EEVblog Electronics Community Forum

Electronics => Projects, Designs, and Technical Stuff => Topic started by: andig on October 20, 2014, 05:30:44 am

Title: Require pointers for a 300 Watts Linear amplifier with 500KHz Power Bandwidth
Post by: andig on October 20, 2014, 05:30:44 am

This was posted on the TI e2e forum and no reply within 24 hours. I guess I posted in the wrong forum section. I just joined EEVblog and thought you guys are great. So...here is a repost.

Hi All,

Forgive me if I am posting to a wrong forum. Moderators please move if required.

I need to figure out a low cost DIY solution for soft ferrite power loss testing. Below is what I have in mind:

A toaster converted into a temperature controlled oven - I don't know what will be the max temp possible -- with thermocouple connected to the Raspberry Pi via ADC
A Function Generator using Raspberry Pi+DDS Chip
A Linear Power Amplifier with about 300 watts output power upto 500Khz
VAW measurement via True RMS High Frequency Current and Voltage Transducer -> I2C for Raspberry Pi-( Yet to Figure Out )
I am clueless on the Linear power amp. Input will be a single frequency at a time, sweeping from low to high. However the load is completely inductive.  Class-A  with a tracking DC-DC (sort'a not envelope tracking, since input magnitude is fixed) converter to keep efficiency High (read Thermal Loss Low) is what I was pondering upon. But then I am not experienced in this subject. I am also concerned about VSWR protection for the amplifier output.

I can handle the programming jobs but I need reference for the amplifier part.

So my questions for the amplifier are:



• Homebrew 10KHz to 500KHz, 200 - 300 watts possible ? Being only experienced in embedded system and power electronics am I inviting nightmares ?
• Any circuit references ? Can I use Audio amplifier class AB as a starting point. Do I risk oscillating tweaking the compensation ?
• Can a Power OP-AMP like the OPA549 be used at all with some boosting or paralleling ? Since the load and the input signal is constant once setup for each measurement cycle and the OP-AMP supports current programming, can I use a MSP430 to sense and balance currents between parallel units actively
• If yes with a GBW of 0.9MHz would it handle 500KHz LINEARLY ? What about the very low impedance inductive load.

Consider me a newbie on amplifiers and ambitious too.

For the entire project and specifically the amplifier please comment, suggest, criticize scare, and enlighten me.

Regards

Sougata

P.S.: For those who are pondering why I am upto this: I will lose $200 on bet if I don't do this and risk losing 10 times if I try and fail.

Title: Re: Require pointers for a 300 Watts Linear amplifier with 500KHz Power Bandwidth
Post by: nctnico on October 20, 2014, 09:11:17 am

Don't fixate on GBW. What you need to consider is the slew rate. There are lots of opamps with a high GBW but the slew rate is insufficient to output full bandwidth at maximum output swing. I still have a plan to use a bunch of LT1210 amplifiers in parallel to make a high power amplifier.

Title: Re: Require pointers for a 300 Watts Linear amplifier with 500KHz Power Bandwidth
Post by: mzzj on October 20, 2014, 10:06:43 am

What kind of load impedance are you expecting?How much current and voltage?

(modified)Class AB audio amplifier circuit can get you close but  expect to sort out few problems...
Quite many audio amplifiers do ~300khz to 8 ohm with some extra losses.

Title: Re: Require pointers for a 300 Watts Linear amplifier with 500KHz Power Bandwidth
Post by: MK on October 20, 2014, 11:34:21 am

you need to define the impedance for the 500 KHz 300W amp, then you will be using fast devices in the ouput stage, oscillations are quite likely, so you will lose the magic smoke out of a few sets of power transistors before you are done.

Again, true rms up to what frequency? you need to know the upper limit there, that will dictate the approach you take.

Title: Re: Require pointers for a 300 Watts Linear amplifier with 500KHz Power Bandwidth
Post by: T3sl4co1l on October 20, 2014, 12:14:33 pm

Easy, resonate the ferrite coil with a capacitor, drive with an inverter at the resonant frequency.  If anything, the small square wave component will be more representative of final use (assuming you're doing this for a switching application, in the end).

Tim

Title: Re: Require pointers for a 300 Watts Linear amplifier with 500KHz Power Bandwidth
Post by: moffy on October 20, 2014, 11:11:40 pm

I agree with T3sl4co1l. You don't even have to use resonance, just drive with a suitable square wave.

Title: Re: Require pointers for a 300 Watts Linear amplifier with 500KHz Power Bandwidth
Post by: mzzj on October 21, 2014, 05:50:52 am

Quote from: T3sl4co1l on October 20, 2014, 12:14:33 pm

Easy, resonate the ferrite coil with a capacitor, drive with an inverter at the resonant frequency.  If anything, the small square wave component will be more representative of final use (assuming you're doing this for a switching application, in the end).

Tim

I think OP wanted variable frequency sweep?  :-/O

Title: Re: Require pointers for a 300 Watts Linear amplifier with 500KHz Power Bandwidth
Post by: Richard Head on October 21, 2014, 06:16:22 am

Ferrite losses are generally specified with a sinewave excitation signal using a toroid. Obviously if you use a square wave there will be higher frequency components that will add to the fundamental frequency losses and introduce errors.
I too would go with the resonant method using a high Q capacitor.