Topics

HP 9825 power supply crowbar mod could have saved disaster #HP9825


 

On this last day of 2020, which was definitely a rotten year, my dear HP 9825 made a bad noise, let some weird smell out, and the display went all crazy. The power regulator transistor in the 5V supply had decided to transform itself into a short, which did result in the 5V being 13V all of a sudden. That probably caused widespread damage to the machine. Rik Bos had warned us many years ago in a classiccmp thread that the 9825 was unprotected, and that such things might happen. And it did, so much so that HP apparently implemented a cure by adding a crowbar circuit on the almost similar 9835 power supply he has. Sheepishly, I will now have to remake that circuit and add it to my other 9825s... For others that are better at following warnings, let me reproduce the thread below, slightly edited for readability. Have a good 2021 everyone!
Marc

It's this mod:



Which is something like this schematics (except it looks at both the 12V and 5V, so it has two zeners, and shunts the unregulated 20V that powers both)



2016-08-13 Thread Rik Bos

For what it's worth a small warning about the HP 9825 series computers.

The power supply doesn't have a crowbar(over voltage protection), so a

transistor failure in the Psu can be catastrophic.

On the other hand the two 9835's I have, which uses the same form factor and

almost the same power supply layout are HP modified with crowbars added.

It seems to be good practice to add some ov-protection to the HP 9825 supply

because the switching transistor and 723 voltage regulators don't have the

eternal life.

And there no certain prediction in how they fail, short or open circuit, I

found out the hard way several years ago.

 

-Rik

2016-08-13 Thread Paul Berger

I wonder if the crowbars might have been a later revision of of the

power supply.  After the accident with my 9835A I had to completely

rebuild the +12V regulator both the 723 and the pass transistor as well

as a few passive components.  I did not see any evidence of crowbars,

and  the power supply in my 9835A seems to match the diagram in the 9835

service guide.  If I was to add crowbars to this power supply I would be

temped to monitor the voltages and crowbar the raw DC when an

overvoltage is detected, in my case it was the -20V raw that did all of

the damage.

 

Paul.

2016-08-13 Thread curiousmarc3

Thanks for the info. Any schematics of the modification?

Marc

2016-08-13 Thread Rik Bos

Marc,

 

Building a crowbar is easy, just take a Zener a little higher than the psu

voltage eg 5.2V for 5V rail put a resistor of 1k in series take a Thyristor

big enough to take about 150% of the schort current and connect it to the

zener through a small resistor. I suppose you can do the math  ;)

Be sure to put a crowbar on both the +12V and +5V and you could consider one

on the -12V rail but that voltage doesn't have a pass transistor but a 320K

voltage regulator which has a reasonable protection for over voltage.

The -5V is connected to the -12V through a zener.

Tony reverse engineered the schematics, you can download them a the HP

Museum website or become a member of the HPCC and order the CD with all

Tony's diagrams (a lot of eexcellent work).

 

-Rik

2016-08-14 Thread Curious Marc

Thanks! Would you have pictures of it and where it sits in the machine by any chance?

Marc

2016-08-15 Thread Rik Bos

 

https://flic.kr/p/L8SNPJ

Done.

-Rik

 

 

 

 


Martin Trusler
 

Commiserations and I hope that the damage is repairable! 
Martin


On 1 Jan 2021, at 07:27, CuriousMarc <marc.verdiell@...> wrote:

On this last day of 2020, which was definitely a rotten year, my dear HP 9825 made a bad noise, let some weird smell out, and the display went all crazy. The power regulator transistor in the 5V supply had decided to transform itself into a short, which did result in the 5V being 13V all of a sudden. That probably caused widespread damage to the machine. Rik Bos had warned us many years ago in a classiccmp thread that the 9825 was unprotected, and that such things might happen. And it did, so much so that HP apparently implemented a cure by adding a crowbar circuit on the almost similar 9835 power supply he has. Sheepishly, I will now have to remake that circuit and add it to my other 9825s... For others that are better at following warnings, let me reproduce the thread below, slightly edited for readability. Have a good 2021 everyone!
Marc

It's this mod:

<28968826376_474bfc89b0_c.jpg>


Which is something like this schematics (except it looks at both the 12V and 5V, so it has two zeners, and shunts the unregulated 20V that powers both)

<thyristor-scr-overvoltage-protection-crowbar-circuit-01.svg>


2016-08-13 Thread Rik Bos

For what it's worth a small warning about the HP 9825 series computers.

The power supply doesn't have a crowbar(over voltage protection), so a

transistor failure in the Psu can be catastrophic.

On the other hand the two 9835's I have, which uses the same form factor and

almost the same power supply layout are HP modified with crowbars added.

It seems to be good practice to add some ov-protection to the HP 9825 supply

because the switching transistor and 723 voltage regulators don't have the

eternal life.

And there no certain prediction in how they fail, short or open circuit, I

found out the hard way several years ago.

 

-Rik

2016-08-13 Thread Paul Berger

I wonder if the crowbars might have been a later revision of of the

power supply.  After the accident with my 9835A I had to completely

rebuild the +12V regulator both the 723 and the pass transistor as well

as a few passive components.  I did not see any evidence of crowbars,

and  the power supply in my 9835A seems to match the diagram in the 9835

service guide.  If I was to add crowbars to this power supply I would be

temped to monitor the voltages and crowbar the raw DC when an

overvoltage is detected, in my case it was the -20V raw that did all of

the damage.

 

Paul.

2016-08-13 Thread curiousmarc3

Thanks for the info. Any schematics of the modification?

Marc

2016-08-13 Thread Rik Bos

Marc,

 

Building a crowbar is easy, just take a Zener a little higher than the psu

voltage eg 5.2V for 5V rail put a resistor of 1k in series take a Thyristor

big enough to take about 150% of the schort current and connect it to the

zener through a small resistor. I suppose you can do the math  ;)

Be sure to put a crowbar on both the +12V and +5V and you could consider one

on the -12V rail but that voltage doesn't have a pass transistor but a 320K

voltage regulator which has a reasonable protection for over voltage.

The -5V is connected to the -12V through a zener.

Tony reverse engineered the schematics, you can download them a the HP

Museum website or become a member of the HPCC and order the CD with all

Tony's diagrams (a lot of eexcellent work).

 

-Rik

2016-08-14 Thread Curious Marc

Thanks! Would you have pictures of it and where it sits in the machine by any chance?

Marc

2016-08-15 Thread Rik Bos

 

https://flic.kr/p/L8SNPJ

Done.

-Rik

 

 

 

 


 

 

 

W68

 

GND

This should work, thyristor is a BTW68 (30A TO-0247AC) Zener types aren’t really important values are.

The diode blocks the revers voltage from the 12V line incase of a 12V overvoltage.

The thyristor needs about 1V 50mA to fire, this circuit works when VCC gets about 5.8-6V or if the 12V rail comes above 13.5-14V.

I’ll make a little example PCB so it becomes easy to implement.

 

-Rik

 

Van: VintHPcom@groups.io <VintHPcom@groups.io> Namens CuriousMarc
Verzonden: vrijdag 1 januari 2021 08:28
Aan: VintHPcom@groups.io
Onderwerp: [VintHPcom] HP 9825 power supply crowbar mod could have saved disaster #HP9825

 

On this last day of 2020, which was definitely a rotten year, my dear HP 9825 made a bad noise, let some weird smell out, and the display went all crazy. The power regulator transistor in the 5V supply had decided to transform itself into a short, which did result in the 5V being 13V all of a sudden. That probably caused widespread damage to the machine. Rik Bos had warned us many years ago in a classiccmp thread that the 9825 was unprotected, and that such things might happen. And it did, so much so that HP apparently implemented a cure by adding a crowbar circuit on the almost similar 9835 power supply he has. Sheepishly, I will now have to remake that circuit and add it to my other 9825s... For others that are better at following warnings, let me reproduce the thread below, slightly edited for readability. Have a good 2021 everyone!
Marc

It's this mod:



Which is something like this schematics (except it looks at both the 12V and 5V, so it has two zeners, and shunts the unregulated 20V that powers both)



2016-08-13 Thread Rik Bos

For what it's worth a small warning about the HP 9825 series computers.

The power supply doesn't have a crowbar(over voltage protection), so a

transistor failure in the Psu can be catastrophic.

