SS final bias

[WA1GFZ]

I just read the FTDX 5000 runs 10 amps of bias current while in class A. Does anyone know the bias current while in class AB? Also anybody out there with an HF rig that has FETs in the final. I would like to know what it runs for final bias current.
I just got my 1200 watt MRI board working at full power 160 through 6 meters and wonder where to set the bias. My board uses 4 SD2932 dual FETs the FTDX5000 uses the SD2931 which are single FET in a pack. Both use the same die.
The FTDX rates its final IMD3 at -31dB but at 100 watts while rates the output at 200 watts. 2 SD2931s will easily do 300 watts at .5 amp bias. The SD2932 is similar to the MRF151G

[WBear2GCR]

Whoa! Class A for an RF final? Wazzup wit dat? Seems like it would be frightfully inefficient?

10 amps at what voltage? Hot too...

I have no idea what the answer is, but I would think you could figure that out from the curves on the datasheet - of course it may matter WRT IMD where it is actually biased. Again, I dunno nutin'. Just interested.

                        _-_-bear

[W1AEX]

Frank, the 4 FET finals in my Flex 5K draw a bit more than 4.0 amps in class AB. As I recall, when I ran the software routine to put the finals into class A they drew right around 10 amps at idle.

Rob W1AEX

[WA1GFZ]

Thanks Rob,
That is good information. I'm running 2 amps into 8 FETs. My IMD is not that great. When I first powered the module it was up around 8 amps but the heat sink couldn't handle the dissipation so cranked it back. My 300 watt strip works pretty well running 1 amp per FET. I also run the driver in class a at 1 amp. IP3 in that stage is better than -40dB. There isn't a lot of data out there on FET bias in class AB. My simulation shows 1 amp per FET looks good. I will be building a giant heat exchanger.
Rob, I'm running 48 volts what does the 5K run on the final?

[W1AEX]

Not only is there limited data, but from what I've observed, there's not a lot of attention being paid to driver and PA bias setting with a few rigs rolling off the assembly line from at least one major plastic radio manufacturer (cough cough Yaesu cough cough).

The Flex 5K PA runs at 13.8 VDC. When the final stage is running in Class A the driver stage also is also adjusted, but not sure if it is moved all the way into Class A. When they were pushing this feature (it is no longer supported because too many hams didn't understand how to implement it) they suggested limiting peak output to about 30 watts to produce the cleanest signal.

By the way, it looks like Flex will be pulling the covers off its newest offering at Dayton this weekend. They are calling it a "game changer". The clues leaking out of the marketing office hint at the following areas of change in the architecture:

completely revamped networking
digital starting right at the antenna connector
major advances in signal processing

The best guess at this point is that they are applying the direct sampling technology that they have been supplying to the military into their next generation ham hardware. The interface will most likely be gigabit ethernet instead of firewire or USB 3.0 and the advances in signal processing hints at a new software platform. No one knows for sure, but Flex is definitely excited about it. We'll see what shows up at Dayton.

Sounds like you are closing in on the final adjustments and setup of the HPSDR. I'll be looking forward to hearing it!

73,

Rob

 

[WA1GFZ]

Rob,
Last night I remembered I have a rack mounted blower stashed away. I'm going to set the heat sink on the output holes and move some big time air through the fins. This might allow me to get on the air with it at full power. Yes driver really needs to be class A to have a clean output. Yup you want to reduce pwer to around 25% running class A.
Looks like the 300 watt strip running class A will drive the 4 board final. then it should be possible to run 1500 watts output class A if I want, or just talk louder to qro.

Flex game changer my bung hole. They are just going to copy HPSDR again. We have gigabit ethernet now. We just have not moved the jumper to go to full speed yet. I have dual receiver modules and 100 MHz handles it fine. Again Flex copies us but now they don't open source the software. Flex one way street.

