开云体育

So I've been given v15 beta some hammer just keying out news stories for 30 mins or so in iambic A into a dummy load (so it is outputting RF) and it is MUCH better in terms of incorrect dits and dahs.?

They still happen and it seems to be more often when transitioning from one to the other. Yes it happens sometimes in a stream but more often when going from a dit to a dah or vice versa.?

Latency seems OK but then every so often I was getting temporary lock up when I'd key and the radio goes silent or emits a continuous low level tone for some seconds.?

It then made a pop and starting showing G on the screen which I've not seen before. This required a restart.?

Definitely much better than 14 though.?

I know Hans you said you have nit looked at the cluncking yet however that is making the radio almost unusable for actual getting on air as it makes it really difficult to determine how many dits have been sent - particularly S and H, V and 4 etc as the clunks make hearing the side tone accurately really hard.?

Screenshot of the H

Peter,

Your 56V Zeners may not be protecting your PA transistors from the L14 spike as well as you hope.

We tend to think of Zeners as being a hard clamp at their spec voltage, often neglecting the fact that they also have a series resistance.

In the case of the 1N4758A, a 56V 1W Zener, that resistance is about 110 ohms at the spec Zener current of just 4.5 mA.

When both banks of PA transistors turn off, there is something like 750 mA +/-? flowing through L14. Let's call it 1A for simplicity.That current will be immediately diverted to the Zeners. As the Zeners are not perfectly matched, the current will not split equally between? them. Again for simplicity, let's assume an equal split.

A half amp through 110 ohms produces a voltage of 55V and that adds on top of the Vz of 56V, 111V in all.

At that point the 100V Vds-max spec of the TN0110 is being violated, but just barely. Things may be better or worse than this, depending on various factors, but in any case this is cause for concern!? Also, the Zener instantaneous power dissipation at that moment is (56V + 55V) x 0.5A? = 55.5W!

The bottom line is that once again you may have damaged the PA transistors and are stressing the driver chip (and the Zeners, too!).

?A diode across L14 may be the better solution to the HV spike.

73, JZ KJ4A?

On Sat, Feb 3, 2024, 1:42?AM Peter Li <chinasaurli@...> wrote:

Hi all, a variety of updates here. Bear in mind that my QDX may not be representative; still interested to hear what kind of heating others observe.

I finally got some photos with my old cheap Flir One thermal camera:?
The photos have descriptive captions, but here are my observations from the first 5 which are a time series:

1. After running a few minutes in idle or receive (~155 mA current draw @ 12V) there are some hot spots. Most notable is the board area between the BNC antenna connector and the power barrel jack; that area has thermal vias for IC2 (5V regulator) on the other side of the board. The 24-bit ADC also get a bit warm.
2. After running 60s WSJT-X "tune" into a dummy load at full power (12V) at 14.074 MHz. (I removed all heatsinks for this test, and if you recall my finals are TN0110 standing up with short leads.) Obviously the finals heat up.
   You can also see that sadly my heating imbalance has returned; the FETs closer to the center of the board (Q9/11) are hotter than those at the edge (Q8/10). I confirmed that the gate voltages are also imbalanced again, although not as badly as I saw before (but see below...).
   Finally, you can see that after 60s the board area between the FETs has already heated up to roughly match the temperature of the outer (cooler) finals. This agrees with my earlier observations, and I think what Dean saw too. It seems interesting given that this area is supposed to be the BS170 heatsink.
3. After resting QDX for 60s, finals are cooler.
4. After running 30s (half as long) at 28.074 MHz the finals are quite hot. The imbalance is also worse, but it is now the other two FETs (Q8/10) that are hotter. I confirmed that the gate voltage imbalance is also worse and opposite the result above for 20m. Again (after only 30s this time) the board area between the FETs is about as hot as the cooler finals.
5. After resting QDX for 60s, finals are cooler. The rest of the thermal photos are hopefully self-explanatory.

I then put the heatsinks back on, with thermal paste. I also put the black heatsink back onto the area between the FETs as it seems that empirically this area could use cooling. I took some thermal pictures of test runs in this configuration, but they're hard to interpret as the heatsinks interfere with the readings. My estimate is that with the heatsinks the finals stay about 5-10 degrees C cooler.

I had another good session on 10m with this configuration. I couldn't resist testing some active cooling; see photos with fans installed. The first is a 30x30x7mm raspi 5V fan standing up to blow directly onto the FETs. This did a good job of keeping things much cooler, is pretty quiet, and I didn't see any signs of electrical interference. The fan runs at about 200 mA @ 5V as advertised. The last photo is with a 30x30x4mm side-blower fan; this fits nicely into the space and also blows directly onto the finals, similar power consumed, but it's a bit louder.

Hi all, a variety of updates here. Bear in mind that my QDX may not be representative; still interested to hear what kind of heating others observe.

I finally got some photos with my old cheap Flir One thermal camera:?
The photos have descriptive captions, but here are my observations from the first 5 which are a time series:

1. After running a few minutes in idle or receive (~155 mA current draw @ 12V) there are some hot spots. Most notable is the board area between the BNC antenna connector and the power barrel jack; that area has thermal vias for IC2 (5V regulator) on the other side of the board. The 24-bit ADC also get a bit warm.
2. After running 60s WSJT-X "tune" into a dummy load at full power (12V) at 14.074 MHz. (I removed all heatsinks for this test, and if you recall my finals are TN0110 standing up with short leads.) Obviously the finals heat up.
   You can also see that sadly my heating imbalance has returned; the FETs closer to the center of the board (Q9/11) are hotter than those at the edge (Q8/10). I confirmed that the gate voltages are also imbalanced again, although not as badly as I saw before (but see below...).
   Finally, you can see that after 60s the board area between the FETs has already heated up to roughly match the temperature of the outer (cooler) finals. This agrees with my earlier observations, and I think what Dean saw too. It seems interesting given that this area is supposed to be the BS170 heatsink.
3. After resting QDX for 60s, finals are cooler.
4. After running 30s (half as long) at 28.074 MHz the finals are quite hot. The imbalance is also worse, but it is now the other two FETs (Q8/10) that are hotter. I confirmed that the gate voltage imbalance is also worse and opposite the result above for 20m. Again (after only 30s this time) the board area between the FETs is about as hot as the cooler finals.
5. After resting QDX for 60s, finals are cooler. The rest of the thermal photos are hopefully self-explanatory.

I then put the heatsinks back on, with thermal paste. I also put the black heatsink back onto the area between the FETs as it seems that empirically this area could use cooling. I took some thermal pictures of test runs in this configuration, but they're hard to interpret as the heatsinks interfere with the readings. My estimate is that with the heatsinks the finals stay about 5-10 degrees C cooler.

I had another good session on 10m with this configuration. I couldn't resist testing some active cooling; see photos with fans installed. The first is a 30x30x7mm raspi 5V fan standing up to blow directly onto the FETs. This did a good job of keeping things much cooler, is pretty quiet, and I didn't see any signs of electrical interference. The fan runs at about 200 mA @ 5V as advertised. The last photo is with a 30x30x4mm side-blower fan; this fits nicely into the space and also blows directly onto the finals, similar power consumed, but it's a bit louder.

There is very little chatter about the QCX-Mini more about QMX issues.? I wanted to take a moment and acknowledge the outstanding performance I received this past weekend for Winter Field Day.? ?Due to poor weather conditions in NNJ that weekend I opted to work from home with the Mini's for 40 & 20 Meters.? I was able to get 180 QSO's over a 15 hour operation period.? For the most part all CO's I heard I was able to establish a QSO.? So BIG PROPS to the QCX-Mini product line.? Still rock solid.?