On the other hand the two 9835's I have, which uses the same form factor and

almost the same power supply layout are HP modified with crowbars added.

It seems to be good practice to add some ov-protection to the HP 9825 supply

because the switching transistor and 723 voltage regulators don't have the

eternal life.

And there no certain prediction in how they fail, short or open circuit, I

found out the hard way several years ago.

 

-Rik

2016-08-13 Thread Paul Berger

I wonder if the crowbars might have been a later revision of of the

power supply.  After the accident with my 9835A I had to completely

rebuild the +12V regulator both the 723 and the pass transistor as well

as a few passive components.  I did not see any evidence of crowbars,

and  the power supply in my 9835A seems to match the diagram in the 9835

service guide.  If I was to add crowbars to this power supply I would be

temped to monitor the voltages and crowbar the raw DC when an

overvoltage is detected, in my case it was the -20V raw that did all of

the damage.

 

Paul.

2016-08-13 Thread curiousmarc3

Thanks for the info. Any schematics of the modification?

Marc

2016-08-13 Thread Rik Bos

Marc,

 

Building a crowbar is easy, just take a Zener a little higher than the psu

voltage eg 5.2V for 5V rail put a resistor of 1k in series take a Thyristor

big enough to take about 150% of the schort current and connect it to the

zener through a small resistor. I suppose you can do the math  ;)

Be sure to put a crowbar on both the +12V and +5V and you could consider one

on the -12V rail but that voltage doesn't have a pass transistor but a 320K

voltage regulator which has a reasonable protection for over voltage.

The -5V is connected to the -12V through a zener.

Tony reverse engineered the schematics, you can download them a the HP

Museum website or become a member of the HPCC and order the CD with all

Tony's diagrams (a lot of eexcellent work).

 

-Rik

2016-08-14 Thread Curious Marc

Thanks! Would you have pictures of it and where it sits in the machine by any chance?

Marc

2016-08-15 Thread Rik Bos

 

https://flic.kr/p/L8SNPJ

Done.

-Rik

 

 

 

 


Paul Berger
 

Rik,

Shouldn't the anode of the SCR connect to the unregulated +20V?  That appears to be how it is connected on the picture you initially posted of the crowbar circuit.

Paul.

On 2021-01-01 8:00 a.m., Rik Bos wrote:

 

 

W68

 

GND

This should work, thyristor is a BTW68 (30A TO-0247AC) Zener types aren’t really important values are.

The diode blocks the revers voltage from the 12V line incase of a 12V overvoltage.

The thyristor needs about 1V 50mA to fire, this circuit works when VCC gets about 5.8-6V or if the 12V rail comes above 13.5-14V.

I’ll make a little example PCB so it becomes easy to implement.

 

-Rik

 

Van: VintHPcom@groups.io <VintHPcom@groups.io> Namens CuriousMarc
Verzonden: vrijdag 1 januari 2021 08:28
Aan: VintHPcom@groups.io
Onderwerp: [VintHPcom] HP 9825 power supply crowbar mod could have saved disaster #HP9825

 

On this last day of 2020, which was definitely a rotten year, my dear HP 9825 made a bad noise, let some weird smell out, and the display went all crazy. The power regulator transistor in the 5V supply had decided to transform itself into a short, which did result in the 5V being 13V all of a sudden. That probably caused widespread damage to the machine. Rik Bos had warned us many years ago in a classiccmp thread that the 9825 was unprotected, and that such things might happen. And it did, so much so that HP apparently implemented a cure by adding a crowbar circuit on the almost similar 9835 power supply he has. Sheepishly, I will now have to remake that circuit and add it to my other 9825s... For others that are better at following warnings, let me reproduce the thread below, slightly edited for readability. Have a good 2021 everyone!
Marc

It's this mod:



Which is something like this schematics (except it looks at both the 12V and 5V, so it has two zeners, and shunts the unregulated 20V that powers both)



2016-08-13 Thread Rik Bos

For what it's worth a small warning about the HP 9825 series computers.

The power supply doesn't have a crowbar(over voltage protection), so a

transistor failure in the Psu can be catastrophic.

On the other hand the two 9835's I have, which uses the same form factor and

almost the same power supply layout are HP modified with crowbars added.

It seems to be good practice to add some ov-protection to the HP 9825 supply

because the switching transistor and 723 voltage regulators don't have the

eternal life.

And there no certain prediction in how they fail, short or open circuit, I

found out the hard way several years ago.

 

-Rik

2016-08-13 Thread Paul Berger

I wonder if the crowbars might have been a later revision of of the

power supply.  After the accident with my 9835A I had to completely

rebuild the +12V regulator both the 723 and the pass transistor as well

as a few passive components.  I did not see any evidence of crowbars,

and  the power supply in my 9835A seems to match the diagram in the 9835

service guide.  If I was to add crowbars to this power supply I would be

temped to monitor the voltages and crowbar the raw DC when an

overvoltage is detected, in my case it was the -20V raw that did all of

the damage.

 

Paul.

2016-08-13 Thread curiousmarc3

Thanks for the info. Any schematics of the modification?

Marc

2016-08-13 Thread Rik Bos

Marc,

 

Building a crowbar is easy, just take a Zener a little higher than the psu

voltage eg 5.2V for 5V rail put a resistor of 1k in series take a Thyristor

big enough to take about 150% of the schort current and connect it to the

zener through a small resistor. I suppose you can do the math  ;)

Be sure to put a crowbar on both the +12V and +5V and you could consider one

on the -12V rail but that voltage doesn't have a pass transistor but a 320K

voltage regulator which has a reasonable protection for over voltage.

The -5V is connected to the -12V through a zener.

Tony reverse engineered the schematics, you can download them a the HP

Museum website or become a member of the HPCC and order the CD with all

Tony's diagrams (a lot of eexcellent work).

 

-Rik

2016-08-14 Thread Curious Marc

Thanks! Would you have pictures of it and where it sits in the machine by any chance?

Marc

2016-08-15 Thread Rik Bos

 

https://flic.kr/p/L8SNPJ

Done.

-Rik

 

 

 

 


 

Hi Paul,

I did it by head, but that's also a good solution.
Even better I think because it shorts the circuit before the regulators. I'll change the diagram and PCB.

Thanks,

-Rik


Kuba Ober
 

What an ending to the year – I’m sorry that you lost one of those beautiful machines. On the other hand – 13V is a lot to regulate 5V from. It might not have saved anything even with a lower input, but that must sure be a good hand warmer in colder climates :)

It probably would be possible to find a BGA packaged FPGA and level shifters to fit on DIP adapters to recreate the function of all the custom logic chips, just to show that supply who’s the boss, but that would sure be a bit of effort.

Happy New Year, everyone!

1 jan. 2021 kl. 2:27 fm skrev CuriousMarc <marc.verdiell@gmail.com>:

On this last day of 2020, which was definitely a rotten year, my dear HP 9825 made a bad noise, let some weird smell out, and the display went all crazy. The power regulator transistor in the 5V supply had decided to transform itself into a short, which did result in the 5V being 13V all of a sudden. That probably caused widespread damage to the machine. Rik Bos had warned us many years ago in a classiccmp thread that the 9825 was unprotected, and that such things might happen. And it did, so much so that HP apparently implemented a cure by adding a crowbar circuit on the almost similar 9835 power supply he has. Sheepishly, I will now have to remake that circuit and add it to my other 9825s... For others that are better at following warnings, let me reproduce the thread below, slightly edited for readability. Have a good 2021 everyone!


Achim Buerger
 

I wonder whether such a crow bar would have prevented the sudden death of my 9825T some years ago. Back the, a rather small capacitator in the PSU exploded and burned for while. After that, the CPU and most of the other boards where scrap.


 

Probably yes, it would shorted the +20 line and low the fuse. Saving the machine.

-Rik


Achim Buerger
 

Another question is, how would you detect, that the crowbar has been activated? Just a blown fuse could have many causes.


 

It keeps blowing the fuse ;)
Getting the main fuse blown, should be a clear signal to check the power supply.

-Rik


Dave McGuire
 

On 1/1/21 7:00 AM, Rik Bos wrote:
This should work, thyristor is a BTW68 (30A TO-0247AC) Zener types aren’t really important values are.
The diode blocks the revers voltage from the 12V line incase of a 12V overvoltage.
The thyristor needs about 1V 50mA to fire, this circuit works when VCC gets about 5.8-6V or if the 12V rail comes above 13.5-14V.
I’ll make a little example PCB so it becomes easy to implement.
Do you plan to share this design? I'd like to build a few and install them in the 9825s here.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA


Achim Buerger
 

I could also use 4 or 5 of them. Maybe we can start a croud funding to produce some PCBs?


