EEVblog Electronics Community Forum
General => General Technical Chat => Topic started by: allanw on July 03, 2010, 03:32:13 am
-
(larger version of pics if you click)
So, I decided I needed an actually decent function generator. I won an HP 8116A 50 MHz function/pulse generator that only said "tested to power on" in the description. Oddly, there were no pictures of the unit while actually powered on. I should have thought more about the implications of that. The final price I paid was $170 shipped. This seemed like a great deal because I saw other completed auctions where it sold at $400, $600, and even one at $1000. The one I got even had option 001, which had some extra features. Here's seller's pic:
(http://i.imgur.com/MahSk.jpg)
When I got the unit (which was in great cosmetic shape), it displayed error code E11 on the screen. So obviously the seller was hiding this by not showing any powered-on pictures. Of course, I did make sure I had a service manual before I went and spent over $100 on a piece of old test equipment. The only really difficult to replace parts are the three custom HP IC's. So, I did think it was worth the risk if it turned out to be a dud. Trying to fix it myself would be a fun adventure too. I opened it up:
(http://i.imgur.com/JV8rJ.jpg) (http://i.imgur.com/qvnif.jpg)
Pretty old school tech. The engineering isn't the best (soldered wires directly to the boards instead of using proper headers, random parts soldered on the backside, jumper wires), but it's pretty decent. The service manual describes E11 as: "The autovernier flip-flop is not setting or resetting. Either the main board A1 is not connected, or there is a failure in latch U103, NOR gate U102, buffer U302, or Q102." Seems pretty promising to be able to fix myself, as these IC's can be easily sourced. So I inspected the circuitry surrounding these chips, and found this strangely colored resistor network:
(http://i.imgur.com/AkWiW.jpg) (http://i.imgur.com/GpIM0.jpg)
To be continued...
-
Oooh, I love good test equipment repair stories....
That resistor network looks a little charred.
Scott
-
The resistor network R112 part # is 750-81. The resistance value is given as 4x470 by the schematic. It appears this is a valid substitute (following Digikey's chain of replacement links):
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=4308R-1-471LF-ND
Going to order one of these and swap out the resistor network and hope everything works after that. This error code is pretty high up the list of error codes, so chances are there'll be a couple more after I fix this...
(http://i.imgur.com/MHKmz.png)
-
I did a quick search of ebs118n and didn't find anything. I would guess the failed resistors happened because of a failure in whate they are connected to?
-
Yeah not sure why the schematic says EBS116N. The components list says it's an MC10116P. That IC seems hard to obtain, but not impossible. edit: Actually, really cheap from ebay. Probably not the issue though.
I don't think the fix is as simple as replacing that resistor network. Some of the pins I can easily reach still are 470 ohms. Another issue I found is that the -5.2V supply is only -0.36V. The other supplies are all good though. Trying to debug their supply voltage generation now. Not sure if it's because there's too much current draw. The manual says that there's a current sense that turns that supply positive if too much current is being drawn, and it's still slightly negative. It's a neat circuit actually.
Going to try to disconnect the load and see if it's still the wrong supply.
Here's the relevant stuff from the service manual if anyone's interested. I'm not necessarily looking for someone to spend a ton of time to do this for me, more like just documenting what I'm doing.
+5.1V supply
A reference voltage of 5.1 V is obtained from the +15 V regulated
supply, using Zener diode VR1 (6.2 V),R11 and R12. U1C compares
the +5.1 V supply with this reference voltage and drives the
regulator transistor Q1, via driver transistor Q2, until there is zero
difference.
If the current drawn from the +5.1 V supply is excessive, a distinct
voltage drop develops across R2. U1D detects this and its output
switches toward the negative supply. This forward-biases diode CR8,
switches off Q2 and Q1, and hence the +5.1 V supply is withdrawn.
+5V supply
The same principles of operation apply to the +5 V voltage
regulator, U2C and current sensor, U2B. The 5 V reference is
obtained from the 5.1 V reference via R13.
-5.2V supply
The -5.2 V reference is obtained from the 5 V reference using U2D as
an inverter with a gain of 1.04. The voltage regulator U1B and the
current sensor U1A operate as above except that the comparator
output is normally negative and switches positive to withdraw the
supply.
