HFU HF Underground

Technical Topics => The RF Workbench => Topic started by: Stretchyman on January 05, 2019, 1850 UTC

Title: Simple SiC TX
Post by: Stretchyman on January 05, 2019, 1850 UTC
Having designed what I think is the limit of my skills (40W Gan, programmable DDS) I've decided to build a much simpler Tx to reduce cost and increase interest! (Hopefully...?)

Having recently discovered CARDINAL CPP oscillators (thank's to someone on here, Cheers!!) it greatly simplifies design.

I now have a Simple SiC TX with ONLY 17 components.

I've found a UK company to wind Mod Tranny's for me and will get the cases milled at source for a decent finish.

Frequency changing is as simple as plugging in a new oscillator and as they're <$5 I think it's a good way to go.

I've had one on the air and no complaints of any sort, rock steady and very efficient with class D Modulator and class E RF stage using C3M CREE SiC mosfet.

Should be ready in a month or so and will advertise on eBay but please get in touch directly and save yourself some $/.

ALL the Best and Happy New Year.

Stretchy.



Title: Re: Simple SiC TX
Post by: Monophonia on January 09, 2019, 1808 UTC
Being able to order those oscillators from Digikey on frequency is a plus, If I ever make something that is truly fixed frequency I may explore those.
Title: Re: Simple SiC TX
Post by: Stretchyman on January 10, 2019, 0719 UTC
Just counted up the amount of components it has replaced, =41!

Current design has programmable DDS which SW needs to be written for etc (at cost as I didn't write it myself), that and a filter and an amplifier and a squarer to drive the driver, lots of bits!

For anyone wanting to build a simple TX just use the 'usual' class E design replacing the 74HC240 with a CPP oscillator and the FET driver of your choice.

I'm trying to find a source for cheap 'unprogrammed' oscillators (DUAL ones preferably), ANYONE?

DK seem to charge more for those than programmed ones, which is weird. Programmers aren't cheap either but will get one if I can get the OSCs cheap!


Str.
Title: Re: Simple SiC TX
Post by: Echoqueen on June 12, 2019, 0315 UTC
For anyone wanting to build a simple TX just use the 'usual' class E design replacing the 74HC240 with a CPP oscillator and the FET driver of your choice.

Hi!

I can only source TC4422? Are those drivers ok for 7 MHz?

Thanks
EQ
Title: Re: Simple SiC TX
Post by: JimIO on June 12, 2019, 0401 UTC
I think so. I had looked into it and ordered some from china on ebay but I never got them. 

   http://www.ece.ualberta.ca/~ee401/parts/data/TC4421.pdf
Title: Re: Simple SiC TX
Post by: redhat on June 12, 2019, 0550 UTC
I have not had good luck with 4422 above medium wave.  4452's work, but you'll need the to-220 versions to get rid of the heat.  The DIP-8 versions will smoke...been there.

+-RH
Title: Re: Simple SiC TX
Post by: Stretchyman on June 12, 2019, 0930 UTC
Why not use something a tad more modern?
 
NCP81074ADR2G.

It's what I use in this design.

Please forget the ones you mention as they're old, slow and obsolete. FET drivers have moved on a lot. Even smpsu's are running at 4 MHz now!

Str.
Title: Re: Simple SiC TX
Post by: Stretchyman on June 12, 2019, 1244 UTC
Images here;

https://www.ebay.co.uk/itm/133036900946?ssPageName=STRK:MESELX:IT&_trksid=p3984.m1555.l2649

Check out the PCB, small, simple & class E so rather efficient.

You wont find a simpler or more efficient design, rugged too!

Quite happy to sell bare (or populated) PCB's, if you want to sort out your own modulation scheme.

Please ignore the inflated eBay prices, much cheaper if you buy direct!

Str.
Title: Re: Simple SiC TX
Post by: redhat on June 12, 2019, 1755 UTC
Please forget the ones you mention as they're old, slow and obsolete. FET drivers have moved on a lot. Even smpsu's are running at 4 MHz now!

I take it you forgot how to read again?  He said...

"Hi!  I can only source TC4422? Are those drivers ok for 7 MHz?"

The parts you mention are not usable by the average hobbyist without reflow equipment.  The ones I mentioned are available in through hole and also a TO-272 smd package which is with care practical for most folks.

+-RH
Title: Re: Simple SiC TX
Post by: Stretchyman on June 12, 2019, 1906 UTC
The older devices aren't very good, just generate a lot of heat and aren't very fast either.

The NCP's are available in a SOIC 8, 'bout half the size of a DIP 8 so fairly easy to solder with the naked eye. X4 eyepiece is however the preferred method, no reason to be shy of small things!

Look, I'm nearly 60 years old, my eyesight ain't that good! I guess the main issue is most folk don't have a decent soldering iron?

METCAL being the tool of choice.

Str.
Title: Re: Simple SiC TX
Post by: Echoqueen on June 12, 2019, 2121 UTC
HI folks

Ok, undestood. Thanks for the advice.

