50W Amp problem


K4ALE
 

I built a 50 W Amp this weekend.  It tested fine.  I hooked it up to a QCX Mini.  Both in 40m.  I went through the testing.  All did well.  I then drove it with the QCX Mini.  It did great for about 10-15 minutes, and then suddenly went to 0 watts out.  No smoke.  No smell.  Power light and transmit light still work.  Just no power.
I presume one of the finals (Q1 or Q2) failed. 
How do I confirm this?  What else should I do to diagnose this problem.
PS, this thing is great.  I hope I can get it back on line.
Bevin
K4ALE
Lynchburg, VA USA


ae1ct1998 <AE1CT1998@...>
 

Did you set the currant to the proper level?


On Tue, Dec 7, 2021, 18:31 K4ALE <bevin@...> wrote:
I built a 50 W Amp this weekend.  It tested fine.  I hooked it up to a QCX Mini.  Both in 40m.  I went through the testing.  All did well.  I then drove it with the QCX Mini.  It did great for about 10-15 minutes, and then suddenly went to 0 watts out.  No smoke.  No smell.  Power light and transmit light still work.  Just no power.
I presume one of the finals (Q1 or Q2) failed. 
How do I confirm this?  What else should I do to diagnose this problem.
PS, this thing is great.  I hope I can get it back on line.
Bevin
K4ALE
Lynchburg, VA USA


Jim Strohm
 

Assuming a push-pull amp, if one final died you should still have half output.  A friend homebrewed an amp and fried one final.   It limped along at half power for months.    

If it’s devices in parallel, it’s common that if one fails, the rest zipper themselves to death in very short order.  

73
Jim N6OTQ 

Sent from my quenched-gap spark transmitter. 

On Dec 7, 2021, at 5:31 PM, K4ALE <bevin@...> wrote:

I built a 50 W Amp this weekend.  It tested fine.  I hooked it up to a QCX Mini.  Both in 40m.  I went through the testing.  All did well.  I then drove it with the QCX Mini.  It did great for about 10-15 minutes, and then suddenly went to 0 watts out.  No smoke.  No smell.  Power light and transmit light still work.  Just no power.
I presume one of the finals (Q1 or Q2) failed. 
How do I confirm this?  What else should I do to diagnose this problem.
PS, this thing is great.  I hope I can get it back on line.
Bevin
K4ALE
Lynchburg, VA USA


K4ALE
 

Current was set at full anticlockwise/counterclockwise minimum. Output from QCX Mini is a bit high, perhaps.
I think it is push-pull....


K4ALE
 

One more symptom.  When I disconnect the PTT, the output of the QCX Mini goes through.  When PTT is connected, it does not.  Power to the Amp does not block the through-put of the QCX Mini's signal.
Bevin


ae1ct1998 <AE1CT1998@...>
 

The amp has a 3db attenuator built in so it shud take the mini output with no problem. I've built about 10 50 watt amps with no problem and I've had so.e sent me for fixing and I've found several guys have installed the binocular transformer backwards and others had bad solder joints.


On Tue, Dec 7, 2021, 19:05 K4ALE <bevin@...> wrote:
Current was set at full anticlockwise/counterclockwise minimum. Output from QCX Mini is a bit high, perhaps.
I think it is push-pull....


Evan Hand
 

On Tue, Dec 7, 2021 at 05:31 PM, K4ALE wrote:
How do I confirm this?  What else should I do to diagnose this problem.
Hi Bevin,

I would start by doing the bias adjustment again.  If there is no current through the IRF510s, then they are most likely toast.  Before totally giving up on them, verify that the gate voltage is varying with the R5 adjustment.  You can also check that the bias regulator is putting out 5 volts when the PTT line is high from the QCX.

Happy Hunting for the problem.
73
Evan
AC9TU


K4ALE
 

Thanks to all.  Will keep working on it and reporting back.
1.  I am certain that the binocular transformer is correctly constructed and installed and there is good continuity through it, and confirmed again tonight.
2;. I will confirm lack of current through the IRF510s (which is my first guess); thanks for the suggestion about the bias regulator (I hadn't thought of that.)  I have ordered a pile of replacement IRF510s. 
Again, thanks for all the suggestions.
Bevin


K4ALE
 

All, esp Evan:
IC1/78L05Z bias regulator putting out exactly 4.97 vDC (tested from 12.5 to 16.5 vDC iinput.)  However, nothing at all is coming out of R5. [E.g. 0 vDC at pin 2, where it should be coming out, according to the schematic.] Therefore, no bias change at Q1 and Q2 when R5 is "adjusted," which bias pins both show 0 vDC at all times (including with RF incoming.)  Voltage on Q1 and Q2 at the drain is same as supply voltage, which I believe is correct downstream from T2 bifilar toroid.

So, can I dare to think that the R5 variable resistor has gone open, thereby depriving Q1 and Q2 of bias, and thus shutting down the output entirely?

Remember, I can follow a schematic and stick the leads of a multimeter on the pins shown on a pinout for a device, but otherwise have almost no idea what I am doing.  I figure [at about the Forrest Gump level of comprehension] that the drain needs the full voltage, the RF from the QCX and the bias voltage go to the gate, the source goes to ground, and the amplified RF siginigal goes out of the drain to the LPF and out to the World.  I think I am missing the bias voltage, which seems to die off in R5.

You are Helping the Handicapped/Challenged here [with apologies to those who are really handicapped....]
Bevin


Evan Hand
 

Bevin,

At this point, we need to verify that it is not a short in the IRF510s that is pulling down the voltage, or an open connection.  The gates of the IRF510s should be open circuits without power.  We need to determine if the failure is R5 or the IRF510s, or the transformer connections.  I would turn off the power and measure the resistance to ground of both of the gates of the IRF510s (should be the same) and adjust R5 to see if it changes.  Then work your way closer to R5 until you do get a change in resistance.  Of course, if there is a short, then My guess would be the IRF510s have failed.  One gate failure would disable both devices as they share a common bias circuit.  Tracking back should determine if it is an open transformer connection or an R5 failure.

If you determine that you need to remove the IRF510s, then I would check the bias circuit before installing the new MOSFETS.  You can power the amp and key it using the QCX practice mode that will not feed RF to the amp.  With this setup be sure that you can turn on and off the voltage to the gate pins with the PTT line, and that you can adjust the voltage with R5.  Once that checks out you can then install the IRF510s again.  Remember to do the short circuit tests in the manual to verify that the heat sink insulators are keeping the drain tab isolated from the heatsink.  This is in section 5 page 40 of the manual.

So to answer your question directly, with the tests we have done prior to the email you do not know what is causing the bias voltage to be lost.  R5 could have failed, but it is more likely that you have either a failed MOSFET or an intermittent solder connection in T1.

Let us know what you find.
73
Evan
AC9TU


John Laidler
 

What voltage were you running the QCX from? My QCX+ drives the 50W amp reliably but I only use a 12 volt supply for the radio. The amp has a seperate 20 volt supply. 


ae1ct1998 <AE1CT1998@...>
 

I run both radio and amp at 13.8vdc and get around 30 watts out.


On Wed, Dec 8, 2021, 08:36 John Laidler <rjohnlaidler@...> wrote:
What voltage were you running the QCX from? My QCX+ drives the 50W amp reliably but I only use a 12 volt supply for the radio. The amp has a seperate 20 volt supply. 


K4ALE
 

Evan,
Many thanks, again.
I did a quick check before I went to work.  I will repeat the test when I get home to confirm, but I think I have a dead short from pin 2 of the 78L05Z to the gate of both IRL510s (Q! & Q2).  BUT, I am also suspicious that there is a short in the trifilar transformer which was not there before (carefully tested during construction and it worked for a while). 
More on this, I think I got it.
Bevin


Evan Hand
 

Bevin,

Pin 2 of the 79L05Z is at ground, so 0 ohms to the gates could be caused by the R5 setting.  Change R5 to the maximum bias setting and see if the gate to the ground on pin 2 of the regulator goes up.  A reading of the gates to ground with R5 at the max bias setting should read about 4.7k Ohms.  You are looking for a variable resistance on the gates as R5 is changed.

73
Evan
AC9TU


K4ALE
 

Evan, et als.,
I have tested until I am cross-eyed.  From the schematic, pin 1 of the 78L05Z connects to one end of the 4.7 trim pot.  The other end is at ground.  The middle pin (the wiper pin) goes from dead short to ground at full anticlockwise to 2.02 k ohms at highest and back to short at full.  I think the pot is squirrelly and shot.  I also think the two IRF510s are also probably shot, because the gates are not working at all and pin 3 (source) shorts to ground, and I don't see that path on the schematic, except through C6 and C7 (which do not test as shorted.).
So, I will get my hands on a collection of 78L05Zs, IRF510s, proper winding wire, and 4.7k trim pots and rebuild the whole network from regulator through the power transistors.  Unless you have a better suggestion.

Thank you for your patience and great help.  I have learned a lot tracing all this down.
Bevin
K4ALE


Evan Hand
 

On Wed, Dec 8, 2021, at 08:38 PM, K4ALE wrote:
I have tested until I am cross-eyed.  From the schematic, pin 1 of the 78L05Z connects to one end of the 4.7 trim pot.  The other end is at ground.  The middle pin (the wiper pin) goes from dead short to ground at full anticlockwise to 2.02 k ohms at highest and back to short at full.  I think the pot is squirrelly and shot.  I also think the two IRF510s are also probably shot, because the gates are not working at all and pin 3 (source) shorts to ground, and I don't see that path on the schematic, except through C6 and C7 (which do not test as shorted.).
So, I will get my hands on a collection of 78L05Zs, IRF510s, proper winding wire, and 4.7k trim pots and rebuild the whole network from regulator through the power transistors.  Unless you have a better suggestion.
Brevin,
The data you just provided indicates that the IRF510 gates are not shorted, as they connect to the R5 wiper and that resistance to ground is changing with adjustment of R5.  If the IRF510s were shorted, pin 2 of R5 would stay at zero ohms.  What could be happening is that IC1 output is shorted.  That would put ground on both sides of R5 and give the results that you are seeing.  In a prior posting, you did state the output of IC1 was 4.97volts.  This does conflict with the current data.

You could do a quick resistance check on the IRF510 drains (middle pin) to ground.  Those should be an open high resistance.  if it is not, then yes, you will need to replace the IRF510s. 

I opened my 50w PA and checked the resistances around R5
Pin 3 is 0 ohms
Pin 2 is 2.41k ohms
pin 1 is 2.94k ohms

That is what I would expect as there is some resistance in the output of IC1 even without power (I stated it should be an open circuit, which is in error).  I did not change R5 as I did not want to upset the bias setting.

Three possible paths forward, all of which will require the removal of the board from the case.
1 - Assume that IC1 is bad.
     a - remove IC1 from the board and verify that now R5 pin 1 measures 4.7k ohms to ground
     b - if R5 measures correctly then replace IC5 and test bias circuit with power applied, the QCX in practice mode, and the PTT line turned on. 
             This should be able to be done with the board out of the case as long as you keep the tabs of the IRF510s from shorting to any ground connections.
             Always use a dummy load when testing, even if you do not expect RF on the output.

2 - Assume that the IRF510s are bad.
     a - Remove both IRF510s (these are best replaced as a pair from the same lot.  That is not required, just reduces bias differences between devices).
     b - Test the bias voltage on both gate pin tabs before putting the IRF510s back in.  You will need the QCX to turn on the PTT line.  Again, make sure the QCX is in practice mode.

3 - Assume there is a wiring or solder problem.
     a - Do the continuity test suggested in the assembly manual
     b - Reheat the coil connections
     c - redo the resistance tests.

When you reinstall the board into the case remember to do the short circuit tests in the manual to verify that the heat sink insulators are keeping the drain tabs isolated from the heatsinks.  This is in section 5 page 40 of the manual.

The above are just suggestions.  Please take proper care when doing the tests with applied power.
73
Evan
AC9TU


Timothy Fidler
 

Bevin

you made a comment about getting more  IRF510s - be aware that the only mafr of  this IRFs approved by Hans is Vishay.  He tried various other mafr's output and he was not as happy with the results.  All mfr's dies for the semiconductors vary a bit and they are not intended for or tested as RF power amps so it is hardly surprising that some are better than others.

Don't plug or unplug the  keying cable  from the Qcx Mini to the PA with power on both unless you have a  470 R resistor in the hot lead of that cable and even then it is not a good idea.  The respective  O point of the microprocessor  gets destroyed sometimes when you do this and nobody has  a better fix. The additional resistor in now way affects the control signal but stops  voltage / current being pumped into the  micro'  OP port.   I believe the issue may relate to independent grounding of the   PA and the  Qcx's supply but that's not proven (clearly the zero volts  can be explicitly tied to prevent this if one wishes). 



Hans Summers
 

Hi TIm, all
 
you made a comment about getting more  IRF510s - be aware that the only mafr of  this IRFs approved by Hans is Vishay.  He tried various other mafr's output and he was not as happy with the results.  All mfr's dies for the semiconductors vary a bit and they are not intended for or tested as RF power amps so it is hardly surprising that some are better than others.

To be clear - what I said was, buy from a reputable distributor. This was the outcome of my own investigations. In the case of IRF510s from Digikey, the only available brand is Vishay (which also happens to be what I am using in the QRP Labs kits). During testing I did also work with original IR-branded parts from Farnell UK and those were fine also. 

What IS a problem, is when you buy "IRF510" from eBay, Amazon, AliExpress, etc. In many (most, even) cases, what you will get is some kind of generic TO220-cased MOSFET which works sort of. I believe these are super-mass-produced and then the manufacturer just slaps various labels on the part, for sales purposes: IRF510, IRF520, IRF530 etc. For many applications you would never know the difference. But in an RF amp the capacitances are of importance. This "generic part", in my testing, behaved more like an IRF530 than a '510. 

So to summarize: my recommendation is, get your IRF510 from a reputable distributor (Digikey, Mouser, Arrow, RS, Farnell, WhatHaveYou), not from online marketplace sales. 

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


K4ALE
 

Evan,
This is hugely constructive and I sincerely appreciate your time and attention to put this together with me. 
Refreshed this morning, I will tackle this again this evening after work as outlined above.
You should be pleased to know that I took each of your prior posts, copied the operative parts, and made a checklist to proceed.
As to the IRF510s:  if a component is available from Digikey or Mouser, that is where I get them.  I have the Vishay parts in my cart at Digi Key in this instance.

The more I mess with this, the more I appreciate the details Hans has taken to design these products, and respect the knowledge required to service, or at least understand the interplay of the components. It's really magic, you know...
Bevin
K4ALE