Need help finding a VCO/PLL on the chip

[videopho]

A device or component that can put out a square-wave TTL signal between 20khz-100khz, 50% duty cycle.
The output frequency tuning would be part of the external R/C circuit built around it. i.e. variable pot or voltages.
Frequency sweeping would be required as part of my application.

 

[William Gaatjes]

:Squeezing and rubbing hands together like an eager salesman:

Dear forum member, i have just what you need :

The timerblox family from Linear technology.

If you can solder an IC with a SOT 32 casing, you will be fine.

The LTC6990 is a voltage controlled oscillator.
http://www.linear.com/product/LTC6990

The LTC6992 is a pulse width modulation ic with adjustable frequency through Rset.
http://www.linear.com/product/LTC6992

 

[videopho]

I'm looking into those two devices at the moment.
My only option is using the resistance adjustable for tuning the Fo output.
Thanks

 

[William Gaatjes]

Just an idea i had with this nifty chip and might come in handy for those people who need it :

The LTC6992 is particularly handy when you want to have an analog control signal between 2 electrically isolated circuits.

http://www.fairchildsemi.com/pf/H1/H11L1M.html
http://www.fairchildsemi.com/ds/H1/H11L1M.pdf
http://www.fairchildsemi.com/ds/6N/6N137.pdf

If you combine the LTC6992 with an optocoupler such as H11L1M or 6N137 and connect a current source or separate logic gate together with a low pass filter on the output pin of the optocoupler, you have a circuit that is insensitive to temperature drift over a large range. And there is no need for calibration.

And you can transfer an analog signal because you convert the analog signal into a digital signal with the LTC6992, transfer the digital signal with the H11L1M or 6N137 and then convert the digital signal back into an analog signal again by use of the low pass filter. Just use a pwm frequency signal above 200kHz.

Because the H11L1M and the 6N137 has an open collector output, you do need to make a small low voltage drop 20mA current source to give the H11L1M or the 6N137 a sort of push pull output.
or use a pullup resistor together with a separate logic gate.

A simple low voltage drop current source can be made :
Please note that the artist forgot to draw the small emittor resistor to make this circuit more reliable.

An improved version :

For great explanations how it all works :
http://www.circuit-fantasia.com/circuit_stories/combined_list_of_circuits.htm

 

[videopho]

My company buyers are unable to find the LT part you recommended?
Do you recommend any other alternative parts?

 

[William Gaatjes]

The good old 4046 cmos pll has a VCO onboard.

Pdf :
http://www.nxp.com/documents/data_sheet/HEF4046B.pdf


Otherwise you can attempt to make a vco from discrete components. I made this quick and dirty schematic for you to test and simulate in LTspice.
It is based on a sawtooth generator. The capacitor is charged with a certain current from a voltage controlled current source. The higher the voltage, the higher the current, the shorter the charge time and thus the higher the frequency. The Dflipflop at the end divides the VCO output frequency by 2, to make a symmetrical square wave output.
I only had 1 coffee, so i might have forgotten something in the formula.

 

[MrDudeMan]

Why not use a microcontroller? There are way more adjustments and the code to produce something like this would be incredibly simple.

onboard A2D to read voltages from whatever you want
onboard PWM generator to make any waveform you want

I do this all the time for a dollar or less and 30 minutes of programming. It is vastly superior to using analog components because it is more scalable and configurable, especially the frequency sweep requirement. If you are at all interested in determinism, this is basically your only option.

The accuracy is entirely dependent on your clock source, but since you didn't mention accuracy parameters, I'm assuming that isn't important.

I will provide the code to get you started if you'd like to use a microcontroller.

 

[TecHNooB]

you should probably use a micro like dudeman says. the speeds you want are really low.

 

[videopho]

Why not use a microcontroller? There are way more adjustments and the code to produce something like this would be incredibly simple.

onboard A2D to read voltages from whatever you want
onboard PWM generator to make any waveform you want

I do this all the time for a dollar or less and 30 minutes of programming. It is vastly superior to using analog components because it is more scalable and configurable, especially the frequency sweep requirement. If you are at all interested in determinism, this is basically your only option.

The accuracy is entirely dependent on your clock source, but since you didn't mention accuracy parameters, I'm assuming that isn't important.

I will provide the code to get you started if you'd like to use a microcontroller.

Pm'd

 

[Modelworks]

The 555 chips can easily do this and are available from many manufacturers and are very cheap too. !
Search google for 555 VCO

 

[MrDudeMan]

The 555 chips can easily do this and are available from many manufacturers and are very cheap too. !
Search google for 555 VCO

Timers aren't going to get anywhere near the precision he needs without a microcontroller, so there's no point in using one since the microcontroller can handle the output on its own.

He needs 39.350KHz and other similarly precise frequencies.

 

[Modelworks]

Timers aren't going to get anywhere near the precision he needs without a microcontroller, so there's no point in using one since the microcontroller can handle the output on its own.

He needs 39.350KHz and other similarly precise frequencies.

I can generate any frequency needed with a 555, just need to select the right one from the correct manufacturer. We did it long before there were MCU that could do it. MCU are not really ideal for generating precise frequencies unless you spend quite a bit on ones with 32 bit timers, the majority are going to be 16 bit which when used with PWM generation that those MCU use loses a lot of precision. The 555 benefits from low capacitance on the outputs, where MCU have high capacitance. If you look at pulses generated from an MCU they typically have a sloped rise and fall time that is very hard to remove, where the 555 has a nearly perfectly vertical rise and fall time.

I happen to have a 555 that connects to a flip flop and generates 50% duty cycle anywhere from 100hz to 747KHZ with accuracy to 2 decimal points.
It is a signal generator I built and use all the time.
People are far too quick now to jump to a MCU for everything when good old analog circuits are sometimes better.

 

[MrDudeMan]

MCU are not really ideal for generating precise frequencies unless you spend quite a bit on ones with 32 bit timers, the majority are going to be 16 bit which when used with PWM generation that those MCU use loses a lot of precision.

Wrong. You can cascade timers to get as much precision as you want and it scales perfectly with frequency. Faster clock + more timers = more precision. I use a microcontroller that costs $1.35 and has an event system that can create 128 bit timers. The precision is linear with clock frequency and the range is unlimited with interrupt vectors.

The 555 benefits from low capacitance on the outputs, where MCU have high capacitance. If you look at pulses generated from an MCU they typically have a sloped rise and fall time that is very hard to remove, where the 555 has a nearly perfectly vertical rise and fall time.

Are you kidding me? I'm not sure if you've looked at a microcontroller that was developed after 1995, but the typical output cap is < 5pF. If you really care, you can buffer it, so that doesn't even make any sense to argue about anyway. I looked at four 555 timer datasheets and couldn't find one that had less than 2pF, so this is not even a valid argument.

I happen to have a 555 that connects to a flip flop and generates 50% duty cycle anywhere from 100hz to 747KHZ with accuracy to 2 decimal points. It is a signal generator I built and use all the time.

And? I have a clock source that is verifiably more configurable and more accurate using a microcontroller. I, too, use it all the time, and it has a range of 1E-6 Hz to 1E7 Hz. I'd love to see you find a God clock (industry standard term referencing highly stable clocks) that is not trimmed internally using digital circuits.

What do you do when you need a different wave shape? I spend 30 seconds generating the points in Excel and then I'm done. What do you do when you need exactly 42.5% duty cycle? I edit 2 characters in the source code.

An analog solution won't even work in his application unless you get crazy with the control circuit. He needs frequency sweeps with specific time intervals. That isn't even a minor complication on a microcontroller, but it's a full blown design project with purely analog components.

I can generate any frequency needed with a 555, just need to select the right one from the correct manufacturer. We did it long before there were MCU that could do it. ... People are far too quick now to jump to a MCU for everything when good old analog circuits are sometimes better.

