Clock & Timing - Low Noise Frequency Synthesizers

TOM FLOYD: Hello everyone and a welcome to today's Clock and Timing webinar.

I'm Tom Boyd.

I'm your moderator for today's session and with me today is Bobby Matinpour, product marketing manager.

Welcome Bobby.

BOBBY MATINPOUR: Thank you for having me again, Tom.

TOM FLOYD: No problem, welcome back.

And before we jump in our conversation today for those of you just joining us live, just one quick housekeeping tip to keep in mind.

As always, we really want our webinar today to be interactive.

We would encourage you to ask a lot of questions.

So for those of you logged in to the Go to Webinar interface, feel free to shoot me any questions that you have during our session today.

I'll make sure to ask those on your behalf.

And with that said, I think we can go ahead and get started, so take us away.

BOBBY MATINPOUR: Thanks again for having me to come and talk about the clock and timing products.

Today we'll start the webinar with a small discussion about the communication infrastructure market to give us a little bit of motivation for this discussion and talk about its trends and its impact on clock and timing products.

We'll jump into synthesizer basics and applications and talk about how these trends in the market specifically impact the synthesizers and local oscillators.

Then we will review National's clock and timing solutions and more specifically PLLs and synthesizers and describe the new LMX2541 family of products, its advantages, architecture and performance and then we'll conclude.

It's probably very apparent to most of us that data communication is picking up, especially on the wireless handsets.

You've probably seen the increasing usage of streaming multimedia.

We're getting more and more used to relying on our mobile cell phones or other mobile devices to access our news and even download shows and content from the TV.

And if it's not us, definitely our children are starting to use these devices more and more in different and unique ways, video calling, interactive gaming, and in other ways where these devices are going to impact the usage of data over the wireless infrastructure in the next few years.

On the wired side also you will see a trend in increasing usage.

One reason is to support the wireless infrastructure, because that information that we send from our cell phones, the data that goes from the cell phones to the towers has to get communicated through the wired infrastructure.

On top of that, we have increased social networking Web sites like YouTube, like Facebook and also YouTube.

So the video content is growing and that's impacting data center traffic as well.

So you will see that basically the point is that the current infrastructure, the way that it is deployed now, is not going to be sufficient for the next generation or the new applications that are coming online.

As a result, the carriers, the service providers, are accelerating their effort to upgrade their infrastructure and that itself results into new equipment.

New equipment results into higher performance components and that has impact as well in clock and timing and synthesizer products.

TOM FLOYD: Now, you had talked about this trend a little bit in the last webinar as well.

BOBBY MATINPOUR: Right.

TOM FLOYD: So, is what you're saying today is that it's not only a trend that's impacting clocking products, for example, but in this case...

BOBBY MATINPOUR: It's synthesizers as well as data converters and amplifiers which we discussed a little bit on the previous webinar.

TOM FLOYD: Right, okay.

BOBBY MATINPOUR: But today we're going to focus on the synthesizers.

TOM FLOYD: Okay, got you.

BOBBY MATINPOUR: Good question.

So a frequency synthesizer is used mostly as a local oscillator in RF transceivers in a radio.

And when I'm talking about a radio I'm talking about high-performance radio, not the transceivers that are fully integrated transceivers used in consumer applications like cell phones.

I'm talking about cellular base stations and repeaters.

These are the stations that actually receive the data from cell phones.

We're very familiar with those.

In addition to that, we're talking about point-to-point RF and microwave links.

These are the links used to bring data into remote areas, bring connectivity to these areas, sometimes used for governments and municipalities or private businesses.

When there's rough terrain it's not easy to lay cable or fiber, point-to-point RF and microwave links are used.

Also microwave backhaul is another method to support the base station infrastructure in such areas where fiber or cable does not exist.

They use a very high-speed microwave link to transmit from their wired network to the infrastructure, the wireless infrastructure.

TOM FLOYD: Okay, and are there any other applications or places where these can be used as well?

BOBBY MATINPOUR: Certainly, today we're going to focus about wireless infrastructure.

TOM FLOYD: Okay.

BOBBY MATINPOUR: However, the PLLs and synthesizers can be used as single clock output generators.

They can be used to clock high-speed or gigasample-per-second ADCs.

So there are other uses for these devices as well, but today we'll focus on how they are used in the wireless infrastructure.

TOM FLOYD: Okay.

BOBBY MATINPOUR: Very good.

Here I'm showing a very generic super heterodyne radio, receive and transmit, where local oscillators are used.

Here you will see an LMK4000 right in the middle that's taking a clock output either from an FPGA or from a SERDES, cleaning up the clock and distributing it to various analog components, such as ADCs, DACs and also into the local oscillators.

So you can see here that the local oscillator they used to help up-convert and down-convert to and from RF the information that's being received or transmitted.