(http://i.imgur.com/qOg73.png)
-
Shorted power supply rail might be caused by a shorted tantalum capacitor. I'd disconnect the load, substitute a dummy load if necessary, and see if that helps. If it does, check every tant connected to that rail.
-
I disconnected their -5.2V generation circuit and used my power supply to put one in but it still gives the same error code. So this might get difficult...
-
Question : why you had not remove from the PCB the resistor network , even for simple testing ,
so to verify that even at this shape , it does work ?
-
It would require a lot of effort to remove this main board from the system chassis in order to desolder the network. The other boards come off very easily.
-
Oh wait, I'm just retarded. The chassis was designed so you could access the bottom of the main board.
Seems like there's some black flux-y stuff near the resistor network and the chip next to it, U100.
(http://i.imgur.com/h69f4.jpg)
The bottom stuff was after I started scraping. Not sure what all the build-up is. The PCB near U100 is slightly charred but near the resistor network I'm able to scrape off most of the black stuff.
Also, the leads of the IC are randomly bent. Maybe someone has already tried replacing this IC? That would confirm the theory that someone has already replaced this IC. The top markings of the IC match HP's part number, but the original chips should be marked with MC10116P.
Question: should I threaten negative feedback? Here's the auction:
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=400130356312&ssPageName=STRK:MEWNX:IT#ht_3149wt_1137
"This unit powers up nicely and appears to be in good used condition. New, this unit would cost thousands!"
Now, do you guys think that "powers up nicely" means that it shouldn't have a really obvious problem that occurs when it's powered up? Should I talk to the seller and tell them this? Chances are that they know it had a problem and tried to hide it.
-
Question: should I threaten negative feedback? Here's the auction:
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=400130356312&ssPageName=STRK:MEWNX:IT#ht_3149wt_1137
"This unit powers up nicely and appears to be in good used condition. New, this unit would cost thousands!"
Now, do you guys think that "powers up nicely" means that it shouldn't have a really obvious problem that occurs when it's powered up? Should I talk to the seller and tell them this? Chances are that they know it had a problem and tried to hide it.
What a mess! :o
That looks like someone used buckets of flux to solder, overheated the joint a bit, and never cleaned off the residue. I'd attempt to desolder the chip, clean up the mess with alcohol and a toothbrush, use solder wick to clean and re-tin the pads, and check for shorts. The only other time I've seen a mess quite like that is when someone used wax or glue to hold the chip in and soldered over it, but the bent leads probably rule that out.
If it were me I'd definitely contact the seller and see if I could negotiate at least a partial refund. Sellers live on their feedback, but judging from this one you might not get too far. They've built up a good bit of negative feedback and it doesn't seem to bother them yet, but it'll catch up with them shortly as their already sort of borderline. They'll probably just point out the 'as is' bit and tell you you're out of luck, and count on you not wanting to shell out for return shipping and a refund. You might be best off contacting Ebay to see if you have any recourse. Getting them involved can show the seller you're serious, or if Ebay sides with them it could confirm that you're out of luck. It's a crapshoot really, but the good news is that you might still be able to make it run.
Hope that helps. :)
-
I notice it has those little top-hat heatsinks on the can transistors - I harvested a couple of hundred of these from old HP equipment (totally busted unfortunately).
It does look like flux - and a lot of it. I don't think it is the solder mask discolouring from overheating. I've seen equipment repaired with plumber's solder, and the acid flux ruins the tracks quite quickly.
-
I believe 99.9% , that the resistor network , when it got overheated , melted the original soldering,
and the leads lost contact with the PCB .
Later on , some one tried to restore the solder with out the proper tools.
As long the solder restored , even by this manner .. and the device did not recover full fanctionality,
it got sold on ebay.
You have only two things to do .
1) Remove the resistor network for testing.
2) Check the PCB lines for continuity , between any resistor network pad and the destination of it.
If you do everything by the book , you could possibly restore the damage at this point.
Beware to double check for any diode , that possibly connects on the resistor network,
its 100% damaged too.
-
This definitely looks like a prior repair. I agree with the prior posters to desolder the part completely and clean out the excess flux and solder. Test the resistor network before replacing.