People are far too quick to cite the "I used to walk to school in the snow with no shoes" paradigm of circuit design. Yes, we made it to the moon with vacuum tubes, but that doesn't mean it's still the right way to do it. I get paid to design analog circuits, so I understand they have benefits. Using them for timers, especially at low frequencies, is not one of their benefits. Microcontrollers were practically designed for this.

 

[WhoBeDaPlaya]

N Dallas huh? Swing by Beltwood Parkway / Maxim Drive. We'll set you up

 

[Modelworks]

Wrong. You can cascade timers to get as much precision as you want and it scales perfectly with frequency. Faster clock + more timers = more precision. I use a microcontroller that costs $1.35 and has an event system that can create 128 bit timers. The precision is linear with clock frequency and the range is unlimited with interrupt vectors.

All of that is fine for a one off design, but adding an MCU to a design that needs to be programmed before it can be used complicates production, you have to get the manufacturer to pre-program the MCU or you have to do it in factory which can be very costly.

Are you kidding me? I'm not sure if you've looked at a microcontroller that was developed after 1995, but the typical output cap is < 5pF. If you really care, you can buffer it, so that doesn't even make any sense to argue about anyway. I looked at four 555 timer datasheets and couldn't find one that had less than 2pF, so this is not even a valid argument.

Again, when you start adding more components in addition to the MCU you are increasing cost . Can an MCU do the job, yes , is an MCU the best choice, should always be considered after looking at analog solutions.

And? I have a clock source that is verifiably more configurable and more accurate using a microcontroller. I, too, use it all the time, and it has a range of 1E-6 Hz to 1E7 Hz. I'd love to see you find a God clock (industry standard term referencing highly stable clocks) that is not trimmed internally using digital circuits.

The point I was making is that the 555 series is more capable then people realize, they are quick to write it off because it takes a little more time to design the circuitry and often just jump straight to MCU.

What do you do when you need a different wave shape? I spend 30 seconds generating the points in Excel and then I'm done. What do you do when you need exactly 42.5% duty cycle? I edit 2 characters in the source code.

And how are you going to push that change in code to thousands of devices in the field ? Just about every product I have seen that uses a VCO doesn't need it changed once it is in production, so being able to change waveform doesn't matter.

An analog solution won't even work in his application unless you get crazy with the control circuit. He needs frequency sweeps with specific time intervals. That isn't even a minor complication on a microcontroller, but it's a full blown design project with purely analog components.

The application was never specified, it could have been for a function generator all the way up to satellite systems. Full design project it isn't, any competent engineer can turn out an analog solution . Another problem with MCU vs analog is environment. MCU do not handle some environments well and while some MCU are designed for specific environments , sometimes you can't get one in to fit the needed specifications. Analog can be designed with just about any needed specification and without having to have a manufacturer design a new chip. It wouldn't be as simple as a single chip, but it would implement the design far cheaper than having a company produce a custom designed chip.

People are far too quick to cite the "I used to walk to school in the snow with no shoes" paradigm of circuit design. Yes, we made it to the moon with vacuum tubes, but that doesn't mean it's still the right way to do it. I get paid to design analog circuits, so I understand they have benefits. Using them for timers, especially at low frequencies, is not one of their benefits. Microcontrollers were practically designed for this.

The issue I was pointing out was that if you look at the current trend from new engineers to the field they are having a hard time accepting that analog circuits can do what the digital ones can and sometimes far cheaper. I work with a lot of new engineers and repeatedly see people clamoring for a specific IC to do the work when just a few analog parts can do it, yet they never consider them and instead re-work the entire design just so they can use a packaged IC or want to use an MCU for everything. One big one I see repeatedly is people using solid state relays to switch large loads, implementing huge heat sink, fans, and thermal control, when a relay, transistor and diode would work. They consider the mechanical relay inferior because it requires external parts and isn't some sort of package with pins sticking out.

I use MCU daily from the old 8bit and up and currently design FPGA with multiple cpu cores running . I agree analog can't do everything but it can do far more than people give it credit for.

 

[Modelworks]

N Dallas huh? Swing by Beltwood Parkway / Maxim Drive. We'll set you up

Maxim is one of my favorite companies, always been willing to help when needed and make good products as well. I have dealt with a lot of companies and the majority give smaller production companies the cold shoulder, but not Maxim, they provide good support all around.

 

[MrDudeMan]

All of that is fine for a one off design, but adding an MCU to a design that needs to be programmed before it can be used complicates production, you have to get the manufacturer to pre-program the MCU or you have to do it in factory which can be very costly.

Again, when you start adding more components in addition to the MCU you are increasing cost . Can an MCU do the job, yes , is an MCU the best choice, should always be considered after looking at analog solutions.

I can do what he wants with a microcontroller and two decoupling capacitors. The capacitors cost a few pennies and the microcontroller is $1.35 in quantities of 1. To be clear, this is a total solution which includes the part you continue to ignore - the frequency sweep.

The point I was making is that the 555 series is more capable then people realize, they are quick to write it off because it takes a little more time to design the circuitry and often just jump straight to MCU.

555 timers aren't nearly as capable as you're making them out to be. Please provide a simple and elegant solution that will sweep the frequency coming out of a 555 timer between precisely 39.350KHz and 40.000KHz by taking .1KHz steps at exactly 2 second intervals. I actually asked him what his requirements are instead of making blanket statements about analog versus digital, but I'll continue to debate this because using a microcontroller made sense even before I knew the details.

And how are you going to push that change in code to thousands of devices in the field ? Just about every product I have seen that uses a VCO doesn't need it changed once it is in production, so being able to change waveform doesn't matter.

First and foremost, my original statement was directed at your signal generator comment. Signal generator implies more than a square wave at 50% duty cycle. Yours is stuck with rough duty cycle adjustments over a tiny range of frequencies. That was the point you ignored since there is obviously no way to refute it.

The OP never asked for a VCO. It was offered as a solution by WG. I commented about microcontrollers because this is a terrible application for a VCO. It's funny, though, that you said a VCO never needs to be changed once it's in production. That couldn't be further from the truth, but given that you weren't applying it to any particular situation, there's really no point in debating it.

Regarding your other comment, it's called a firmware update. Regardless of the quantity, you are wrong to cite this as a reason not to use a microcontroller because it's a solved problem. Since you felt like bringing it up, though, I'd love to hear your solution for releasing a fix or an update to your pure analog solution once it's in the field. Oh, that's right, you can't unless all of your customers own an oscilloscope and a soldering iron. There's a reason the average house has over 70 microcontrollers or CPUs.

The application was never specified, it could have been for a function generator all the way up to satellite systems. Full design project it isn't, any competent engineer can turn out an analog solution.

Please provide a frequency sweep solution with pure analog components in less than an hour using less than $2 worth of components with a board footprint of 1 sq. in. or less. That's very generous compared to how long it will take me to do the same thing with a microcontroller, so "any competent engineer" should be able to do it if it's as easy as you say. Hey, I could be wrong, so here is your opportunity to prove it.

Another problem with MCU vs analog is environment. MCU do not handle some environments well and while some MCU are designed for specific environments , sometimes you can't get one in to fit the needed specifications. Analog can be designed with just about any needed specification and without having to have a manufacturer design a new chip. It wouldn't be as simple as a single chip, but it would implement the design far cheaper than having a company produce a custom designed chip.

There are so many generalizations in the above quote that I'm just going to ignore it because it has nothing to do with the topic.

The issue I was pointing out was that if you look at the current trend from new engineers to the field they are having a hard time accepting that analog circuits can do what the digital ones can and sometimes far cheaper. I work with a lot of new engineers and repeatedly see people clamoring for a specific IC to do the work when just a few analog parts can do it, yet they never consider them and instead re-work the entire design just so they can use a packaged IC or want to use an MCU for everything.

Since you obviously haven't considered it, they might be wondering if you're an old dog who can't learn new tricks, which is similar to what you're saying about them. What your colleagues may see that you simply haven't yet is scalability and time to market. Also, you're flat out wrong to say analog circuits can do everything digital circuits can do with the obvious implication being that they'd do it just as well. That's like saying digital circuits can do everything analog circuits can do, which obviously isn't true.

One big one I see repeatedly is people using solid state relays to switch large loads, implementing huge heat sink, fans, and thermal control, when a relay, transistor and diode would work. They consider the mechanical relay inferior because it requires external parts and isn't some sort of package with pins sticking out.

Assumptions, assumptions, and more assumptions. If you want to continue with the blanket statement game, then vacuum tubes are far superior to your solution. Also, comparing a solid state relay to a mechanical relay is ludicrous. They are completely different in many ways and can not be used interchangeably. To name a few ways that solid state relays are superior to mechanical relays: no output bounce, no resistance degradation, much faster operation, much longer lifetime, and they are totally silent. Similarly, they have disadvantages, so again, your blanket statements are ridiculous. This has even less to do with the topic at hand, but I couldn't ignore it because it's rife with inaccuracy.

I use MCU daily from the old 8bit and up and currently design FPGA with multiple cpu cores running . I agree analog can't do everything but it can do far more than people give it credit for.

I don't recall saying analog circuits don't deserve credit. I rely on the fact that analog circuits are useful to feed my kids. I'm refuting your dismissive attitude toward microcontrollers, especially for the application at hand, because your reasons for preferring a pure analog solution are not sound.

 

[Modelworks]

I can do what he wants with a microcontroller and two decoupling capacitors. The capacitors cost a few pennies and the microcontroller is $1.35 in quantities of 1. To be clear, this is a total solution which includes the part you continue to ignore - the frequency sweep.

Doing a frequency sweep with a VCO is kind of the point in using a VCO. You still have to program the MCU either at the manufacturer or the factory, that isn't cheap and updating can be problematic.

555 timers aren't nearly as capable as you're making them out to be. Please provide a simple and elegant solution that will sweep the frequency coming out of a 555 timer between precisely 39.350KHz and 40.000KHz by taking .1KHz steps at exactly 2 second intervals. I actually asked him what his requirements are instead of making blanket statements about analog versus digital, but I'll continue to debate this because using a microcontroller made sense even before I knew the details.

I went with the information given, if the OP supplied more information then I would have used that information. If someone said they needed a VCO to produce sound for a device would you still have said use MCU ? I wouldn't use a 555 in the design you describe, I would use a 7556.

First and foremost, my original statement was directed at your signal generator comment. Signal generator implies more than a square wave at 50% duty cycle. Yours is stuck with rough duty cycle adjustments over a tiny range of frequencies. That was the point you ignored since there is obviously no way to refute it.

A 555 feeding a flip flop isn't a rough duty cycle, a radiation hardened version produced a lot of the clock signals that were used in the space shuttle flight computers. I wouldn't call 10hz to 750KHZ a tiny range of frequencies.

The OP never asked for a VCO.

Read the subject line again.

It was offered as a solution by WG. I commented about microcontrollers because this is a terrible application for a VCO. It's funny, though, that you said a VCO never needs to be changed once it's in production.

Why would I change the setting on a VCO that is designed with specific specifications to do a specific task after the product is shipped ? VCO are used in millions of products and never are changed after production. If the EE designed it wrong or with the wrong code or specifications then yes it would have to be changed.

Regarding your other comment, it's called a firmware update.

How are you going to get that update on thousands of devices without having to implement more hardware to allow the user of the devices to upgrade the firmware ? Not everyone has internet access or access to a computer or the knowledge of how to perform a firmware update. We have some radios right now that use VCO located in very remote areas, Kenya to be specific, that have no chance of ever being updated with firmware. If we had designed something with a MCU that required firmware updates it would be like a lot of the hardware that has been trialed there, discarded junk.