Now, the criteria that we use to specify the performance requirements for a local oscillator are very critical in the system.

In particular error-vector-magnitude or integrated noise is a system-level spec that is translated down to local oscillators and other components.

And the amount that's budgeted for that load is generally very limited.

So this is a key criteria for local oscillators that they need to meet.

Another one is phase noise at particular offsets.

What I mean by that is at particular offsets like 600 kHz, 800 kHz, 1 MHz, these are specially specified based on standards and determined by the adjacent channel blockers.

Another one is also on the transmit side these phase noise offsets are determined by the amount of emissions that are dictated by the regulatory body.

TOM FLOYD: Now, can you explain a little bit more about what you mean by EVM budget for LL as well?

BOBBY MATINPOUR: Yes, so what happens is designers have a total budget for the EVM.

TOM FLOYD: Okay.

BOBBY MATINPOUR: And basically they delegate that or budget that for various components.

The power amplifier and the local oscillator, in general, get a large portion of that, that very large portion of that small amount.

And, for example, if the designer can use a local oscillator that uses less of that budget allocated, like if you budgeted let's say 2% and this oscillator can tolerate only 1% or add only 1%, there's another percent left to delegate to the power amplifier making the design of the power amplifier easier and you can run it more efficiently.

So there's a lot of trade-offs that could be done, but the local oscillator has a great play in how that EVM turns up at the end of the design.

TOM FLOYD: Okay, got you.

BOBBY MATINPOUR: Good question, again, thank you.

I also mentioned the spurs and harmonics is also a big one, harmonics being a little bit easier to deal with since you can filter it out easily.

But spurs created by a local oscillator are really critical and, again, these are limited by the emission masks that are dictated by the regulatory body.

So depending on what standard the radio is being used in, these specs will change.

Lock time is another issue that is less and less of an issue right now in the nextgeneration standards such as LTE and WiMAX.

It was more used in the frequency hopping GSM networks and it's unlikely that there's a lot of active design going on right now for these networks.

So that's also a specification to be considered.

TOM FLOYD: Okay.

BOBBY MATINPOUR: So we talked about EVM being a very important issue and you probably remember if you saw our previous webinar we talked in detail about jitter and its impact and its integration bandwidth.

So it's very important to be specific about the integration bandwidth used to specify a jitter number or an EVM number.

In particular, to set the low-end and high-end of integration there's some dependence on that and the overall system design.

So right now carrier and clock recovery loop bandwidth is very important inside the bottom end of the integration bandwidth.

And depending on if it's in a mobile environment, multipath affects or Doppler, again, impact how low integration -- closer to the carrier you have to worry about noise.

F50 or frame link has also a great play in that, and for LTE and WiMAX standards or next-generation infrastructure these are getting more and more difficult.

So the low end of the integration bandwidth is moving from a range of let's say kilohertz range, like tens of kilohertz down to hundreds of a hertz.

And as you can see on the phase noise plot, the yellow curve, as you go closer to the carrier or lower offsets, the amount of noise tends to pick up.

So that's a very critical issue.

On the high end of integration bandwidth, for at least (inaudible) applications, it's very simple, it's dependent on the channel of bandwidth used.

So the communication channel that determines what the high end of the integration is going to be.

So this is critical when you compare one synthesizer versus another or you look at an EVM number or a jitter number, because that is immediately impacted by how you set your integration balance.

So we're going to dig in a little bit deeper into particular requirements for receiver

and transmitter.

EVM, as I mentioned, for a receiver is very important. This is, again, the in-band integrated noise.

Integration bandwidth, as I mentioned, is typically from the carrier recovery loop bandwidth all the way up to the channel bandwidth.

Phase noise at particular offsets is mostly dictated by the standards and the regulatory body.

So if you look in the bottom left, let's say we're trying to receive a call from someone.

Let's say you're at the edge of the cell and you're trying to receive a call, and the base engine's trying to detect your call.

If there's other people closer to the base station and they're transmitting at the time, that transmission is going to come across as interference to the information that your cell phone is sending.

So in this case what will happen is the local oscillator phase noise can get superimposed on the interferers, and the tails of the phase noise can end up interfering and coming into the band of interest.

So it's very critical to have particular specifications for the amount of phase noise allowed at particular offsets, and that's dependent on the channel spacing.

So for GSM, for example, you see here about 600 MHz, 600 kHz and 800 kHz spacing.

And these are all, again, very standards dependent.

So for other standards it will be very different.

TOM FLOYD: So can high phase noise result in things like dropped calls also?

BOBBY MATINPOUR: Potentially, right.

Generally these equipment are certified to make sure that they meet the requirements before they're deployed.

But if afterwards something goes wrong, a synthesizer starts misbehaving, you could actually have a dropped call because now, in the presence of interference and blockers, you will have a very small signal-to-noise ratio and the call cannot be received properly.

