LDMOS new future AMPS?

[warnberg]

Anyone doing any experimentation with these?  LDMOS.. starting to see plans, schematics, etc... popping up using LDMOS.

NXP Unbreakable BLF578XR: http://www.nxp.com/pip/BLF578XR.html

David

[KJDS]

I've done a lot of work with LDMOS, anything specific you want to know?

[uncle_bob]

Hi

LDMOS has been around for a long time. There are a number of people making it. It's been the goto item for PA's for at least a decade.

Bob

[KJDS]

Unless you have access to a decent RF simulator, including planar EM, and the ability to tune the design looking at output P1dB and efficiency at the same time, then it's unlikely you'll get close to optimum performance.

[kazam]

I would like to suggest Keysight ADS for non-linear simulations and Ansys HFSS for final tuning of matching circuits in simulation.

It's very difficult to hit optimum impedances without these tools. You're way out in the Smith chart.

/K

Unless you have access to a decent RF simulator, including planar EM, and the ability to tune the design looking at output P1dB and efficiency at the same time, then it's unlikely you'll get close to optimum performance.

[dmills]

The need for advanced simulation depends very much on where you are working....

For example, for a broadband HF set, where you clearly cannot do conjugate matching (except possibly on one band), and where the highest working frequency is very low compared to the (typically 600MHz) frequency that the broadband devices top out at, you really don't need ADS and friends (Being able to plot stability circles is useful still).
In the bands near DC the design is very much like that of an old school mosfet amplifier except that overdrive is an even bigger device killer, and especially in deep class C you need to watch the negative swing on the gate (Also watch this if getting clever with harmonic terminations).   

Now if I was trying for a narrow band design up in the UHF or above, that would be different, especially if I wanted to do a Doherty or similar architecture, up there the near DC method of just slugging the gates with a resistor to make the overall gate load approximately resistive will no longer fly.

I have played with LDMOS a bit.

73 Dan.

[uncle_bob]

Hi

If I'm doing this in the basement (even my basement), I'm not going to fire up my own copy of ADS with full EM and model things up. I can buy a room full of amps for what the license will cost me. At work, absolutely, I'm not going to do a product that I release to the production line without simulation. I'll probably also go through a few rounds of this and that before the final design gets to the shop floor. For a basement project .... stick to frequencies that still qualify as "DC". For the higher frequency stuff eBay is very much your friend. As a matter of fact, I'd do a search there before I'd do an HF design. There are all sorts of odd amp blocks that turn up dirt cheap from time to time. Modifying one of them saves an ocean of time and money compared to a scratch build.

Bob

[kazam]

I haven't used LDMOS below 900MHz so can't comment on that. Even at 900 I would say it's not trivial. Especially if you're looking for exotic topologies like Doherty. But why would you need something like that? Usually you go Doherty for efficiency reasons in back-off and then peaks are biased in Class C so non-linear. And PAR will be high because why would you otherwise choose Doherty. So that's a BTS amp if you ask me. You need DPD which is probably out of reach for most people.

Class C, narrowband you can probably hit OK. Be prepared to move caps for a while though. You will need RF test equipment of course. Some way to measure performance. What are your interests? I can think of a dozen performance targets that you could weigh against each other depending on what you're trying to achieve.

Definitely, if all you need is an amp there are easier ways than doing your own from scratch.

It's great fun to play around with though!

[uncle_bob]

Hi

My interpretation of the original request is a "play for fun / stay out of trouble" approach.

Bob

[kazam]

Agreed.

I guess I'm just wondering why someone would go to LDMOS unless they were interested in RF.

[dmills]

They have their virtues down near 'DC' as well, by which I mean the HF band, and you don't really need simulation down there where broadband transformer matching and gate swamping resistors rule the land. 

The HPSDR crew are doing interesting things with predistortion so Doherty is not an altogether unreasonable thought, and SSB has PEP/Average in excess of 4:1 in most cases which makes it a goer.

Regards, Dan.

[vk6zgo]

We cooked a lot of LDMOS devices in 1kW PAs at 433MHz in an ISM application,until we learnt not to trust the equipment manufacturer.

The designers made a mess of things,------ it wasn't the LDMOS device's fault
Their unbalance loads were too small,& would cook up,leading to the demise of the devices.
We couldn't redesign the loads,so we needed to tune each of the paralleled amps to be within 1.5dB(or better),of any other in the operating frequency range-------the manufacturer did not "come within a bull's roar" of that.

Once we tuned the things up correctly,they were reliable.
(Great fun,pushing chip caps around the PCBs with toothpicks).

