Author Topic: How to increase TCXO output?  (Read 4401 times)

0 Members and 1 Guest are viewing this topic.

Offline rcbuckTopic starter

  • Frequent Contributor
  • **
  • Posts: 355
  • Country: us
How to increase TCXO output?
« on: April 30, 2019, 12:43:48 am »
I have a 16.384 MHz TCXO which has the standard 0.8 volt Pk-Pk clipped sine wave output. I need to increase that level to ~3 volts to drive a SN74LVC1G14 inverter. I tried a single transistor common emitter amplifier but didn't have any luck with that. I have decided the best way to handle this is a high speed comparator. That would give me a nice square wave output.

I am going to use the STMicroelectronics TS3011 part. If I am interpreting the data sheet properly, it will work. On page 10, Figure 9, the upper trip point appears to be slightly more than 1 mV over the reference level. The lower trip point is slightly less than 1 mV below the reference. So If I set the reference to about 300 mV the TCXO should easily drive the comparator.

Is my thinking correct as to how the comparator works?
 

Online edavid

  • Super Contributor
  • ***
  • Posts: 3464
  • Country: us
Re: How to increase TCXO output?
« Reply #1 on: April 30, 2019, 12:54:58 am »
If it's a 1-off circuit, I suggest just AC coupling it.  Use a 10K pot to bias the inverter input to the right place.  It may not meet absolute worst case specs, but it should work.

Otherwise, you could use an unbuffered 74LVC inverter as a 10X amplifier.
 

Offline rcbuckTopic starter

  • Frequent Contributor
  • **
  • Posts: 355
  • Country: us
Re: How to increase TCXO output?
« Reply #2 on: April 30, 2019, 03:04:08 am »
I may want to build more than 1 unit. If I remember correctly, I think I tried the inverter biasing circuit in a past project and it made the inverter run hot. I would also be somewhat concerned about reliability of this method due to temperature excursions.

I've never had any luck with 10X inverter amplifiers. The output level tended to follow the input level. Perhaps I had the circuit wrong. I just used a 10 meg resistor as the feedback element.

Do you foresee any issues with using the TS3011?
 

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 7678
  • Country: ca
Re: How to increase TCXO output?
« Reply #3 on: April 30, 2019, 03:52:22 am »
74LVC1G14 worked for me, (but not for your 100kHz oscillator), more discussion here:
https://www.eevblog.com/forum/projects/level-translate-the-output-of-a-tcxo-crystal-oscillator-module/msg1550348/#msg1550348
74LVC1GX04 looked like it has enough gain.

An AC-coupled comparator with some hysteresis would work too.
« Last Edit: April 30, 2019, 03:41:25 pm by floobydust »
 

Online edavid

  • Super Contributor
  • ***
  • Posts: 3464
  • Country: us
Re: How to increase TCXO output?
« Reply #4 on: April 30, 2019, 02:51:25 pm »
I may want to build more than 1 unit. If I remember correctly, I think I tried the inverter biasing circuit in a past project and it made the inverter run hot.
Since it's a Schmitt trigger inverter, that's not an issue.

Quote
I've never had any luck with 10X inverter amplifiers. The output level tended to follow the input level. Perhaps I had the circuit wrong. I just used a 10 meg resistor as the feedback element.
I've never had any problem with them  :-//

Quote
Do you foresee any issues with using the TS3011?
Not really, it's just overkill.
 

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 7678
  • Country: ca
Re: How to increase TCXO output?
« Reply #5 on: April 30, 2019, 03:27:08 pm »
It looks like the (sine) output of a XO does not like a dynamic load? OP, check your module's datasheet.
Clipped Sinewave to CMOS Conversion Circuit, Crystek App Note "... the TCXO will be pulled by the output transitions..."

This may mean using a comparator with hysteresis feedback to the oscillator output would cause trouble. I've never seen data on the "clipped sine-wave" oscillator circuit which seems common in the oscillator module industry.
Is it truly buffered from the crystal, or just the raw unbuffered gate output.

edit, found some answers:
Why do TCXOs have clipped sinewave outputs?
"Most ceramic LCC packaged TCXOs are only available with clipped sinewave outputs. There are many reasons why this is done, even though it frustrates many of the circuit designers who want CMOS output levels.