TOM FLOYD: Okay.

BOBBY MATINPOUR: So that, yes, it could.

TOM FLOYD: Okay, good to know.

BOBBY MATINPOUR: And to relate it really to the real world environment (inaudible).

TOM FLOYD: I can relate to the dropped call thing, yes.

BOBBY MATINPOUR: I think we all can.

TOM FLOYD: Unfortunately, every once in a while.

BOBBY MATINPOUR: So on the transmitter at low requirements, EVM, again, is very important.

Integration bandwidth, as I mentioned, is going to be around the carrier recover loop bandwidth to the channel bandwidth.

This is kind of the rule of thumb.

Spurs and phase noise, again, at particular offsets are important, here for a different reason.

Let's see, on the bottom left you see an output of a radio.

In this case you see the signal is very large because it's coming out of the power amplifier.

And I'm highlighting the adjacent channels and there's a limited amount of emissions that you can have into the adjacent channels.

You're really confined to your own channel and there's some amount allowed, but if you go beyond that, that's going to be out of the spec.

And this is, again, determined by the standards.

So, on the right you'll see that if you have excessive phase noise on the local oscillator or unwanted spurs that can easily translate immediately onto the output power.

The phase noise can cause what they call spectral regrowth, where the information now kind of in a sense leaks into the adjacent channels.

And again, if you do that, then the reception of the information on the other channels would be limited, so you would be out of spec.

And again, these are serious issues that have to be avoided and that could be caused by poor design of the local oscillator.

And spurs are the same way.

There's also a mask for the amount of spur emissions in-band and out-of-band and anything that is on the yellow can easily translate onto the signal and amplify it through the transmit path.

TOM FLOYD: You touched on this little bit, I'm noticing the word regulation on the slide as well.

BOBBY MATINPOUR: Right.

TOM FLOYD: So who actually monitors or polices the transmitter emissions?

BOBBY MATINPOUR: So the standard bodies define the regulations or define the specs for each standard, so what is it, wideband CDMA, CDMA, GSM.

These are ratified, excepted, and everybody has got to adhere by it.

And the local governments for where this equipment is deployed, they have their own regulatory agencies.

Here we have FCC.

So the equipment, after they're built, they get tested, certified, and then deployed.

TOM FLOYD: Okay.

BOBBY MATINPOUR: So basically, before they deploy, they have to be certified.

TOM FLOYD: Okay.

BOBBY MATINPOUR: Nobody wants actually have a problem during that process, so it's very important that people pay a lot of attention to local oscillator, other high-performance analog design in the radio, to make sure to avoid any hiccups, delay in deployment, or additional rework for equipment.

TOM FLOYD: Okay, that makes sense.

BOBBY MATINPOUR: Yes, very good question.

Now, what we discussed in the past few slides were really applied to all standards.

Now, as we move into the next-generation standards such as LTE and WiMAX, this challenge actually gets a little bit aggravated.

You know, the difficulties grow.

We mentioned earlier that the carriers are trying to really find a way to pack more data through the existing infrastructure they have or in the existing spectrum, especially spectrum is very limited.

Each carrier has certain frequency bands that they have paid dearly to get and now they want to try to pack more data through that existing infrastructure, existing spectrum.

To do that, the trend is that they're moving into the high order of modulation.

On the bottom you see on the left that I'm showing a QPS gate type of modulation.

There's an IQ consolation diagram.

You see four distinct binary states, so a receiver will have to distinguish between these four distinct binary states.

As you move over to a high order of modulation like 64 QAM, you can see that the constellation gets very crowded.

Now, I'm showing the size of the circle that dictates its distinct binary state, as you have more noise this will grow.

And in the case of the high order of modulation 64 QAM you can see that these distinct states are very close to one another, so the possibility of making a wrong call, mistaking one state for another, is higher.

So right now, as you see on this slide, the system noise is too high for a 64 QAM type of architecture.

But to get around that you would need to clean up the environment, and you can do that by using very clean clocks and local oscillators to clean up the environment so that you can reduce the amount of bitter rate that you will have in the system.

So, again, that translates to a higher data rate, less errors, less dropped calls and so on.

TOM FLOYD: Got it.

BOBBY MATINPOUR: So right now this is an example of showing how the specifications or noise specifications of a clock and the load requirements of the previous generation of radio standards are not going to be sufficient for the next generation because the requirements are so much more.

TOM FLOYD: Okay.

BOBBY MATINPOUR: Another challenge with LTE and WiMAX systems is going to be that they're using OFDM modulation.

So this consists of closely spaced sub-carriers and they're actually almost overlapping, and we're talking about over a thousand sub-carriers spaced sometimes as closely as 10 kHz away, especially in the mobile WiMAX and LTE environment.