[uncle_bob]

We cooked a lot of LDMOS devices in 1kW PAs at 433MHz in an ISM application,until we learnt not to trust the equipment manufacturer.

The designers made a mess of things,------ it wasn't the LDMOS device's fault
Their unbalance loads were too small,& would cook up,leading to the demise of the devices.
We couldn't redesign the loads,so we needed to tune each of the paralleled amps to be within 1.5dB(or better),of any other in the operating frequency range-------the manufacturer did not "come within a bull's roar" of that.

Once we tuned the things up correctly,they were reliable.
(Great fun,pushing chip caps around the PCBs with toothpicks).

Hi

.... and indeed you run the risk of getting the "un tuned" amps when you shop surplus. There is no ideal one size fits all solution.

Bob

[vk6zgo]

Bob,they weren't surplus,the amplifiers & built-in exciter were a purpose-built order from the dear old People's Republic.
Apparently,they normally used the amps on 470MHz,as they were roughly tuned up for there.
The exciter had an "exciting" fault,too---the PLL would let go & slide up to a default frequency around 470MHz.

433MHz in Oz is within a Ham Band,but some low power devices are allowed in that band.
Our use of the frequency was tolerated by ACMA,as long as our screening didn't leak radiation above the level of those devices.

[uncle_bob]

Bob,they weren't surplus,the amplifiers & built-in exciter were a purpose-built order from the dear old People's Republic.
Apparently,they normally used the amps on 470MHz,as they were roughly tuned up for there.
The exciter had an "exciting" fault,too---the PLL would let go & slide up to a default frequency around 470MHz.

433MHz in Oz is within a Ham Band,but some low power devices are allowed in that band.
Our use of the frequency was tolerated by ACMA,as long as our screening didn't leak radiation above the level of those devices.

Hi

Oh, I very much understand that you got "not so great" stuff brand new from the factory.

That same fine factory with it's dubious approach to building amps probably kept on going. That or somebody else popped up with the same "just put it together and ship it" approach. Those amps will indeed show up on eBay. Even as "factory built" amps, they will still have the problems you describe when surplussed out. (assuming they lasted that long ...).

Bob

[vk6zgo]

Bob,they weren't surplus,the amplifiers & built-in exciter were a purpose-built order from the dear old People's Republic.
Apparently,they normally used the amps on 470MHz,as they were roughly tuned up for there.
The exciter had an "exciting" fault,too---the PLL would let go & slide up to a default frequency around 470MHz.

433MHz in Oz is within a Ham Band,but some low power devices are allowed in that band.
Our use of the frequency was tolerated by ACMA,as long as our screening didn't leak radiation above the level of those devices.

Hi

Oh, I very much understand that you got "not so great" stuff brand new from the factory.

That same fine factory with it's dubious approach to building amps probably kept on going. That or somebody else popped up with the same "just put it together and ship it" approach. Those amps will indeed show up on eBay. Even as "factory built" amps, they will still have the problems you describe when surplussed out. (assuming they lasted that long ...).

Bob

Yeah,Bob ,I got into this as a "help out from time to time" thing.
The original bloke fell ill,& myself & another Tech got "sucked into the maelstrom".

Tuning amps by sliding chip capacitors around with a toothpick whilst watching the swept response was something I hadn't done for about 20 years at the time (& I certainly hadn't missed it! ).

Googling,we found the test board setup for the devices used,& our boards looked like two of these stuck together,with splitters at the input,& combiners at the output.
These units were in turn paralleled,& the same done with the output of that,& so on.

It certainly looked like they thought putting the caps in the same place would make an amp with repeatable performance.
Maybe they grabbed a "Rent-A-EE",& as soon as he had done the initial work,chucked him back on the street!

Still,I guess if you could pick these things up cheap,& didn't mind tuning them,the individual amps would be quite useful.

[uncle_bob]

Bob,they weren't surplus,the amplifiers & built-in exciter were a purpose-built order from the dear old People's Republic.
Apparently,they normally used the amps on 470MHz,as they were roughly tuned up for there.
The exciter had an "exciting" fault,too---the PLL would let go & slide up to a default frequency around 470MHz.

433MHz in Oz is within a Ham Band,but some low power devices are allowed in that band.
Our use of the frequency was tolerated by ACMA,as long as our screening didn't leak radiation above the level of those devices.

Hi

Oh, I very much understand that you got "not so great" stuff brand new from the factory.

That same fine factory with it's dubious approach to building amps probably kept on going. That or somebody else popped up with the same "just put it together and ship it" approach. Those amps will indeed show up on eBay. Even as "factory built" amps, they will still have the problems you describe when surplussed out. (assuming they lasted that long ...).

Bob

Yeah,Bob ,I got into this as a "help out from time to time" thing.
The original bloke fell ill,& myself & another Tech got "sucked into the maelstrom".

Tuning amps by sliding chip capacitors around with a toothpick whilst watching the swept response was something I hadn't done for about 20 years at the time (& I certainly hadn't missed it! ).

Googling,we found the test board setup for the devices used,& our boards looked like two of these stuck together,with splitters at the input,& combiners at the output.
These units were in turn paralleled,& the same done with the output of that,& so on.

It certainly looked like they thought putting the caps in the same place would make an amp with repeatable performance.
Maybe they grabbed a "Rent-A-EE",& as soon as he had done the initial work,chucked him back on the street!

Still,I guess if you could pick these things up cheap,& didn't mind tuning them,the individual amps would be quite useful.

Hi

There are a lot of these places that don't even know what an EE is. They tear a gizmo apart, look up the parts, and start putting the same parts together. Before you say this is something new or specific to one country or the other .... I first ran into it back in the 60's on ... wait for it .... DOD purchased spares made in the good old USA.

Bob

[Theboel]

I working with tis LDMOS family for sometimes mainly use for FM broadcast transmitter, some of them run 24/7.
What I found so far:
1. There is no way use a "enough heatsink" there is more than just open catalog find the XXX thermal capacity put fan and nature will help You.
You must really think about heat dissipation very seriously "DO NOT CHEAP" so far the best are water cooling but oveeeeeeer size heatsink is welcome.

2. You can not run it up to 1,2kw without some serious effort about spurious, harmonic and etc, the sweet point are 800w for continue service with about 45 to 46 volt very clean and stable DC supply.

3. You must pay extra attention about output transformer You use. the heat "generated" in output transformer is nasty thing to deal with.
a thermal image or at least infra red thermometer will help to fine tuning Your design.

Maybe our experience is different but this is mine and I do it every day and before somebody shoot me I am talking about 20/7 to 24/7 service in 23 degree centigrade room.

[dmills]

Yep, thermal design is everything with these high power density devices.

Do actually put a manometer across the fan and check against the fan curves to see what air flow you are actually getting, blowers can be much better then fans with some heatsink designs, and soldering the part to the heat spreader has much to recommend it (The heat spreader does not need to be particularly large to be effective however).

You **MUST** mill the heat sink and heat spreader faces flat, simple extruded heatsink looks flat until you look closely.

I am surprised by heat in the output magnetics being a problem in the FM BCB, I always figured that air cored transmission line transformers were the way to fly there, with at most very low permeability magnetics.

73 Dan.

[Theboel]

Yep, thermal design is everything with these high power density devices.

Do actually put a manometer across the fan and check against the fan curves to see what air flow you are actually getting, blowers can be much better then fans with some heatsink designs, and soldering the part to the heat spreader has much to recommend it (The heat spreader does not need to be particularly large to be effective however).

You **MUST** mill the heat sink and heat spreader faces flat, simple extruded heatsink looks flat until you look closely.

I am surprised by heat in the output magnetics being a problem in the FM BCB, I always figured that air cored transmission line transformers were the way to fly there, with at most very low permeability magnetics.

73 Dan.

Hi Dan,
excellent idea for manometer, I slowly changed air cooler to water cooler system the efficiency is better caused I can used less Air condition, system is cleaner no dust accumulate in air filter and etc.

in FM transmitter usually they do not use magnetics only from very first generation of LDMOS the manufacture usually use teflon insulator low impedance coaxial as transformer and the temperature of this coaxial sometimes can go up to 80 centigrade and some manufacture do not test the pallet extensively sometimes You need to re soldering the cap.

I also found the technique use by W6PQL really work and also a small heatsink on the top of LDMOS can decrease my job to replace a broken LDMOS in my transmitter.
if You Working with 70 transmitter thats a good news 

 

[warnberg]

http://www.w6pql.com/1_kw_sspa_for_1_8-54_mhz.htm

I was referring to this posting in particular..

[dmills]

80C in a teflon coax transformer would not worry me particularly, providing it was measured at rated SWR in the low Z condition.
The days when 'too hot to touch' was a good indication of a problem with power sand are long gone.

And yea, water cooling cures many ills, but can be a bit of a pain in deployment, I am thinking of things like condensation if the cooling loop water gets below the equipment room dew point and the condensate drips all over the control card (Coherent Laser, I am looking at you with respect to the passbanks in old school ion laser supplies).

I would also point out that in some ways a hot heatsink is less worrying then a cool one, as that sometimes means there is excessive thermal resistance die to heatsink.

Regards, Dan.