 

The HP designed crowbar appears to have a small capacitor between the SCR gate and ground. I also see one in the example crowbar circuit in the Art of Electronics (10 nF) to prevent triggering due to short spikes. I wonder if that would be a desirable addition to this circuit?


jbemond
 

Marc and others,

Thank you for the warning, I will look into it carefully. This shitty year... I did however allow myself to repair / modify some power supplies, but that didn't prevent some machines from crashing, I repaired most of them, but there are still three old things to finish: Computer Devices inc Miniterm 1203S, Texas Instrument Silence 743, Kaypro 4 (Video card out of order, I'll get a new one... if you have that lying around).

JB

Le 01/01/2021 à 08:27, CuriousMarc a écrit :

On this last day of 2020, which was definitely a rotten year, my dear HP 9825 made a bad noise, let some weird smell out, and the display went all crazy. The power regulator transistor in the 5V supply had decided to transform itself into a short, which did result in the 5V being 13V all of a sudden. That probably caused widespread damage to the machine. Rik Bos had warned us many years ago in a classiccmp thread that the 9825 was unprotected, and that such things might happen. And it did, so much so that HP apparently implemented a cure by adding a crowbar circuit on the almost similar 9835 power supply he has. Sheepishly, I will now have to remake that circuit and add it to my other 9825s... For others that are better at following warnings, let me reproduce the thread below, slightly edited for readability. Have a good 2021 everyone!
Marc

It's this mod:



Which is something like this schematics (except it looks at both the 12V and 5V, so it has two zeners, and shunts the unregulated 20V that powers both)



2016-08-13 Thread Rik Bos

For what it's worth a small warning about the HP 9825 series computers.

The power supply doesn't have a crowbar(over voltage protection), so a

transistor failure in the Psu can be catastrophic.

On the other hand the two 9835's I have, which uses the same form factor and

almost the same power supply layout are HP modified with crowbars added.

It seems to be good practice to add some ov-protection to the HP 9825 supply

because the switching transistor and 723 voltage regulators don't have the

eternal life.

And there no certain prediction in how they fail, short or open circuit, I

found out the hard way several years ago.

 

-Rik

2016-08-13 Thread Paul Berger

I wonder if the crowbars might have been a later revision of of the

power supply.  After the accident with my 9835A I had to completely

rebuild the +12V regulator both the 723 and the pass transistor as well

as a few passive components.  I did not see any evidence of crowbars,

and  the power supply in my 9835A seems to match the diagram in the 9835

service guide.  If I was to add crowbars to this power supply I would be

temped to monitor the voltages and crowbar the raw DC when an

overvoltage is detected, in my case it was the -20V raw that did all of

the damage.

 

Paul.

2016-08-13 Thread curiousmarc3

Thanks for the info. Any schematics of the modification?

Marc

2016-08-13 Thread Rik Bos

Marc,

 

Building a crowbar is easy, just take a Zener a little higher than the psu

voltage eg 5.2V for 5V rail put a resistor of 1k in series take a Thyristor

big enough to take about 150% of the schort current and connect it to the

zener through a small resistor. I suppose you can do the math  ;)

Be sure to put a crowbar on both the +12V and +5V and you could consider one

on the -12V rail but that voltage doesn't have a pass transistor but a 320K

voltage regulator which has a reasonable protection for over voltage.

The -5V is connected to the -12V through a zener.

Tony reverse engineered the schematics, you can download them a the HP

Museum website or become a member of the HPCC and order the CD with all

Tony's diagrams (a lot of eexcellent work).

 

-Rik

2016-08-14 Thread Curious Marc

Thanks! Would you have pictures of it and where it sits in the machine by any chance?

Marc

2016-08-15 Thread Rik Bos

 

https://flic.kr/p/L8SNPJ

Done.

-Rik

 

 

 

 



 

No problem, I'll add it to the PCB.

-Rik


Jack Rubin
 

I got my boards back from OSHPark today. I couldn't find a BZV60-C4V3 anywhere, so I used a BZV85-C4V3. It seems that all three diodes could be laid out as DO-34 or DO-35 footprints to make substitutions easier. With the change to BZV85-C4V3, all components are available at Mouser.

My boards are Rik's first rev without the bypass cap. I will just wire one in. Rik specified a metal film cap for this component. Any reason not to use a standard MLCC bypass cap here? 

We're about to move to New York City temporarily to stay with my daughter while her first child is born; we'll be gone for six weeks during which time I won't have access to tools or my 9825 so no chance to build or test while we're away. Hopefully someone else will build one soon.

--
Jack
www.computerarium.org


 

No, no problem.

-Rik


Dave McGuire
 

On 1/12/21 12:24 AM, Jack Rubin wrote:
I got my boards back from OSHPark today. I couldn't find a BZV60-C4V3 anywhere, so I used a BZV85-C4V3. It seems that all three diodes could be laid out as DO-34 or DO-35 footprints to make substitutions easier. With the change to BZV85-C4V3, all components are available at Mouser.
My boards are Rik's first rev without the bypass cap. I will just wire one in. Rik specified a metal film cap for this component. Any reason not to use a standard MLCC bypass cap here?
We're about to move to New York City temporarily to stay with my daughter while her first child is born; we'll be gone for six weeks during which time I won't have access to tools or my 9825 so no chance to build or test while we're away. Hopefully someone else will build one soon.
I'm very much interested in building a couple of these, now that I have an operational 9825. (actually three!)

Are you going to stop by here on your way to NYC, Gramps?

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA


 

I’ve placed a small order for 10pcs of the latest version, I need 5 the others are available for 8,- a piece.

They will be available sometime next week.

 

-Rik

 

Van: VintHPcom@groups.io <VintHPcom@groups.io> Namens Jack Rubin
Verzonden: dinsdag 12 januari 2021 06:25
Aan: VintHPcom@groups.io
Onderwerp: Re: [VintHPcom] HP 9825 power supply crowbar mod could have saved disaster #HP9825

 

I got my boards back from OSHPark today. I couldn't find a BZV60-C4V3 anywhere, so I used a BZV85-C4V3. It seems that all three diodes could be laid out as DO-34 or DO-35 footprints to make substitutions easier. With the change to BZV85-C4V3, all components are available at Mouser.

My boards are Rik's first rev without the bypass cap. I will just wire one in. Rik specified a metal film cap for this component. Any reason not to use a standard MLCC bypass cap here? 

We're about to move to New York City temporarily to stay with my daughter while her first child is born; we'll be gone for six weeks during which time I won't have access to tools or my 9825 so no chance to build or test while we're away. Hopefully someone else will build one soon.

--
Jack
www.computerarium.org


David Stokes
 

Hello Rik,

I would like 3 of those remaining, can settle the cost via PayPal, to include postage to my address in Italy.

Best regards,

David Stokes




On 12 Jan 2021, at 11:57 am, Rik Bos <hp-fix@...> wrote:



I’ve placed a small order for 10pcs of the latest version, I need 5 the others are available for 8,- a piece.

They will be available sometime next week.

 

-Rik

 

Van: VintHPcom@groups.io <VintHPcom@groups.io> Namens Jack Rubin
Verzonden: dinsdag 12 januari 2021 06:25
Aan: VintHPcom@groups.io
Onderwerp: Re: [VintHPcom] HP 9825 power supply crowbar mod could have saved disaster #HP9825

 

I got my boards back from OSHPark today. I couldn't find a BZV60-C4V3 anywhere, so I used a BZV85-C4V3. It seems that all three diodes could be laid out as DO-34 or DO-35 footprints to make substitutions easier. With the change to BZV85-C4V3, all components are available at Mouser.

My boards are Rik's first rev without the bypass cap. I will just wire one in. Rik specified a metal film cap for this component. Any reason not to use a standard MLCC bypass cap here? 

We're about to move to New York City temporarily to stay with my daughter while her first child is born; we'll be gone for six weeks during which time I won't have access to tools or my 9825 so no chance to build or test while we're away. Hopefully someone else will build one soon.

<image001.jpg>

--
Jack
www.computerarium.org