HP8640B in Brittany

[Renaud F5ZR]

Creating myself a "new" lab (i.e. one with old instruments) I read all I could about RF generators.
Of course the mouth watering HP8640B got all my attention. There is so good advice all around. Here I read twice all the post HP8640B Signal Generator by Allan, very inspiring.

So I very recently bought not one but two HP8640B, (yes I know that's glutonny...).
Of course, and that's what I wanted, they are faulty.
The oldest was british made in 1978, the "young" one is american maybe ten years less as the SN is scratched and not readable but IC's are dated 1987.
Visible faults are the usual for them. Broken gears for both and loss of brushes only on the older. (As long as I can see!)
Having found that the US one (coming from a french lab) is complete and with only a thin crack in one gear I decided to plug it. It is working above 250 MHz. Nothing out on lower ranges though the frequency is displayed.
From what I understand it means probably that the first ECL divider is faulty which is also rather common from what I read.

This is where I stand. Any advice very welcome and will get all my appreciation of course.

Renaud

[Radio Wrangler]

You're probably going to need a few more 8640s as parts donors.

The power supplies run hot. Rectifier diodes are a common failure. The OBIC (on-board IC) displays dissipate 1 W per digit so they run hot too.

Biggest failure is user error. If you lift one, your hand curls naturally round the side-frame onto the back panel and clicks the counter reference frequency switch to 'External' and the counter no longer works until the switch is set back to 'internal'

You would be amazed at the number of 8640 that were sent from R&D to the equipment maintenance department for repair..... in the factory which built them. It caught me out too

David

[Renaud F5ZR]

Hello David. Thanks.
Yes, having read really all I could find, I do know I will have troubles.
That said, once opened, the beauty of the building is incredible. I'm also a mechanics (restoring pre WWII cars) so I'm an admirer!
I believe I'm lucky to have got that one in such a good state. It was in a professional french lab. They discarted it to buy a synthetised one. Maybe because of the divider defect. Anyway it is at least cosmetically as "factory leaving". All the commands are smooth, band switch, cavity, even the "Fine tune" works.
I verified that Mouser have the MC10EL31DG to replace à la N2GX the HP EECL.
First I'll have to verify that the fault is really there anyway.

This morning, I hooked the output to my R&S URV5 meter and 50 ohm load.
Got a -1,5 dB for the 0 dBm I adjusted yesterday on the HP's meter but nothing on that meter. Contact cleaner in the meter's switch and plenty of movements cured it.
The URV5 agrees perfectly with the 10 db's attenuator range.

Here's the only mechanical defect I could find. I'll live with it till I must fix it.

[Radio Wrangler]

Maybe a metal band around the cylindrical portion of the rightmost gear in your photo of the differential will prevent it breaking further.

It was many years before synthesised generators had evolved sufficiently to be able to replace the 8640.

David

[Renaud F5ZR]

Originally Posted by Radio Wrangler

Maybe a metal band around the cylindrical portion of the rightmost gear in your photo of the differential will prevent it breaking further.

Yes, I was thinking something like that. Maybe a clamping ring if I can find a very small one.

It was many years before synthesised generators had evolved sufficiently to be able to replace the 8640.

David

And still more noisy I believe.
Renaud

[Renaud F5ZR]

To confirm what I believed about U12 (EECL first divider in the signal path) I swapped the board A10A2 with the one of my other 8640 (UK made 1978). This older I bought first last fall coming from London. I have not plugged it since bought but there's at least an A10A2 working board as now the US 8640 is working down on all bands.
There was some design changes in the 10 years between those.
Images down:
UK from 1978 first and US circa 1988 (made by HP in 1982!). And different references too, 1820-0736 and 1820-2642. Maybe same chip but just a cheaper package? Hope so!
Renaud

[factory]

The ICs above & right are the same number, but lower cost package, the one on the right has a different location number though.

No idea about those with different numbers, maybe lower cost package, but improved in some other way.

David

[Renaud F5ZR]

Thanks David.
The two boards are completely different in fact. Not even the same number of ic's.
My service manual is:
"HP_8640B_Maintenance_Manual 1986" but it refers to the older generator with the posh IC's. I don't have one for the more recent which is faulty.

[Renaud F5ZR]

Here are the two boards.
If someone knows about this new design thanks for telling.
Renaud

[factory]

They have managed to reduce the number of ICs, by combining some of the single divide by 2 ICs into fewer dual divide by 2 ICs, not sure about the higher speed ICs, but this has been done with the lower speed ICs, haven't got access to the later manual at the moment.

David

[Radio Wrangler]

Motorola's ECL families were offering the required speed and noise characteristics by the 1980s, so it became feasible to do a cost reduction re-design compared to HP's in-houde fab. Also, manufacture of in-house designs got contracted out. The earlier board with the solid groundplane has a 1981 date code.

Note that some of the dividers had to run continuously, irrespective of band, to pre-scale for the counter.

David

[Renaud F5ZR]

Originally Posted by factory

The ICs above & right are the same number, but lower cost package, the one on the right has a different location number though.

No idea about those with different numbers, maybe lower cost package, but improved in some other way.

David

Thanks David but the first divider U12 changed its reference so I'll have to find the right manual for the more recent as I could need it somewhere else anyway
I was said that Artek has it for my board epoch.

A date for the second board would be nice David! (Radio Wrangler)
Thanks,
Renaud

[Radio Wrangler]

There were several changes and I didn't keep any notes, so it's all out of memory. I didn't work on the 8640, but I used them extensively and my desk was about 30 metres from the 8640 production line!

I collected enough cosmetic reject bits to build myself an 8640 cavity. I was intending it for a communications receiver local oscillator with motor drive controlled by a synthesiser.

David

[Renaud F5ZR]

Originally Posted by Radio Wrangler

Note that some of the dividers had to run continuously, irrespective of band, to pre-scale for the counter.
David

I read again that which iI didn't know and is interesting.
The counter on the 8640 (with it's original faulty divider board) would give no output on ranges under 250 MHz but the correct frequencies would be displayed on the counter. It means obviously that all the dividers work on this side. So the fault can't be on the divider chain but somewhere between the oscillator and the filter bank. But how for all the lower bands? Aren't they independent?
I'll have to investigate!
Renaud

[factory]

It turns out there are at least four versions of the A10A2 RF divider board, I've combined date code info from pictures of boards on the web and the messy manual pdf from 1991, the actual info the OP needs is in the 16th Feb 1984 manual changes pdf, not sure where I found that pdf, but I've added it to the files section on HPAK groups.io;

A10A2

08460-60023 D-1339-4, D-1441-4, E-1712-4
08460-60092 service exchange by HP, rebuilt of above 08460-60023

08460-60370 A-2118-4, B-2314-10
(change 5, s/n 2118A to 2403A, from manual changes dated 16th Feb 1984)

08460-60383
(change 27, s/n 2404A, s/n 2408A, from manual changes dated 16th Feb 1984)
(s/n 2520A to 2828A)

08460-60483
(s/n 2835A and above)

David

[factory]

The relevant pages from the 1984 manual changes for the 08460-60370 A10A2 board;
A10A2 08640-60370 circuit changes pg8.pdf
A10A2 08640-60370 circuit changes pg23.pdf
A10A2 08640-60370 circuit changes pg24.pdf

David