For our one IRF510, a good size for our heatsink is half of the 24 sq inches that Allison uses for two devices, and with 1 inch fins.

Lots of good heat sink material in old PC motherboards and power supplies. ?I found a motherboard chipset heatsink that looks ideal, will trim it to size using my $10 Harbor Freight angle grinder with a cut-off wheel. ?For use at 12v most people seem to be getting by fine with the stock heatsink, but it is marginal.

I mentioned heat a couple times in my last post. ?That is not your primary problem, the IRF510 is not hot when it first goes into oscillation.

Allison's post is worth going over thoroughly if you want to know about using IRF510's at RF. ??/g/BITX20/message/22597? She knows what she is talking about. ?She says that if the heat sink is at 50C (122F) ?then the die is much hotter, and close to failure. ?These parts have a very small active region on the die and poor heat transfer characteristics. ?So a heat sink that is more than comfortably warm is too hot. ?She says her 2 IRF510 amp has a heatsink that's 4x6 inches with fins that are an inch tall. ?So figure 4x3=12 sq inches and 1 inch fins. ?Ours is about 1 sq inch and 1/4" fins. ?I won't be using the stock heat sink with power supplies beyond 12v. ?Don Cantrell's blog post about heat sinking directly to a chassis wall might be a good way to go, but there must be insulation between the two as the IRF510 tab is hot with 12v and all the 7mhz RF destined for the antenna.

Allison's recommendation for large heat sinks conflicts with John Backo's warning to watch out for capacitive coupling from the drain pin and ground. ?(Note that there is a drain pin, plus the tab that connects to the heat sink is also connected to the drain.) ?Allison doesn't mention that, but does mention short traces. ?The Bitx40 board is not optimal in that regard but should be fine, and does have the very short connection from IRF510 source pin to ground that she singles out.

51.2 ohms is great. ?Anything between 25 and 100 ohms is ok when running the final from 12v. ?Not an issue if you are only attempting to receive, I doubt any VHF oscillation would get through the 40m lowpass filter between the IRF510 and the dummy load.

And it does indeed sound like John Backo's oscillation. ?Gate should be close to 0 volts whenever in receive mode. ?Suggested suggested the cause was capacitive coupling from the IRF510 drain (the tab that is attached to the heatsink) to ground that is causing this. ?If you can you might try keeping any other objects (such as a metal chassis) away from that heatsink till this problem is solved.

> ?Pin 1 of relay k1 goes to 0.9v at pseudotransmit.

Yes, should be low unless transmitting with PTT pressed. ?What voltage is at K1-1 when just receiving without this pseudotransmit problem?

> Pseudotransmit happens more frequently once it's been triggered once

Likely a heat issue, which can significantly change the IRF510 characteristics

> I did grab a couple more irf510's, I've heard this usage is a bit hard on them.

No. ?Lift the gate pin (or cut that trace) and put a 10 ohm resistor in series with the gate.

Could be that you just have an ?overly-sensitive/out-of-spec ?IRF510, and that swapping it out will fix this. ?But best to look hard for other causes first.

Jerry, KE7ER

Seems this is already fixed.

Patrick Strasser-Mikhail 2017-02-27 19:48:

The wiki seems to have a bug:

/g/BITX20/wiki will produce wrong links.

Working:
/g/BITX20/wiki/ (note the slash at the end!)
/g/BITX20/wiki/Home

I've reported that in the beta group via
/g/beta/message/13072

73 de Patrick OE6PSE

Cor, thanks for your reply. I found it on Amazon.de for €36.33 as paperback and for €25.32 as kindle edition.

I think I will order it, but I am not sure, which version I should take. Kindle is cheaper and has the advantage that I have it on the screen of my PC or my tablet with the posibility of text search. Paperback on the other hand does not need a wire to the power socket or a battery ;-)

Will sleep over it.

Thank you both for the feedback. Yes, when I said into transmission, that was probably inaccurate. There is a similar sound on the speaker as when I switch to transmit, but when I look at the output, I don't see the RF transmission I expect. I am measuring power use for the whole thing at once. This is all when hooked up to the dummy load (it's 51.2 ohm, not 50, but this is close enough? I built it in a metal coffee can). I have not tried transmitting, or touching the heatsink 'transmitting' [pseudotransmit henceforth] on the magloop (so I'm not sure if it's different. I found this high current mode when trying to get an accurate frequency measurement..).

The gate (pin1) seems to be at 7--10v when it goes to pseudotransmit, from 0v while just receiving. I'm not super sure on that one though, as getting leads in there makes it go even without my hands nearby. For clarification, from before, when I touched it, I was not also touching any other part of the circuit, or plugged in devices. Nor was I using the PTT button. Touching was also via two layers of electrical tape. I believe the 3A draw, as the heatsink got hot quickly. Pin 1 of relay k1 goes to 0.9v at pseudotransmit. Pseudotransmit happens more frequently once it's been triggered once, even when the device has been shut off (may be a heat thing, or some stray capacitance memory?).

I did grab a couple more irf510's, I've heard this usage is a bit hard on them.

As I follow it, the suppressor resistor you are suggesting is from gate (pin 1 of IRF510) to ground? Is this (http://www.geocities.ws/radio107mhz/ for the grenade RF section. Some of the pictures on this site do not load.) the kind of fix (1k resistor) the one you were referring to? Would you recommend the 10k drain-gate resistor too? In message??#22548?you say 10 ohms, this seems very low to me, is this correct? From that webpage, the 'miniboots 5W PA' also has a setup from gate to ground, but it is a parallel of 2k7 with 10r in series with a 1n4740. There is also the section 'Protect your MOSFET final stage' which details a zener + 10 ohm. Is this attacking the same problem? I get the impression it is more to limit regular usage than accidental triggering.

I'm going to go ahead and give the 1k gate-gnd a go while I wait for any replies :)

Thanks,

Andrew (VA2QCH)

Jerry I am sorry you are having the same low power problem. It was making me and my test equipment look really bad.

There are three things I did to make it work. #1 Set the bias with no audio to 100mA then adjust R136 to get 1A long aahhh. You should see your watt meter go up at this point. #2 Replace the IRF510. #3 swap out the D20 dual diode for two 1N4148 through hole diodes. I am still waiting for the proper SMD replacements to be put in the mail. Finally, Farhan said this--- "i am not certain the bat54s are the culprit. in two low output cases that i came across on the bench, the 10 ohms in the emitter lead of the 2N2219 had cooked up.

The wiki seems to have a bug:

/g/BITX20/wiki will produce wrong links.

Working:
/g/BITX20/wiki/ (note the slash at the end!)
/g/BITX20/wiki/Home

I've reported that in the beta group via
/g/beta/message/13072

The "Building Guides" link on the WIKI directs back to the home page
rather than the "Building Guides" page.

The "Building Guides" link on the WIKI directs back to the home page
rather than the "Building Guides" page.

Mike N.

K5ESS

_._,_._,_
------------------------------------------------------------------------

I'm intending to do the same. ?The 40m bandpass filter at L1,L2,L3 will block 10mhz. ?So was going to just put a wire across from the top of C1 to the top of C7. ?With no bandpass filter and assuming stock low-side vfo injection at 12-10=2mhz, we would get an image from any 12+2=14mhz station, but should work well enough.

Jerry, KE7ER

Hello Jack,

may I ask you for a code validation?

I wanted to allow me to increase the tuning speed by pressing the button connected to A3.

So I made a modification to the doTuning function.

Seems to work, but I am not sure, if this has some flaws the way I did it. Could you please take a look at it and tell me what you think?

As you know, I am not yet familiar with coding Arduino sketches.

I attached my test sketch. Hope attaching workes.

BTW, there are warnings for every printLine1 and printLine2

E:\bitx40\raduino_new_library\raduino_new_library.ino:666:34: warning: deprecated conversion from string constant to 'char*' [-Wwrite-strings]

? ? ?printLine1("Raduino v1.01");

I fugured out, that there must be some reason why it does not want a string as paramater for the function. So I changed all this calls for example from

printLine1("Raduino v1.01");

to

? strcpy(c, "Raduino v1.01");

? printLine1(c);

I saw in the code that the frequency upper and lower limit is preset in software. I am sure it could be set for your band permissions if someone wanted to. But how far out can it receive with it's hardware limitations? I would like to receive 10 Mhz for WWV clock for frequency calibration.

Tried the calibrate function using the firmware in a unit as shipped end of January. ?When I move a station up the dial by 1khz, the offset shown is -0235000. ?But at least when I am done calibrating it really is moved up the dial. ?Not sure if it successfully read back from EEPROM after a power down and retained the new calibration, didn't try that out.

I then updated the Raduino with the firmware from??using the adjustments for calls to the new Si5351 library. ?Totally different behavior with the github code. ?Offset for calibrate() is now conveniently displayed in Hz. ?But when I release the calibrate switch, the station is still where it was before. ?That's consistent with my reading of the code from github: ? The variable "cal" is local to function calibrate(), and no attempt is made to use it by doTuning() or setFrequency(). ?No call to si5351_set_correction(). ?

The github code executes this line in calibrate() just before writing to the EEPROM:

??cal = (cal * 10000000l) / frequency;

That looks nonsensical to me, though it may help explain the weird offset values that get displayed during calibrate() in the code as shipped.

I've wasted a couple days on this, and I suspect a dozen others have done so as well. ?Would be good if the code most people are using (the code in units getting shipped) could be posted to github. ?Meanwhile, I guess I write my own calibrate() and give up trying to understand what this stuff was supposed to do. ?The calibration procedure should include a step where we zero-beat the vfo to the bfo so we can measure where the analog bfo is. ?(That step won't be necessary if the Si5351 also drives the bfo.) ? Then after lining up with a station of known frequency as the current code does, calculate the vfo error in ppb for a call to Si5351.set_correction(). ?That makes the correction proportional to the Si5351 vfo frequency, not just an offset, and thus remains correct across multiple bands.

I now have the Bitx40 ordered on Jan 30 mostly working. ?But seeing just a few mW of 7mhz out to dummy load when whistling into the microphone, and R136 does not help. ?May need to ask John Smith for advice.

Jerry, KE7ER

On Sat, Feb 25, 2017 at 01:39 pm, Jerry Gaffke wrote ?(in thread?Re: CW or not to CW?)

The "Building Guides" link on the WIKI directs back to the home page rather than the "Building Guides" page.?

Mike N.

K5ESS

¿ªÔÆÌåÓý