HP/Agilent Infinium 548xx scope - Power Supply Schematic !

[Conventional Wisdom]

Hello folks,

You know what I miss?  I miss the way HP and others used to include detailed service information, and even full schematics, in their product manuals.  Ah, the good old days.  Now they act like a simple power supply is a closely guarded secret.

I found this forum last year while searching for repair data for my 54845A scope, after its power supply stopped working.  Mainly what I found out is that there are a lot of people in the same boat, who could repair the darn thing if only they had a schematic.  If you're one of them, I have good news:  I'm an engineer, I'm retired, I'm stubborn and I don't like secrets.  So...

I reverse-engineered the entire power supply, drafted a complete schematic, and I'm placing it in the public domain here.  It's my gift to all you folks who take the time to help each other out.

This isn't some sophisticated, digitally-managed modern marvel; it's a very conventional 25-year-old, 300-watt design with a wide-range PFC front end, an isolated flyback converter for standby and bias supplies, a two-transistor forward converter with post-regulated outputs, and some protection and control circuitry.  All the semiconductors are generally non-special and still available.

First, the obligatory warnings: This information is provided WITHOUT WARRANTY OF ANY KIND and you use it entirely AT YOUR OWN RISK.  There's 380 volts DC in that chassis, even in the standby state, so if you're not completely comfortable with troubleshooting and repairing high-voltage, line-powered assemblies, seek qualified help.  Also, there may be differences between this schematic and your particular unit.  Be sure to read the notes on sheet 1.

Other notes...

1.  The PFC stage and standby/bias flyback supply are always operating whenever AC power is applied.  To enable full operation, the scope connects about a 200 ohm resistance between J9 pins 3 and 11.  The supply can also be enabled for bench-testing by applying 200 ohms between J4 pins 1 and 2.  (The design of the control input is weird:  If the resistance is too low, or zero, the supply will not turn on.)

2.  The overcurrent threshold for the +5 V output is configurable according to the resistance the scope presents between J9 pins 3 and 10.  In my 54845A, the resistance is 1210 ohms.  It may be different in other models depending on the expected nominal current, or it may vary during operation.  I did not investigate further.

3.  There are separate overvoltage detector circuits for each output, and an additional detector monitoring J9 pin 7, with a nominal threshold of +2.64 V.  I did not investigate the source of the voltage on that pin, but it measured only 220 mV during operation.  If an overvoltage condition is detected on this pin or on any output, the supply shuts down until either AC power or the enable input is removed then reapplied.

4.  There is no output current limiting.  If an overcurrent condition is detected on any output or at the DC-DC converter input, the supply shuts down for about 1 second then attempts to restart.

Of course, I can't resist offering a few editorial comments about the design.

1.  The enable input circuit seems unnecessarily complex, but maybe there was some reason for it that escapes me.  I assume that the 15 mA constant current source is intended to provide wetting current for the mechanical relay contact that enables the supply, but a simple resistor to VCC would have done that.

2.  If the intent of the R-C network in the reset circuit (R292-C239-CR216) was to provide a simple power-on reset function, it doesn't work, because the comparator is inverting, and since  C239 is initially discharged, the low-active output (RESET_N) is deasserted, not asserted, at startup.

3.  Connecting the summing node of an op amp (U203 pin 2, sheet 9/D3) directly to the external wiring harness at J9 makes it highly susceptible to noise and is a poor design practice.

On the other hand, the actual power-conversion circuitry seems well designed and robust.

Incidentally, the actual failure of my power supply was the result of some poor workmanship on the part of whoever repaired it before I acquired it.  The daughterboard is mounted to the main board using right-angle pin headers, and about half of the plated-through holes in the main board had been damaged through negligence.  If you need to remove the daughterboard, make sure you thoroughly unsolder each pin so it's loose in the hole.

Good luck!  C.W.

*** THERE IS A NEWER SCHEMATIC ***  See further down the thread.

[mankan]

Incredible first post!

And thanks, I have a working 54820A scope and a spare parts donor with a faulty power supply. Hopefully they have used the same PSU design in the entire product range. This might be very useful in the future.

[Wallace Gasiewicz]

Thanks for the PS Schematic. 
Did you also put in a new Hard Drive? Lots of them fail. 
Mine has been upgraded to SSD.

[Conventional Wisdom]

Mankan, thank you for the kind words.

Wallace, I haven't had to replace the drive yet, but the SSD is a good idea.  This thing takes a LONG time to boot.  I expect that the challenge will be to find one with the 2 mm IDE interface.

Like everyone else within the last few years, I did have to grind open the DS12887A to wire in a fresh lithium cell.  Amazing how many instrument manufacturers didn't imagine (or care) that their products would be used for more than 10 years.

[G0HZU]

Wow, great work thankyou!

I have a HP 54825A and it has the factory upgrades to make it run better. However, the PSU has been unreliable so this info is really useful!

The physically smallest electrolytic caps in my PSU had all gone high ESR and I've replaced them. There's also some sort of status flag on one of the headers that would trip every so often and you have identified this as a brownout pin. Thanks again for this!

