• You can now help support WorldwideDX when you shop on Amazon at no additional cost to you! Simply follow this Shop on Amazon link first and a portion of any purchase is sent to WorldwideDX to help with site costs.

Chinese MiniPA-100 Kits

G4XRZ

New Member
Jan 31, 2022
6
3
3
65
Just adding another pic, of the inside...
Hello Rick and thanks for the great info. Can I please ask that you explain more about your heatsink arrangements ?
Is that FR4 board on top of an aluminium heatsink with the 9120 attached direct to aluminium via a cut-out ? - or some other arrangement. I would be very interested. Also, I have read many reports of the large wattage feedback resistors cooking.
Can you please advise what values your resistors are, and also what the value the resistor is across the first input transformer ?. (I am having difficulty making out the actual band colours (old age haha :) ) - ah, correction - different glasses ;), I believe the resistor across the input tx/splitter looks like 22 Ohms and the feedback resistors 100 ohm ? - is that correct sir ?
Also what bias current did you use please ?
I specifically bought a couple of the 70W versions (around 18 Euro only) which are - or appear to be - identical apart from the high wattage resistors and of course the PA transistors. The switching fets of dubious origin on the 70W version I will throw in the parts bin and I also specifically purchased genuine used MRF9120 units as some sellers are providing non genuine parts (although yours definitely looks like a used and fully genuine part ! - in fact it is identical to the two 1used 920 units I have. - so you got a really good seller !).
Apologies for all the questions. I am very impressed with your results.
73
Joe G4XRZ
 
Last edited:
  • Like
Reactions: Shadetree Mechanic

Rick Miller

Member
Apr 1, 2015
27
40
23
53
Hi Joe, I used a 'CBS-4' package from ICA Manufacturing as my starting point. https://www.icamanufacturing.com/cbs-packs/cbs4m.

I then cut out a hole in the PCB to allow the PA-100 kit to fit into the hole. Thus the 9120 mounts directly to the aluminum heatsink below. I decided to do this to maintain proper airflow through the heatsink and around the bandpass filter board. It's over sized for sure, but I can run FT8 all day long without worry.

As for the feedback resisters getting hot, that would most likely mean there's oscillation happening. When I get home tonight I can take some pics and show you a couple tricks to help keep things stable. I've built 4 of these now, and the only one that was stable from the start was my first one. Since then I've learned a couple of tweaks...

The bias... The first one I built was easy to set to 1A quiescent current, however the next one would not, no matter how much bias I gave it. It would produce power, but I'm sure it wasn't linear. I went out and purchased some new 9120's and they had the same issue. Then I decided to up the bias voltage to see what would happen, that did the trick. So I replaced the 7805 voltage regulator with 7808 and that seems to have fixed the issue. However, this is not spec so I'm thinking these 9120's are somehow different than originals.

Next, the design allows for different styles of transistors. If you're using the 100W/9120's then there some extra PCB material that isn't really needed. I cut the traces off and simply ground the ends. For me this seemed to help stabilize things. Also, though I haven't had this issue, I hear if you use an insulator under the PCB it can cause oscillation as well, don't do that. The PCB should be directly mounted to the heatsink.

When I get a chance to take some pictures I'll show you how I cut the unneeded material away. You are correct on the resistors, 100 ohm for the feedbacks, and 22 ohms across the input transformer.
 

G4XRZ

New Member
Jan 31, 2022
6
3
3
65
Hi Joe, I used a 'CBS-4' package from ICA Manufacturing as my starting point. https://www.icamanufacturing.com/cbs-packs/cbs4m.

I then cut out a hole in the PCB to allow the PA-100 kit to fit into the hole. Thus the 9120 mounts directly to the aluminum heatsink below. I decided to do this to maintain proper airflow through the heatsink and around the bandpass filter board. It's over sized for sure, but I can run FT8 all day long without worry.

As for the feedback resisters getting hot, that would most likely mean there's oscillation happening. When I get home tonight I can take some pics and show you a couple tricks to help keep things stable. I've built 4 of these now, and the only one that was stable from the start was my first one. Since then I've learned a couple of tweaks...

The bias... The first one I built was easy to set to 1A quiescent current, however the next one would not, no matter how much bias I gave it. It would produce power, but I'm sure it wasn't linear. I went out and purchased some new 9120's and they had the same issue. Then I decided to up the bias voltage to see what would happen, that did the trick. So I replaced the 7805 voltage regulator with 7808 and that seems to have fixed the issue. However, this is not spec so I'm thinking these 9120's are somehow different than originals.

Next, the design allows for different styles of transistors. If you're using the 100W/9120's then there some extra PCB material that isn't really needed. I cut the traces off and simply ground the ends. For me this seemed to help stabilize things. Also, though I haven't had this issue, I hear if you use an insulator under the PCB it can cause oscillation as well, don't do that. The PCB should be directly mounted to the heatsink.

When I get a chance to take some pictures I'll show you how I cut the unneeded material away. You are correct on the resistors, 100 ohm for the feedbacks, and 22 ohms across the input transformer.
Hello Rick and thank you ever so much for the detailed reply. I see now, on looking more closely, that as you say the large PCB material piece is cut away so that the minipa board can be inserted in the cut-out. Very neat and a great way of all round ground plane access. Is the 'pcb' from the CBS-4 pack (basically) simply a piece of FR4 single or double sided board ? - or is it more complex. It seems to be a plain pcb ground plane with the 4 cut-outs in it's original form ?
I am guessing it is single sided copper clad FR4 ?
I have suitable heatsinks in stock luckily, and stacks of single and double sided FR4 PCB material. Interesting regarding the oscillation. I considered that may be the case due to similar designs using lower much lower wattage units without ill effects, but with proper construction and layout techniques as you have done. I was also, just this afternoon, reading an article about the 9120 starting to oscillate which was cured by cutting the t0220 traces, again, exactly as you have discovered. Lovely work. The Bias is noted, thanks. Out of interest, the ones with the tricky bias ?, even though you varied the Gate voltage did the quiescent current still end up at a similar level to the others ? - around 1A ?.
Very interested to see the modifications you have done Rick. All info is appreciated. You are the first person I have come across that has a stable design putting out the correct power. It can clearly be seen in the image that a heck of a lot of care and attention to detail has gone into the construction.
Oh, what is the input drive level you are using (oops, noticed you said 5W in an earlier post- doh , So am I correct in thinking that the gain is fairly flat across the bands from 80 to 10 ? - or are you altering the power as you go up the bands?. Also, have you ever had chance to look at the output on a spectrum analyser. I know you said you hadn't originally, but perhaps you have had an opportunity since ?. Also possibly a two tone test with a scope ?. I have a tinySA and suitable high power attenuators so that will be interesting at a later stage. The two tone is fairly easy. There is a great little PC soundcard util called 'sweepgen' available from the author here - you only need the free version - http://www.satsignal.eu/software/reg-audio-tools.html Great for two tone testing.
The case for mine was on the shelf :) .. it was a surplus old ATX PSU. The heatsink will fit very nicely inside with enough room for an LPF unit. The LPF unit you have looks like the ones from AliX / Ebay. ? - I was considering one of those as it is actually cheaper than making one !. Huge thanks for everything sir. Joe G4XRZ - a Brit but residing in Portugal.
 
