JUMA PA1000 Linear Amplifier

[4z4rb]

it looks ok worth trying

[OH7SV]

Frank,

Thank you for a good advice. I purchased shielded Teflon wire from the johnswireshop.com
Here is one link http://stores.ebay.com/Johns-Silver-Tef ... 34.c0.m322

The wires look good and he price is preferable. I made inital mechanical measurements and estimated the impedance with AppCAD software.
We also measured the transmission line impedances. Inital tests show that these shielded wired are good at least in HF applications. See the picture.

PTFE shielded wires, updated with measured impedances

73 Matti OH7SV

[hb9dds]

Hello Ju + Ma

Very interesting amplifier. My wishlist:

- Fully Automatic
- Automatic Antenna Tuner built-in
- No knobs (looks old stylish) in the front panel. Push Buttons instead.
- just 1 display but bigger/wider
- SO2R
- Two inputs available
- 4 output's
- RS 232 port for PC control
- Fully protected

Like the Expert 1K-FA

Daniel, HB9DDS

[VK4AKV]

Hi Guy's,
Congratulations on putting together such a fine looking professionally finished amplifier using the current technology of freescale fets etc.

It is great to see some defelopment in the HF spectrum using the new generation Freescale Fets.

I would love to have one of these amplifiers sitting in my shack next to my PRO 111.

I have a 2M amp using 2 of the 300 freescale fets on 48V and it is very good, but I was always saying that what was needed was a new fet amp for HF using these modern bits.

Seems like it is here and has most of the requirements very well covered (no ATU is not really an issue) If this one goes well then we will see one with 2 Fets next

Keep up the good work

73
Kevin
VK4AKV

[vk4dd]

Hi,

Great respect for the positive work Juha Niinikoski and Matti Hohtola are doing.
It has been a joy to read about their projects and designs.
I am very impressed by how the kits are put together and it looks like perfection into the smallest detail.

The PA1000 from JUMA is I think the first of its kind and that means a lot of new experiences and techniques have to be pioneered. It is always good to prepare for a fall back scenario in case changes are needed. That is why I personally would have choosen to separate the RF section containing the big fet from the rest of the PCB. Time will tell if every thing on one PCB is the idea of a genius or that it has been a step to far. It certainly reduces a lot of wiring and complexity.

FCC seems to have a new requirement for max 15dB gain. You might want to look into that if you want to export to the USA.
The EU also have some requirements for power factor. That is the main reason why the high power solid state seems to go switched mode.

Parallel gate and drain. Have you considered using two fets and just parallel the gate and drain?
You would put less stress on the devices and potentially get 1200W out in the same configuration. Cleaner and Cooler.
I have have designed a PCB for the MRFE6VP6300H with parallel gate and drain using a single fet.
But push pull would be better. I kind of liked this part because it survived all nasty tests even shorting the drain to gnd.
It was very impressive to see how strong it was. See also http://vk4dd.com/forum/viewtopic.php?f=22&t=197
Free PCB are available for experimenters until stock runs out.

Best 73 Ron
VK4DD / PA4K

[OH7SV]

Hello Kevin and Ron,

Nice to hear that you are interested in Juma PA1000. I appreciate all your comments.

Ron, I understand your comment on the separate RF section. The reason to the design is that I am seeking simplicity and reproducibility because I like it and I don't like the popular hassle constructions. I understand the risk of the construction but it is worth of doing that as Albert Einstein is saying "Any fool can make things bigger, more complex, and more violent. It takes a touch of genius-and a lot of courage-to move in the opposite direction". Simple is harder to do but it gives better prize.

FCC requirement of max 15dB gain sounds strange. I have to look on that. What about a preamp with a gain of 20dB?

The EU requirement for power factor is actually EMC directive standard which defines max conducted harmonics emissions to the mains network. Direct rectification with a big stack of capacitors is not the best approach to that. Most probably the power supply will be replaced with a switched mode one.

The Freescale LDMOS. Most of the applications notes are for narrow band (fmin/fmax=1/1.1) VHF constructions and the output circuits are more or less resonating at the operating frequency. The 1.2kW can be reached with the flywheel effect in nonlinear FM. That is not possible in HF wide band linear operation. The bandwidth in HF is huge (fmin/fmax=1/17 or even 1/30 with 50MHz). Your suggestion to use two fets in parallel is a good idea. It seems that we can get 1kW single tone (CW) from one Freesacle MRFE6VP61K25 at wideband construction but most probably a sufficient linear operation is limited approx 1dB below (800W) with a single mosfet packet. The drain breakdown voltage (125V) limits the supply voltage to 50V...55V.

I have also tried BLF578. That fet is not so good for 1kW single mosfet HF amplifier. The gate voltage is even more limited for accidental overdrive. Additionally the drain breakdown voltage is only 110V which can be easily exceeded in a mismatch condition. I broke one BLF578 mosfet during a test when the drain voltage exceeded

The black art of output matching. It is very important to understand that there are no push-pull transistors. A transistor collector or a mosfet drain can only pull! In wide band operation without a "flywheel" the output circuit must handle the push action with the conducting transistor during OFF state of the other transistor. There are very few designers who understand this principle and people copy mistakes from each others. I have found only one article in the net which is telling the point. The writer is a clever guy from Chile, Manfred XQ6FOD. See the paragraph "Output architectures of conventional class AB push-pull amplifiers " in this link http://ludens.cl/Electron/mosfetamps/amps.html

I have been working with this fact for a long time and it seems I'm getting close to a good HF solution in kilowat range.

W'll keep you informed...

73 Matti OH7SV

[vk4dd]

Hi Matti,

From what I understood is that the gate is not the problem with the BLF578 (you can calculate what voltage it can handle and what associated power level that is). More likely the failure would be thermal. The BLF578 is not near saturation at 1 KW and keeps going, so you end up blowing up the drain source channel if you happen to have a power spike. You would be aware of P\power spikes or bad ALC off course. But many hams blow up fets like this depending on ALC to control power. Hams have to learn that big fets need attenuator. I think you know all that so this remark is not for you. Could it be that's why the FCC came up with this new 15dB rule for big amplifiers?

Quote from K6AER: The FCC 50 watts minimum drive requirement is for amplifiers out putting 1500 watts. The actual requirement from the FCC is the amplifier gain is limited to 15 dB or less. Most of the K3 users are into CW so IMD is not as important to them as the audio crowd.

You can ask K6AER about this 15dB rule from the FCC he highlighted this topic in a discussion about the Elecraft KPA500. That's how I heard about it. A preamp is a different thing and is not in this category power amplifiers.

Thank you for the interesting link to Manfreds work. Yes that second transformer in the supply has also an other function.
It reduces the DC offset in the RF transformer. That's will max flux in RF transformer, and off course the associated losses. may be that's why you end up with better efficiency?

Have you seen the polyfet application notes? http://www.polyfet.com/tbplt.htm
Have a look at TB212 they use this technique and they use a 100 Ohm attenuator in a feedback circuit.

I am not an expert in RF I only started to look seriously into broadband amplifiers since a year or so.
Before that I was more interested in tuned amplifiers. But I have learned a lot from reading and calculating the design notes, going through the measurements and simple experiments with my VNA. When I finally decide to drop in a fet than I want it to work. Designing and experimenting is one of the best aspects of amateur radio.

Its good to exchange ideas and learn something from each other.

I will continue to follow the juma webpages and forum. Hope that it will be all smooth sailing from now on.

73 Ron
VK4DD

[vk4dd]

Hi Matti,

Allow me to give you some more ideas. There is no need to have exactly a 50 Ohm match just behind the transformer on all bands. You can use the band filter and tweak the impedance of say the 6m and the 160m filter a bit .
Example.
If your transformer has a lower impedance of say 30 Ohms on 6M or may be 40 Ohms that you could also use the harmonic filter to do the rest of the matching. Include the track to the filter in your model and make it happen. Its not hard to design a harmonic filter which transfers 25 Ohm in to 50 Ohm out. Keep in mind that lower impedance means more current and as such more relay contacts needed. But that should not be a problem either.

It is just an other opportunity to take control over matching. It can be used to increase efficiency / IMD. What matters is that the fet is happy and runs clean and efficiently.

If you are planning to use a switcher than you could also think about lowering the voltage at lower output levels to increase efficiency. Less heat means better cooling, means longer life of the finals. All you need is some control over the voltage of the PS. I think its now a good time to stop loading you op with ideas

73 Ron.
VK4DD

[OH7SV]

Initial encouragement by Freescale in the choice of the device was strong but experience has revealed problematic levels of harmonic energy at the output. Freescale now acknowledge this problem to be significant when using these devices at HF. They have quoted H3 at -7dBc and H5 at -13dBc being their experience. These figures are bad enough to require use of a power dumping diplexer at the output or development of some frequency adaptive class F filtering which seems rather impractical in a broadband amplifier. Extensive discussion with Freescale resulted in them saying, "Development of a wholly acceptable amplifier for this frequency range (1.8 - 54MHz) will be very difficult to achieve." I am wondering whether you have had better experience with harmonic levels or found a way to address the issue satisfactorily?

I think there is no fundamental problem in LDMOS which produce harmonics in HF more that other semiconductors. Thus I think that MRFE6VP61K25H can be used down to DC if desired. I made these JUMA PA1000 measurements with the present output transformer to a 50 ohm dummy load without LPF-filters. Bias is 1A per mosfet. See the clean drain voltages at 3.5MHz.

JUMA PA1000 clean drain voltage swings at 3.5MHz

Naturally the drain voltages will start to clip when the mosfets start to saturate and there are odd harmonics in the signal as in any class-AB amplifier.
See picture below.

JUMA PA1000 drain voltage swings starting to saturate

These kind of clean symmetrical signals need a proper phase reversing transformer from drain to drain to maintain a good voltage swing in drains.
A MOSFET drain or a BJT collector can only pull, they are not capable to push. Actually push-pull BJT transistors or MOSFETs do not exist

Here is one example of bad voltage swing in drain without proper phase reversing transformer from drain to drain.

Bad drain voltage swing (This is not PA1000)

In a narrow band amplifier the situation is different because the tuned ouput circuit takes care of the push function, like in the "good old class-B tube amplifiers".
In a real broadband amplifier the max output power will be lower compared to a tuned amplifier because of the known reasons.
What do you say?

73 Matti OH7SV

[hb9dds]

Hi Matti and all other "SWL" here

I'm wondering, when the Project "PA1000" is really finished and ......... available?? Let us know.

Daniel