When this pin tripped on mine it would halt the scope and a message about a failed fine interpolator would show on the screen in red text as below:

Fine interpolator gain cal failed: Service is required

This didn't make any sense because the fault was actually in the PSU. Something was upsetting it and thanks to your info I now know something in the PSU was falsely tripping the brownout detection in the PSU.

[Chris56000]

WOW!

. . .What superb professional layout and circuit diagrams you have produced Sir!

I carry out this sort of work myself and I've not been able to reproduce something as neat myself, even with my Abacom SPlan 8, and that is pretty good software in itself!

Could you tell me what you used to produce this please? – your work is better than H.P.'s themselves!

Chris Williams

[Conventional Wisdom]

Hi Chris,

Thank you for noticing!  I've experimented a lot to achieve the "1960s Aerospace" aesthetic, which I grew up with, and which I consider to be the gold standard.

For documentation-only projects like this, I use AutoCAD LT 2000 and draw my own symbols as I need them.  This gives me the flexibility to add accurately-scaled mechanical details, notes, etc. in whatever form I choose.  It just doesn't generate a netlist or provide any of the automatic connectivity/cross-reference/checking features that an actual CAE package does.  The font is the superb TrueType Routed Gothic (thank you, Darren Embry!), which was traced from original Leroy inking templates.

I'm waiting to see if anybody spots the two errors I've found since posting it .  They're trivial, but I'll upload a corrected version later.

[Gerhard_dk4xp]

Now please repeat that for the Vega 450 power supply of my R&S ZVB-8 VNA.    1/2 :-)
That thing is severely over-engineered. There are even Vega-450s on ebay, but
all with different output modules.

---

I have "repaired" my 54750A sampling scope. When its Lithium battery is empty,
it does not boot at all and it says nothing on the screen, maybe the greyish
background. You need T10 and T15 screwdrivers, remove the feet on the backside
and 6 TORX screws, then you can peel away the back plate.
Top slot on the backside is the power supply, then the CPU and the graphics card.
CPU&Graphics are connected with a short flat cable and can be removed only together.
On the CPU is a 3V Lithium cell Sanyo / FDK  CR17450SE or the version with 2 wires.
It took me some searching to get a few. It is impossible to solder wires to the
battery version without them. There is a warning on the battery not to solder but
I tried it anyway. The contacts did not take tin and I did not want to go too far.
With the new battery the 54750A boots again but complains that mainframe
calibration is lost. Calibration is easy, just follow the instructions on the screen.
You'll need a 500 MHz and 2 GHz signal generator with 10 MHz ref output. I used
an SMIQ-4. The 10 MHz must be connected throughout the calibration, the HP
manual is not precise there. On the back side of the mainframe is a RAM protect
switch that must be flipped during cal.

----

I also have a 54846B 2.4 GHz / 8 GSPS scope that had an unreliable hard disk.
Since it has a DVD on the backside, I inserted a Knoppix Linux boot DVD and
it really booted Linux. It is weird to see open office on a scope screen.
I used dd (disk dump) to save the hard drive contents over the network, as long
as the disk was still cool. You could use ddrescue / dd_rescue if the hard disk
has errors right from power up and save as much as possible. I replaced the
disk with one from a Dell laptop, IIRC. The new disk is slower, probably because
of the sector interleaving being too ambitious for the old CPU. But no lost sectors
anymore.

regards,  Gerhard

[bostonman]

That is impressive you traced it all. A few years ago I was working on my Agilent 54831M Acquisition Board and had to try tracing several components on a multilayer board with blind vias.

By any chance, do you know if this is the same power supply used in the Agilent 54831M?

[Conventional Wisdom]

Gerhard:  Sorry, I don't have a failed Vega 450 that I need to fix (or a working one, for that matter).  Thank you for the notes on the 54750A and the 54846B.

bostonman:  Unfortunately, no.  The Service Guides for 54810A/15A/20A/25A/35A/45A/46A show p/n 0950-3252 for this power supply.  The Service Guide for 54830/31/32/33 shows a different power supply p/n 0950-4191, with a different form factor, higher power (440W vs. 300W) and an additional +3.3V output.

For reference, the label on this power supply reads:

International Power Systems Inc.
MODEL:  NX300-6003
        9290600100
DATE:   9741N  REV E

These boards were 2-sided, so tracing them was tedious, but straightforward.  I had to peel off some of the rubbery glue, and experience has taught me that it saves time to just unsolder some of the ICs and reinstall them afterwards, rather than doing a lot of continuity probing to find out where component-side traces go.  Tracing a multilayer board is, of course, much harder, but with the additional challenge of blind vias?  Yeesh.

[bostonman]

Someone on here suggested I use a wire brush to trace connections.

It worked quite well and got me to the general location which I then used a probe to get the exact location.

It wasn't easy, but I managed to trace a few circuits in question.

[asis]

If you're one of them, I have good news:  I'm an engineer, I'm retired, I'm stubborn and I don't like secrets.

Hi,

Exactly 20 years ago I did the same thing, letting the bird out of the cage.
Since then, I have reverse engineered dozens of complex circuits for various equipment.
Unfortunately, altruism and selflessness did not make me richer.
The only thing that brought me satisfaction was the experience gained and understanding of how everything works.
You are very close to me.

Thank you.
-
I am grateful to this forum and its creators for the opportunity to communicate with all of you.

[bostonman]

I am grateful to this forum and its creators for the opportunity to communicate with all of you.

Not to deviate, but I fully agree on being grateful for this site. I owe some complicated repairs to this site that otherwise would have been too complicated to solve without input from others; one of them being my Agilent 54831M scope.

I am also grateful for the general knowledge I've gained. It amazes me that I post a question, and within a short time see not only many users viewed the message, but almost always provide a response in a short time.

Sometimes I sense people butt heads a bit, but it's impressive how users from around the world all come together to help others who they never met.

[Conventional Wisdom]

Hello asis,

Great job on the Delta power supply schematic.  The package reference diagrams are a nice detail.

"The only thing that brought me satisfaction was the experience gained and understanding of how everything works."

Absolutely true.
[old-fart-mode]
Formal education is a fine thing, but I think it's also important to learn by example.  I've been reverse-engineering electronics since I was a pre-teenager, sometimes for a practical purpose such as repair or improvement, but more often out of simple curiosity.  I was fortunate to grow up in an area and at a time when there were nearby shops overflowing with aerospace and military surplus at dirt-cheap prices.  I'd buy interesting items to take home and learn everything I could about how they were designed -- everything from PCBs bristling with TO-39 transistors, silvered-mica capacitors and pulse transformers to a suitcase-sized digital voltmeter with a 3-digit Nixie readout, all-relay logic and a stepper switch sequencer -- and then I'd use the parts to build something else.

Obviously, I didn't earn any money doing it, but I can say with confidence that all that "play time" gave me the perspective and knowledge to be a far better and more valuable "out-of-the-box" thinker later on.  I'm sure there are many here who share these sort of experiences, and in an age where writing a little Arduino code constitutes invention, it is gratifying to me to see so many young people here down in the trenches with the soldering irons.
[/old-fart-mode]

The TOP200 on your schematic caught my eye, so I had to look it up.  See what I mean about curiosity? 

[Conventional Wisdom]

Just a quick update to close the loop:

This rev-2 schematic corrects some minor goofs (e.g., missing SHUTDOWN_N signal at 9/D2) and renames the output voltages to match the values in the service manual (+12.2V instead of +12V, etc.).

[999masks]

thanks for the post. Really appreciate your work i jnow it is very time consuming.
I have a question if you guys can give me any suggestions i will appreciate it.

I have Agilent infinum 54855DSO which i accidentally shorted the ATx connector i belive 12V output.
after that the oscilloscope stopped powering on. Mine has Celestica 5100 power supply with 4 power modules.
The current issue
if i remove the connector from power supply that goes to
analog power generation board and plug the power cord,
it will power on and boot. At this state I measured all the
voltages from power supply which are good. As soon as i
connect that plug (this is not a power cable) it wont turn on. I
measured the voltages from those wires, basically the power
supply has 4 different power modules. 2x3.3V 1x5V and-12 ,
0, 12V the connector that i was referring, carries the same
voltages as their output at ATX connector (maybe POWER
GOOD signal?) basically from that plug if i remove the pin
that connects to -12,0 +12v =24v (any side) the oscilloscope
will power on and boots. At this state the neither power
button or any button works on the front panel. I am asking
if you can help me to troubleshoot or provide the model
schematic or similar. thanks in advance

[microbug]

thanks for the post. Really appreciate your work i jnow it is very time consuming.
I have a question if you guys can give me any suggestions i will appreciate it.

I have Agilent infinum 54855DSO which i accidentally shorted the ATx connector i belive 12V output.

[...]

This really belongs in a new thread.


I am really impressed by the clarity and beauty of the schematic in the original post. If only major CAD companies cared as much about such things.

[ARF]

It's good that I found this article. Thanks for the diagram.

I decided to make a complete re-cap of the power supply of my 54845A. There is no point in waiting for problems and I decided make it in advance.
The 470uf/450V capacitor has the specific contacts with 22.5mm step. I was not able to buy such a new brand capacitor, but I was able to buy a 680/400V. I was sure that this power supply does not have a PFC, but in the diagram I see that it is there and there will be 380V on this capacitor. I'm wondering how safe it is to use a 400V capacitor.

[Conventional Wisdom]

I understand your dilemma.  The original cap was discontinued some time ago, and I couldn't find a drop-in replacement, either.

While it is common to choose an electrolytic with a voltage rating higher than the expected voltage, presumably to reduce stress and maximize life, I'd feel OK about the 400 V cap here, because:

1)  The 380 V is pretty well regulated and unlikely to rise above 400 V for an extended time, and

2)  Although manufacturers generally no longer specify surge voltages the way they used to be (I remember when both WVDC and SURGE were printed right on the capacitor case), I have to expect that there's still a reasonable margin over the working voltage.

IMHO in this PFC circuit it's more important that the replacement cap's high-frequency ripple current rating be at least as high as the original's.  I welcome others' comments.

Good luck - CW