So the issue of superposition of a local oscillator onto the sub-carriers or onto the information is going to even be a bigger issue here.

Because right now you see that as the phase noise of the local oscillator gets superimposed on each sub-carrier, the overall signal-to-noise ratio dramatically drops.

So the requirements are even more stringent in this case, and you notice earlier I mentioned about moving to 64 QAM you have to reduce the amount of noise.

So the SNR requirements are even higher, now you've got the issue of subcarrier, which makes it a little bit more difficult to avoid the issues caused by local oscillator phase noise.

So local oscillator phase noise is even more of an issue in LTE and WiMAX systems.

TOM FLOYD: Okay.

BOBBY MATINPOUR: Now, having these trends in mind and requirements, at National we have developed a new set of clock and timing products, both on the clocking side and synthesizer, to address these issues and enable LTE and WiMAX wireless infrastructure development.

Devices such as LMK4000 that we discussed in the previous webinar, this is a high-performance, ultra-low noise clock jitter cleaner.

And LMX2541, which is our new local oscillator or frequency synthesizer, combined together they make a complete timing subsystem that enables LTE and WiMAX base stations and radio transceivers.

In the bottom I'm showing an example of a three-channel RF transceiver.

It could be a radio unit or a remote radio unit that's being linked to a base spanning unit with a (inaudible) and off-sited link.

In this particular case the Scan 25100 is recovering the clock, feeding it into an LMK4000.

LMK4000 does the jitter cleaning and distribution of high performance low noise clock to various components, including the LMX2531 or 2541, which would be the local oscillator.

So here's an example of how they're using it in the system.

TOM FLOYD: So should I take from this that there's an actual subsystem that we offer customers?

BOBBY MATINPOUR: Not in a subsystem in a package.

TOM FLOYD: Okay.

BOBBY MATINPOUR: That would actually be a little bit difficult for usage of customers because they want to place various components in the proximity of other components so that you really don't want to have them all in the same package.

But yes, as a two component subsystem.

Currently we're actually working on developing a subsystem board where these parts are on the same board together, it'll be easy to actually see how they work together, look at the footprint of the layout and our recommendations.

But in the meantime, customers can easily grab two of the evaluation boards, one for LMK4000, for example, and one for LMX2531 or LMX2541.

And basically use them together and connect them to each other and that will give them pretty similar results in having a subsystem for testing purposes.

TOM FLOYD: Okay, we just had a question that came in as well.

BOBBY MATINPOUR: Sure.

TOM FLOYD: And the question is, is the PLL applicable for cell phones?

BOBBY MATINPOUR: We'll talk about the LMX -- you know, traditionally PLLs were used very widely in cell phones and currently a lot of PLLs and synthesizers for cell phones have been integrated in a part of a fully integrated transceiver.

However, we have seen cases where for next-generation like LTE and WiMAX and mobile Internet devices or handhelds or cell phones, where due to the higher performance requirements, some people, they've started looking at discrete PLLs or synthesizers.

And I'll talk a little bit more about that.

LMX2531 would be a good choice because of its low current consumption.

LMX2541 is really highly targeted at the base station market and it has a higher power consumption.

And that depends on the choices that the customer wants to make.

TOM FLOYD: So it really depends on what the customer wants to do with it.

BOBBY MATINPOUR: Right, so we will talk a little bit more about that.

TOM FLOYD: Okay.

BOBBY MATINPOUR: 2531 might be a good choice for application in handsets for next-generation, the ones that require a little bit more flexibility on the frequency range, and the ones that do require higher performance to enable up to 100 MB per second if needed.

TOM FLOYD: Okay.

BOBBY MATINPOUR: Good question, thank you.

TOM FLOYD: Yes.

BOBBY MATINPOUR: So we're going to talk a little bit more about the clock and timing products.

Today, as I mentioned, we're talking about synthesizers and PLLs, and more specifically most of you are probably familiar with the Platinum PLLs, this is LMX2306, 16, and 26, and also LMX2485, 86, and 87.

Also we will talk a little bit about frequency synthesizers, the first generation of frequency synthesizer PLLs, integrated VCO, LMX2531 family.

This is a device that was released over three years ago, so a very mature product and it's running in volume production with a lot of major OEMs.

LMX2541 is the one we'll talk a little bit more about today.

This is the new product.

This is an ultra low-noise frequency synthesizer, mostly targeted at WiMAX and LTE base stations and radio wireless infrastructure.

On the persistent clock conditioners, these are devices that do a clock jitter cleaning, generation, distribution.

We talked in detail about those in the first clock and timing webinar.

I recommend that you take a look at that if you haven't already.

LMK4000 is the new device that we've released recently and it's a clock jitter cleaner with cascaded PLLs.

And this device would be a good fit with LMX2541 for, again, next-generation LTE and WiMAX platforms.

On the PLLs, more specifically the LMX2485, 86, and 87 family of devices, these are ideal devices to be used with an external VCO ranging from 5 to 7.5, 50 MHz to 7.5 GHz.

The key criteria here that we like to focus on is the low in-band spurs of -55 dBc.

As you can see below, we've taken a competitive analysis at 2.4 gigs and you will see that the traces, the red, the dark red and the light black traces are National products.

And in comparison with our other competitors, several of them here are looked at.

You will see a great advantage on the spurious response, and this is the key criteria that makes this device popular for usage, whether it's a discrete design or a local oscillator module.

The 2531, again, I mentioned this is a frequency synthesizer.

This is a PLL integrated with an ultra low-noise VCO, all on chip.

This is a very mature device introduced over three years ago.

This has a very high performance PLO.

It's a programmable fourth-order delta signal modulator that enhances the spurious response.

So, again, we're bringing over 15 years of experience of fractional PLL design into these devices here.

This covers a wide frequency range of 553 to 2.8 gigs.

We have new devices now that go up to 3.1 GHz.

Other features, again, one of the great features of this device is it not only provides great performance but it does it at very low current consumption, and this is ideal for next-generation mobile Internet devices or handsets, handhelds using LTE and WiMAX where performance is required or the designer requires flexibility.

And it's very difficult to integrate multiple VCOs onto an integrated transceiver.

So having a local oscillate synthesizer with low current at high performance externally available to hit different bands is very useful.

TOM FLOYD: Now, does our competition also offer these features?

BOBBY MATINPOUR: Our competition has devices similar to this.

TOM FLOYD: Okay.

BOBBY MATINPOUR: In general, I would say that we are in the forefront when it comes to performance.

Also, what's very critical is, as I mention on the slide here, one of the key features of these devices other than performance is the fact that it has level integration.

We have integrated multiple low-noise LDOs internally to the device.

It not only helps us to isolate the digital analog blocks internally to give us the great performance, but also takes away from all the external components that the customer would have to use if they're using competitors' products.

So using a competitor's product you will have to use external low-noise LDOs.

It makes the footprint of the design larger, adds cost and complexity to the implementation, and risk.

And the other advantage would be also on the output of our devices.

We do not require a lot of external components.

Most of our competitors require a kind of complicated L&C network to do the matching on the output and that's not required with ours.

So there are other advantages as well on the feature side, not only on performance, but also on the features and the external component usage.

TOM FLOYD: Okay, thanks for that.

BOBBY MATINPOUR: A very good question, thank you.

Here I'm showing on the left a list of original devices that were offered with this family of products.

You will notice there's a hi-band and low-band.

In this device you can access the output of the VCO immediately or go through a divider, do a divide by two and that will be the low-band.

This gives it a little bit of flexibility.

And on the right are all the frequency bands that we added on later based on customer requests.

And on the bottom you see the two higher frequencies all the way up to 3.1 gigs that will be released shortly.

So now we're going to talk a little bit about the new device, the LMX2541 ultra low-noise synthesizer.

This device is really targeting the key criteria for next-generation infrastructure.

The low integrated noise is a key differentiator of this device.

This is done using an ultra low-noise PLL.

In integer mode the figure of merit of this device is going to be -224 dBc/Hz and in fractional mode -223 dBc/Hz.

Now, this figure of merit is well described in all our data sheets and app notes.

But to just put it in perspective for you, relative to our LMX2531, which was at the time it released and even up to now industry-leading in performance, this device has 12 dB better performance in the PLL area.

At 12 dB not only from this LMX2531, our own first-generation device, but also to newly released parts from competitors that this still has 12 dB better of PLL noise.

And in the world of phase noise 2 dB is a big number, 12 dB is huge.

So this adds a lot of advantage to the customer when using this part, reducing the total EVM that it experienced from the local oscillator.

Another key feature is the phase detector rate.

This device is designed to go all the way up to 100 MHz, the phase detector rate.

And the rule of thumb to use there is as you increase the phase detector rate every factor of two gives you 3 dB in the in-band PLL noise reduction.

Competitors, they are typically on the order of 40 to 50 MHz, so this just doubles the phase detector rate in comparison.

One good thing about this device is it can also be used as a stand-alone PLL.

So if a customer wants to really use this super PLL, super high-performance PLL inside this device with their own VCO, they can just bypass the VCO and use the PLL.

Another key spec here is the low spurious emissions.

This is, again, using the programmable fourth order sigma delta fractional PLL engine, bringing in all our experience over the last 15 years designing fractional PLLs into this device.

We also get excellent frequency coverage all the way up to 4 gigs using this device.

We went all the way up to 4 gigs to be able to address frequency bands for WiMAX at 3.5 and 3.8 gig and also satellite communication, there are some bands there.

This frequency coverage is enabled by using six different devices and very flexible dividers.

So this is a much more flexible device than our 2531.

TOM FLOYD: Now, will the stand-alone PLL be a new device also?

BOBBY MATINPOUR: Not at this point.

At this point we recommend people to use it and just bypass the VCO.

Whether we do a stand-alone device as a PLL, it's still to be decided.

TOM FLOYD: Okay, another question, more of a request that came in, is it possible to find out more about who our competitors are by name and part number?

BOBBY MATINPOUR: Yes, I will have more information available shortly that will compare LMX2541, 2531, and many other competitors that are -- well, let's say about three top competitors to clearly highlight how these parts vary and the value that the 2541 brings, specifically for LTE and WiMAX next-generation base stations.

TOM FLOYD: Great, okay.

BOBBY MATINPOUR: Yes, we'll have that.

This is a block diagram of the 2541.

If you're familiar with our 2531 you will see that the block diagram generally is fairly similar.

The differences are that the VCO, we're using the wider tuning range VCOs, allowing us to cover a range of about 1.92 gig to 4 GHz using six distinct devices.

And now, when you pair that up with a very flexible divider that we have included in this, you will be able to go all the way down to 50 or 30 MHz of frequency coverage.

In the 2531 we had only a divide by two option.

In this case we have a divide by from 1 to 63 odd and even divides, all resulting in 50% duty cycle.

So all of the divide values are useful to be used as a driving (inaudible) or a local oscillator.

But the key criteria, again, on this device, relative to competition and to our own LMX2531, is the ultra low-noise PLL that gives you a tremendous advantage.

TOM FLOYD: Okay, we just had another question that came in.

BOBBY MATINPOUR: Sure.

TOM FLOYD: And the question is, are these parts built on a process that can be used in space applications?

BOBBY MATINPOUR: This product, as it's designed right now, will not be able to be used in space applications.

It will not pass the radiation testing.

However, we do, on an ongoing basis, modify devices for those applications.

So this is a good question and I think they should get in communication with our Higher L group to get details on what existing products from the clock and timing portfolio they're working on for space applications.

TOM FLOYD: Okay.

BOBBY MATINPOUR: Thank you.

To even further highlight the value of a 2541 I've shown here a phase noise plot, or let's say it's not a measurement but just a drawing to highlight the point I want to make about PLL noise.

You see on the gray area just the LMX2531 and you will see that the PLL noise in this case at a frequency of 2 GHz, you will see that we are right around about - 90 dBc per hertz.

The 2541, what will happen is as you reduce the PLL noise it allows you to open up the loop filter to beyond from the case of 10 kHz all the way down to maybe 50 kHz and gives you that in-band reduction of phase noise by a factor of 12 dB.

Now, jitter or EVM is the area under the curve and this has a huge impact on that number.

So 12 dB reduction in PLL noise in general what we are measuring translates into 3 to 4x reduction in the milliradians or percentage of EVM.

So, on top of the 12 dB better PLL noise versus the 2531 competitors, one of the areas of improvement is in the flicker noise or 1/f noise.

That's an area of noise that you will see below 1 kHz offset and we're seeing an order of 17 dB improvement there.

So, again, that translates into a smaller area under the curve, hence a lower integrated noise number.

TOM FLOYD: So you touched on this a little bit previously, so just to kind of recap.

Are we ever saying that the LMX2541 is a better device than the 2531 or it really just depends on what the customer (inaudible)?

BOBBY MATINPOUR: You have a very good point, right.

So 2541 is a higher performance device than 2531, targeted for base station applications for LTE, WiMAX, and the next generation.

Even the UMTSW CDMA platform that they're designing right now, they're designing it to be future kind of proof for LTE.

So for all those designs, 2541 is a more suitable choice.

But 2531 brings a lot of performance for a very low current consumption.

So for customers that are designing a radio transceiver that has tough, stringent power consumption requirements, 2531 obviously is a better choice.

For mobile Internet devices or handsets for next-generation LTE and WiMAX that require higher performance or bigger flexibility, again, 2531, because this low current consumption is the right choice.

So like you said, it's very much dependent on what the criteria or what the specifications are.

TOM FLOYD: Okay.

BOBBY MATINPOUR: Especially on the power consumption side.

TOM FLOYD: Okay, a couple more questions that just came in and the first question is, when will the 2541 be enabled on code loader and other clock design tools?

BOBBY MATINPOUR: 2541, right now we have a code loader that works with it.

I'm not sure how complete it is, but shortly.

I would say the code loader version, if it's not available right now, I do know we're sending alpha samples of the device as of now.

So I'm assuming there is a code loader working because that's how the customer would program it.

On the clock design tool, I would say in the next two to three months.

TOM FLOYD: Okay, great.

Next question is, as LDOs are integrated, could the device be powered from a DC converter, or is an additional low LDO required to power the device?