Regardless of the quantity, you are wrong to cite this as a reason not to use a microcontroller because it's a solved problem. Since you felt like bringing it up, though, I'd love to hear your solution for releasing a fix or an update to your pure analog solution once it's in the field.

You don't update the solution because you don't need to. There is no chance of an uncaught firmware error or something the manufacturer forgot to include in the datasheet, it works when you make it and every time after that.

Oh, that's right, you can't unless all of your customers own an oscilloscope and a soldering iron. There's a reason the average house has over 70 microcontrollers or CPUs.

Again, I am not saying that a MCU doesn't have uses, but EE should not be jumping to them as the first choice without considering other options.

Please provide a frequency sweep solution with pure analog components in less than an hour using less than $2 worth of components with a board footprint of 1 sq. in. or less. That's very generous compared to how long it will take me to do the same thing with a microcontroller, so "any competent engineer" should be able to do it if it's as easy as you say. Hey, I could be wrong, so here is your opportunity to prove it.

If you can provide the exact specifications I can design a solution around it, but I can't do it with generalizations.

Since you obviously haven't considered it, they might be wondering if you're an old dog who can't learn new tricks, which is similar to what you're saying about them.

I have quite a few people who are thankful that I showed them that in electronics there are many ways to do things and each has their trade offs, that a EE doesn't focus on a specific segment of technology but looks at the entire field.

What your colleagues may see that you simply haven't yet is scalability and time to market.

What they see is someone who can consider all the options and doesn't resort to assuming that there is only one right way to do something.

Also, you're flat out wrong to say analog circuits can do everything digital circuits can do with the obvious implication being that they'd do it just as well. That's like saying digital circuits can do everything analog circuits can do, which obviously isn't true.

Where did I say that ?
I said "Analog can be designed with just about any needed specification and without having to have a manufacturer design a new chip. "
And when I said it I was referring to circuits that analog and digital have in common, not that you can do a Cray supercomputer in analog.

Assumptions, assumptions, and more assumptions. If you want to continue with the blanket statement game, then vacuum tubes are far superior to your solution.

You seem to have a hard time debating something without trying to ridicule the other person, debating a topic has no place for emotions.

Also, comparing a solid state relay to a mechanical relay is ludicrous. They are completely different in many ways and can not be used interchangeably. To name a few ways that solid state relays are superior to mechanical relays: no output bounce, no resistance degradation, much faster operation, much longer lifetime, and they are totally silent. Similarly, they have disadvantages, so again, your blanket statements are ridiculous. This has even less to do with the topic at hand, but I couldn't ignore it because it's rife with inaccuracy.

Again with the ridicule.
Comparing two items "relays" with each other is far from ludicrous, they share the same purpose being "relays" and yes they can be used interchangeably and that is done every day. Using a solid state relay to switch 15000W of lighting when a relay can do the same and has a 5 million switch life cycle is what I am referring to, to do that with solid state relays requires heat sinks and proper cooling, with a relay it requires nothing but the relay.

I don't recall saying analog circuits don't deserve credit. I rely on the fact that analog circuits are useful to feed my kids. I'm refuting your dismissive attitude toward microcontrollers, especially for the application at hand, because your reasons for preferring a pure analog solution are not sound.

I did not say MCU are not able to the job but that other options exist. I have used MCU every day for 20 years in projects, currently working with Altera ( if that isn't digital I don't know what is), so I am far from dismissive but I always consider all the options, not just resorting to the 'grab a chip to do it' mindset.

 

[MrDudeMan]

Since you are obviously reading impaired, let's take it one comment at a time. Edited the frequencies to be even steps of .1KHz.

Please provide a simple and elegant solution that will sweep the frequency coming out of a 555 timer between precisely 39.400KHz and 40.000KHz by taking .1KHz steps at exactly 2 second intervals. I actually asked him what his requirements are...

I went with the information given, if the OP supplied more information then I would have used that information.

Remember you have less than 2 dollars and 1 sq. in. of board space. Go.