Prototype HP/Agilent GPS based refererence clocks generator

[mmagin]

Hmm.  Some aspects of this are reminiscent of the oldest of the GPSDOs I own.
The front panel says Z3804A Oscillator and Distribution Amplifier, and *IDN? says
HEWLETT-PACKARD,59551A,Prod. Eng.,3618-A

It also suffers from the 1024 week bug, and on the inside it looks like they grafted a GPS board onto a product that was previously just a nice OCXO and 10 port 10mhz distribution amplifier.  Also I can't seem to get the holdover uncertainty below about 19-20 us.  (My newer Samsung-rebranded Z3805A settles down to 800-900 ns)  Nothing about it is as prototype as yours, but it feels like a smaller volume thing for a particular customer.

On the other hand, lots of 10mhz outputs and a line voltage supply, so I use it.  Also IRIG-B output, which I bought a cute little Datum Timeview display for.

[TheSteve]

Sounds about right, this has got to be one of the first. In your case I assume the SCPI commands work properly to set the location and such. This one wouldn't be too bad if it just took forever to find its location at power up but being they set to hold position only it will never lock on its own. I figure I can remove the 28C64 that has the data and try to edit it manually or add a microcontroller that modifies the position packet sent by the 58501A to the GPS receiver. Or perhaps at some point I'll find more SCPI commands that work, but that seems unlikely.

[mmagin]

Yeah.  Resetting it to defaults and then setting the time I got things to do a position survey and get into a correct state.  It works mostly like the newer HP GPSDOs you can find documentation for on the net.  Although the serial port occasionally becomes unresponsive (like now) and there is another serial port which is a mystery to me.

I would take some pictures of the insides if I didn't have it in operation :/

[TheSteve]

I had looked for a second serial port with no success. I'll have to check again, being such an odd/old unit for all I know there is another one that will give me some management options.

[TheSteve]

Well some progress was made - the stock power supply is fixed. It was a royal pain in the butt and a replacement off ebay would have been cheap but there is that whole principle thing.
It was mostly dead but worked once briefly - so I was thinking bad caps as it is from the 90's. I checked the caps in circuit with an LCR meter and they all met spec. I figured I'd resolder the board and when I did the odd connection sizzled a little and smelled bad - leaky capacitor bad. Sure enough a few of the caps had started leaking and the nastiness they leaked was conductive. Again though they meet capacitance and ESR specs just fine, even after being removed. I replaced a few caps and gave it a try, at this point it worked great again. So I fully reassembled it and it promptly blew the fuse and the main switching power FET. It was long discontinued but I found a local seller on ebay that had a few and met up with him to get them. I then replaced every single cap, leaky or not and replaced the FET. It was totally dead still. I removed several parts to check and finally found a bad MPS2222A. With the MPS2222A replaced the supply is again working fine.
btw the leaky caps are made by Nichicon and were spec'd at double or triple the working voltage so I am a little surprised they are failing like this. The power supply is still sold brand new many places and is far from cheap. It is a Powerone MAP55-4003.

As for the unit itself the location is clearly hard coded in the 58501A's 28C64 EEPROM. Next Digikey or Mouser order I'll get a PLCC32 socket so I can try reprogramming the eeprom myself(we'll see if it has checksums). I have also ordered a Motorola Oncore UT Plus GPS off ebay to replace the horrid one that is in the unit. The original uses RS232 and the new one uses TTL but the 58501A mainboard happens to have jumpers for the GPS TX/RX lines to switch between RS232 and TTL.
If I power the unit on, let it initialize the GPS, then manually reprogram the location in the GPS to my coordinates and then reconnect the jumpers back to the 58501A it will lock as long as it is receiving satellites. At times it locks 5 but quite often it is only 1 or 2. Its sensitivity is clearly in the crapper.

I also looked for a second serial port without success and found no other commands that work.

[mmagin]

Mine has the same supply.  It was working until I accidentally dropped things a couple inches on the bench and the physical shock killed it.  Seemed like it should not have died so easily, but maybe it was merely precipitating other problems.  In my case, I waited until one reasonably priced showed up on ebay.

I've been getting better at diagnosing switching supplies though.  Didn't give up on the supply in an HP 8648B until I discovered a transformer winding had opened up!  Sadly those seem to be unobtainable for reasonable prices.  I'm tempted to modify it a bit and just put suitable DC power supplies in a separate box.  It's not such a high end signal generator that I'd feel bad about the modifications.

[robert_]

Nichicon PL?
Same as Chemicon LXF and Panasonic HF* , these love to leak after 20years+ because the sealing rubber degrades. Doesnt matter if used or not, they also will leak in their original bag, given enough time.

Chemi-Con LXY is a good replacement for those, KY might be a bit low ESR for some badly-designed PSU of that time.