Last edited:
  • Like
Reactions: Shadetree Mechanic

Rick Miller

Member
Apr 1, 2015
27
40
23
53
Hello Rick and thank you ever so much for the detailed reply. I see now, on looking more closely, that as you say the large PCB material piece is cut away so that the minipa board can be inserted in the cut-out. Very neat and a great way of all round ground plane access. Is the 'pcb' from the CBS-4 pack (basically) simply a piece of FR4 single or double sided board ? - or is it more complex. It seems to be a plane pcb ground plane with the 4 cut-outs in it's original form ?
I am guessing it is single sided copper clad FR4 ?
It's just a plain single sided PCB
I have suitable heatsinks in stock luckily, and stacks of single and double sided FR4 PCB material. Interesting regarding the oscillation. I considered that may be the case due to similar designs using lower much lower wattage units without ill effects, but with proper construction and layout techniques as you have done. I was also, just this afternoon, reading an article about the 9120 starting to oscillate which was cured by cutting the t0220 traces, again, exactly as you have discovered. Lovely work. The Bias is noted, thanks. Out of interest, the ones with the tricky bias ?, even though you varied the Gate voltage did the quiescent current still end up at a similar level to the others ? - around 1A ?.
The ones that wouldn't bias up would only show around 30ma max no matter where you set the pot. I had to bring the voltage up to nearly 5v before they would start pulling current.
Very interested to see the modifications you have done Rick. All info is appreciated. You are the first person I have come across that has a stable design putting out the correct power. It can clearly be seen in the image that a heck of a lot of care and attention to detail has gone into the construction.
Oh, what is the input drive level you are using ?, and also, is the gain fairly flat across the bands from 80 to 10 ?. Also, have you ever had chance to look at the output on a spectrum analyser ?. Also possibly a two tone test with a scope ?.
I was running a yaesu ft-817 into it, on SSB that amounts to about 10W peak. On AM I have to bring the power down to about 1-2.5W though. The gain appears to be quite flat across 80-10 showing nearly identical output. However, I've ran it on 160 and the gain goes through the roof. It does work as long as you bring your input WAY down.

I recently bought a NanaSA, but I haven't yet had the chance to look. I did have it on a scope, I was only looking at the output of my FT8 and it did appear quite clean.
The case for mine was on the shelf :) .. it was a surplus old ATX PSU. The heatsink will fit very nicely inside with enough room for an LPF unit. The LPF unit you have looks like the ones from AliX / Ebay. ? - I was considering one of those as it is actually cheaper than making one !. Huge thanks for everything sir. Joe G4XRZ - a Brit but residing in Portugal.

Awesome! You'll have to share pictures of your build. The LPF was indeed an ebay purchase, and they seem to work great especially for the price. No way I could build one at that price point. On my bigger amps, the LPF costs almost as much as the power amp to build.

Again, I'll add some photos this evening. I am posting from my desk here at work and as such should get back to work, lol

Nice to make your acquaintance sir! -- KK7XU
 

G4XRZ

New Member
Jan 31, 2022
6
3
3
65
It's just a plain single sided PCB

The ones that wouldn't bias up would only show around 30ma max no matter where you set the pot. I had to bring the voltage up to nearly 5v before they would start pulling current.

I was running a yaesu ft-817 into it, on SSB that amounts to about 10W peak. On AM I have to bring the power down to about 1-2.5W though. The gain appears to be quite flat across 80-10 showing nearly identical output. However, I've ran it on 160 and the gain goes through the roof. It does work as long as you bring your input WAY down.

I recently bought a NanaSA, but I haven't yet had the chance to look. I did have it on a scope, I was only looking at the output of my FT8 and it did appear quite clean.


