|
Hi,
?
I notice a downward drift in measured power output and wanted to know if it is typical.? I am using a QRPguys 12W dummy load with diode rectifier and capacitor envelope follower, plus a DVM.
?
For example, on 30m I measure 22.4V at the output of the envelope follower, which translates to about 5.2W.? If I key down continuously, the measured voltage drifts down to about 21.5V within 30 seconds, for an output of about 4.8W.??
?
I don't observe as much drift on other bands, but those bands already start at a lower power output to begin with, and less heat generated presumably.? For example, on 20M, I measure 17.4V for about 3.17W, and after 30 seconds, it has only drifted down 0.1V to about 17.3V.??
?
I suspected changes to the diode and capacitor as the dummy load heats up, so I repeated the measurement with a different QRP rig, a SW-3B.? I adjusted the input voltage to measure the same initial 22.4V.? There is very little drift in 30 seconds, and only about 0.1V after about a minute.??
?
I'm not too concerned about the tiny loss of power (if there really is any).? I'm more concerned that I may not have put the BS170's flush over the PCB heat sink and sufficiently tightened the washer over the top.? If what I am observing is evidence that I'm not getting the heat sinking I should, it might crop up as a problem down the road when doing WSPR and FT8.?
?
73
Mike N9OHW?
|
Some additional information:
?
On the QMX on 30m where I measure the drift down in power from 5.2W to 4.8W, the initial current draw is 1.16 Amps.? On 20m, where I don't observe this level of drift, and the power is about 3W, the initial current draw is 0.74 Amps.??
?
On the SW-3B rig measured as a reference, when the output is 5.2W (22.4V on the DVM), the current draw is 0.76 Amps.? It is a more efficient PA.??
?
So the drift seems to follow high current draw of > 1 Amp.? Just looking to see if this is a typical behavior.
?
73
Mike N9OHW
|
A slight decrease in output power in extended operation is normal for any rig that uses MOSFET finals. Their resistance increases with temperature, reducing the gain of the final stage. That makes them inherently resistant to thermal runaway. If stable output power is important, the circuit needs to include output measurement and adjustment to gain; none of the QRP Labs rigs bother with that, nor do most other amateur radio designs that use MOSFETs.
That makes them different from bipolar transistors, where the effective resistance drops with rising temperature; those have to be stabilized to avoid damage to the transistors. I can see that effect in my Elecraft K2, though it does use thermal stabilization in its design so power output only rises by a couple of percent, not nearly enough to put any components in danger.
Some additional information:
?
On the QMX on 30m where I measure the drift down in power from 5.2W to 4.8W, the initial current draw is 1.16 Amps.? On 20m, where I don't observe this level of drift, and the power is about 3W, the initial current draw is 0.74 Amps.??
?
On the SW-3B rig measured as a reference, when the output is 5.2W (22.4V on the DVM), the current draw is 0.76 Amps.? It is a more efficient PA.??
?
So the drift seems to follow high current draw of > 1 Amp.? Just looking to see if this is a typical behavior.
?
73
Mike N9OHW
|
This goes to a question I've been meaning to ask, for the BS170s in the PA, is there any reason (providing there is room) why one could not use transistor clip on heat sinks? I tried something similar with a Pixie just to see how it would work. I used a strip of aluminum with heat shrink holding it to the transistor body. Did the job very well.
--
Wayne S.
KF7RCM
|
Thanks for the feedback.? I understand a MOSFET will increase in resistance as it heats up and many QRP rigs don't compensate for that.? My SW-3B also has a MOSFET final, but its power output drift on keydown is much less, given the same power output, but about 2/3 the current.? It is using an IFR510.? I haven't opened it up to see how (or whether) it sinks heat.??
?
What I wondering is if the rate of drift down I see is typical for ~5 Watts at ~1.1 Amps.? This is a 9V build.? I'm not confident I did a great job of flattening the BS170's flat side onto the PCB to get maximum surface area for heat transfer.??
?
In a video, Hans mentioned that the washer and screw aren't for heat sink, but just to press the transistors down onto the PCB to transfer heat there.? I guess this makes sense, the area of contact on the round parts of the transistors with the washer is small.??
?
Thanks and 73
Mike N9OHW
|
I wonder if adding thermal paste to the board at the transistors would be advisable to potentially aid in heat dissipation. Has anyone tried this with noticeable success?
|
@ Robert KD9YUY I wondered the same. I followed the instructions to a "T", but it bothered me that this step seemed to be missing and just figured that there was a reason. I also wonder since the edge of the PCB seems to have the ground plane exposed, if adding thermal paste would assist in conducting some of the heat transfer to the radio case for heat dissipation.?
?
Of course, I always regret the necessity to handle circuits with this white goo on it.
?
|
On 14/08/2024 11:17, Robert KD9YUY via groups.io wrote: I wonder if adding thermal paste to the board at the transistors would be advisable Robert, In the past there has been much discussion about PA heatsinking. I run low power, just read the posts, I feel no true conclusion was made. Some said, for example, paste will make very little difference. Others have gone to extremes with heatsinking and ventilation. In groups like this we have widely differing opinions really each one has to decide for themselves. I do not want to worry you but your comments about power dropping are the first I recall. I'd think someone else would have noted when they have reported power output to one or even two decimal points. 73 Alan G4ZFQ
|
On Wed, Aug 14, 2024 at 07:17 AM, Robert KD9YUY wrote:
I wonder if adding thermal paste to the board at the transistors would be advisable to potentially aid in heat dissipation
I built the QDX in the past and I blew out the finals a couple of times, so I paid attention to thermal management when I built my QMX+. I put some good quality thermal paste (Grizzly, IIRC) between the PCB and the MOSFETS.? I snugged the nut down, but not too tightly.? After I was done, I took a thermal video of a cycle of CW about 30 seconds long while connected to a dummy load.? What I noticed was that the MOSFETS heated up (to a little above 50 degrees C).? The reflectivity of the metallicized part of the PCB is not really a good place to measure temperature with infrared, but to me it did not appear that the heat was transferred into the board material.? I was going to upload this movie, but it was too large.? I will try to redo the video to keep it short and upload a better version later.
?
My take on it is that all I did was improve the conductivity to the copper square on the board.? That would provide some thermal capacitance, but probably does not improve the situation in steady state.
?
I do have plans sometime when I have time, to take a piece of 16 mil copper (0.016 inches thick) and fabricate a U-shaped heat sink that would fit under the MOSFETS and rise up along the two sides of the square without the MOSFET leads. This will have to be precision work to stay away from the surface mount components nearby and the MOSFET leads.? If I do this, I will use some thermal epoxy to attach a thermistor and measure the temperature of the new heatsink over time.? I would also measure the temperature of the MOSFET cases.??
?
As an aside, I have not noticed any issues with the temperature of the finals on the QMX+ yet.??
Note that the QMX+ has SWR sensing, and this allows you to help yourself out when you are using it by watching the SWR.? The QDX did not have this, and I expect that I am doing a much better job of managing my SWR with the QMX+.
|
On my QMX, the longest transmissions I've had were with JS8, but I haven't noticed any drop in output over longer transmissions into an antenna. I tend to set my power supply for less than the rated voltage for all of my QRP Labs rigs. I built my QMX for 12 volts, but I usually run it at 11.5 volts and frequently at just 8.7 volts. That keeps everything cool and steady.
--
73, Dan - W2DLC
|
Thanks for the replies.? I'm not totally convinced that this is only due to the BS170's heating up.? It seems to follow the higher current draw of the 9V build, and the band that happens to draw the most current on this rig.? There are other components that will have a larger I*R drop and heat generation when at this higher current at 9V.? ?
?
Out of curiosity, I performed the same measurements on various QRP rigs I have, and I also notice a slow measured power droop upon long keydown.? It is less pronounced on other QRP rigs, which all draw about 0.6 to 0.7 Amps, much less than the > 1 Amp of the 9V QMX.? When this occurs on the QMX, it is mostly noticeable on 30m, which happens to have the highest power and current of the bands, 5.2W at 1.16 Amps.? When I back off on the supply voltage to make it comparable to other bands at about 4W and 0.9 Amps, the power droop is less.??
?
At the most, it is about a 0.3 dB droop in measured power after a minute of keydown, which is negligeable at the RX side.? On CW, the duty cycle makes it a non-issue.? But I do occasionally use WSPR and FT8 now that it is so easy to use with iFTx and the iPhone, and I'd like to be sure I've done what I can to properly heat sink.??
?
I now back off on supply voltage to about 8V when using WSPR and FT8 to reduce the current draw and heat generation.
?
Thanks and 73
Mike N9OHW
?
|
Hi Mike,
It may not just be the BS170¡¯s heating up. It could also be the gate driver heating up as well, not to mention the entire rig heating up, especially if it¡¯s in an enclosure with no airflow. Your slight loss of power correlates well with higher power out on certain bands. I¡¯d take Shirley at her word regarding this.
Tony AC9QY
Thanks for the replies.? I'm not totally convinced that this is only due to the BS170's heating up.? It seems to follow the higher current draw of the 9V build, and the band that happens to draw the most current on this rig.? There are other components that will have a larger I*R drop and heat generation when at this higher current at 9V.? ?
?
Out of curiosity, I performed the same measurements on various QRP rigs I have, and I also notice a slow measured power droop upon long keydown.? It is less pronounced on other QRP rigs, which all draw about 0.6 to 0.7 Amps, much less than the > 1 Amp of the 9V QMX.? When this occurs on the QMX, it is mostly noticeable on 30m, which happens to have the highest power and current of the bands, 5.2W at 1.16 Amps.? When I back off on the supply voltage to make it comparable to other bands at about 4W and 0.9 Amps, the power droop is less.??
?
At the most, it is about a 0.3 dB droop in measured power after a minute of keydown, which is negligeable at the RX side.? On CW, the duty cycle makes it a non-issue.? But I do occasionally use WSPR and FT8 now that it is so easy to use with iFTx and the iPhone, and I'd like to be sure I've done what I can to properly heat sink.??
?
I now back off on supply voltage to about 8V when using WSPR and FT8 to reduce the current draw and heat generation.
?
Thanks and 73
Mike N9OHW
?
|
Shirley Dulcey KE1L
Daniel Conklin