HP6012A 1kW Power Supply Repair

[Jay_Diddy_B]

Hi group,

I am going to share the steps I took to repair a HP6012A 1kW Power Supply.

Before I get into the details, these power supplies are off-line switching supplies. A large portion of the circuit is connected directly to the line. Read and understand all the warnings in the HP Service Manual before attempting repair of these units.

The nature of this circuit, is that if you don't repair everything, it will fail catastrophically again. It is very important to find all the damaged components.

Some interesting details about this power supply can be found in the HP Journal article:

http://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1981-08.pdf

This power supply can deliver over 1kW for output voltage between 20V and 60V.

This is the load map:



I bought this unit at local store for $56.50 Canadian dollars in very poor condition.

In the center of the unit is a tunnel which houses the two MOSFET switch assemblies and a down programmer. The down programmer is used to discharge the output capacitors if the output voltage is turned down.
This is the top cover of the tunnel:



I removed the two screws which hold the cover to the chassis and pulled out the three boards. There was the largest collection of dust that I have ever seen in a piece of equipment:



I used a vacuum to remove the dust as I dissembled the unit. Here is a picture of the edge connectors:



I am already thinking that this dust build up has caused inadequate cooling and possible failure of the MOSFETs.
Here is a picture of the original MOSFETs. They were made by IR but are marked with a HP part number:



This is a picture of the component side of the board before cleaning:



After cleaning the board I found some traces that had been fused. You can also see that R10 is damaged:






I studied the schematic in the Service Manual, and assumed that the damage to the driver circuit was probably caused by MOSFET failure:




To Be continued...

Regards,

Jay_Diddy_B

[Jay_Diddy_B]

Continuation

Q3 was visibly damaged:



I also checked the MOSFETs with my DMM. I look for the body diode, suspecting that they were damaged I opened one up:



The MOSFET shows that the source lead is open.
I believe that these MOSFETs are IRF450s, this from the size of the silicon die, the application and what was available when the unit was designed.


I checked the diodes in the driver circuit and found some bad parts. I removed the bad parts:



I then started to rebuild the circuit board. I replaced the bad parts and replaced the fused tracks with wires:





I installed two IRF450 MOSFETs using thermal grease:




To be continued....

Regards,

Jay_Diddy_B

[Jay_Diddy_B]

Continuation..

In addition to the damage to the MOSFET board, the on/off switch was broken. (I think that this happened when the unit was thrown away.) Here is a picture showing the board with the broken switch removed. I used chipquik low melting point solder to remove the switch without damaging the board:




I tried to find a new replacement for the switch, the original manufacturer was Schadow (or ITT-Schadow) unable to find one I pulled one out of a HP6034L parts units that I have:



With the unit all put back together, I plugged the unit in and it now works:






I am missing the top cover for the chassis, if anybody has one please send me a PM.



Regards,

Jay_Diddy_B

[radhaz]

Great work. It's amazing how long people will use fan cooled gear and not think of the dirt building up inside.

[pomonabill221]

WOW what a STEAL!   Good that you were able to repair it.
WHAT a MESS though.... WAY too much dust and of course, no air flow.
Thanks for the great pictures of what you went through.

These supplies are really built... strange the stat didn't protect the finals from failing, but who knows what actually made it fail.
I am sure it feels good to be back though!

[nctnico]

It is weird it failed. There are temperature sensors on each heatsink. I have 2 of these and in one the ventilator didn't run when I bought it. After replacing the ventilator it works fine.

[Jay_Diddy_B]

Hi group,

I know this is an old thread, but I want to keep all the information together.

I repaired another HP6012A today, and it was pretty much the same story.

Here is the schematic of the MOSFET board:



I found that R10 was burnt, Q1 and Q2 were open circuit, VR1 was short.
Here is a photograph with the parts removed:



In addition to the damaged parts, some of the gate drive tracks had fused:



I installed the new parts:



And repaired the damaged tracks with 30awg wire:



I replaced Q1 and Q2 with IRF450. Notice the 8601 date code, these MOSFETs have been waiting for 30 years.



With these parts replaced the unit works again.

Regards,

Jay_Diddy_B

[megajocke]

Could it be that the gunk is causing leakage current from drain to gate rather than a thermal failure?

It looks like just 10 microamps would be enough to turn the FETs on when the active turn-off circuit is not running. If the gunk goes below 30 megaohms at 300 V DC you'll be in trouble...

[Gyro]

That track going directly from the gates to P1/J1 pin 17 (not used) doesn't look like a great idea. At best, an antenna, at worst, a leakage path. 

There also doesn't seem to be any arrangement for current sharing on the FETs, maybe some Source resistors? Individual gate stopper resistors rather than shared would also be preferable.

[Jay_Diddy_B]

Hi,
In hindsight the design of the MOSFET circuit in the 6012A is not the greatest design.

There is no need for resistors to force the current sharing in this application, because in this application the MOSFETs RDSon has a positive temperature coefficient.

Separate gate resistors would be a good idea.

Here is the schematic that was used in later power supplies including the 6012B:



J1 is a test connector.

Regards,

Jay_Diddy_B

[Gyro]

There is no need for resistors to force the current sharing in this application, because in this application the MOSFETs RDSon has a positive temperature coefficient.

Yes, you're quite right in switching mode of course!

Just as an aside (not relevant in this application) I believe modern mosfets are less predictable in their temperature coefficient than older geometries when it comes to linear operation. At a certain point in their transconductance curve they can turn from PTC to NTC. Probably why they need to be de-rated quite heavily when operated as linear pass elements. I've found a reference to it here... http://www.onsemi.com/pub_link/Collateral/AND8199-D.PDF

[nctnico]

For use in linear applications you need MOSFETs which are designed to be used in linear applications. They are made differently (lateral structure).

[Gyro]

...Which are getting very hard to find these days, apart from 'boutique' audio ones. 

(Sorry Jay_Diddy_B, dragging your thread OT)

[Jay_Diddy_B]

...Which are getting very hard to find these days, apart from 'boutique' audio ones. 

(Sorry Jay_Diddy_B, dragging your thread OT)

I believe the original transistors were audio audible transistors. They went 'Bang !!'   

Fortunately, I have a reasonable stock of original, from the 80's, IRF450.

Regards,

Jay_Diddy_B