Awesome! You'll have to share pictures of your build. The LPF was indeed an ebay purchase, and they seem to work great especially for the price. No way I could build one at that price point. On my bigger amps, the LPF costs almost as much as the power amp to build.
Hello Rick. Thank you again. Excellent information again. I am so glad I found this forum and your good self. It is also really nice to make your acquaintance sir. What a super hobby we have. I will look closely at those LPF units. I agree on price. I looked at building my own and the price soon becomes prohibitive. I will have to check the parts 'department' (meaning the many boxes and tubs ;)) as I am fairly sure that I may have some Amidon T80-2 and T80-6 units, but even then the caps can cost an arm and a leg. Out of interest, there is a super filter design and response plotter called 'Elsie' if you haven't come across it. Really first class. You can use it for any type of filter you desire. The full VNA plots are given. Really nice.
I look forward to seeing more images.
Yes, I will definitely post some photos when I get cracking on it. I am in the process of building an mcHF transceiver from Kriss (M0NKA) The amp will work nicely with that. A good friend also donated his Xiegu G90 and it would be very nice with that rig also. I have a couple of TS-50 units in boxes but they are already 100W output.
I have been out of the hobby for about 17 years due to other things but am slowly getting back in the saddle. I still have most of my old gear and parts. To see what is available now compared to 17 years ago is simply mind blowing. Hand Held spectrum analysers that are perfectly capable for our hobby. Hand held VNAs. 8Ghz power meter modules for a few euros. Amazing DDS units. A whole new world compared to the late 70's / early 80's.
73 Rick. Joe G4XRZ (or when I HAVE to use it, my reciprocal / temp callsign in Portugal is CT7/G4XRZ. My key hand would go numb with that.
 
Last edited:
  • Like
Reactions: Shadetree Mechanic

kopcicle

Sr. Member
Feb 17, 2016
2,079
3,335
273
BTW: I love how the MRF300AN Datasheet from NXP give a 27Mhz Reference Circuit Layout and parts list. LMAO. It's like they know what's up.
Well, 13.560 is a well known and well used ISM frequency used for industrial inductive heating and sputtering deposition. It should come as no surprise that 27.120 is also a well known ISM frequency used primarily for Radio Frequency Welding of specific fabrics.
 

kopcicle

Sr. Member
Feb 17, 2016
2,079
3,335
273
My next project will use a pair of MRF300AN/BN transistors running 50V. I started assembly this weekend but am waiting on components for the bandpass filter which will take over a month to get here. In the meantime I will get the pallet assembled and tested.

I'm quite interested in this device, anyone here have experience with the MRF300's?

I changed my mind on the 2879C's after reading the specs and seeing a few videos online. I'll just save those transistors for a future repair or some such.
Well, I kinda suck. I got a tremendous start and then life gobbled up time, space, motivation, and several small, unidentified critters that make the small parts on my bench disappear. Normally the critters disappearing wouldn't be a bad thing but I would occasionally find their hoard and return it to the project in question. This time they just vanished. More on this some other time.

What I'm about to show you is something of a hybrid construct. It contains the bias circuitry of NXP's MRFX1K80H design board, The traditional layout of twin device, transformer coupled boards that would have contained MRF455,2SC2290,2SC2879 et all, and the physical requirements of the MRF300AN/BN TO-247 package.
I left plenty of room for the output transformer. I also considered taking the DC off the output transformer and directly choke feeding the devices.
If you scale the artwork to the actual size of the TO-247 package you should have room for everything.

So, artwork. I have the .svg files somewhere and even gcode if you are that kind of masochist. :- )
MRF300_art.jpg


YMMV
 

Rick Miller

Member
Apr 1, 2015
27
40
23
53
Hello Rick. Thank you again. Excellent information again. I am so glad I found this forum and your good self. It is also really nice to make your acquaintance sir. What a super hobby we have. I will look closely at those LPF units. I agree on price. I looked at building my own and the price soon becomes prohibitive. I will have to check the parts 'department' (meaning the many boxes and tubs ;)) as I am fairly sure that I may have some Amidon T80-2 and T80-6 units, but even then the caps can cost an arm and a leg. Out of interest, there is a super filter design and response plotter called 'Elsie' if you haven't come across it. Really first class. You can use it for any type of filter you desire. The full VNA plots are given. Really nice.
I look forward to seeing more images.
Yes, I will definitely post some photos when I get cracking on it. I am in the process of building an mcHF transceiver from Kriss (M0NKA) The amp will work nicely with that. A good friend also donated his Xiegu G90 and it would be very nice with that rig also. I have a couple of TS-50 units in boxes but they are already 100W output.
I have been out of the hobby for about 17 years due to other things but am slowly getting back in the saddle. I still have most of my old gear and parts. To see what is available now compared to 17 years ago is simply mind blowing. Hand Held spectrum analysers that are perfectly capable for our hobby. Hand held VNAs. 8Ghz power meter modules for a few euros. Amazing DDS units. A whole new world compared to the late 70's / early 80's.
73 Rick. Joe G4XRZ (or when I HAVE to use it, my reciprocal / temp callsign in Portugal is CT7/G4XRZ. My key hand would go numb with that.
I must apologize, I will not be able to get any decent pictures tonight. My daughter is coming over tomorrow night and I'll have her bring her camera. My phone just cannot seem to get any detail close up.

