Pre-Distortion of Power Amplifiers: MAX2009, MAX2010

[gszczesz]

apd dpd control predistortion

Anyone have any experience with these chips: MAX2009 and MAX2010 (**broken link removed**) ???

Anyone tried to implement open-loop pre-distortion of power amplifiers? In my experience the power amplifier IP3 signature changes too much over temperature and time to implement any open-loop pre-distortion effectively... Has anyone had any success??

Greg

 

[springf2000]

i donnot use the max2009 as the pre-distorter, the correction only 3 dB, i use the
rf-DPD, the correstion is about 12-15dB at the common condition, the worest condition , the correction is about 10dBc

 

[gszczesz]

RF Digital pre-distortion is a closed-loop system where you monitor the output signal in order to correct the IM3 components.... It doesn't care about what is creating the IM3 or how, it just measures it and creates the inverse. This ofcourse will be much better then open-loop analog pre-distortion..

But the question is, does open-loop analog pre-distortion incrementally improve performance (will it give you the 3dB ABOVE your RF-DPD) or is RF-DPD improvement limitted by something else, regardless of what the initial IP3 of the amplifier is?

Greg

 

[biff44]

Open loop predistortion is nothing new. You generate an intermod product that is equal in amplitude but 180 degrees out of phase with the power amplifier's intermod. Usually the phases stay fairly stable over temperature, but the amplitudes get screwed up.

Things to get you closer over temperature:
1) if you have a multistage power amplifier, design it so that the output stage is contributing the most to the intermod. You accomplish this by having the mid stages have higher IP3 than you would normally design for. You do not want some weird mix of output/mid stages generating the intermod, because over temperature that will be non-repeatable and non-monotonic vs temperature from unit to unit.

2) You control the amplitude of the predistorter generated intermod product by passing it thru a voltage variable attenuator (PIN diode for instance) that is controlled by a room temperature set point resistor and a temperature varying thermistor or sensistor. With a little empirical work, you can get the curves to match.

As I recall, I was able to get around 6-8 dB of 3rd-order-product suppression over -20 to 80 deg C.

 

[gszczesz]

I tried using the MAX2009 and I also could get 6dB improvement, for a while anyways. The room temperature itself could be compensated (i.e. look up table for the MAX2009 controls). The problem occured with the self-heating effects of the main amplifier. Depending on what is being transmitted, the self-heating residue could last a long time. This became very evident when using 2 carriers close together modulated by pseudo-random modulation. The beat frequency of the two carriers and their pseudo-random codes generates a different self-heating memory effect that could either make the MAX2009 usefull or tottally useless.... Very frustrating to work with. Because the amplifiers self-heating memory could dominate the IP3 signature, it was impossible to tune the open-loop predistorter for a good correction..... This is for base-station transmitters like the Motorola/Freescale MRF21120R6 LDMOS and such.

To compound the problem, the LDMOS they sell have internal IP3 cancellation schemes that make the IP3 signature very sensitive to bias and temperature. A two-tone test show the IP3 get nulled out at a very particular bias setting. IP3 being nulled out doesn't mean ACPR-1 being nulled out, but it does make open-loop pre-distortion tricky if not impossible. So the question is, is the work invested in trying to make open-loop pre-distortion on an already optimized, self-heating LDMOS amplifier worth it, or is digital pre-distortion limitted by something that doesn't care what IP3 you start of with (i.e. by the resolution of your DAC, or something like that) ??

Greg

 

[springf2000]

you can use the open loop RF_DPD!
certainly y can use the APD

 

[biff44]

Yes, you have to accurately sense the temperature of the offending device. If you do not have adequate thermal conduction to the heat sink, you are going to have trouble sensing the temperature somewhere other than the device itself. Maybe try glueing a leaded thermistor to the device's package?

 

[michaelzhang]

The cost of DPD is much higher than the cost of APD, so I think you'd better use APD though the performance of DPD is better. According to a new research report, the APD can approve ACPR more than 10dB.

 

[aniltunwal]

Hi

Anyone can guide how to get optimum performance using max2009?? I am not getting any improvement in IM3. whtever i change is causing the IM3 worst. Please help.

 

[gszczesz]

That all depends what you are trying to do...

Are you running it closed-loop or open-loop?
What kind of device are you trying to compensate? Is it a big base-station like LDMOS, or something smaller?
Finally, is the modulation scheme only phase-modulation or does it include amplitude modulation as well??

For example, if you're doing open-loop compensation of base-station LDMOS that has both AM and PM modulation, the main source of distortion will be the non-linear capacitor, and as such you probably want to start of with just hooking up the phase pre-distorter. You than set some pseudo-random generator and monitor the IM3 components as you tweak the two knobs of the phase distortion. When you get an optimal performance, you set that as your operating point.

 

[aniltunwal]

I am using the LDMOS big base station amplifier. I am trying with open loop. right now i am optimizing the MAX2009 parameter with 2tone CW signals from sig Gen.

 

[gszczesz]

While 2 Tone testing can be helpful, you probably want a pseudo-random signal generator.
The problem is that most LDMOS these days include built-in open-loop IM3 nulling, and your two-tone test may show very good results even without the pre-distortier. Once you send a modulated signal, the adjasent channel noise will include more than just IM3 products, and you will see a much worse response. Here's where the MAX2009 can really benefit.

For your situation, you're best to only use the phase pre-distorter, which makes tuning easier since you don't have to compensate for the gain variation during tuning of the gain pre-distorter. This means you only have to worry about 2 knobs. The gain section may buy you more though, it all depends how much effort you want to put into it. to tune the gain pre-distorter, you may want to use a VGA to compensate for the insertion loss quickly during tuning.

 

[aniltunwal]

Hi thanks

i will try this on moday. i were getting good IMD when i checked the PA without predistortion but as soon as i connected predistorter it gets worst than before. So i thought it may be due to unoptimized predistorter parameters.

 

[gszczesz]

The MAX2009 is open-loop correction, which means you can expect it to make distortion worse before tuning, but better after tuning. This is because the distortion has to be tuned to match your amplifier. If it is not tuned, than it won't match it and instead of cancelling it, it will only add to it. So tuning is essential.

 

[aniltunwal]

Hi

The GCS and PBIN parameters are affecting more in my case. but still i could not see more than 4dB improvement in IMD. is this the real improvement from the MAX2009?

this is the im[provement when i am giving simple CW 2 tones.
any suggesstions...

After some when the amplifier is running for some time (half an hour), IMD gets worse. what can be the reason?
I thought it could be the memory affect but it doesnt improve if put fan on the fins. any suggesstions??

 

[gszczesz]

The limitted improvement in IMD is probably because the amplifier already implements some form on IM3 cancellation. You really should be looking at ACP3 using a modulated signal. Seeing 3dB improvement on an already cancelled IM3 is probably expected. Once you look at the ACP3, you should be able to tune it for 6dB improvement. I've seen 10dB improvements typically, but it's hard to maintain....

The amplifier's nonlinearity has some thermal dependence, and is why you see a shift in performance once the amplifier is running for a while. Ultimatelly, this limits the open-loop improvement. Ideally you have a thermal sensor in the amplifier and have a look-up table to tune MAX2009. What I would suggest maybe is to find the optimal settings for a warm amp and a cold amp, and then compromise in the middle???