[TheSteve]

Mine has the same supply.  It was working until I accidentally dropped things a couple inches on the bench and the physical shock killed it.  Seemed like it should not have died so easily, but maybe it was merely precipitating other problems.  In my case, I waited until one reasonably priced showed up on ebay.

I've been getting better at diagnosing switching supplies though.  Didn't give up on the supply in an HP 8648B until I discovered a transformer winding had opened up!  Sadly those seem to be unobtainable for reasonable prices.  I'm tempted to modify it a bit and just put suitable DC power supplies in a separate box.  It's not such a high end signal generator that I'd feel bad about the modifications.

If you do decide to try to fix the original I'm happy to help. Most of the parts it uses are pretty common. The PCB is delicate though and easy to damage. At this point it likely needs all new caps. You can buy a power supply out of china that will work for so cheap though, it is almost hard to justify.

I paid 41 USD plus shipping for the 58501A so for me it is worth the time and is a cheap/fun project. I also didn't want the power supply to "beat" me...

[mmagin]

Mine has the same supply.  It was working until I accidentally dropped things a couple inches on the bench and the physical shock killed it.  Seemed like it should not have died so easily, but maybe it was merely precipitating other problems.  In my case, I waited until one reasonably priced showed up on ebay.

I've been getting better at diagnosing switching supplies though.  Didn't give up on the supply in an HP 8648B until I discovered a transformer winding had opened up!  Sadly those seem to be unobtainable for reasonable prices.  I'm tempted to modify it a bit and just put suitable DC power supplies in a separate box.  It's not such a high end signal generator that I'd feel bad about the modifications.

If you do decide to try to fix the original I'm happy to help. Most of the parts it uses are pretty common. The PCB is delicate though and easy to damage. At this point it likely needs all new caps. You can buy a power supply out of china that will work for so cheap though, it is almost hard to justify.

I paid 41 USD plus shipping for the 58501A so for me it is worth the time and is a cheap/fun project. I also didn't want the power supply to "beat" me...

One winding of the transformer was definitely open.  Do you have any suggestions for such problems, aside from a painful rewinding job?  It was my impression that the transformers in most off-line switching supplies were custom parts.

[TheSteve]

I removed the 28C64 EEPROM today and installed a PLCC32 socket(what a pain!). The EEPROM does have a bunch of data in it, it also appears to be checksum protected. I changed two bytes to edit what I believe is the fixed elevation value ($698F) as a test and it then ignored the EEPROM. On the plus side with the EEPROM removed or blank the unit powers up and seems to work. It will use the last position stored in the GPS receiver and acquire/lock within 10 minutes. So while it would be nice to figure out everything in the EEPROM and alter the fixed location perhaps it is best to just leave it blank. I also checked that if the EEPROM is blank it will never write any data to it even after it locks and switches to position hold mode. Lastly with the EEPROM removed/blank/corrupted etc there are still no further SCPI commands that work.

Not sure what else is all in the EEPROM, there is quite a bit of data considering it runs without it - I may be missing some useful setting data for the OCXO, that would be a shame to lose.
I have attached the file in case anyone is interested in taking a look. The fixed coordinate data I know is stored in it is:
37.35865472 -79.17463639 270.23

[mojoe]

I just looked up the GPS coordinates that your unit was hard coded to. Very interesting location.

Harris has two facilities in Lynchburg, VA. One of them is for training and tech support. I have been there a few times myself. As I understand it, the history goes something like this: The GE division that sold radio systems was a joint venture with Ericsson. GE later sold it's share to Ericsson, which was later sold to MA-COMM, which was later sold to Tyco, who very quickly sold it to Harris. Our EDACS system at work has all five names on various pieces of equipment.

The coordinates you listed places the location on the campus of Liberty University. Perhaps GE/Ericsson was doing some R&D in cooperation with the university.

[TheSteve]

I was thinking maybe the entire facility was HP engineering back in the 90's and it was sold to the university at some point. I think it is neat you can see the equipment shed where the antenna would have been mounted.

[uncle_bob]

I just looked up the GPS coordinates that your unit was hard coded to. Very interesting location.

Harris has two facilities in Lynchburg, VA. One of them is for training and tech support. I have been there a few times myself. As I understand it, the history goes something like this: The GE division that sold radio systems was a joint venture with Ericsson. GE later sold it's share to Ericsson, which was later sold to MA-COMM, which was later sold to Tyco, who very quickly sold it to Harris. Our EDACS system at work has all five names on various pieces of equipment.

The coordinates you listed places the location on the campus of Liberty University. Perhaps GE/Ericsson was doing some R&D in cooperation with the university.

Hi

The GE land mobile radio group in Lynchburg goes back to the WWII era. For many decades it was number 2 in that business to Motorola. With the advent of trunked systems and then cell phones, they became less important to the GE empire. They slowly spun off from them.

Bob