I will post them as soon as I can...
 

Shockwave

Sr. Member
Sep 19, 2009
3,785
3,286
273
Very interesting. Where have I been? A LDMOS in a TO-247 package. Now that's something else. Only issue I could see, after having spent 14 years replacing blown high power TO-247 (400-500V) Mosfets in audio amplifiers (for cars, KW+ size amps) is that you would want to make use of copper heatsinks (if you can), copper heat spreaders, whatever you can to help with this parts heat from that type of package (relatively small). It should be fairly efficient, however 320W per device (rated power), I would ensure it can spread that heat out quick if needed.

Looks like Input Capacitance is not too crazy high, should be real easy to drive.
I'm in complete agreement with using a copper spreader, with this package, anytime the dissipation is around 250 watts or more. The package stinks as far as thermal resistance is concerned. The copper spreader makes a big enough improvement, that you can easily see it in the bias. The amount of thermal compensation and drift in the bias current, is far less with copper.

The main reason you see test circuits for 27 MHz. has nothing to do with CB. The band contains ISM frequencies on some of the "A" channels, where RF Generators are allowed to operate. They are used in applications that range from Gas Chromatograph Mass Spectrometers, to RF heating units. They are not intended to radiate an over the air signal, but many are easily modified to make high end linear amplifiers.
 

G4XRZ

New Member
Jan 31, 2022
6
3
3
65
It's just a plain single sided PCB

The ones that wouldn't bias up would only show around 30ma max no matter where you set the pot. I had to bring the voltage up to nearly 5v before they would start pulling current.

I was running a yaesu ft-817 into it, on SSB that amounts to about 10W peak. On AM I have to bring the power down to about 1-2.5W though. The gain appears to be quite flat across 80-10 showing nearly identical output. However, I've ran it on 160 and the gain goes through the roof. It does work as long as you bring your input WAY down.

I recently bought a NanaSA, but I haven't yet had the chance to look. I did have it on a scope, I was only looking at the output of my FT8 and it did appear quite clean.


Awesome! You'll have to share pictures of your build. The LPF was indeed an ebay purchase, and they seem to work great especially for the price. No way I could build one at that price point. On my bigger amps, the LPF costs almost as much as the power amp to build.

Again, I'll add some photos this evening. I am posting from my desk here at work and as such should get back to work, lol

Nice to make your acquaintance sir! -- KK7XU
Hello Rick sir, I hope you don't mind me asking but have you had time to take any more detailed images please ? I would be fascinated to see more detail of what you have accomplished. Best regards and 73 G4XRZ Joe
 

Rick Miller

Member
Apr 1, 2015
27
40
23
53
Sorry for taking so long to respond. I had a hell of a time getting a good picture. I've attached a pdf file that I have marked up so you can see exactly where I cut.
 

Attachments

  • pa100-cut.pdf
    101.3 KB · Views: 55

dxChat
Help Users
  • No one is chatting at the moment.
  • @ Roadstar:
    good morning to all and Happy Thanksgiving to all:giggle:
  • dxBot:
    Greg T has left the room.
  • @ BJ radionut:
    solar storm report
  • @ k3zl:
    I am installing a G5RV and wanted to know if there is a way to enhance prior to installation.