B. Neubig (here, page 5) that it was a resonant circuit, which is capacitive at nominal frequency"
The resonance frequency of L1 & C12 is 15.651 MHz so at 10Mhz they act like inductance not capacitance. They start to act like capacitance above 16MHz.
« Last Edit: May 10, 2020, 11:26:37 am by sorin »
 

Offline JeanFTopic starter

  • Regular Contributor
  • *
  • Posts: 143
  • Country: fr
Re: Oscillator design
« Reply #6 on: May 10, 2020, 05:14:33 pm »
Quote
In my opinion the schematic is not correct because Q1 is connected as short circuit (on AC). It cant inject anything into R7 because C7 act as a short circuit at 10MHz. Q1 don't work also because is incorrectly biased, the base of Qi should be connected to ground not to + 6.8V, it is a PNP transistor, the base should be negative compared to the emitter. Maybe you forget to draw some resistors or you put the C7 on the wrong place?

Thank you for your comment. I checked again very carefully all the connections around Q1 and I couldn't find anything at all different from what I've drawn in the schematic...  I've checked the pinouts once more and all seems correct. I've measured (DC) 6.63 V at the emitter of Q1, 6.64 V at the base and 3.68 V at the collector. And 2.66 V at the left side of R7. Anything else in particular that I could verify?

Quote
The resonance frequency of L1 & C12 is 15.651 MHz so at 10Mhz they act like inductance not capacitance.
whoops... my mistake. Indeed I got the capacitive and inductive sides mixed up. It turns out that C12 is 220p not 22p, I've been fooled by its "220" marking. I used to think that the last digit is always a multiplier, but that seems to not always be true. So the resonance is in fact at 4.9 MHz. At the same time I spotted another error, C5 is 1ยต not 100n.

By the way, Q1 is a 2N2907A, Q2 to Q5 are 2N2369 and Q6 is a BD138.
I'll edit the schematic.

Note : I've uploaded here a .pdn file (opens with Paint.Net) containing photos of top and bottom that were aligned, and additional layers where I drew components and traces, if anyone feels curious to have a look. It's a bit painful really because all the traces are covered by shiny solder and I couldn't get a good contrast, so I had to draw over them manually; and almost all the resistors are mounted vertically so the values don't appear on the picture.
 

Online mawyatt

  • Super Contributor
  • ***
  • Posts: 4119
  • Country: us
Re: Oscillator design
« Reply #7 on: May 11, 2020, 01:03:03 am »
The purpose of Q1 as mentioned is an automatic level control. As this signal swing increases in peak to peak amplitude across Q2 and Q3 the signal at Q1's base will go more negative and cause more collector current flowing into R7. This current will look more like a clipped sine wave pulse and C7 acts as a low pass filter. The average current thru R7 causes the voltage across R24 to increase which reduces the current in Q2. Since the transconductance is proportional to current, Q2's transconductance decreases which decreases the signal swing amplitude across Q2 and Q3. By controlling the current the overall linearity of Q2 during the complete cycle swing is improved and this helps with stability and harmonic generation from the oscillator core which is Q2.

Another way of thinking about this is that the gain of Q2 must be exactly 1 integrated over the entire cycle otherwise the signal amplitude will either be decreasing (<1) or increasing (>1), the feedback keeps the gain "centered" and well behaved over the cycle rather than following limit cycles typical of some oscillators.

So this behaves as a negative feedback level control to stabilize the output amplitude and improve performance.

Anyway this is how I understand the circuit to behave, hope this helps.

Best,
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 
The following users thanked this post: JeanF

Offline sorin

  • Frequent Contributor
  • **
  • Posts: 279
  • Country: de
Re: Oscillator design
« Reply #8 on: May 11, 2020, 06:37:33 am »
Im confident that te shematic i wrong, it cant work.

 

Offline JeanFTopic starter

  • Regular Contributor
  • *
  • Posts: 143
  • Country: fr
Re: Oscillator design
« Reply #9 on: May 11, 2020, 10:35:00 am »
mawyatt, thank you for this explanation. Thanks to ArthurDent and moffy, I was understanding the circuit in the same way, but your wording is much better than what I could have done myself...

sorin, thank you for taking the time to do this. I'm not pretending my diagram is 100% correct ; I just don't understand what's wrong. I already spent a very long time checking and checking again, and I found nothing. It's possible that a mistake is indeed hiding in front of me in plain sight. That's not too critical for me, at this point in my life electronics is still a hobby and I was more interested in the general working principles of this circuit than in getting a military-grade schematic of it, and all the previous replies (thanks again everyone) made sense. I think I'm going to focus on other things now, but as I said the pictures of the board are available; if someone says "well there must be a resistor between this point and this point, and also this wire that you drew here isn't correct..." I'll be happy to get the circuit out of the drawer and look again.
 

Offline sorin

  • Frequent Contributor
  • **
  • Posts: 279
  • Country: de
Re: Oscillator design
« Reply #10 on: May 15, 2020, 10:44:39 pm »
I was wrong, your schematic is correct.
 
The following users thanked this post: JeanF

Offline moffy

  • Super Contributor
  • ***
  • Posts: 2218
  • Country: au
Re: Oscillator design
« Reply #11 on: May 16, 2020, 12:52:01 am »
To Sorin:

Very nicely said. :)
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf