Topics

SI chip overheating - lack of heat-sink #problem


A Schwarz
 

Hi Everyone;


First I wan to say that the QCX unit is great value and has potential....

There are a couple of things that I would like to see changed:

For the PA use one IRF-730 transistor or similar device and change the Q6, which turns of the finals to a medium current device in a TO-220 enclosure. Three devices in parallel is trouble, because you can not guarantee that they will be loaded equally without a loading resistor as used in analog power supplies. In this application a loading resistor will limit the power out and make the PA inefficient for battery use.
.
The second point is the lack of a heat-sink of the SI chip that can quit warm and should be sinked to the ground plane or externally by a clip that dissipates heat.
My unit failed because of the heat and as I was trying to repair it by removing the chip the traces came loose because of heat damage.(see pic below).
I would like to buy a new unit but I am afraid that I spend a lot of time assembling it, and it will fail again because of these shortcomings!



All the best and 73;

Alex - VE7DXW


Alan G4ZFQ
 

There are a couple of things that I would like to see changed:
For the PA use one IRF-730 transistor or similar device
Alex,

You are guaranteed to get replies but they are likely to be negative.
The PA is designed that way, a single device will not work due to high gate capacitance requiring more drive, more complexity, expense.

Q6, which turns of the finals to a medium current device in a TO-220
I think is adequate, IF a BS170 fails that's what usually blows it.

The second point is the lack of a heat-sink of the SI chip.
Failure of this is very rare, heat has never been said to be the cause. They are never heat-sinked in any radio.

And any change seems unlikely, it is an established design, many thousands sold.

73 Alan G4ZFQ


A Schwarz
 

Hi Allen;

And many experience the same issue, but nobody complains because it is not a lot of money... and they think it is their fault! Your units end up in the garbage because the are not reliable, which is sad. With a little more effort you could change that and really provide an alternative to overpriced radios!

Thank you very much for your answers!

Alex - VE7DXW


Alan G4ZFQ
 

And many experience the same issue, but nobody complains because it is not a lot of money... and they think it is their fault!
Alex,

While the PA fault is not uncommon it is only reported by a relatively small number of people. Thousands, not sure how many, have been sold.
Hans has several times explained his position on this and obviously feels it makes little difference to his sales which are still strong.
In any case it gives at least some of the constructors experience in trouble shooting and component replacement:-)

As far as I recall only a very few Si5351s have failed. One of mine did but not because it ran hot. It could have been a result of mistreatment, a year later it failed and did get hot then!

73 Alan G4ZFQ


A Schwarz
 

Hi Everybody

I am a bit disappointed about this and can now move on to better things, so much potential and work wasted because of a few parts... and a poor design of a final stage. I like to be cheap too, but if you want to build a customer base reliability has to be there first.

Needless to say that I won’t buy any QRPLaps kits until these issues have been addressed.

 

All the best;

 

Alex

 

PS: better get a cooling clip on your SI chip!

 


N3MNT
 

If you must have a radio that is immune to mistreatment, you might want to look at an Elecraft KX2 or KX3 .Several hundred hours on four QRP-LABS radios zero failures.


James Daldry W4JED
 

Goodbye

On 10/16/20 3:57 PM, A Schwarz wrote:

Hi Everybody

I am a bit disappointed about this and can now move on to better things, so much potential and work wasted because of a few parts... and a poor design of a final stage. I like to be cheap too, but if you want to build a customer base reliability has to be there first.

Needless to say that I won’t buy any QRPLaps kits until these issues have been addressed.

 

All the best;

 

Alex

 

PS: better get a cooling clip on your SI chip!

 


John
 

Stick on heat sinks are available for about  $0.99 on Ebay

Sure would be easier to apply one rather than to avoid using a QCX!

A suggestion. If you're going to remove and scrap through hole chip, just cut the leads off at the body and individually remove the lead with heat.

John K5MO


On Fri, Oct 16, 2020 at 1:19 PM A Schwarz <ve7dxw@...> wrote:
Hi Everyone;


First I wan to say that the QCX unit is great value and has potential....

There are a couple of things that I would like to see changed:

For the PA use one IRF-730 transistor or similar device and change the Q6, which turns of the finals to a medium current device in a TO-220 enclosure. Three devices in parallel is trouble, because you can not guarantee that they will be loaded equally without a loading resistor as used in analog power supplies. In this application a loading resistor will limit the power out and make the PA inefficient for battery use.
.
The second point is the lack of a heat-sink of the SI chip that can quit warm and should be sinked to the ground plane or externally by a clip that dissipates heat.
My unit failed because of the heat and as I was trying to repair it by removing the chip the traces came loose because of heat damage.(see pic below).
I would like to buy a new unit but I am afraid that I spend a lot of time assembling it, and it will fail again because of these shortcomings!



All the best and 73;

Alex - VE7DXW


Mike Besemer - WM4B
 

Funny… mine was one of the first QCX+ delivered and it’s been transmitting WSPR every 10 minutes since the day it was built without a lick of a problem. 

 

Feel free to buy what you like; more QRP-Labs for me!

 

Mike

WM4B

 

From: QRPLabs@groups.io [mailto:QRPLabs@groups.io] On Behalf Of A Schwarz
Sent: Friday, October 16, 2020 3:57 PM
To: QRPLabs@groups.io
Subject: Re: [QRPLabs] SI chip overheating - lack of heat-sink #problem

 

Hi Everybody

I am a bit disappointed about this and can now move on to better things, so much potential and work wasted because of a few parts... and a poor design of a final stage. I like to be cheap too, but if you want to build a customer base reliability has to be there first.

Needless to say that I won’t buy any QRPLaps kits until these issues have been addressed.

 

All the best;

 

Alex

 

PS: better get a cooling clip on your SI chip!

 


John
 

Bye Felicia!!!

On Fri, Oct 16, 2020 at 3:57 PM A Schwarz <ve7dxw@...> wrote:

Hi Everybody

I am a bit disappointed about this and can now move on to better things, so much potential and work wasted because of a few parts... and a poor design of a final stage. I like to be cheap too, but if you want to build a customer base reliability has to be there first.

Needless to say that I won’t buy any QRPLaps kits until these issues have been addressed.

 

All the best;

 

Alex

 

PS: better get a cooling clip on your SI chip!

 


Hans Summers
 

Hi Alex

I very much doubt it is possible to design a Class E IRF-730 PA that produces 5W from 80m to 17m inclusive. There is nothing wrong with three BS170 in parallel, many QRP designs have done this, it works and it is reliable. I've had almost 1000 QSOs on QCX kits and never blown a BS170 yet. 

Bear in mind that parallel BJT transistors are prone to thermal runaway. Any imbalance means one device does more work, which makes it hotter, and as it gets hotter its resistance goes down then it does even more work, gets even hotter, and the cycle repeats until soon failure by overhearing occurs. MOSFETs are the opposite: if one device does an unfair share of the load it will heat up and its resistance will increase, which decreases its share of the load and cools it. Therefore there is an inherent load balancing. 

The QCX+ PA stage is efficient, low cost and reliable. You can probably kill it if you try hard enough, but it is not a commonplace event. There's no need to change it. 

As for the Si5351A overheating there must have been some other fault in your radio. I've never seen an Si5351A get hot like that. Your board is scorched black! Something else has gone very wrong there! In ordinary use there is no need for heatsinking this component. Even if you buy an Elecraft KX2 and look at their Si5351A you'll see it is not heatsinked.

11,254 QCX/QCX+ kits have been sold to date in just over 3 years. I have received large amounts of feedback, overwhelmingly positive, and many minor changes have been incorporated into PCB revisions as the manufacturing batches progressed, as well as lots of improvements and new functionality in the firmware. The QCX+ and forthcoming QCX-mini both grew out of feedback. But simply put: there are no fundamental significant flaws in the design. If there were, I'd know about them by now. 

73 Hans G0UPL
http://qrp-labs.com


Jim Allyn - N7JA
 

Alex, I hate to be the bearer of bad news, but almost certainly the problem you have is assembly error - something YOU did.  Probably a solder bridge or something.  I don't currently have a QCX, but I have been watching this forum for, I don't know, probably a year or more, and this is the FIRST time I have seen anybody post about an overheated Si5351.  There are well over ten thousand QCXs in the field, and this just isn't a problem.  I suppose it is possible that the Si5351 was installed wrong at the factory, that would make it overheat, but other than that, it is almost certain to be an assembly error.  Also, Class E output stage has been used in zillions of QRP radios, and it is reliable.  You might want to read up on Class E amplifiers - it's an odd one but a good one!

If you decide you just want to give up on your QCX and toss it out, toss it my way!


Shirley Dulcey KE1L
 

The rated maximum power consumption of the Si5351A with three active outputs is 35 mA. (The QCX uses all three outputs; CLK0 and CLK1 provide the quadrature clocks for the receive mixer, and CLK2 provides the transmitter LO.) Even if you turn the voltage up above the nominal 3.3V to the maximum of 3.6V, that's only about 125 milliwatts. The typical power consumption under the same conditions is specified as 22 mA. The QCX does not contain a 3.3V regulator; the Si5351A power rail is 5V minus two diode drops so it's somewhere in the neighborhood of 3.6V.

There is no heat problem there. The chip can easily dissipate that much heat. The Si5351A won't overheat unless there is something wrong elsewhere in the radio, such as one of its outputs being shorted.

On Fri, Oct 16, 2020 at 7:20 PM Jim Allyn - N7JA <jim@...> wrote:
Alex, I hate to be the bearer of bad news, but almost certainly the problem you have is assembly error - something YOU did.  Probably a solder bridge or something.  I don't currently have a QCX, but I have been watching this forum for, I don't know, probably a year or more, and this is the FIRST time I have seen anybody post about an overheated Si5351.  There are well over ten thousand QCXs in the field, and this just isn't a problem.  I suppose it is possible that the Si5351 was installed wrong at the factory, that would make it overheat, but other than that, it is almost certain to be an assembly error.  Also, Class E output stage has been used in zillions of QRP radios, and it is reliable.  You might want to read up on Class E amplifiers - it's an odd one but a good one!

If you decide you just want to give up on your QCX and toss it out, toss it my way!


Mike
 

Has anyone else observed an overheating Si chip ?
I did'nt realise that The manufacturers offer a version with a heatsink to
groundplane opion.
I have seen quite a few SI5351 failiers, but all due to either shorted output
pin, overvolting the I2C or overvolting the entire chip.
fwiw Mike

On 16 Oct 2020 at 10:19, A Schwarz wrote:

Hi Everyone;

First I wan to say that the QCX unit is great value and has potential....

There are a couple of things that I would like to see changed:

For the PA use one IRF-730 transistor or similar device and change the Q6,
which turns of the finals to a medium current device in a TO-220 enclosure.
Three devices in parallel is trouble, because you can not guarantee that they
will be loaded equally without a loading resistor as used in analog power
supplies. In this application a loading resistor will limit the power out and
make the PA inefficient for battery use.
.
The second point is the lack of a heat-sink of the SI chip that can quit warm
and should be sinked to the ground plane or externally by a clip that
dissipates heat.
My unit failed because of the heat and as I was trying to repair it by
removing the chip the traces came loose because of heat damage.(see pic
below).
I would like to buy a new unit but I am afraid that I spend a lot of time
assembling it, and it will fail again because of these shortcomings!

All the best and 73;

Alex - VE7DXW






@CurtisM
 

OM Alex

Glad you clarify issue is chip removal not operation. I built a large surface mount kit once, and to remove an IC we were taught to flood the part with extra solder until it swam, then it could be slid off under heat. Maybe 50 years ago we had these springy heat sinks, but they aren't usually needed for silicon parts.

As others explain, those low ohm resistors used in emitters of combined bipolar devices aren't required for mosfets, and combining bs170 devices are lovely.

This is the most assembled transceiver kit maybe in human history, trust us its been assembled by humans from teenagers to near centurians. We can get a little sensitive from criticism when its a bit outside the lines. I hope you can address your issues. This rig is amazing you won't find a value anything like this elsewhere. Please stay with our community.

73 curt wb8yyy


R. Tyson
 

Hmmmm !  Your chip is overheating.  Thousands of these kits have been built and used and no one has reported a similar problem. Does that suggest something to you ? Might have been worth while posting photos and asking about the problem, that way you may have learnt what you did wrong.
Customer base reliability, as you call it. has already been well established by the thousands of people who come back again and again to buy further kits. They have found out that they are extremely good and work extremely well. Any problems are normally down to an error by the constructor.
I have built several, different types of kits from QRP-Labs and they all work perfectly and the instructions are the best to be found anywhere.

Reg                        G4NFR


Steven Dick
 

I did my own analysis of the design, out of curiosity, and the design is
fine. The most likely failure in this proven design is a shorted output
pin resulting in overheating of the chip. A floating input pin could
also cause the chip to draw excessive current but this is more unlikely.
As a simple precaution. It might be a good idea for builders to inspect
the pins on the SI5351A with a magnifier to make sure there are no
hairpin wire shorts or unsoldered pins before applying power to a board.
These are highly unlikely but a visual inspection is simple enough to
do on this pre-soldered chip. The entire board should also be visually
inspected with a magnifier for any issues before applying power.

-Steve K1RF

------ Original Message ------
From: "Mike" <@pencoys>
To: "A Schwarz" <ve7dxw@...>; QRPLabs@groups.io
Sent: 10/17/2020 2:27:04 AM
Subject: Re: [QRPLabs] SI chip overheating - lack of heat-sink #problem

Has anyone else observed an overheating Si chip ?
I did'nt realise that The manufacturers offer a version with a heatsink to
groundplane opion.
I have seen quite a few SI5351 failiers, but all due to either shorted output
pin, overvolting the I2C or overvolting the entire chip.
fwiw Mike

On 16 Oct 2020 at 10:19, A Schwarz wrote:

Hi Everyone;

First I wan to say that the QCX unit is great value and has potential....

There are a couple of things that I would like to see changed:

For the PA use one IRF-730 transistor or similar device and change the Q6,
which turns of the finals to a medium current device in a TO-220 enclosure.
Three devices in parallel is trouble, because you can not guarantee that they
will be loaded equally without a loading resistor as used in analog power
supplies. In this application a loading resistor will limit the power out and
make the PA inefficient for battery use.
.
The second point is the lack of a heat-sink of the SI chip that can quit warm
and should be sinked to the ground plane or externally by a clip that
dissipates heat.
My unit failed because of the heat and as I was trying to repair it by
removing the chip the traces came loose because of heat damage.(see pic
below).
I would like to buy a new unit but I am afraid that I spend a lot of time
assembling it, and it will fail again because of these shortcomings!

All the best and 73;

Alex - VE7DXW











--
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus


Richard G4TGJ
 

It seems that in any long running successful endeavour someone always comes late to the party thinking they know better.
--
73
Richard
G4TGJ


geoff M0ORE
 

RadCom, the monthly journal of the RSGB, has this month an article on building a frequency generator using the Si5351A chip. The circuit is exactly the same as used in the QRPLabs synth module. Given the few connections, there is not more you can do with it. The RadCom article has a suggested PCB ready for home building and no mention of heat problems.

Like the previous comments, it could be excessive load on one or more of the outputs.

On 17/10/2020 12:19, Steven Dick wrote:
I did my own analysis of the design, out of curiosity, and the design is
fine. The most likely failure in this proven design is a shorted output
pin resulting in overheating of the chip. A floating input pin could
also cause the chip to draw excessive current but this is more unlikely.
As a simple precaution. It might be a good idea for builders to inspect
the pins on the SI5351A with a magnifier to make sure there are no
hairpin wire shorts or unsoldered pins before applying power to a board.
 These are highly unlikely but a visual inspection is simple enough to
do on this pre-soldered chip. The entire board should also be visually
inspected with a magnifier for any issues before applying power.

-Steve K1RF

------ Original Message ------
From: "Mike" <mike@...>
To: "A Schwarz" <ve7dxw@...>; QRPLabs@groups.io
Sent: 10/17/2020 2:27:04 AM
Subject: Re: [QRPLabs] SI chip overheating - lack of heat-sink #problem

Has anyone else observed an overheating Si chip ?
I did'nt realise that The manufacturers offer a version with a heatsink to
groundplane opion.
I have seen quite a few SI5351 failiers, but all due to either shorted output
pin, overvolting the I2C or overvolting the entire chip.
fwiw Mike

On 16 Oct 2020 at 10:19, A Schwarz wrote:

 Hi Everyone;

 First I wan to say that the QCX unit is great value and has potential....

 There are a couple of things that I would like to see changed:

 For the PA use one IRF-730 transistor or similar device and change the Q6,
 which turns of the finals to a medium current device in a TO-220 enclosure.
 Three devices in parallel is trouble, because you can not guarantee that they
 will be loaded equally without a loading resistor as used in analog power
 supplies. In this application a loading resistor will limit the power out and
 make the PA inefficient for battery use.
 .
 The second point is the lack of a heat-sink of the SI chip that can quit warm
 and should be sinked to the ground plane or externally by a clip that
 dissipates heat.
 My unit failed because of the heat and as I was trying to repair it by
 removing the chip the traces came loose because of heat damage.(see pic
 below).
 I would like to buy a new unit but I am afraid that I spend a lot of time
 assembling it, and it will fail again because of these shortcomings!

 All the best and 73;

 Alex - VE7DXW















--
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus







Hans Summers
 

Hi all

As I looked again at Alex's original image he attached to his post, I realized something is wrong, and we have been overlooking this significant detail. The circled black-scorched location in Alex's photo is not where the Si5351A chip is. This is a QCX board (not QCX+). The location of the Si5351A chip (on the top side of the board) is 15mm away from the circled scorched location. See the attached photo, in which I have drawn the actual location of the Si5351A chip in yellow. You can see that it is somewhat away from the circled heat-damaged area. 

The black scorched pins actually belong to the I-Q balance potentiometer R27, 500-ohm 24-turn trimmer. The blackened area of PCB in the photo is just above these scorched pins and there are NO components on the other side of the PCB here, just groundplane... which, by the way, is intentional - the bunch of tracks cross the board from one edge to the other, dissecting the board in half along the bottom side, connecting digital signals from the various controls, to the microcontroller. It was my intention to provide a good continuous ground plane opposite these so as to avoid any RF interference. 

In other words, whatever caused this heat damage, was most likely related to soldering, it cannot possibly be anything to do with the Si5351A which is not in the right position to cause such damage anyway. 

I'd encourage Alex to try again with the QCX+. Any residual worries about the PA should be alleviated by the fact that in the QCX+ the three BS170 PA transistors and the MPS751 are now both bolted with their flats against the exposed PCB groundplane as heatsink. This was not a necessity, just a nice improvement to make it even more robust. The design is at least 80% efficient; the power dissipation rating of BS170 transistors is 830mW but at 5W and 80% efficiency each would be dissipating only half this, so the devices are within specification, even for 100% key-down WSPR operation; ordinary CW is less than half this again. The MPS751 has a 2A rating which is well in excess of what it needs, and a 625mW power dissipation, it is also well within its specification (this device dissipates power mainly for a few milliseconds at key-down and key-up, not continuously). The PA design is within specification, with good performance, robust and low cost. 

I am reminded again of my high school physics teacher Mr Ploughman who once told us something along the lines of (regrettably I forget the exact words): 

"anyone can build a bridge across a river given enough concrete. An engineer is someone who can build it safely, reliably and elegantly at a tenth the cost of the rest of us."

73 Hans G0UPL
http://qrp-labs.com

On Sat, Oct 17, 2020 at 2:59 PM geoff M0ORE via groups.io <m0ore=tiscali.co.uk@groups.io> wrote:

RadCom, the monthly journal of the RSGB, has this month an article on building a frequency generator using the Si5351A chip. The circuit is exactly the same as used in the QRPLabs synth module. Given the few connections, there is not more you can do with it. The RadCom article has a suggested PCB ready for home building and no mention of heat problems.

Like the previous comments, it could be excessive load on one or more of the outputs.

On 17/10/2020 12:19, Steven Dick wrote:
I did my own analysis of the design, out of curiosity, and the design is
fine. The most likely failure in this proven design is a shorted output
pin resulting in overheating of the chip. A floating input pin could
also cause the chip to draw excessive current but this is more unlikely.
As a simple precaution. It might be a good idea for builders to inspect
the pins on the SI5351A with a magnifier to make sure there are no
hairpin wire shorts or unsoldered pins before applying power to a board.
 These are highly unlikely but a visual inspection is simple enough to
do on this pre-soldered chip. The entire board should also be visually
inspected with a magnifier for any issues before applying power.

-Steve K1RF

------ Original Message ------
From: "Mike" <mike@...>
To: "A Schwarz" <ve7dxw@...>; QRPLabs@groups.io
Sent: 10/17/2020 2:27:04 AM
Subject: Re: [QRPLabs] SI chip overheating - lack of heat-sink #problem

Has anyone else observed an overheating Si chip ?
I did'nt realise that The manufacturers offer a version with a heatsink to
groundplane opion.
I have seen quite a few SI5351 failiers, but all due to either shorted output
pin, overvolting the I2C or overvolting the entire chip.
fwiw Mike

On 16 Oct 2020 at 10:19, A Schwarz wrote:

 Hi Everyone;

 First I wan to say that the QCX unit is great value and has potential....

 There are a couple of things that I would like to see changed:

 For the PA use one IRF-730 transistor or similar device and change the Q6,
 which turns of the finals to a medium current device in a TO-220 enclosure.
 Three devices in parallel is trouble, because you can not guarantee that they
 will be loaded equally without a loading resistor as used in analog power
 supplies. In this application a loading resistor will limit the power out and
 make the PA inefficient for battery use.
 .
 The second point is the lack of a heat-sink of the SI chip that can quit warm
 and should be sinked to the ground plane or externally by a clip that
 dissipates heat.
 My unit failed because of the heat and as I was trying to repair it by
 removing the chip the traces came loose because of heat damage.(see pic
 below).
 I would like to buy a new unit but I am afraid that I spend a lot of time
 assembling it, and it will fail again because of these shortcomings!

 All the best and 73;

 Alex - VE7DXW















--
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus