Re: sBitx DE finals and heat sink temperature plots
Allison,
respect your experience and views but i totally disagree with anyone saying that there is no thermally stable fet or power transistor.Whether they are an engineer or not.
just my own opinon.and work experience. I would take a lower gain component over a fried component every day. Hope u all get ur fried finals problems figured out.Truthfully, Ashhar designed a beautiful transceiver.When shipping from overseas, things happen stuff comes loose.
For $499 I wouldnt expect any? problems like fried finals .especially if its the second version with improvements.Someone is doing something wrong. like tinkering where they shouldnt, not setting biasing correctly, or adding stuff that doesnt belong, or not truelynot matching the unit to the antenna correctly.
I have bought many units off Ashhar, bitx40,ubitx3,ubitx6, several times. never had any problesm with any of them except when i tinkered with them to "upgrade the software, and or display and such".
been building my own rf power amps for over 20 years.And if i experienced this with mine, i would be going a different route.
But again, Allison, no offense,. just my own opinon.and not of staff and management.HI HI.
73
David
ac9xh
|
Re: sBitx DE finals and heat sink temperature plots
G Todd,
Simple solution do like I do for the over 100W stuff MRF151, MRFE101, BLF184, BLF188
The only down side is the need for 50V power.? All of them required thermal stabilized bias.
MRF255, 100W push pull and good cooling flange mount 13V part.? Theremal stabilized
bias as well.
Or 8 IRF510s on 6M putting out 220w at 28V.? ?Oh, that can't work they?
are not flange ground.? SiO/SiN insulators do well.
There are no such thing as "?more thermal stable fet", going bipolar?
is really nuts as most bipolar are 10-11DB max gain where even the lowly
IRF510 is 13-16Db gain.
THere there is the QRPL 10W amp two stage amp with 36DB of stable gain
good from 1.8 to 10 and useful at 6M (gain is 4db don but full power).? Not
bad for a nominal 12V amp.
--
Allison
------------------
Please use the forum, offline and private will go to bit bucket.
|
Stuart,
HFSignals does sell the v6 Raduino board separately here:
It is 1/4 of the cost of the uBitx full kit.
73
Evan
AC9TU
|
Re: sBitx DE finals and heat sink temperature plots
conductivity of heat is the most important thing for power amplifiers, but also im my own opion is a design that allows the finals to be mechanically bolted to heatsinks and transceiver casings WITHOUT the insulating material. Meaning the designers design the transceiver or Amplifier with the finals with common? connections without the need for insulating wafers like mica or anything. conductive grease helps alot,but if the finals can actually be physically connected to the heatsinks without ANY insulating material, it will solve alot of problems. I own 3 RF power amplifiers and only 1 of the uses this practice. its an older amplifier of about 25 years.
comparing it to the new ones, it seems to me the design is simpler and the heatsink plus the transceiver casing adds to the control of heat . NO worries if you are fully insulating the semiconductor with separating wafers of mica or whatever, This particular amp has a heatsink that also clamps onto the front of the finals to help dissipate heat plus the conductive grease. Its also a genric Amp.
Maybe someone needs to go a different path to seek out a more thermal stable fet? or maybe a small board allowing POWER TRANSISTORS? I mean if all these fets are burning up, then a new WHEEL needs to be created.have the option to wire into the finals an experimental board that allows different fets or power transistors.might have to use a converter to raise the voltage to allow a different thermally stable semiconductor?Plenty of different parts on the market.like the mrf422,mrf486,or the mrf455 which is bipolar transistor,2-30 mhz, 60 watts, 12 votls. these and other rf power semiconductors might cost more, but you are installing parts proven in the field many times over.
just my take of all your problems.
73 David ac9xh
On Saturday, July 15, 2023 at 09:54:11 AM EDT, Gordon Gibby <docvacuumtubes@...> wrote:
got the aluminum nitride insulators ordered via aliexpress hope they get here sometime!! Gordon
On Sat, Jul 15, 2023 at 9:29?AM Gordon Gibby via <docvacuumtubes= [email protected]> wrote: Thanks, Allison!? ?Great recommendation.? ?I think I'll try to find some of the?aluminum nitride insulators.
Another SCARE this morning with the radio.? Just starting a QSO on 40 meter CW and suddenly dropping dits and loss?of power.? Lots of stuff going wrong all at once.? ?Then basically zero power.? And of course, in my befuddlement I hit the SPLIT somewhere also.....to add insult to injury.
An hour later I finally figured it out -- the Drain solder joint of the RIGHT IRFZ24N had quit.? ?Basically zero DC and zero AC on the drain.? ?Left hand side working perfectly.? This exact same thing happened to the LEFT side months ago!? ?There wasn't much solder on those connections; instead of being convex?surface they were indented down into the hole....? ? Got the soldering iron -- I'm in the vacation home, less equipment ...of course a solder bridge, finally got it all soldered and cleaned up. and POWER AGAIN!!?
I'm guessing the expansion/contraction driven by the hot temperatures of the final tend to work poor soldering joints loose, to have this happen to BOTH of the final mosfets over time?? ?And, I had this apart again working on things 2 weeks ago which put torque on those connections, also...? ?Anyway, something to watch for.? ?my? QTH is in a terrible valley beween?mountains and my end-fed antenna is sad here also, but managed a cw qso with a POTA volunteer to prove it was working right.? ?
If I get some aluminum nitride insulators, I'll wait for a good time and pull this apart and check the surface of the heatsink also and get this done right.? ??
Ashhar has really made some good improvements.? ?The experience of cw qsos (once I got the relay mess cleaned out) is getting to be really really good.? ?Very little click/thump.? ?The semi-full break in is SO SO SO SO MUCH BETTER than my 7300.? ??
Gordon kx4z
On Sat, Jul 15, 2023 at 9:13?AM ajparent1/kb1gmx < kb1gmx@...> wrote: Some notes:
BEO dust is toxic!? Just don't!
I have used?Aluminum Nitride as its not toxic and works well...
One place personally used it the OZ1PIF amp and I found the thermal
conductivity excellent.
I was never a fan of the mica used for the DE, also they were used DRY
without thermal conductive paste.? Add to that a heatsink with terrible
surface quality (rough) and bad stuff happens.
--
Allison
------------------
Please use the forum, offline and private will go to bit bucket.
|
Re: sBitx DE finals and heat sink temperature plots
got the aluminum nitride insulators ordered via aliexpress hope they get here sometime!! Gordon
On Sat, Jul 15, 2023 at 9:29?AM Gordon Gibby via <docvacuumtubes= [email protected]> wrote: Thanks, Allison!? ?Great recommendation.? ?I think I'll try to find some of the?aluminum nitride insulators.
Another SCARE this morning with the radio.? Just starting a QSO on 40 meter CW and suddenly dropping dits and loss?of power.? Lots of stuff going wrong all at once.? ?Then basically zero power.? And of course, in my befuddlement I hit the SPLIT somewhere also.....to add insult to injury.
An hour later I finally figured it out -- the Drain solder joint of the RIGHT IRFZ24N had quit.? ?Basically zero DC and zero AC on the drain.? ?Left hand side working perfectly.? This exact same thing happened to the LEFT side months ago!? ?There wasn't much solder on those connections; instead of being convex?surface they were indented down into the hole....? ? Got the soldering iron -- I'm in the vacation home, less equipment ...of course a solder bridge, finally got it all soldered and cleaned up. and POWER AGAIN!!?
I'm guessing the expansion/contraction driven by the hot temperatures of the final tend to work poor soldering joints loose, to have this happen to BOTH of the final mosfets over time?? ?And, I had this apart again working on things 2 weeks ago which put torque on those connections, also...? ?Anyway, something to watch for.? ?my? QTH is in a terrible valley beween?mountains and my end-fed antenna is sad here also, but managed a cw qso with a POTA volunteer to prove it was working right.? ?
If I get some aluminum nitride insulators, I'll wait for a good time and pull this apart and check the surface of the heatsink also and get this done right.? ??
Ashhar has really made some good improvements.? ?The experience of cw qsos (once I got the relay mess cleaned out) is getting to be really really good.? ?Very little click/thump.? ?The semi-full break in is SO SO SO SO MUCH BETTER than my 7300.? ??
Gordon kx4z
On Sat, Jul 15, 2023 at 9:13?AM ajparent1/kb1gmx < kb1gmx@...> wrote: Some notes:
BEO dust is toxic!? Just don't!
I have used?Aluminum Nitride as its not toxic and works well...
One place personally used it the OZ1PIF amp and I found the thermal
conductivity excellent.
I was never a fan of the mica used for the DE, also they were used DRY
without thermal conductive paste.? Add to that a heatsink with terrible
surface quality (rough) and bad stuff happens.