The main issue is not the size of the components but the fact that I cannot buy the new-and-faster ones here. They are unobtanium for me right now.

There are some ixys that are mentioned elsewhere in the forum that maybe I could get.

Cheers.
Title: Re: Simple SiC TX
Post by: OgreVorbis on June 12, 2019, 2307 UTC
HI folks

Ok, undestood. Thanks for the advice.

The main issue is not the size of the components but the fact that I cannot buy the new-and-faster ones here. They are unobtanium for me right now.

There are some ixys that are mentioned elsewhere in the forum that maybe I could get.

Cheers.

I know from personal experience that the 4422's are no good above 5 MHz. The NCPs I am about to test in a few days.
If you can't source those, then get the IXDN614's. I know these work at 7 MHz if the board is done right.
Title: Re: Simple SiC TX
Post by: Stretchyman on June 13, 2019, 0648 UTC
I can send you some, no problem.

You can buy anything from anywhere surely?

As long as you're on the same planet!

Where are u bud?

Str
Title: Re: Simple SiC TX
Post by: redhat on June 13, 2019, 1005 UTC
Granted, a lot of the suitability will depend on how large of a gate structure your trying to drive...but I've found with most of the SiC and conventional mosfet below 1000pF work fine to 7 MHz with the '614s and '4452's  I eventually settled on the 614's largely because the tab is tied to ground, and this simplified mounting and heatsinking.

Stretchy, your rambling are unfounded.  The IXDN614's have nearly identical switching times to your specs of dust, and are easier to replace.  I also worry about the longtime durability of a part that does generate heat and can't get rid of it.  For the same reasons I had DIP-8 versions of the 4452's melt down while the tabbed versions were fine.

Sometime in the future a new PA layout is in order...2 RF fets for 1.5KW, 4 for 2.5KW...carrier that is.  5KW in 5 rack units, for the power modules anyway....wow.

+-RH
Title: Re: Simple SiC TX
Post by: Stretchyman on June 13, 2019, 1834 UTC
Well I must apologize!

I have an unfounded attitude to make things small!

The 'Specs of dust' don't get that hot BTW...

Regards.

Stretchy.
Title: Re: Simple SiC TX
Post by: OgreVorbis on June 14, 2019, 0028 UTC

Sometime in the future a new PA layout is in order...2 RF fets for 1.5KW, 4 for 2.5KW...carrier that is.  5KW in 5 rack units, for the power modules anyway....wow.

+-RH

What FETs are you talking about? What frequency? That's an insane amount of power. Is this some design you are currently working on?
Title: Re: Simple SiC TX
Post by: redhat on June 14, 2019, 0715 UTC
Same ones I'm currently using, C2M0080120D's and C2M0045170D's.  Of course its all theory until you build one... 

A big part of the problem in getting above 1KW is the thermal management.  Large bonded fin heatsinks and fan capacities over 200CFM become mandatory, especially for gear meant to survive operation in outdoor summer conditions.

It will be quite the feat when it is running.

+-RH
Title: Re: Simple SiC TX
Post by: OgreVorbis on June 15, 2019, 0524 UTC
Same ones I'm currently using, C2M0080120D's and C2M0045170D's.  Of course its all theory until you build one... 

A big part of the problem in getting above 1KW is the thermal management.  Large bonded fin heatsinks and fan capacities over 200CFM become mandatory, especially for gear meant to survive operation in outdoor summer conditions.

It will be quite the feat when it is running.

+-RH

https://www.fischerelektronik.de/web_fischer/en_US/K%C3%BChlk%C3%B6rper/D/L%C3%BCfteraggregate/index.xhtml (https://www.fischerelektronik.de/web_fischer/en_US/K%C3%BChlk%C3%B6rper/D/L%C3%BCfteraggregate/index.xhtml)

These are a bit pricey, but it's a good company and they have everything.
Title: Re: Simple SiC TX
Post by: redhat on June 15, 2019, 2106 UTC
I found a local company that does custom bonded fin stuff for reasonable prices.  I'll see how that works out.

+-RH
Title: Re: Simple SiC TX
Post by: Josh on June 15, 2019, 2137 UTC
No pelts or is that too much current?
Title: Re: Simple SiC TX
Post by: redhat on June 16, 2019, 0524 UTC
You mean peltier junction modules?  I haven't looked into it, but I would guess too much power consumption for the drop in temperature you get.

+-RH
Title: Re: Simple SiC TX
Post by: Josh on June 16, 2019, 1808 UTC
Shame there isn't something that can absorb all that heat and turn it to lectricity.
Title: Re: Simple SiC TX
Post by: ChrisSmolinski on June 16, 2019, 1849 UTC
Shame there isn't something that can absorb all that heat and turn it to lectricity.

Curse you, Laws of Thermodynamics!
Title: Re: Simple SiC TX
Post by: JimIO on June 16, 2019, 2046 UTC

   https://www.autozone.com/cooling-heating-and-climate-control/heater-core/toyota/corolla
Title: Re: Simple SiC TX
Post by: Josh on June 19, 2019, 1854 UTC

   https://www.autozone.com/cooling-heating-and-climate-control/heater-core/toyota/corolla

While I'm sure our brother Jimmy is being droll, water cooling might be an option, taking expense in funds and current as well as complexity/reliability issues into consideration.
Title: Re: Simple SiC TX
Post by: ChrisSmolinski on June 19, 2019, 1913 UTC

   https://www.autozone.com/cooling-heating-and-climate-control/heater-core/toyota/corolla

While I'm sure our brother Jimmy is being droll, water cooling might be an option, taking expense in funds and current as well as complexity/reliability issues into consideration.

Years ago I worked with an engineer originally from Egypt, he worked at their broadcasting department, I think mostly with the MW transmitters. He mentioned how they had some Soviet supplied transmitters that used steam cooling for the tubes, which was actually an excellent way to do it, initial thoughts aside.
Title: Re: Simple SiC TX
Post by: redhat on June 20, 2019, 0537 UTC
Water cooling is a good option for stationary service but for field use is totally impractical.  If you have a good source of cold process water from a well or something much less than ambient it would be worth doing In some situations.  It may in Some cases allow you to use fewer power devices for a given output power.  I considered this for 1kw on up as it would allow two devices to reach 3kw carrier, but only if you could keep the chill block temperature under 80 degrees F.

+-RH
Title: Re: Simple SiC TX
Post by: Josh on June 20, 2019, 2032 UTC
Perhaps hsf units with heatpipes as commonly used in the pc world could be a compromise?
Title: Re: Simple SiC TX
Post by: Stretchyman on June 22, 2019, 1124 UTC
A decent design should generate very little heat.

As you seemed determined to use outdated drivers which wil generate 5W of heat each trying to drive 4nF of gate C you will need an enormouse heatsink.

Sticking with the C2M0080120D should help as they have less than 1nF of gate C to drive.

Just to explain whats going on here.......

To maintain high efficiencies with class D or E the switching devices must be turned on and off very quickly.

Any time spent in switching between on or off leaves the switching devices in a resistive region and will generate heat.

So if it's off it's open circuit (no current drawn) and if it's on the only heat genearated is Current squared X RDS(on).

So with the C2M0080120D (RDS(on) is only 80mR! super low, so with lets (say) 10A flowing thru' it will only generate 8W of heat.

So not much!

The trick is to use high voltages and as these devices have a VDS of 1200V, you can use (say) 150V with no issues.

So we know we can push 10A and use 150V, so that's 1500W of potential power.

However unless the design is over 95% efficient, your going to generate a fair amount of heat.

Even 5% is 75W!

I dont think any design is going to work with just 2 devices.

Looking at the current 'Trend' maybe 8 fror a 1KW carrier, there's certainly no need to use 24 11N90's, that's a bit silly now....

Anyway you seem to have hijacked my thread as I was trying to promote simplicity, not high power, I'll start another thread...

Str.

Title: Re: Simple SiC TX
Post by: redhat on June 22, 2019, 2146 UTC
My current test bed has logged may broadcast hours using just two RF devices running at 600-700W carrier and about 165VDC of B+ before the modulator.  The problem with adding more devices is you multiply the number and quantity of circuit strays, and this in turn winds up limiting your operating frequency.  This is the primary reason those frankenstein rigs with dozens of fets barely work on 80 meters, and not at all on 40 meters.

If you want big power, you need big fets, which have higher input capacity.  What I'm doing now was not practical prior to the introduction of SiC devices due mainly to limited Vds ratings and input capacity fo conventional fets.  The thermal management problem is the last hurdle.  The current PA runs about 93% efficient at 6.9 MHz.  Add to that the losses in the modulator, around 4-5%, total DC to RF efficiency is on the order of 88%.  At carrier that is approximately 30W per device, and it scales from there.  At 1500W and average modulation, its around 77W per device and the thermal headroom starts to get thin.  Maximum case temperature at 77W for the C2M0080120D is about 100 degrees C.  The aluminum oxide insulators we use have a thermal resistance of about 0.315 degrees C per watt, so expect a 25 degree C rise right there.  A reasonable heatsink temperature is about 50-60C, so we would need a heatsink capable of dissipating 260W of heat and exhibiting a rise of only 10 degrees C.  This means we need a heatsink with a rating of 0.038 degrees C per watt, which is tough, but attainable under forced air with a bonded fin heatsink.

I'll try some of your exalted fet drivers this week and see how they do.  If history is any indication, I expect them to slide off the board when trying to drive one of my fets.

+-RH
Title: Re: Simple SiC TX
Post by: Stretchyman on June 22, 2019, 2200 UTC
Good news and Good Luck!

At least you know what your doing and more realistically what you're up against.

The NCP'S will drive the C3M0280090D's just fine on 40m. However for a 1KW carrier your going to need more than 2!

For the crazy powers you are generating and the large gate C's (Qg's) the NCP'S will fry for sure driving high gate C FETs.

Sounds like you're best sticking with what you got!

Str.