* CMOS output would add significant power dissipation to the TCXO IC. The resulting temperature gradients on the IC and in the package would limit the ability to achieve the best TCXO compensation.

* The wide range of possible CMOS loads would alter the compensation design as internal thermal gradients result in the TCXO not meeting specifications.

* CMOS outputs result in power supply and ground transients when the output is changing levels. This noise would adversely affect the phase noise performance of the TCXO.

* CMOS outputs generate much larger EMI/RFI signals that can result in difficulties meeting the USA FCC and other countries radiated energy limits from the final system. The clipped sinewave, being both low amplitude and mostly sinusoiddal, results in low signal levels and less harmonics.

* Many of the applications where these TCXOs are used have ASIC inputs that accept the clipped sinewave and convert to CMOS logic levels internally.

See our FAQ about converting clipped sinewave levels to CMOS. It is easy to do and low cost."

Converting Clipped Sinewave Output to CMOS
« Last Edit: April 30, 2019, 05:28:52 pm by floobydust »
 
The following users thanked this post: thm_w, edavid, Jacon

Offline rcbuckTopic starter

  • Frequent Contributor
  • **
  • Posts: 355
  • Country: us
Re: How to increase TCXO output?
« Reply #6 on: May 01, 2019, 06:31:30 am »
Just for reference, the part I am using is a Murata XTCLH16M384THJA2P0. The data sheet that Mouser links to does not cover this part. It is a 0.2ppm part and I had to contact Murata directly to get the data sheet. It is attached here.

I have ordered the TCXO from Mouser along with some inverters, comparators, and clock buffer parts. I will post results of my tests, probably this weekend or early next week.
 

Offline rcbuckTopic starter

  • Frequent Contributor
  • **
  • Posts: 355
  • Country: us
Re: How to increase TCXO output?
« Reply #7 on: May 03, 2019, 06:53:28 pm »
Tests have been completed with the TCXO output driving different parts. The Vcc supply for the TCXO and buffers is 3.26 volts. The output of the TCXO measures about 1.2 volts Pk-Pk unloaded. When driving any of the 3 buffers, the output drops to about 1.0 volts.

Using a SN74LVC1G14 as the buffer with a 10 meg ohm feedback resistor, the output from the buffer is about 3.0 volts Pk-Pk. The circuit configuration I used is the one floobydust shows in post #5. I received some NC7SZ04 parts in my Mouser order but didn't go to the trouble of breadboarding one of them. They should perform the same as the SN74LVC1G14 did.

Using a PL133-37 fanout buffer the level at all 3 outputs is 3.12 volts Pk-Pk.

Using a TS3011 comparator with the inverting input reference at 1.2 volts, the output is 3.24 volts Pk-Pk. As an additional test, I changed the TS3011 Vcc voltage to 5.0 volts. I also switched the high end of the reference pot to 5.0 volts and reset the reference voltage to 1.2 volts. With this configuration, the output of the comparator is 5.0 Pk-Pk.

These measurements are of course, subject to test equipment tolerances. I am usinga Tektronix TDS2014B scope for the signal measurements. The DC voltage measurements are with a Mastech MS8264 which is supposed to be about +/-0.5% accurate. The meter is roughly 10 years old.

The bottom line is that any of the buffer circuits will work in a 3.3 volt environment. For my project I will use the TS3011 with a Vcc of 5.0 volts. Everything in the project is running on 5 volts except for the oscillator. The other logic chips following the buffer specify a logic high input of 0.7 x Vcc. I will use the TS3011 output of 5.0 Pk-Pk as the driving source. Using a level of ~3 volts would be cutting it too close for comfort.
 

Online edavid

  • Super Contributor
  • ***
  • Posts: 3464
  • Country: us
Re: How to increase TCXO output?
« Reply #8 on: May 03, 2019, 08:08:00 pm »
Using a SN74LVC1G14 as the buffer with a 10 meg ohm feedback resistor, the output from the buffer is about 3.0 volts Pk-Pk. The circuit configuration I used is the one floobydust shows in post #5. I received some NC7SZ04 parts in my Mouser order but didn't go to the trouble of breadboarding one of them. They should perform the same as the SN74LVC1G14 did.

Those two parts are very different.

The SN74LVC1G14 is a Schmitt trigger, so the feedback resistor does nothing.

The NC7SZ04 is a (buffered) inverter, so will act as an amplifier, without hysteresis.

Quote
The other logic chips following the buffer specify a logic high input of 0.7 x Vcc. I will use the TS3011 output of 5.0 Pk-Pk as the driving source. Using a level of ~3 volts would be cutting it too close for comfort.

Why don't you use the NC7SZ04 with a 5V supply?

 

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 7678
  • Country: ca
Re: How to increase TCXO output?
« Reply #9 on: May 03, 2019, 08:38:37 pm »
If your comparator circuit has hysteresis feedback connecting to the oscillator output, then it better be a pretty large resistor. TCXO oscillator modules are usually rated 10k//10pF load maximum, so a 10k resistor there would be non-linear loading of the TCXO I think.
 

Offline rcbuckTopic starter

  • Frequent Contributor
  • **
  • Posts: 355
  • Country: us
Re: How to increase TCXO output?
« Reply #10 on: May 03, 2019, 11:47:48 pm »
Quote
Why don't you use the NC7SZ04 with a 5V supply?
I tried that with the SN74LVC1G14 but could only get about 0.2 volts more out of it than when it was running on the 3.3 volt supply. Probably something to do with the TCXO drive level not being high enough.

Quote
If your comparator circuit has hysteresis feedback connecting to the oscillator output......
I don't know what the internal circuitry of the comparator looks like. It does have internal hysteresis according to the data sheet. The TCXO output level into the comparator was about 1.0 volts Pk-Pk. It is probably loading it somewhat as the wave form isn't square on the negative portion of the cycle. I am only interested in counting pulses so as long as there is no ringing I am good.
 

Online edavid

  • Super Contributor
  • ***
  • Posts: 3464
  • Country: us
Re: How to increase TCXO output?
« Reply #11 on: May 04, 2019, 02:07:25 am »
Quote
Why don't you use the NC7SZ04 with a 5V supply?
I tried that with the SN74LVC1G14 but could only get about 0.2 volts more out of it than when it was running on the 3.3 volt supply. Probably something to do with the TCXO drive level not being high enough.

That doesn't really make sense.  If the drive level is high enough to make the Schmitt trigger switch, you should get full output.  Was the input AC coupled?

In any case, the NC7SZ04 would behave differently.
 

Offline rcbuckTopic starter

  • Frequent Contributor
  • **
  • Posts: 355
  • Country: us
Re: How to increase TCXO output?
« Reply #12 on: May 04, 2019, 05:29:52 pm »
Quote
That doesn't really make sense.
I agree, but the that is what I was seeing on the scope. The output from the TCXO is AC coupled with a 1 nF cap. The data sheet says the output has to be AC coupled.

I breadboarded the NC7SZ04 this morning using a 1 meg ohm feedback resistor. The NC7SZ04 Vcc was connected to the 5 volt power supply feed. Screen shot is attached. As you said, it behaves differently. The input pin is showing 3.36 V Pk-Pk and the output pin is showing 5.36 V Pk-Pk. The power supply DC levels are actually 3.26 volts and 4.99 volts. So the Pk-Pk values may be picking up random noise to create the higher readings. The 3.36 volt input reading is probably due to the feedback resistor.

The scope leads are both in the 10X mode. If I change the input probe to 1X, it kills the output from the NC7SZ04. That is why I am going to use the comparator. The 1.0 volt Pk-Pk output from the TCXO is just barely enough to drive the buffer. If any stray capacitance were to show up in the PCB traces, it could drop the level enough that the buffer isn't triggered. Of course, if you hand picked the buffer you may be able to get around that issue. But why take the chance?
 

Online edavid

  • Super Contributor
  • ***
  • Posts: 3464
  • Country: us
Re: How to increase TCXO output?
« Reply #13 on: May 04, 2019, 09:28:48 pm »
The scope leads are both in the 10X mode. If I change the input probe to 1X, it kills the output from the NC7SZ04.

A 1X scope probe usually has a capacitance of about 130pF.  At 16MHz, the reactance is 75ohms, obviously a heavy load on the TCXO.  I don't think it's a meaningful test.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf