My poor mans SMU - The Agilent 66311B

[nctnico]

Feet can be bought from Ebay for less money then Keysight is asking. Isn't there some generic knob which will fit?

[SKPang]

I've ordered the knob from Keysight UK. £2.16 + VAT with free delivery.

Paid by credit card and thats it. No minimum order. Should be here this week.

[ve7xen]

So I've done some additional testing/calibration, and my box does great as a power supply, well within spec per the verification procedure after calibration. However the current sink seems to have failed short. Though this would seem to short across the output rails, so perhaps it is part of the control circuitry for the 'downprogrammer' that's the problem.

In trying to track this down I'm finding this is by far the worst piece of HP test equipment to work on I've ever seen. All the component designators are hidden under the components, many of the test points called out in the service manual are unreachable with normal probes, and no schematic or component placement diagram seems to be available. It looks like I'm going to have to remove the A1 board to do further testing, which is also a much more arduous chore than most HP gear.

Ah well, at least it makes a great power supply if nothing else.

Keep in mind that while it can sink current, it is not programmable. The current sink will be around 1 to 2.5 A (depending on voltage at output terminals), regardless of programmed current limit.

Thanks I will look into this when I return home next week. From my reading of the verification procedure it seemed that it was programmable, and the block diagram does seem to show a constant current loop around the downprogramming regulator, so I'm not so sure that's the case, but you're probably right. I was seeing the 66311B put my 3.5A power supply into CC mode at 3.5A, while the display on the 66311B would show the setpoint on the current readback, which seemed a bit weird to me.

@stuartk Not willing to coordinate, but if you are I'm in for a couple of items .

[macboy]

So I've done some additional testing/calibration, and my box does great as a power supply, well within spec per the verification procedure after calibration. However the current sink seems to have failed short. Though this would seem to short across the output rails, so perhaps it is part of the control circuitry for the 'downprogrammer' that's the problem.

In trying to track this down I'm finding this is by far the worst piece of HP test equipment to work on I've ever seen. All the component designators are hidden under the components, many of the test points called out in the service manual are unreachable with normal probes, and no schematic or component placement diagram seems to be available. It looks like I'm going to have to remove the A1 board to do further testing, which is also a much more arduous chore than most HP gear.

Ah well, at least it makes a great power supply if nothing else.

Keep in mind that while it can sink current, it is not programmable. The current sink will be around 1 to 2.5 A (depending on voltage at output terminals), regardless of programmed current limit.

Thanks I will look into this when I return home next week. From my reading of the verification procedure it seemed that it was programmable, and the block diagram does seem to show a constant current loop around the downprogramming regulator, so I'm not so sure that's the case, but you're probably right. I was seeing the 66311B put my 3.5A power supply into CC mode at 3.5A, while the display on the 66311B would show the setpoint on the current readback, which seemed a bit weird to me.
...

I'm looking at the user manual, "USER’S GUIDE Agilent Model 66111A Fast Transient DC Source Agilent Model 66311B/D, 66309B/D Mobile Communications DC Source", Agilent Part No. 5964-8125, Page 144 "Supplemental Characteristics". The note given for the sink current states:

1 The sink current decreases linearly from 2.8A @ 0 V to 1.2 A @ 15 V. Sink current does not track the programmed current.

Also on page 26,

Figure 2-1 also shows a single range ? two quadrant capability. This means that the dc source is capable
of sourcing as well as sinking current over the output voltage range from zero volts to the rated voltage.
This negative current sinking capability provides fast downprogramming of the output of the dc source. It
can also be used to sink current from a battery charger, thus providing battery charger test capability. The
negative current is not programmable, and varies linearly from approximately 1.2 amperes at the full
rated voltage, to approximately 2.8 amperes at zero output voltage.".

This really is too bad.

[mtdoc]

It is too bad about the lack of programmable current sinking.

Yesterday I replaced the loud fan with a quiet Vantec temp controlled fan. Much better!

I also ordered the Phoenix connector from Digikey. I think I'll hold off on front terminals.

I noticed NI makes a driver for this unit.  I wonder if anyone has developed a LabView VI for this?  The Keysight software looks nice but just way too expensive.

[TheSteve]

So it seems that Keysight Canada has a minimum $50 order.   
I'm not certain what shipping is

The parts I need are:

Feet: 5041-8801 $19 CAD (I'm assuming that it contains 4 feet, not 1 foot although you never know) The fellow at Keysight told me he didn't know if it was 1 foot or 4 feet as the parts don't show up on their system. He said it's an Agilent product and referred me to Agilent who said that it's a Keysight device and part and they don't sell parts....

Knob:33120-87401 $3.67

Is anyone coordinating a group buy to get over the minimum?

I'm certainly happy to pay for a share of the order + shipping + handling + pain and suffering

Stuart

Once you hit 50 dollars the shipping is free from Keysight Canada.
This is the foot package I bought for something else but ended up using with the 66311B - check out ebay # 151996686789
The metal flip out legs work very nicely, and you need the locking tabs.

[stuartk]

Yesterday I replaced the loud fan with a quiet Vantec temp controlled fan. Much better!

Out of curiosity, what do you do with the 3rd wire (yellow) on the Vantec fan?

[mtdoc]

Yesterday I replaced the loud fan with a quiet Vantec temp controlled fan. Much better!

Out of curiosity, what do you do with the 3rd wire (yellow) on the Vantec fan?

Coiled it and zip tied it out of the way. Could also be cut off but just in case I re-purpose the fan at some point.  The yellow wire is for RPM detection on a computer motherboard I believe. It is not involved in the temp control.

[samofab]

Mine arrived today. It was described as "doesn't power on". Two fuses later (one soldered) and some TLC and it works as new. Sometime in its history it fell hard on the front panel. Miraculously, there was no permanent damage. Only transformer retaining sheet metal was slightly bent.

[SKPang]

The knob has arrived. They shipped it in a massive box.

The unit looks a bit nicer now. 

[gslick]

The knob has arrived. They shipped it in a massive box.

The unit looks a bit nicer now. 

You only paid £2.16 + VAT with free delivery for the knob in an antistatic bag in that huge box? How much did they lose on that transaction? 

[SKPang]

Yes, a total of £2.59 I think the shipping cost is more then that.

[Muxr]

I am going to have to 3D print mine since Agilent won't sell me one.

[MountainEEV]

Mine just arrived today and appears to be intact.  Powers on without errors but fan is not spinning at all.   I have done limited testing so far as I'm waiting for phoenix connectors to arrive.

[nctnico]

The 0.05% accuracy spec seems to be driven by the limited voltage setting resolution and probably also the large output current but I wanted to know how stable it is over a longer time period like one hour with a very light load attached. I took two measurements. The first one is at 5V where the measurement is started after the 66311B has been on for 15 minutes, the second measurement is at 500mV and includes the 66311B heating up. The measurement is taken with a 2200uf capacitor across the DVM inputs.

At 5V the variation is less than 165uV:


At 500mV the variation is less than 125uV:


At least the drift doesn't get significantly better or worse with higher or lower output settings. Still I wonder if the temperature drift could be improved by putting a small heater on top of the voltage reference in the 66311B or that the drift is mainly due to temperature dependant offset voltage drift in other components.

[TheSteve]

Looks like I have an 66319D coming. It has dual outputs and the volt meter option. It can also do battery emulation. Should be fun to compare it to the 66311B to see what the differences are. I will also post the firmware.

[ve7xen]

So a lesson in getting too overzealous with repairs...

Today I decided to poke around the front panel board and try and get a bit more brightness out of my VFD by running the filaments hot for a few minutes. While probing around looking for the filament supply, I accidentally shorted the last two pins in one of the groups on the VFD - something at +12V and a driver output or micro pin or something. Killed the whole front panel, just junk on the display.

Of course I couldn't put it down and think clearly or be patient, so i attempted to use a heat gun to desolder the VFD and inspect the circuits on the other side, with the intent of replacing whatever part I'd fed 12V up the backside of. Of course the heat gun is not the correct tool for the job, and I now have a ruined FFC connector and several random parts unintentionally desoldered - but the VFD remains.

If anyone has a picture of the top side of the front panel PCB, without the VFD installed, or a schematic, or even a BOM with part designators, that would be extremely helpful. I don't think there is component level information available on this box, but maybe there are other models using the same front panel? Also knowing what 8-pin FFC connector would work would be nice, though there is an unused one on the mainboard I can steal.

[stuartk]

I know how you feel. A few days ago I trashed a $60 TFT touch panel and a 100 pin Pic32MX while probing voltages not realizing that my multi meter probes were in the current slot instead of a voltage slot. 
Desoldering the pic and 5 blown SMD IC's was a real joy

I think it's worthwhile to fire off an email to the guy on ebay we bought these units from. For all we know he might have some junk units too bad to sell that may have parts. He seems to have acquired dozens of them anyways. You should post a pic of the damage.

[ve7xen]

And the sad story continues...

After painstakingly removing the VFD properly, I began the process of determining where the parts that fell off might live, there are no silkscreens, and which parts were likely damaged and would need to be replaced.

The board consists primarily of an Intel TN87C196 16-bit microprocessor (OTP) and two Allegro UCN5812 VFD drivers. There is also a MAX773 DC-DC converter (to generate the 12V rail), and some other miscellaneous components. I decided to power up the front-panel board from a bench PSU and try to determine which part was bad. This was easy enough, it's just a 5V rail. At 1A, the supply was still in CC mode at about 2V so I figure there must be a short. Not finding anything obviously getting hot other than the DC-DC converter, which made me suspect one of the 5812s (which I would have guessed anyway), I decided to proceed with resoldering the components that fell off.

The ICL7660 was one of the desoldered parts, and studying its pinout it was pretty easy to determine which space it went in due to the boost capacitor's placement. It was replaced. As I did not want to fully reverse engineer the circuit, I thought I would power the board up again to measure around the pads for the dual FET and determine its orientation. In the process of doing so, Mr Bozo struck again, and I reversed the power supply to the board. Unlike many of HP's well engineered products, the board has no reverse power protection, not that it would normally need it. Getting that dreaded feeling, I went to steal the "spare" FFC connector on the main board to connect it back up. I had previously noticed that the main fan doesn't start for 15s or so when the front panel is disconnected, so I could test the board that way.  I now suspect the main controller has been killed too, as when it's connected to the main board, I get the same delayed startup that happens when the front panel was disconnected, so it seems that the main board can't communicate with it anymore (this was, I think, still working after my initial blunder).

So I'm left in a difficult position on repairing this. I might be able to get a new TN87C196, there seem to be some on eBay that are claimed to be new, however getting the code is likely impossible. I think it's nearly certain that it's dead, so my options are more or less to find a replacement front-panel board from a donor unit, or reassemble it and sell it on as for parts. I don't think it's worth ordering the likely parts I killed the first time around at this point, unless someone can give a story where a 1990s microcontroller survived a reverse voltage for 30s.

I'm still going to try repairing the front panel board as best I can to take some high-res photos for anyone else that may come across having to repair one in the future. Stay tuned.

[ve7xen]

So I might have actually been saved on this one...the main power supply to the MCU runs through one of the passives that fell off the board and I hadn't reinstalled yet. I realized that I couldn't have tested the front panel board that way after my initial blunder, since the FFC connector was fried. After replacing the passives in the correct spots and powering up through the main unit, I don't get the long fan start delay any longer, it's a more typical 2-3s. The box still doesn't respond on RS232, which is a bit odd to me, but progress!

After powering up that way, I found the lower-left 5812 gets hot, and it was the part I suspected based on what I'd shorted anyway, so a definite direction now. I will order all the candidate ICs - I think they are all pretty easily obtainable and proceed with a repair attempt. The reverse voltage might have fried one of the other bits, but they're cheap and easy to replace compared to the MCU.

Attached is an annotated photo of the board for anyone who comes across this thread in the future.

I've posted the full resolution of that image, plus some other photos, at http://www.gotroot.ca/66311b . A blog post will probably come eventually, if I get this thing working again.

[ve7xen]

I gather everyone is probably tired of my blunders and repair attempts, but one final post...

tl;dr after 6 hours of work tonight, my 66311B is fully operational and now has front-panel outputs!



The longer story is that the first thing to replace was the FFC cable, since I ripped that. Parlex 100R8-305B worked, though it's slightly shorter than the stock cable and more care was needed when re-attaching the front panel. Yet another thing I don't really like about the mechanical design of this box is that the flex has to go around a fairly sharp bend as it leaves the front panel, which created a kink at the stiffener and made it easy to break.

First things first, I replaced the obviously burnt UCN5812 with a new HV5812. Having never soldered PLCC by hand before, I didn't realize that this package is the spawn of Satan. After some fiddling around I got that installed. Added two bodge wires to the VFD pins in place of the pads I had damaged while removing the dead IC with a poor technique. A lot of hassle would have been saved if I had a hot air station...

At this point I also elected to replace the MAX773 DC-DC driver (which actually creates a 40V supply, not 12V as I had originally thought), because I thought it was also dead since there was no output on the regulated output pins. In hindsight the removed part may be functional, since the DC-DC in this design uses an external drive transistor and the internal outputs are not used.

These being the parts I suspected most strongly, I decided to give the board another try at this point. With the VFD sitting free and some pressure applied, I could now get all segments to light when powered on. There didn't seem to be any indication of activity on the display, just all segments lit. An hour or two of probing later I was pretty positive that the power supplies, drivers, and microcontroller were working properly. So being stupid, I decided to try soldering the VFD back on, just in case some pins weren't making contact and causing the all-segments issue. Of course this failed, and I had to desolder the thing again.

Now we're here:



I should have just replaced the pesky LM339 to begin with, I purchased one and it was cheap and easy insurance. The one on the board seemed to be working, but I replaced it anyway. Immediately the VFD showed signs of life, though I wasn't able to get a proper display until it was soldered down. There wasn't much else left to do. During my previous probing I had already determined that I was pretty sure the transistors and other ICs on the board were working perfectly, so pretty much the last ditch was to power it up and cross my fingers. And lo and behold! It's alive!



So on to the front panel output mod. In my rummaging looking for replacement parts, I had found a pair of these Pomona 72913 jacks in red and black from a long abandoned project:



They were very close to fitting perfectly in the existing front panel cutouts. I did have to enlarge the holes by 1mm or so, but it was an easy fit. Somewhat unfortunate that they are 'safety' jacks, but I felt I would have a hard time mounting the traditional type due to the depth and size of the holes in the plastic. Plus I had them on hand. With the holes enlarged, they mounted easily. I chose to solder directly to the bottom pads of the rear output connector. My first whack at this had the wires interfering with one of the mounting standoffs, so I had to redo it to avoid that. My choice was to bring the 'remote' sense lines from the connector to the front panel and short them directly to the output there. Without adding a switch, this means I won't have remote sense capability, but that was a feature I was unlikely to ever use anyway.

With these jacks there is not much clearance between the chassis and the bottom jack, but it's enough. I would recommend filing the sharp edges off the chassis or adding some rubber or something to avoid damage to the wire.



Finally I replaced the broken switch on the rear panel for calibration inhibit. And now I think I'm done breaking things. Time for an Old Fashioned . Perhaps a fancy knob will be in my future.

Hope this helps or that someone has enjoyed my repair log, even if maybe I should have created my own thread .

[stuartk]

Ve7xen That's a really impressive save

I'll likely organize a group buy for knobs and any other Keysight parts when I'm back to the office in a week..

Don't bother getting the feet from Keysight. I contacted them and they are 19$ Cad per FOOT..  Buy them on eBay instead..

More details to come. You can pm me if you are interested. My apologies if I don't respond for a week. Canada is a bit expensive to ship from, so probably this group buy would be best for USA and Canada

[macboy]

...Having never soldered PLCC by hand before, I didn't realize that this package is the spawn of Satan. After some fiddling around I got that installed. Added two bodge wires to the VFD pins in place of the pads I had damaged while removing the dead IC with a poor technique. A lot of hassle would have been saved if I had a hot air station...

Good work on the successful repair.
Next time that you need to replace a PLCC, remember that you are trying to save the board, not the dead device. Clip all the leads off close to the package, remove the legless black blob, then desolder pins one by one easy as pie. Use some braid the clean up the pads, and you are ready to solder on the replacement, which is a much easier task than removal.

stuartk: I am definitely in for one knob in a group buy.

[ve7xen]

...Having never soldered PLCC by hand before, I didn't realize that this package is the spawn of Satan. After some fiddling around I got that installed. Added two bodge wires to the VFD pins in place of the pads I had damaged while removing the dead IC with a poor technique. A lot of hassle would have been saved if I had a hot air station...

Good work on the successful repair.
Next time that you need to replace a PLCC, remember that you are trying to save the board, not the dead device. Clip all the leads off close to the package, remove the legless black blob, then desolder pins one by one easy as pie. Use some braid the clean up the pads, and you are ready to solder on the replacement, which is a much easier task than removal.

stuartk: I am definitely in for one knob in a group buy.

Thanks for the tip. This is actually what I tried, but using a sharp x-acto knife to cut the leads, which has worked well for me before on (T)SO packages. The cutting pressure was a bit much and a couple pads tore off, so I switched to heat shielding the rest of the board and using the heat gun... I didn't think to try side cutters which there might just have been enough room for. Next time .

I actually found soldering the new part quite challenging also. Because of the way the pins curl under the package, there's quite a large gap at the edge of the part between the pin and the pad. Guess I need more practice, so now to find some other stuff to repair.

[nctnico]

I did some further testing on my 66311B regarding the temperature sensitivity. The reference is pretty much stable after a few minutes and doesn't vary much with a changing room temperature so the conclusion is that the rest of the circuitry causes most of the temperature drift and therefore it doesn't make sense to put a better reference chip in the 66311B. Mine has an MC1403 2.5V reference by the way.