BOBBY MATINPOUR: That's a tough question to answer, because it's really dependent on how noisy the environment the device is going into is going to be.

We have to review that on a case-by-case basis.

It all depends on how large of a ripple exists on the supply, how much noise there is on the supply.

Sometimes just using the internal LDO is going to be sufficient and some other times you will need to have some sort of crude kind of noise filtering done on the outside as well.

It may be that you can get away with an RC or an LC on the supply prior to the device, depending on the noise environment we're in.

If it's a very noisy environment you might need to put in another LDO.

TOM FLOYD: Okay.

Another question that just came in.

BOBBY MATINPOUR: Sure.

TOM FLOYD: Will the 2541 be easy to replace a 2531 in an existing design?

And the second part of that question is, will there be an application to help with this?

BOBBY MATINPOUR: Yes, so the first question.

The footprint of LMX2541 and 2531 are different, so a board spin will be required to replace the one for the other.

An app note on that, this is a great idea, did not think about it up until now, so I would be saying yes, we would be wanting to do some sort of small app note to do that.

However, given that it cannot be directly replaced, it may turn into an app note on how to design in 2541 with some reference to 2531.

TOM FLOYD: Okay, great.

BOBBY MATINPOUR: So now we're going to start looking at some of the data that we have taken on the 2541, some of these performance characteristics.

In this plot you will see 2541 phase noise looking at the frequency, a VCO frequency of -- so the device is locked at 3.6 GHz.

We're running the highest phase detector rate, 100 MHz, and looking at the highest charge bump setting of 3.2 million, really looking at its best possible performance to highlight the PLL noise.

And the PLL is configured for a very wide loop filter so that you can basically reach all the way down to the bottom of the PLL noise floor.

In this case, at 3.6 GHz you're seeing a -111 dBc per hertz, outstanding performance for PLL noise.

And this validates all the information I gave you earlier, especially when you compare it to the 2531 red line on the top you will see the 12 dB improvement in PLL noise and, again, 17 dB improvement on the flicker noise or 1/f noise.

The 2531, unfortunately we do not have a device that goes all the way up to 3.6 gig, so the red line that you see on the plot is actually extrapolated from the lower frequency number to do the comparison.

TOM FLOYD: Another question that just came in, and it's more of a scenario based or kind of an example question.

And the question is, if you had a customer who was designing a CPE system but felt the part was considered on the expensive side, what's the best way to handle that?

BOBBY MATINPOUR: You're talking about -- I'm not sure which device they're in particular talking about.

TOM FLOYD: The 2531.

BOBBY MATINPOUR: 2531 on the expensive side?

TOM FLOYD: Yes.

BOBBY MATINPOUR: Again, it's dependent on what ASB we're talking about and what volumes they are.

A recommendation on the expensive side is to really highlight the value of the device.

In the case of they're comparing it with another competitive solution we need to make sure that they understand that using the competitive solution, you're going to have to use more LDOs.

TOM FLOYD: Okay.

BOBBY MATINPOUR: You're going to have to use external components, Ls and Cs to do the matching.

So some of it is the BOM cost itself, and some of it is actually going to be the amount of R&D effort that the customers have to put into design in a more complex part rather than a more integrated solution.

TOM FLOYD: Okay.

BOBBY MATINPOUR: Another option would be to also look at some of our Platinum PLL products.

That might be a good choice as well.

TOM FLOYD: Okay.

BOBBY MATINPOUR: That will help reduce the cost, and the customer can potentially do their own VCO design externally and get away with a lower BOM cost.

TOM FLOYD: Okay, fair enough.

BOBBY MATINPOUR: Another slide on some performance metrics for 2541, I figured I'd put a couple of scope shots here so that you can see real data.

We're looking at it, again, at integer mode.

The VCO is locked at 3.1 GHz, highest phase detector rate, highest charge form setting.

Again, we're experiencing a PLL noise of -224 dBc/Hz.

This is the normalized figure of merit, and RMS noise here, we're looking at 3.4 milliradians integrated from 1 kHz all the way up to 20 meg.

TOM FLOYD: And can you tell us a little bit more about that unit, about MRADs?

BOBBY MATINPOUR: Oh, milliradians, okay, so if you reviewed the previous webinar you probably heard a lot about RMS jitter.

TOM FLOYD: Right.

BOBBY MATINPOUR: As a measure used, milliradians and RMS jitter is basically the same thing.

TOM FLOYD: Oh, okay.

BOBBY MATINPOUR: Just a different way to look at it.

In fact, if you're able to see on this slide you will see that the RMS noise is specified in milliradians, in millidegrees, in femtoseconds.

So in this case it would have been 173 femtoseconds of RMS jitter.

TOM FLOYD: Okay.

BOBBY MATINPOUR: So very similar.

On the next slide I'm showing actually 2541 configured in a more applicationspecific manner for WiMAX and LTE at about 3 GHz.

So the device, now it's locked at 3 GHz, the phase detector rate, again, is running at the maximum 100 MHz, and we configure the device in fractional mode, optimize the loop filter setting and the charge form setting as you can see is now back down to 2.5 mA instead of 3.2 to optimize the spurious response.

And we set this up for synthesizer spacing, a channel spacing of 250 kHz, and we optimized to be able to maintain a worse case spurs or in this case would be a typical spurs of about -65 dBc with the intention that nothing would be worse than -60 dBc.

The RMS noise that we're seeing is about 4.2 milliradians integrated from 1 kHz to 20 meg.

So this is, again, really application-specific.

The previous slide was just trying to purely show performance.

Here we're saying let's now try and optimize between spurs and phase noise by tweaking the loop filter and going into the fractional mode so you can get a 250 kHz spacing and still running at a high phase deck rate of 100 MHz.

So again, really outstanding performance for this device.

BOBBY MATINPOUR: So now we're going to do a summary of the target applications for our PLLs and Synthesizer products for LMX2485, 86, 87 PLLs.

These are high performance, low power choices PLLs that can be used for discrete Synthesizer designs. This is the case where the customer really likes a little bit of more flexibility or they want to use their external VCO. And also for Local Oscillator Synthesizer modules these are excellent choices. LMX2531 Low Noise Synthesizer, this is PLL, we're going to have a VCO again. This is a very mature product running in volume production and is a great choice for CDMA, UMTS or W-CDMA base stations and radios. Also, given that it has very high performance at low power consumption, this is an idea choice for LTE and WiMAX mobile devices. This is a case where a customer would require a frequency band that is not covered with an integrated transceiver. It's too pricy to go and re-spin that device, but using LMX2531 gives that flexibility to hit a band that they cannot hit at the moment. And also if they need a higher performance to really push it up to the 100Mbps kind of performance, VLT and WiMAX, this will be a great choice. LMX2541 Ultra Low Noise Synthesizer is a great choice and is directly targeted for LTE, UMTS/LTE and WiMAX base stations and radios. What I mean by UMTS/LTE is this is really due to the fact that really no wide-band CDMA or UMTS design are currently done without really looking forward to LTE, so the hardware that's put in place needs to be LTE compatible -- future proofed in a sense. The LMX2541 with this PLL noise performance is going to be an ideal choice for these designs as well. Also looking at new RF and microwave links and backhaul, commonly they use a very complex and high order modulation, all the way up to 256-QAM. And for these types of applications, again, in-band PLL noise is extremely critical, even more than it is for LTE and WiMAX. These will be great choices for LMX2541 to be used as a local oscillator in the radio.

TOM FLOYD: Now, for other applications such as clocking ADCs, which one of these devices would be the best for that?

BOBBY MATINPOUR: Both 2531 and 2541 would be great to clock (bigger?) sample ADCs. 2531, in fact, is on evaluation boards for (bigger?) sample ADCs right now and being used. 2541 brings it a little bit more performance, especially in a close end phase noise side, and also more flexibility. For example, if you get a 2541 device, it locks at 4 gigs. With internal divider you can easily divide it down to 2 gigs or divide by three or divide by four to like 1 gig or 500 megahertz.

So it brings a little bit of more flexibility because of that divider output that's more flexible -- output divider. So, but either one, especially if they are concerned about power consumption, 2531 will still probably be a pretty good choice.

TOM FLOYD: Okay, got it.

BOBBY MATINPOUR: Great. So in summary I'd like to strongly urge that you visit our clock and timing website. We continue to update the content there.

Examine our LMP and LMX portfolio, and especially LMX4000, a new device clock jitter cleaner there, and LMX 2531, LMX2541. That will be shortly up there.

And contact your National sales rep for more information or evaluation board.

Certainly feel free, as always, to contact me.

TOM FLOYD: Great. Well, thanks again Bobby for joining us and coming back.

BOBBY MATINPOUR: Thank you.

TOM FLOYD: And for all of you as well, thank you for listening to this webinar.

A few last things in closing. As always, we have a short 10 question multiple choice assessment associated with this webinar. We'd ask you to take a few minutes to complete that as well. And another really important thing. As always, we have an evaluation associated with today's webinar as well. We'd like to ask you to complete that as well. It's pretty painless. It'll take you less than 5 minutes. But we really do go through all of those and look at all of your comments, your feedback, your suggestions. It's especially really helpful in terms of identifying topics and other things to cover in future webinars in our series as well. And with that said, thanks again for joining us, and enjoy the rest of your day.

From -	Name:	Email:
To -	Name:	Email:
Subject:	Clock & Timing - Low Noise Frequency Synthesizers