'Powers up nicely' does not imply it will work, but good used condition does imply functional, albeit both are vague: how is functional defined? Buttons press? LED works? So, seller is not being untrue, nor is the picture complete. It would be best in the future to ask sellers if a device passes the self test before purchasing, if it does its near certain it works as designed.
This is a high end HP FG, and still a costly device even from equivalent FGs from more economical makers like GWInstek. So, its worth putting effort into repairing it.
Currently, another eBayer has 2 such units for sale starting at $250, seller has 100% feedback and is much more truthful about its status, IMHO. Quite a number of these used ones are damaged, makes one think. Repaired ones are pricey. A lot of them look ratty.
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=250656336655&rvr_id=&crlp=1_263602_263622&UA=WXF%3F&GUID=47561c301260a0269d914691ff8605f9&itemid=250656336655&ff4=263602_263622
OTAH, this guy is something I would look into, and he knows what a buyer should know, and he knows roughly what such a unit is worth too:
http://cgi.ebay.com/HP-8116A-50-MHz-Pulse-Function-Generator-Works-Great-/260625415832?cmd=ViewItem&pt=BI_Signal_Sources&hash=item3cae7c3e98
Oh wait, I'm just retarded. The chassis was designed so you could access the bottom of the main board.
Seems like there's some black flux-y stuff near the resistor network and the chip next to it, U100.
Question: should I threaten negative feedback? Here's the auction:
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=400130356312&ssPageName=STRK:MEWNX:IT#ht_3149wt_1137
"This unit powers up nicely and appears to be in good used condition. New, this unit would cost thousands!"
Now, do you guys think that "powers up nicely" means that it shouldn't have a really obvious problem that occurs when it's powered up? Should I talk to the seller and tell them this? Chances are that they know it had a problem and tried to hide it.
-
Thanks for the suggestions guys.
I tested the resistor network and all the resistors still show up as 470 ohms. One of them is 420 ohms though. It's not obviously broken and these resistors are just used as pull-downs anyway.
Kiriakos, I've tested the resistance of this network using the tracks that it connects to and it's still okay. I'll replace it anyway but I don't think just doing that will fix everything.
I've been testing to see if the IC still works. Here it is: http://www.datasheetcatalog.org/datasheet/motorola/MC10116FN.pdf
The trigger amplifier is a three-stage ECL circuit with Schmitt-
trigger characteristics. Each op-amp stage U100B/D/C has both its
main and complementary outputs pulled down to -5.2 V via 470 R.
The threshold of the circuit is fixed at -1.3 V.
(http://i.imgur.com/tZoZu.png)
U100 is the IC, R112 is that charred resistor network. I probed the voltages and put in the logic levels according to ECL thresholds. It appears the IC is still functioning in that the outputs are still complementary logic. I'm not too familiar with ECL, so I don't know what logic level that the voltages -3.7V, -1.3V correspond to. I'm a little confused by the circuit. I'm guessing it takes an input trigger signal, amplifies it and then turns it into ECL logic for the rest of the triggering circuitry?
If so, then it seems like the input into this circuit is logic low (-3.7V) and the output is logic low (-1.8V), which indicates that this circuit is still working. Any thoughts?
-
Okay, I believe this whole circuit still functions fine. I provided -5.2V to the board without powering up the whole thing. Then I tried a couple of voltages at pin 4, which is the input into the trigger input amplifier circuitry. Then I measured the output. Everything seems to be correct. For voltages below -1.3V, it outputs -1.8V, logic low. For voltages above -1.3V, it outputs -0.8V, logic high. So the circuit is just a comparator. I found a model of the MC10116 and confirmed this behavior by using LTspice to simulate this circuit.
-
So is the thinking that the overheating was caused by a poor
repair job rather than a circuit fault? Maybe someone using a soldering
gun instead of an iron?
Scott
-
I'm thinking that the charred stuff is a red herring. This circuitry is purely used to handle the external trigger input and can't even be accessed by the internal self test so it can't be the cause of the error code anyway.
-
So, the error code means that the auto-vernier flip flop isn't working. So I'm testing this flip flop. http://www.datasheetcatalog.org/datasheet/motorola/MC10231L.pdf
(http://i.imgur.com/0p89U.jpg)
CE1 is the clock controlled by a MCU and R is reset coming from the MCU too. ECL voltage levels are -1.7V low, -0.9V high, and the center threshold is around -1.3V. Measuring the output of this flip flop, ~Q1, I get -1.3V. This shouldn't happen even if the flip flop was never clocked, right? I even tried clocking the flip flop myself with R on and off, but the output voltage was always -1.3V. Pretty bizarre. This is probably the issue, right? Replacements for this IC seem to cost $10 from ebay, so I'd like to make sure...
Also, the unconnected Q1 always reads 0V, but I'm not sure if this is because it needs a pull-down resistor to actually be valid. Their circuit always pulls the outputs down, so I'm going to try measuring Q1 with a pull-down too.
By the way, what do those arrows mean on the logic diagram of the IC? I see those a lot in these ECL circuits.
-
Replacements for this IC seem to cost $10 from ebay, so I'd like to make sure...
I wish I could help with the diagnosis, but it's over my head. :(
I did however find the website of a fellow (http://azur-electronics.com/motorola_ecl.aspx) who's been repairing a lot of HP equipment. You might consider emailing him and seeing if he's got specific advice or spare parts (http://azur-electronics.com/forsale.aspx).
Hope that helps some. :)
-
Measuring the output of this flip flop, ~Q1, I get -1.3V.
My opinion is: This Voltage level (if it is really true static level) can not accept for good designed and well working circuit using ECL. (nearly just like illegal level (always need remember with this kind of circuits that DVM or VM show only some kind of "average")
(also it is good sometimes check levels with high speed oscilloscope so that there are not any unexpected oscillation and with ECL it can be really high frequency.)
-
Found this thread while browsing - did you ever get this fixed?
I'm looking at one with a wonky output - is the service manual something you would be able/willing to share. I've got the operating/programming/"service" manual from the Agilent web site but it doesn't actually cover servicing as there are no circuit diagrams.
-
I have the manuals for those including full schematics. And i have repaired about 8 or 9 of these generators.
First thing to do : check the two zener diodes close to the bnc output connector. They make two bias voltages. What frequently happens is that someone overloads the output and these short out and then the output stage dies....
If that is ok : check the input signal to the drive train for the endstage. Left front ( front panel pointed to you , on the left hand side , check the output of the hybrid closest to you ( the 24 pin chip with heatsink on its back ). There is a testpoint closeby tp5 i believe.
You should have a correct signal there. If not then the shaper hybrid is shot and it is game over as you cannot get these anymore...
If the shaper is ok.. You will need the schematics before i can walk you through it.
Never, ever run the eendstage without the heatsink ! This amplifier has a tnedency to run away thermally. Keeping the transistors attached to the heatsink prevents this from happening.
-
Thanks, might come back to you but that's all useful stuff. Have located schematics on't web
-
If not then the shaper hybrid is shot and it is game over as you cannot get these anymore
Hmmm.... might be the case - this is what the output looks like for sine & triangle:
Fortunately I haven't bought it yet.
-
where did you measure that ? at the testpoint next to the shaper hybrid ? or after the amplifier ?
-
I didn't but the ebay seller did so it will be at the output connector, I would think.
-
Are displayed error codes helpful in telling you if the hybrid IC is shot? I'm thinking about buying a used 3314a but it's displaying error codes EOH and E8.1.
-
Holy bump, batman.
-
Service manual:
http://literature.cdn.keysight.com/litweb/pdf/03314-90021.pdf (http://literature.cdn.keysight.com/litweb/pdf/03314-90021.pdf)
I can't find those errors in the service manual, and frankly I don't think that display is working right with a lone segment lit. I would expect to see E and a two-digit number.
I see this unit on ebay and for $50 + $34 shipping +15% restocking fee if you try to return it. I wouldn't go near it.
Also, posting under an ancient 8116A thread is hardly the place to ask. If you want more opinions you could try opening a new thread with "3314A" in the subject to attract anyone who has some 3314A experience. I don't have one but I did look in the manual for you.
-
3314 is junk ...