Agilent/HP 8648 B/C/D Power Supply

[USMC_Spike]

Have the HP 8648 b/c/d unit with what looks like a fried transformer.

I've searched for the Computer Products company in Hong Kong, but
couldn't find them.  The makers of the transformer.  The P/S is
china made.   

Here is a link to a similar spec transformer, with a the HP part number
0950-2293.  The transformer has four output voltages,
+5V, +15V, -15V, +38V. 

That is the first step anyway,
then it looks like I should replace the
discreet devices on the same board
along with the electrolytic and diodes.

Those are my starting thoughts anyway.

The whole unit stinks and is covered in a
thin layer of burned black soot.  So I've
been busy cleaning it with 99.99% ISO.
It will take a bit longer, but I'm getting
there.

Yes, I have the manual.

Kind of a pisser I'd say. 

[bd139]

Keysight might still sell that part. It'll probably cost more than a new unit though.

I've had HP spares from them a decade after the product was discontinued.

[USMC_Spike]

Hi BD,

Do you mean a NEW unit as in new power supply or or a whole new (used) 8648 b/c/d replacement found in the market
somewhere?

[bd139]

A new transformer from Keysight, if they have it, will probably cost you more than a whole replacement device on the used market. Their spares are crazy high price.

You may be able to bodge something up with a couple of separate toroidal transformers instead. If you can, I'd confirm that everything past the power supply works with a bench supply (or three) first. I lost my RF gen to a pass transistor failure which took out all the logic and the transformer

[alm]

Spare prices and spare parts are usually set when the instrument is new and do not go down, unlike prices on the used market. So a power supply for a $5k MSRP unit might cost $500, even though a decade later on the used market the instrument is only worth $300.

Depending on the current needed for the various rails, I might consider getting a toroidal transformer to supply the high-current (5V digital? +38V?) and supply some low-current rails from extra windings you add around the outside of the transformer. Alternatively, multiple transformers, depending on the mechanical space. You are probably not likely to find an exact replacement outside Keysight. If you feel up to it, you could also consider winding the entire transformer from scratch. But winding with the kind of heavy wire used for high-current windings is not my idea of fun. I have unwound a toroidal transformer partially to get the voltage I wanted, and even that was fairly annoying due to the stiffness (this was a winding that could supply about 3A).

[Armadillo]

It is not possible to rewind yourself?

[deepskyridge]

I have a 8648A that had the same shorted xfmr in the power supply. I ordered a new SMPS from Mouser as a replacement. Has been working great since then.

Gary

[USMC_Spike]

Do you recall what the ratings were or the part number from mouser?

Hmmm thinking here. 

@Armadillo, no, not possible to wind for me.

A former colleague could have, but he died last year.  When I inquired
about the transformer winders that he and his buddy had (a couple of
months back) the family gave it to one of colleagues friends.

It might be possible to have a rewind from elsewhere.

@ Alm, your right it's about $475 from Keysight.

I'll have to check and see what the rail voltages are
but I don't think they show current draw....

BD139, yes they have spares...and some you have to
send them the instrument to install themselves....

[USMC_Spike]

Here is the ebay power supply, looks the same as my fried one:
As I look at that, is it, 2.6 - 1.3A,
1.3 amps.
is 2.6 the version?

or was it 6-1.3A,
Six sections of 1.3 amps each?

[USMC_Spike]

As I'm working to pull it apart,
desoldering the big filter cap,
the transformer then the smaller
parts in the immediate vicinity.

I'll be cleaning some more and
measuring all the parts.

The main cap is 2 pf, with ESR of 15.5Mohms.
I didn't see any burst our outlet, so that makes
me wonder about transformer.

[Gertjan]

Hi USMC_Spike,

I think the sticker on the transformer in the Photo has this meaning:
On 100V Mains the supply draws 2.6Amps, on 240V Mains it draws 1.3Amps from the Mains.

So it doesn't say anything about the secondaries.....
The type number to search for is: NSF96-7630.

Regards, Gertjan.

(edit: typo corrections)

[USMC_Spike]

Hi Gertjan, I tried searching for that number and didn't find it.
check out the power transformer images I did find.

I went back and looked at the board layouts and
found the A2 power supply had only +15V,-15V,+38V
on that board,  then it sent to the mutha' board A3.

Now on A2J2, pin 1 is x...through w4 pin 1 is x
until it reaches, A3J6 pin 1 and it's magically +5V.

So that much be created in the Post Regulator
from pins 1&2 are spliced together at the A3J6.
That also connect through to RS232 and is the +5V digital.

There must be a chip on the post regulator that converts it.

So much for the HP literature. 

[alm]

Does the s/n range of your service manual match your instrument? Did you check the backdating section of the manual?

[USMC_Spike]

ALM,

Interesting.  I just looked and it says the following:
Serial Number Prefixes:
This manual applies directly to the following model/serial perfix
combinations and below:
8648A.......8648B.....8648C
3847A/U..4037A/U...4037A/U

It's suppoesed to be a 8648B,
but the prefix is 3847A....
which could explain why the power supply fried.

I'll post a pic of the Unit's front....I thought it looked
kind of funny there, it might explain why the power supply fried.

But don't take my word for it, here are the pics:

Anyone know if Agilent/HP made transition 8648s
that is B models with the A prefix?

If not, then this might be a case of fraud. Maybe it is happening
to other people too.  Maybe that is why we have power supplies frying
on the HP8648 line.  Unless this is a major screw up from HP.

Can these even power up with the main filter cap at 2pf and ESR of 15.5M Ohms?

Maybe it worked when pics were originally taken and then poof, all the smoke
went out after they switched things around.

Another F-ing Ebay....problem.
Where broken test equipment
gets sold as working.

[alm]

The manual applies to serial prefix 4037A/U and below. Nothing to indicate that a 8648B can not have serial prefix 3847. 3847 means that the last design revision was in 1998 week 47. 4037 stands for 2000 week 37. Note that this is the date of the last substantial design change. Actual manufacturing date could be between 1998/47 and 2000/37.

[USMC_Spike]

Alm, thanks for the explanation.

[carl_lab]

Have you found a schematic of that SMPS for 8647A/8648A/B/C/D/?

AFAIK they all use the same SMPS:
-> https://www.eevblog.com/forum/repair/hp8647a-rf-generator-'504-modulation-cal-reastore-failure'/msg1265269/#msg1265269

[USMC_Spike]

Hi Carl,

No, I've not found the schematic yet.
And thank you for the link to your thread too.

I'm still in the cleaning and getting the
burned smell and particulates out of the boards.

[scopeman]

This response is not about your power supply but on the 1EA 100mW option fix. Unfortunately I was not able to fix my module but was able to find one on ebay. I replaced the module and fixed the problem.

Were you able to fix your power supply issue.

Sorry about the latent response on this. I don't check here as often as I should.

Sam
w3OHM

[Totus]

I've got HP8648 PSU with damaged primary side components (resistors and semiconductors).  Some of them are burnt or exploded so badly it's impossible to identyfiy their values.
Could someone please post a good quality picture (sharp, neutral light) with visible bar codes on primary side resistors. The ones critical for me are R3, R4, R6, R7, R35, R36, R38 and Q5.
The PSU board is described as Computer Products Inc, 510160 Rev F. I've found few pics and videos on internet but they don't help.
Thanks in advance.

[fenugrec]

Just some info for future reference. OP probably moved on to other projects...

The ones critical for me are R3, R4, R6, R7, R35, R36, R38 and Q5.
The PSU board is described as Computer Products Inc, 510160 Rev F.

I'll post a photo later. I have the unit out right now (label on the transformer says NFS96-7630 ), assuming you have the same off-the-shelf power supply as in my hp 8643A.

I can't read most of the silkscreens (hidden under the resistors), if you point out on a photo which ones you need I can verify.
R3
R4
R6
R7 240R
R35
R36
R? right beside and identical to R38
R38 : 2k ? measures 1k9 incircuit
Q5 ; 2N3906

[dkrishnagopal]

I have a Agilent model Signal Generator 8648A. The Power supply part number 0950-2293 (appearing to be familiar to many forum members),  has failed. Where can I get used or new spare power supply. Krishna Gopal, Space Pulse Techno Solutions, India. Can I get a schematic for the Power Supply. It has 240 /110 V0lts AC input and gives 36/+15/-15 Volts output.