HPSDR Hermes will be at Dayton also. 4 inch by 6 inch XCVR board that does everything my set does except for dual RX on dual antennas.
My next cabinet to my system will be the Hermes with an amplifier. The present system will be used for RX only and become a test set I may go to quad RX at some point for beam steering.  There will be a game changer when there is a tunable preselector.

[W1AEX]

That would be great if the rack blower does the job. I saw Bob K1KBW and Bill W1MPY walking around at NearFest with some major chunks of heat sink material. Too bad some of that didn't end up in your trunk!

Actually, the direct sampling Flex CDRX-3200 with gigabit ethernet has been in production and in the hands of the military and various government agencies since the end of 2009, so I don't think they are stealing the concept from anyone else. The thought is that they will be offering a consumer version of that platform. (Other than the military and various government agencies who needs a box with 32 individual direct sampling gigabit ethernet streaming receivers?) Also, the same portions of PowerSDR that have always been open source continue to be open source and in fact the latest version will work fine with Softrock boards and whatever else needs it. I frequently get email from the PowerSDR-IQ guys who have questions about adapting the custom skins I make for PowerSDR. They re-write PowerSDR for use with lots of homebrew hardware and connection interfaces. The Flex hardware's firmware is proprietary, the distributed PowerSDR skins are copy protected, and the firewire drivers, USB drivers and DLL's that are tied to communicating with the Flex interface components are also proprietary and copy protected. So... nothing has changed there other than if third parties wish to use it with their hardware they will need to select SDR-1000 or write their own interface code.

At any rate, it should be interesting to see what they're all excited about! Good luck with the cooling!

Rob W1AEX

[WA1GFZ]

Rob,
I have the heatsinks for the final build once I determine all the changes. This is the first prototype to work out  the design changes to pull it from 80MHz down to 1.8MHz.  I picked up a very nice six foot extrusion from Antek with almost 10,000 square inches of surface area. This will increase the heat sink more than 4 times over the stoch MRI extrusion. Each board only has 550 square inches of area so quite small. I'll be running close to 2500 square inches per board. Also have a slab of copper as a spreader to help things. I now have the same gain on 160 as 6 meters but have not yet driven it to full power PEP. The blower should get me there. BTW when we worked a couple weeks ago the final was biased class B due to RF getting into the shutdown circuit. It ran a lot cooler but still couldn't talk too long. So far all the new transformers are nice and cool.
I will have to drill and tap about 50 holes per board so want to make sure I have it right before I drill and blast.
The final heat exchanger for 4 boards will weigh almost 90 pounds
Flex cut us off from their latest features.
 

[kb3ouk]

Not only is there limited data, but from what I've observed, there's not a lot of attention being paid to driver and PA bias setting with a few rigs rolling off the assembly line from at least one major plastic radio manufacturer (cough cough Yaesu cough cough).

Weren't they the same ones that brought out a rig a few years ago that would do 400 watts barefoot, but had adjustable bias and could be set to run class A at 100 watts? I have no idea what solid state Yaesu stuff is like, the only solid state radio I have of theirs is a 2 meter HT, but their old tube gear seems to have some really sucky bias regulation. Just about every time I retune the final in one (especially the FT-101 series), I have to reach in and tweak the bias to bring it back to where it should be. My bad 901 is the same way, now I'm trying to figure out how to rip it apart to salvage it for a homebrew rig using the 6146s, the VFO, and some other parts out of it.

[WA1GFZ]

Rob,
I should be on Thursday night, the blower works and allows me to run 200 watts carrier out without getting very hot. I still need more pad at the output of the HPSDR because the buffer amp (300 watt strip running 24 volts class A) has too much gain. Carrier output setting of 2 puts me at 200 watts. On AM you really need to set the output higher to take advantage of all the D/A bits. Another 10 dB should do it. I don't think I can set the bias back up where it was yet.

[W1AEX]

Sounds like you've got it tamed. I will most likely be floating around on 160 meters at around 9:30 pm tomorrow night. Might catch you then!

[WA1GFZ]

I'm going to start on 75 but might try 160 later. I appear to have a dip in gain on 75 only every other band the gain is the same.
Sometime after 7:30.

[W3RSW]

I agree with Frank.  HPDSR and Phil's, N8VB, QS1R have been trucking' along for years. Phil finally has released an xmitter board, albeit only 1or 2 milliwatts for the experimenter. One more stage and a Drake TR7 pre/PA brick will get you to 150 pep or so.

Flex obviously has the business locked up and the sales of old tech sound card machines to generate the cash to finally go to direct digital conversion for we, the unwashed.

[WA1GFZ]

The thing is direct digital takes less hardware. Expensive hardware though. The HPSDR Hermes module will do 1/2 watt and have everything on one board.
All you need is a final amp and filters. Too bad Flex doesn't do open source any more.
Somebody actually tried to patent the HPSDR configuration. Good thing we have legal people in our gang to stop it.

[W3RSW]

Yeah Francois, shades of the S-100 buss. 

[WA1GFZ]

200 watts carrier and well over 100% positive peaks.
300 watts carrier things got a bit too warm for the stock heat sink.
All the transformers nice and cool.

[W1VD]

Frank

Listened along last night while working at the bench. Once the audio level was set correctly and some highs were added it sounded pretty good. Did notice an ocassional anomoly on the carrier that appeared unrelated to modulation ... maybe a phase jump? Didn't happen while I had the BFO on but did notice the carrier seemed to have very slight phase noise. Maybe rf getting back in to the exciter?

Anyway, this is a hyper critical report and another listen would be in order to back this up. Will be around tonight and over the weekend if you're interested.     

[WA1GFZ]

TNX Jay,
Everything very JS so there could be some RF getting into the exciter. I have 26dB of pads between the exciter and ERB 300.
Erb 300 is in class A. Then about 6 feet of cable to the floor where final is sitting on a fan. Power leads are also about 6 feet long.
Nothing is grounded. At 200 watts I was hitting about 150%. Later I put the compressor on and changed the EQ slightly. The big power supply is on a different circuit, (washing machine across the basement) so I bet there is a big ground loop. Also it is a constant voltage transformer telpephone battery charger supply, so might be dirty.

[W1AEX]

Frank,

Sorry to have missed the Thursday night 75/160 session(s). I didn't get back until late. I'll be on the lookout this weekend for your wireless! By the way, for your reading amusement, the release data and specs for the new Flex 6000 series are available in a PDF (3 meg) at the link below. There's already some moaning about the price for the Flex 6500 ($3999) which is what most will be looking at. It's a rack mountable direct sampling platform with gigabit ethernet running from 160-6 meters @ 100 watts. Receive runs from .30 MHz to 77 MHz. It will run software called "SmartSDR" which will allow the user to view up to 4 simultaneous full band panadapters while operating.

http://www.flexradio.com/FLEX-6000.pdf

Looks like fun, but it's not for me at this point! I would imagine that Jay will grab one just for measurement purposes!

Rob W1AEX

[ke7trp]

Flex is locked down tighter then a Frogs A$$ now.  No more playing around. Also. The new rigs are bandwidth limited.  The fun is over for running 20 KC wide

C

[WA1GFZ]

Rob,
I saw that on their site. It is basically the same set up as I have but clocked at twice the speed with the same number of bits. It looks like they improved the input filter to the RX. But I'm would like to see their close in clock phase noise. A good 1GHz SAW oscillator might be -105 dBc at 1 KHz.
I'm not sure how well this equates to phase jitter.
Well now everybody is paying for software. Yes, quite over priced for the small amount of hardware but Flex was always over priced. Send it to Rob Sherwood and see how he rates it. Notice 110dB IMD is speced inside the RX bandwidth. That is bogus. I bet it is no better than 100 dB outside, 16 bits.
DDC gives you a few dB over 16 bits.
I might be on Sunday later in the day. It would be cool to have you take a close look at spectrum and IMD. Still only running 2 amps of bias. Time to mount it on a real heat sink and air it out.