Hey everyone,

I'm absolutely new to electronics (and this forum), and I'm pretty lost and desperate for help.
I've looked through some related threads but could not find a clear answer to my question or a solution to my current problem.
I'll try to give as much detail as possible, so sorry for the long post!

I am currently tasked to investigate the impact of shock impulses on quartz crystals used in oscillator circuits. In order to properly focus on the crystal part, I wish to seperate the crystal from the osc. circuit and place it on another board. This way, I can use a single shock impulse to hit multiple crystals simultaneously without affecting the rest of the circuit (that's my idea at least). I do not require the crystal to oscillate at a specific frequency as long as it is relatively stable, so I can compare the before and after and therefore deduce the changes resulting from the shock.

My question is: Is this plug-in principle actually realisable? I am thinking of using terminal blocks and very short cables to "plug in" the osc. circuit to individual crystals that are, as stated above, located on a different board

I've designed a prototype PCB where I directly put in the connectors of the crystal (through-hole type) at the respective terminal block on the osc. circuit. Using a single SN74HC14 schmitt trigger inverter, I wanted to place 3 separate oscillator circuits close to each other on the same PCB, simply to reduce the amount of parts needed. Using this setup however, I do net get any oscillation when using the cystals 2 or 3 (1 is an SMD type which would require me to build a second PCB, which I will do once 2 and 3 function properly).
Could you help me find the problems in my circuit?

Here are the specs:
General assumptions: C_stray = 3 pF, C_in = 10 pF (datasheet, max), C_out = 5 pF (guess) [inverter input/output cap.]
Crystal 2 (Kyocera) @ 20 MHz, C_L2 = 12pF
Crystal 3 (Citizen): @ ~3,6864 MHz, C_L3 = 16-18 pF
For resistors and capacitors, see the attached schematic.
[used crystek.com 's introduction to pierce gate crystal oscillators (pdf) for determining these values]

Lastly, the schematic and PCB layout created using EAGLE are attached below.

Again, sorry for the long post.
Thank you very much for your time! I appreciate every bit of advice.

Best regards,

Max

 

Yes. I often use machine tool (round) IC sockets. They make reliable connections and the 2.54 mm spacing marches the lead spacing of many subminiture leaded crystals when a set of 3 pins is used.

 

Anyone else? Comments on my circuit maybe?

 

Now that you asked, putting three outputs of the 74HC14 in parallel while the inputs are not all connected together is almost certain to not work. You can Wire-OR open collector gates and inverters but not those with devices with active pull-ups. It would be best to have a switch or jumpers to select which oscillator connects to the output, or using a single oscillator use jumpers to select which RC feedback network is to be used.

 

Please define "shock impulses".

 

@DickCappels: thanks for your answer! Can you elaborate on why connecting them to a single output creates problems? Also, do the jumpers not create additional resistances or other disturbances in the circuit?
Lastly, am I right in the assumption that I do not need a specific input (so, a power source) inside the feedback network?

@BR-549: A physical shockwave with high acceleration (up to 10000 G) that fades within a few milliseconds. Imagine screwing your circuit to a table and then smashing that table with a big hammer or something.

 

The outputs would all be "fighting" each other unless they all have the same inputs.

For example let's say you plug a crystal into U1C, but it won't be able to oscillate because its output is looking into the low impedance outputs of the other two Schmitt triggers.

In short, tying together two or more CMOS outputs that are not tri-state outputs is not ok.

The jumpers will introduce some stray circuit elements but that should not be significant.

If I understand your last question, no -you do not need to add anything to the feedback circuit but the 3 pf load capacitance can probably be supplied partly or entirely by the output and input capacitance of the Schmitt triggers. Check the datasheet.

 

Thank you very much, that clarifies it a bit! That might even be the single root of all my problems, so I'll investigate into that.

The last question refers to the beginning and continuation of the oscillation.
The only thing that requires power inside the whole circuit is the schmitt-trigger.If I understood it correctly, the oscillation itself happens as an amplification of natural electrical noise inside the circuit, some other starting signal or an additional power supply is not necessary.
Is the above correct?

 

Essentially correct, it starts itself.

By the way resistors are not needed in series with the unused CMOS inputs tied to ground or Vdd, you can tie them directly without a problem.

 

Nice. You've been a great help! I'll see if this fixes my problem tomorrow and report back.

Good to know! Another student told me this would be good practice just to be safe, but well! Is there any harm in keeping it this way or would I be better off removing them?

 

The extra resistors do no harm.

With TTL logic unused inputs were usually pulled up to Vcc through a resistor. Why resistors were used I don't really remember -something about protecting the input from power supply transients, but now I am skeptical.

 

Hey!

So I tried your suggestion last week without getting results unfortunately, which led me to the thought of trying to test it with a breadboard first and seeing if I actually have the basic circuit right (and testing the plug-in thing). Because of a lack of parts, I only tried it with the setup for the ~4MHz crystal only, but after confirming that it worked (perfectly actually), I tried plugging in the other crystals aswell. The measured frequencies were obviously not equal to the ones specified on the crystal datasheets, but getting some kind of stable oscillation was a great relief.

With this, I designed new PCB circuits and decided to create one for each crystal, simply to eliminate error sources. I've attached two designs: the first is for a through-hole type crystal, the second for an SMD type which needs a connection to ground).

For the first one, I do not get any.oscillation at all. The second one does output some oscillation, but the frequency is all over the place (ranging from 7 MHz up to 230 MHz at times [specified freq. is 12 MHz]).
I checked all connections and the voltage readings also seem to be okay. I really don't know what the problem is and I am starting to get rather frustrated, since this should be a really easy circuit.

I would be extremely grateful if someone could take a closer look and point me to possible error sources again. I will also do some more testing in the meantime.

Best regards,

Max

PS: The pin headers next to the IC are used for the frequency measurement (definitely not an optimal solution, but I was out of BNC connectors and thought this should work properly). Also, to only use one layer of the PCB I placed all the SMD parts on the bottom, which means that pin 1 of the IC is at the top left in the pictures.

EDIT: can't upload files for some reason. Always getting an error that does not specify what is wrong. Will try again in a few minutes

 

Sorry for double-posting, but I was not able to upload the pictures because of some weird undiscriptive error and crossed the time limit for editing in the process.
Here are the images hosted on imgur: http://imgur.com/a/WU9nb

 

If you have a schematic please post it, the layout drawing is difficult to interpret. Is there a power supply decoupling capacitor on the board? Is that a one sided or two-sided board?

As for the oscillator that won't start, your luck might improve if you use an inverter without hysteresis (a non-Schmitt trigger inverter). It might be that the hysteresis, which varies greatly from chip-to-chip, is too great for the oscillation to start or too great to start with a crystal with a low Q. A 74HC04 would be a better bet.

This oscillator can be finicky -if you know the crystal manufacturer's recommended load capacitance, use those values. The values on your schematic appear to be much smaller than I usually see at these low frequencies.

As for the oscillator that appears to oscillate from 7 MHz up to 230 Mhz, It is highly doubtful that a 74HC14 can oscillate in the hundreds of MHz. I am not sure it can do 20 Mhz at 3.3V. The most likely reason for the wild readings has something to do with the measurement. What kind of instrument are you using to measure the frequency?

As an aside, the oscillator you have drawn will oscillate at the crystal's fundamental frequency, and most crystals below about 25 MHz are fundamental crystals, but some crystals, particularly those at higher frequencies are intended to oscillate at a harmonic or "overtone" of the fundamental frequency. With the later a crystal marked "30.00 MH" would oscillate near 9.00 Mhz in this circuit.

For now, it would be best to concentrate on the oscillator that doesn't start. Check the connections both on paper and the physical model. If there is not a ceramic power supply bypass capacitor on the board, add one right at the IC pins. Check the load capacitor values, or just try increasing them to see whether or not that is the problem.

By the way, the TI application note at the link below might give you some helpful insight if you haven't seen it before.
http://www.ti.com/general/docs/lit/getliterature.tsp?baseLiteratureNumber=szza043&fileType=pdf

 

@DickCappels: again, thank you so much for taking your time to help me!
I've attached the respective schematics to the imgur link from above. The terminal block on the top is connected to the power supply, and said capacitor is placed below that. The board is one-sided to ease manufacturing. The terminal block on the bottom connects to the respective crystal.

I am using the Texas Instruments SN-74HC14-D. Is this not a typical 74HC04 unit?

I do know the load capacitances and have calculated the values for the resistors and capacitors accordingly. As stated in my first post, I had to guess for the input- and output-capacitances of the inverter, but since the exact same setup has given me a perfectly stable oscillation at the expected frequency in a breadboard environment, I don't think my calculations are too far off. I was suprised at arriving at such low capacitor values aswell, but then again, a load capacitance of 8 or 12pF is also pretty much on the low end of the spectrum, right?
This pdf guided me in my calculations/assumptions:
https://www.google.co.jp/url?sa=t&r...fR1h7mKMPxVqu7pfA&sig2=799tPqbgIRxSa3f13n2lww

I have asked myself the same question. I am using an oscilloscope with a probe (I connect them to the pin headers), and I do see the possibility of the cable not being in an optimal state, but still - I have used the same oscilloscope to get a perfectly stable oscillation using the same circuit simply on a breadboard, so I think there must be something else.

I have heard of overtone oscillation, but since the crystal is labeled as a 12 MHz crystal I thought this would not be the case.

As for the bypass capacitor, I noted that removing it in my breadboard circuit did not result in any changes. The oscillation did start and continue as before. There is one in my cicuits, although the value is a bit strange (0.1uF, the label on the bag where it was in was wrong :/ ).
I will definitely recheck all the connections and voltages again though.

Also, thank you very much for the pdf, that looks extremely informative. I'll take a good look!

Best regards,

Max

 

(Some text removed for clarity)
I am using the Texas Instruments SN-74HC14-D. Is this not a typical 74HC04 unit?

They are different. The 7414 is an invering Schmitt trigger while the 7404 is an inverter without hysteresis.

I do know the load capacitances and have calculated the values for the resistors and capacitors accordingly. As stated in my first post, I had to guess for the input- and output-capacitances of the inverter, but since the exact same setup has given me a perfectly stable oscillation at the expected frequency in a breadboard environment, I don't think my calculations are too far off. I was suprised at arriving at such low capacitor values aswell, but then again, a load capacitance of 8 or 12pF is also pretty much on the low end of the spectrum, right?

For this frequency range, I think so.

I have asked myself the same question. I am using an oscilloscope with a probe (I connect them to the pin headers), and I do see the possibility of the cable not being in an optimal state, but still - I have used the same oscilloscope to get a perfectly stable oscillation using the same circuit simply on a breadboard, so I think there must be something else.

Good -you can actually see the waveform then. does that mean that you saw a 230 MHz waveform?

I hope you are using a 10X probe or higher with your scope so the cable and input capacitance don't affect the circuit.

I have heard of overtone oscillation, but since the crystal is labeled as a 12 MHz crystal I thought this would not be the case.

As for the bypass capacitor, I noted that removing it in my breadboard circuit did not result in any changes. The oscillation did start and continue as before. There is one in my cicuits, although the value is a bit strange (0.1uF, the label on the bag where it was in was wrong :/ ).
I will definitely recheck all the connections and voltages again though.

I would be surprised it the 12 Mhz crystal were an overtone crystal, but we need to be open to things that should not be but might be.

Those plastic plug-in breadboards are not to be relied upon for high frequency work because of their parasitic reactances and reistances. They might be ok for a quick check of some idea but when high frequency or high impedance circuits get moved to a printed circuit board things can change a lot. Yes, you can sometimes get away without bypass capacitors, but when things don't work right you should at least have some confidence in the power supply.

In your schematic, I only see one oscillator. One of the crystal sockets is connected to the inputs of two inverters which you should understand will not result in oscillation unless there are connections not shown on the schematic <= is this the one you said will not oscillate?

From the preceding my suspicions that the inverter needs to be changed from a Schmitt trigger to a "non-Schmitt" inverting buffer have increased.

 

How could I not have seen that, I just copy-and-pasted the name you wrote....my bad. Of course they are not the same.

Not really, the values were simply all over the place. I'll check again. Also, thanks for the tip about the probe, I will look into that.

You are probably referring to the second schematic. In there, the X-3 connection of the terminal block is simply connected to ground, since the SMD crystal I want to use needs to have a GND connectoon. The labels are changed in the schematic to make it visually more appealing, but the connections should be alright. If you are referring to the JP1 block in the bottom however - this is the pin header I attach the probe to for measuring.

I would like to get back to the problem that irritates me the most, since I somewhat cannot image that the inverter really is at fault (in this example). I will isolate the case just to illustrate what I mean.

I used the circuit in the first example (through-hole crystal @ 3.6864 MHz with C_L = 16-18 pF) and calculated the corresponding resistor and capacitor values based on the pdf I linked above (which, after reading your pdf, seems to take a bit more factors into account for the calculation): I arrived at 22pF caps, a 2k R_S and a 5,6M R_F, assuming 3pF of stray capacitances on the PCB and 5 pF each for the input and output capacitances of the inverter gate.
Using these values, I successfully modeled the circuit on a breadboard. Stable, exact frequency visible in the oscilloscope. Exactly because a breadboard is not a very good long term solution, I was assuming that if it works there, it is guaranteed to work on a PCB when using the same parts (value wise). Or, in other words, the stray capacitances and other disadvantages of a breadboard are no longer present on a PCB, so it should work even better.
With this thought, I changed my PCB design to resemble that of my breadboard circuit and created the new PCB (as you can see in the imgur link). When measuring the performance of the new PCB, I used the same equipment - multimeter, oscilloscope, probe and power supply - but I get absolutely no oscillation.
I get that going higher in the frequency spectrum (say 20 MHz) might require a different inverter, but in this case alone, can the inverter or the probe really be at fault? I just can't imagine that.
I will measure the voltages and check the connections and every possible point again, but if you have any other suggestions or suspicions, please let me know.

Sorry for the long post!

Best regards,

Max