--
Allison
------------------
Please use the forum, offline and private will go to bit bucket.
|
Re: sBitx DE finals and heat sink temperature plots
Thanks, Allison!? ?Great recommendation.? ?I think I'll try to find some of the?aluminum nitride insulators.
Another SCARE this morning with the radio.? Just starting a QSO on 40 meter CW and suddenly dropping dits and loss?of power.? Lots of stuff going wrong all at once.? ?Then basically zero power.? And of course, in my befuddlement I hit the SPLIT somewhere also.....to add insult to injury.
An hour later I finally figured it out -- the Drain solder joint of the RIGHT IRFZ24N had quit.? ?Basically zero DC and zero AC on the drain.? ?Left hand side working perfectly.? This exact same thing happened to the LEFT side months ago!? ?There wasn't much solder on those connections; instead of being convex?surface they were indented down into the hole....? ? Got the soldering iron -- I'm in the vacation home, less equipment ...of course a solder bridge, finally got it all soldered and cleaned up. and POWER AGAIN!!?
I'm guessing the expansion/contraction driven by the hot temperatures of the final tend to work poor soldering joints loose, to have this happen to BOTH of the final mosfets over time?? ?And, I had this apart again working on things 2 weeks ago which put torque on those connections, also...? ?Anyway, something to watch for.? ?my? QTH is in a terrible valley beween?mountains and my end-fed antenna is sad here also, but managed a cw qso with a POTA volunteer to prove it was working right.? ?
If I get some aluminum nitride insulators, I'll wait for a good time and pull this apart and check the surface of the heatsink also and get this done right.? ??
Ashhar has really made some good improvements.? ?The experience of cw qsos (once I got the relay mess cleaned out) is getting to be really really good.? ?Very little click/thump.? ?The semi-full break in is SO SO SO SO MUCH BETTER than my 7300.? ??
Gordon kx4z
On Sat, Jul 15, 2023 at 9:13?AM ajparent1/kb1gmx < kb1gmx@...> wrote: Some notes:
BEO dust is toxic!? Just don't!
I have used?Aluminum Nitride as its not toxic and works well...
One place personally used it the OZ1PIF amp and I found the thermal
conductivity excellent.
I was never a fan of the mica used for the DE, also they were used DRY
without thermal conductive paste.? Add to that a heatsink with terrible
surface quality (rough) and bad stuff happens.
--
Allison
------------------
Please use the forum, offline and private will go to bit bucket.
|
Re: sBitx DE finals and heat sink temperature plots
Some notes:
BEO dust is toxic!? Just don't!
I have used?Aluminum Nitride as its not toxic and works well...
One place personally used it the OZ1PIF amp and I found the thermal
conductivity excellent.
I was never a fan of the mica used for the DE, also they were used DRY
without thermal conductive paste.? Add to that a heatsink with terrible
surface quality (rough) and bad stuff happens.
--
Allison
------------------
Please use the forum, offline and private will go to bit bucket.
|
Re: SBitx V2 things I would like to see
As I recall in March or April, when I saw the device on video and there was no mechanical size anywhere. Then I saved a picture from the video and measured it with a screen meter. Good old days :-) -- Gyula HA3HZ 
|
Re: sBitx DE finals and heat sink temperature plots
Both of you guys are amazing.??
The curves that John produced appear valid.? ?His temperature rise of the heatsink is about the same in both experiments, indicating validity of his measurements.
The time for the DIE to hit its temperature limit is much, much longer when he used the BeO insulator.
I hunted for such an insulator and only found very very expensive items.
I ran some calculations for Aluminum oxide insulators and came up with a disappointing 0.8 degC/W for an available product of 1.78 mm thickness, and TO-220 physical dimensions (had to break out the AP Physics equations to run that calculation with thermal conductivity).?
The data for Aluminum Nitride that Gyula presents appears much more exciting, with conductivity similar to BeO and hopefully less toxic and cheaper.
The only suppliers I could find for TO-220 type aluminum nitride insulators were chinese (alibaba, and a couple of other firms) -- but this is very very interesting.
Switching to a much more highly conductive insulator would appear to dramatically increase the safety of the design for key-down time @ 30W output or so.
I looked at the paper on the inflection point in temperature properties -- VERY interesting.? ?Looking at our usage, there is a reasonable chance that we are in the "good" side of the inflection point at higher powers on the sBitx, although at lower powers we may be on the unfavorable side of temperature properties, but it won't matter as much there because the temperatures/dissipation are so much lower.
Since the development edition uses a BAR to press the devices, it makes it much easier to try some of these much more interestng insulators.
I think the next step would be to acquire some of them.
Great work, guys!
Gordon KX4Z
|
Re: The content of hw_settings.ini is incorrect
#sbitx_v2
I would like to close this topic. Replacing D27 and D28 solved the problem that the swr-power meter can now measure correctly. This is how output power monitoring works. I received the IRFZ24Ns I ordered yesterday. I didn't want to change anything without them. As you can read, their existence was necessary. I measured their characteristics with a tester (which the meter can indicate: connection of legs, input capacity and opening voltage.) First, I measured with a used 9V battery, according to the display it was 7.1V. Since a measurement is not a measurement, I also supplied it with 9.0V from an external power supply according to the display. According to the measurements, I measured input capacitance between 555pF...562pF and opening voltages of 3.1...3.2V from the IOR P925 J 80 0C production product (10 pcs). I only mention as an interesting fact that I ordered the 2N2219A from two dealers, and I measured their characteristics similarly. There was a difference in HFE characteristics. My measurement was 163 from one dealer, and 216 from the other. What is interesting is that the batch marking does not fit on the part from the same manufacturer, only here. After that I returned to Sbitx. Since the arrival of the device, I have only used it in reception, with the exception of the May 4 testing txcal artificial load transmission mode. It was possible to short-circuit one of the FETs with a previously modified hw_settings.ini during the txcal (PSU 12.5V limited to 8 amps), so the replacement was now possible. I soldered the FETs as shown in the picture. I turned 'VR_bias' all the way towards the ground. Power on: ssb mode, drive 1, mic 0 to set the quiescent current of the final stage, then reboot. My power supply has a current monitor and current limiter, so I limited it to 2 Amps. The quiescent current is different when receiving and transmitting. This quiescent current taken during transmission (i.e. in the absence of an input signal) must be increased in the case of 2 FETs by 100-100 mA, i.e. 200mA (maximum 250 mA). The transmission is started by pressing PTT. There is artificial load on the output. My further measurements show that I am satisfied with my output even without changing the original hw_settings.ini. Measured with an external power meter at 80% drive, 80M and 40M= 35W. 30M and 20M = 20W. 17M and 15M= 15W. 12M and 10M= 10W In short, I received a good product, except for the D28, which became the cause of all the problems. I write this only for the edification of others. I have no hurtful intentions. -- Gyula HA3HZ 
|
Re: sBitx DE finals and heat sink temperature plots
I went a bit around the issue of heat conductors.
I came to an interesting conclusion. I found a study material in Hungarian that deals with this topic.
Briefly about the thermal conductivity of materials:
Aluminum /Al, 99.99%/?????? 296 [W/mK?]
Copper /Cu-E, 99.99%/????? 380 [W/mK?]
Silver /Ag/?????????????????????????? 416 [W/mK?]
Ceramic Al-oxide????????????????? 25 [W/mK?]
Ceramic Be-oxide?????????????? 200 [W/mK?]
Mica?????????????????????????????????? 0.58 [W/mK?]
Teflon /PTFE/???????????????????? 0.23 [W/mK?]
"Heat conductive" paste????????????? 0.1 ¡Â 0.12 [W/mK?]
Fr 4 PCB - lateral 70¦Ìm copper????????????? 35 [W/mK?]
After that, I started looking for available materials that can be placed between the FET and the Alu heatsink.
After all, the title theme also revolves around this. Insulators like mica, silicone, and Kapton have low thermal conductivity.
BerylliumOxide MIL is used and is dangerous if broken or cut.
I searched further and found Aluminum Nitride.
According to Wikipedia: Its thermal conductivity is high, up to 321 W/(m¡¤K), electrically insulating.
And there is a dealer where you can buy with or without holes in 0.6 and 1mm thickness.
According to the manufacturer's dealer, Thermal conductivity: 190-260W/m.K
This is a must try. It is affordable and has very good thermal conductivity.
What would you say to this topic?
--
Gyula HA3HZ
|
Re: Paper between finals and heat sink?
Speaking of thermal conductivity please see my message #103820. You might find it eye opening.
Jack, N6LN
|
Re: sBitx DE finals and heat sink temperature plots
There is another failure mechanism to MOSFETs that is not being mentioned here, that of too low a gate to source voltage, this does result in thermal runaway.
Whether that is part of the problem of them dying prematurely in these rigs, maybe someone might care to investigate the possibility (I don't currently have any workbench facilities).
The failure mechanism is explained in the attached pdf from Onsemi.
--
|
Re: sBitx DE finals and heat sink temperature plots
BTW since the question was brought up the idle receive current for the rig including the screen is about 0.8 amps. My linear supply is at 14.0V at idle and drops to 13.6V on 30W CW transmit on 80M. The first test on transmit was at 8.2A, while the second was at 7.46A. After the first test I looked at the idle current on USB PTT and it had increased slightly, so I reset the PA pot to have a 200 mA increase on PTT (mic = 0 and drive = 1) from 1.07A to 1.27A.
Jack, N6LN
|
Re: sBitx DE finals and heat sink temperature plots
This beryllium graph is shocking. I decided to replace the mica insulators with beryllium oxide insulators.The first graph I showed two days ago comparing the sBitx DE continuous CW 30 W key down for 5 minutes, one thermometer chip on the metal tab of an IRFZ24N PA MOSFET and the other thermometer attached to the heat sink a few millimeters away. That configuration was with the old mica insulators. I was pleased to see it run hot without a PA failure. There was an understandable differential in temperature between the MOSFET and the heat sink. After swapping out the mica insulators for beryllium oxide ceramic insulators the difference is astonishing. The thermal conductivity of BeO is about 200 times that of mica. The BeO discs are thicker by a factor of maybe 10 but look at the effect! The heat is conducted so much more efficiently that the MOSFET never gets much above 55 C, whereas with mica it got above 90 C. I'm not sure it's worth changing to beryllium oxide but wow, what an effect. If you do get beryllium ceramic washers remember not to try grinding them. The dust is the worst type of poison because your body can't get rid of it. I snapped one washer in two to get it to fit both MOSFETs.   Jack, N6LN
|
Re: SBitx V2 things I would like to see
Frank, are you looking at the same website that I¡¯m looking at?hfsignals.com?
Right on the front page it tells you the size of the box, which gives you information about the size of the printed circuit board, of which there is a very nice photograph on an interior page, with plenty of standard components that you could easily use to measure anything you wanted.?
Their printed circuit board is there own intellectual property and they aren¡¯t required to give you information about each and every trace
It seems to me they¡¯re providing great deal of information. Furthermore we have an entire website here full of information & a group blog with more.
And we¡¯re pretty brutally honest about the good and bad points.
Gordon Kx4z
toggle quoted message
Show quoted text
On Jul 14, 2023, at 16:32, Frank PH2M <ph2m@...> wrote:
?Nothing about the size of the PCB, sadly neither any close-up pictures of the PCB
73 de Frank PH2M
|
Evan and Mark:? I now have my ubitx v6 back with the original firmware and screen.? The old screen works well but the radio does not receive or transmit.? I tried the BFO calibration again with no success.? I am going to put it on the shelf for now.? I have ordered some Arduino Nano's to try but I believe you are correct, it is probably the ST5351 chip.? I will see if I can buy a new Raduino board.? I noticed that there have been several versions of this board.? I was gong to start with the HF Signals store.? Is there somewhere else I can purchase the correct Raduino board for my ubitx v6??
I thank both of you for trying to help me.? Stuart KN4RGB
|
Re: SBitx V2 things I would like to see
Nothing about the size of the PCB, sadly neither any close-up pictures of the PCB
73 de Frank PH2M
|
Re: sBitx PCB only - laser cut case
I forgot the solution I used for one SSB TRX (analog) was to make?
an opening in the rear for a finned heat sink.? IT had to fitted so the
fins were outside the box for convection cooling.? Since it was on the
rear of the box it was not a visual impact.
The SBITX has heat issues for final that require a substantial heat sink
that should not be compromised.
--
Allison
------------------
Please use the forum, offline and private will go to bit bucket.
|
Re: sBitx PCB only - laser cut case
Line the inner surface with copper foil or conductive paint.
I've done that with wood cases many times to achieve aesthetic results
and still have the needed shielding.? The bare unfinished wood boxes with
hinged tops from various hobby/craft suppliers are a favorite for cost
and ability to achieve a nice finished result.? Self adhesive copper tape
made doing the lining easy.
--
Allison
------------------
Please use the forum, offline and private will go to bit bucket.
|
HA3HZ
John Terrell, N6LN
Frank PH2M