sine wave current booster / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

sine wave current booster

first previous 2 3 4 5 next

Move Thread

LAN_403

GeordieBoy

Tue May 26 2009, 01:28PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

Mn/Zn is a SMPSU power ferrite. Mn/Zn ferrite is what you would use for flyback transformers and high power ferrite transformers running below about 2MHz. It is also typically used for gate-drive transformers. It has high permeability but becomes lossy as you go higher in frequency. For these reasons Mn/Zn is often used for common-mode suppression chokes at the inputs and outputs of SMPSUs.

Ni/Zn ferrite is an RF material. It has a lower permeability and much lower core losses as you go up into the tens of MHz. This is the material of choice for gate-drive transformers around 4MHz and above.

I don't know much about linear amplifier design as I'm really a power electronics engineer so can't elaborate much further on things like transformer design. All I would say is that it is likely things like magnetising inductance, leakage inductance and inter-winding capacitance that determine what transformer works well in a broadband amplifier. That is why certain core types and geometries work better than others.

-Richie,

Sulaiman

Tue May 26 2009, 07:09PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

You may want to consider TWO amplifiers/buffers,
one using 'normal' audio-amp design, up to 100's of kHz
and
an RF amplifier for higher frequencies.

Why?

All the nice theories work below 1 MHz, just good closed-loop lumped-component design.
You can make a really nice PA for the sig-gen that is also an excellent audio PA.
(damping,accuracy etc. etc....)
Then make a PA as if for amateur-radio use
(There's vast ammounts of info/help etc on t'internet and I'm biased, it's a hobby of mine)

e.g. I can easily get 50 Wrms from a pair of IRF510 and a Ni/Zn toroid..and a few other bits=cheap
pretty flat from <<1 MHz to >10 MHz. (not yet stable for some complex loads)
There are some 'proper' designs out there that are well tried and tested.
For c100W, start looking for NiZn ferrite tubes as used on cables

Or just buy a hf linear amp. (eBay etc.)

Turkey9

Tue May 26 2009, 08:19PM

Registered Member #1451 Joined: Wed Apr 23 2008, 03:48AM
Location: Boulder, Co
Posts: 661

Can you push a simple mosfet halfbridge up past 10 MHz? I know it's just a square wave but it could be used for some applications.

GeordieBoy

Tue May 26 2009, 08:39PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

> Can you push a simple mosfet halfbridge up past 10 MHz?

Not easily.

And not with standard switchmode power MOSFETs if you want reasonable efficiency at 10MHz. The major problems are capacitive turn-on losses, cross-conduction and gate polysilicon losses. Using metal-gate RF MOSFETs removes the last problem but getting balanced operation gets more and more tricky as the frequency increases. Switch utilisation also gets worse and worse as deadtime starts to eat into the total period as the frequency increases. That is why the single-ended topologies like Class A, Class C, Class E/F etc become more popular as F goes up.

If you want a stick in the sand, the best i've managed *experimentally* was about 1.5kW at 5MHz with a pair of IRFP460A devices in Class DE. It was very finicky to set up!

-Richie,

Hazmatt_(The Underdog)

Thu May 28 2009, 03:16AM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

Sulamin, I would like to see some pictures of your IRF510 amplifier. I was trying to make the same thing but all I could do was burn up the mosfets! I've been using more rugged mosfets with 18A Id because my RF amp construction methods are really bad.

I also need to find out who is the source for RF chip caps that handle high power. I keep ordering the chip components with the wrong footprint (vastly too small).

MinorityCarrier

Thu May 28 2009, 04:56PM

Registered Member #2123 Joined: Sat May 16 2009, 03:10AM
Location: Bend, Oregon
Posts: 312

The Apex PA10 hybrid amp has gain/power up to 4Mhz. It might make a good interim solution (costly though) until you find something that does all that you want.

Sulaiman

Thu May 28 2009, 05:02PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

I haven't got notes and the pa is 'resting' at the moment.
.
I'll dig it out over the weekend.

IRF510 mosfet are cheap and have low capacitances,...other parameters are not so impressive
horribly non-linear below Vds=10V, poor Rds(on) etc.

Richie...1.5kW at 5MHz with a pair of IRFP460A........That's IMPRESSIVE.

GeordieBoy

Fri May 29 2009, 12:25PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

The IRF510 MOSFET is indeed an incredible little device! A rep I spoke to from International Rectifier told me that it was originally designed to switch filament lamps in automotive applications!

He said he was amazed to hear that people are using them in switched-mode power supplies operating in the high hundreds of kHz, and RF amplifiers operating in the tens of MHz!

The series IRF510, 520, 530 540 provides the RF amplifier hacker with a range of die sizes to trade off output power against maximum frequency. The small die IRF510 is suitable for low-power use up to high frequencies, and the 540 being able to deliver much more power and efficiency but to a lower maximum frequency. With progresses in MOSFET fabrication technology there are likely devices out there now with reduced reverse-transfer capacitances that are even better for RF amplifier use.

I think the only downside of these devices for linear amplifiers is their switched-mode design goals. The linear region is deliberately made as small as possible because you usually want to transition through this lossy region as quickly as possible to maximise efficiency. In linear applications this equates to very high gain in the crossover region. The situation is further complicated by a temperature coefficient on the gate-source threshold voltage that necessitates bias voltage to be temperature compensated. The pinchoff point essentially moves with temperature, and quiescent drain current rises quickly beyond the threshold voltage!

It is theoretically possible to get a couple of kilowatts out of a pair of IRFP460A devices in a half-bridge run in Class DE service. This is a switching amplifier with about 90% efficiency though. This means it has its downsides... It is narrowband, totally non-linear, and entirely intolerant of any load-mismatch!

The Class A amplifier alternatively has terrible efficiency, is broadband, highly linear, and can often tolerate infinite VSWR with ease. (A 20% efficient amplifier already burns up 80% of its input power as heat, so an additional 20% is usually easily dissipated in the event of total reflected output power. The switching amplifier on the other hand, with its 90% efficiency is only designed to dissipate 10% of its input power as heat. A load mismatch increasing this to 100% causes 10 times the temperature rise, and guatanteed thermal death if the amplifier is not protected!)

-Richie,

Sulaiman

Mon Jun 08 2009, 01:14PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

Sorry for the delay ......here is a pic of my 50W hf pa

It's just two IRF510, each on it's own heatsink, each with it's own dc-gate-bias circuit,
push-pull driving a quad-filar output transformer.
The output transformer is a TN23/14/7-4C65 with 4x 9turns 0.8mm copper wire, quad-filar.
Philips 4C65 material is almost identical to Material 61 .... Ni/Zn.
(a T100-61 would have almost identical performance)

In the photo you can just about see the two main heatsinks,
the output transformer and the two dc bias circuits.
Yellow wires are balanced output, green = center-tap.
Below is a 48V 60W smpsu adjusted for 45V dc.
50W rms CW no problem.

I don't want to publish the design yet cuz it's cr@p ... not stable with reactive antenna loads,
more work reqd. but not for a while due to other projects.

Hazmatt_(The Underdog)

Tue Jun 09 2009, 03:53AM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

That's totally different then my construction articles, I'd like to try it out and see how it works.

Are your gate bias networks capacitivily coupled to the input phase splitting transformer?

The trouble I'm having is the articles are suggesting that a large binocular core is used with brass tube, or one or two turns, and I just don't see how that is very practical. The impedance is very low at that point and it's easy to kill the mosfets.

Now what you're running into there with your amp is that the mosfet RF amps need output filtering. Filtering may not be as necessary with CW, but once you start modulating you will really need the filter.

I'll have to try this out with a variety of mosfets, ferrites, etc. I finally got some #43 tubes to play with.

first previous 2 3 4 5 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob