EEVblog #540 - HP35670A DSA Repair - Part 3

[EEVblog]

More progress (or is it?) on the HP35670A DSA repair.
Part 2:
Part 1:

[vaualbus]

The first thingh you could do is replace all the op amp equals to the broken ic.
Secondly If I were you I would try to repair the source circuitary first and than from there see.
The overload protection can be cause by the fact that the two coaxial cable from the attenuator board aren't connect to the analog board.
On the anlog board the voltages are right so I will not worry about that board.
Probably the other ics aren't break at least the analog switches (DGxxx part) aren't beak because they work at five volt for the digital part and probably accept probably more than -19v to input.
The good news  is that the asic is definitly work because they drive the reference voltages as the dacs (on the source the othere is the for the reference voltages).
By now you should try to repost the blog in HP forum this time asking for the analog board problems.
Post also in Agilent/Hp yahoo group.
Wish for a next video.
Best regards, Alberto

[nitro2k01]

That "sine" looks like one or more of the bits, including the most significant one, are faulty (ie held constant). The question is if that fault is in the DAC itself or in the FPGA (or otherwise outside the DAC chip.) There's a DAC near the voltage reference. Is that DAC setting the voltage reference, or is it using it? If the latter, is there a second voltage reference near the output DAC you can try doing something about?

The reversed polarity of the reference TP's are obviously suspicious. Maybe try tracing that back through the components and see if gives you something useful. My theory is that they're both generated from the same reference somewhere, and one is inverted through the opamp. But the actual reference is blown, such that it output a voltage of the opposite polarity compared to the expected value. And then that creates the faulty test point values.

But yeah, this is starting to look like beyond economical repair. Maybe one of the in/outputs of the unit received a large negative voltage which leaked into the negative rail through protection diodes and destroyed everything in its way on the negative side.

[qczech]

Hi Dave.

Maybe it is possible to obtain the schematics from someone who worked on the instrument? Might be worth asking if you have some friends in Agilent.

I've found the Supplemental Operator's Guide and googling the name that appears on the first page I've got this: David Forrest. There is a phone and an email. Might be worth a shot.

Kind regards.

[nitro2k01]

Probably the other ics aren't break at least the analog switches (DGxxx part) aren't beak because they work at five volt for the digital part and probably accept probably more than -19v to input.

The DGxxx switches will have internal protection diodes, so operation outside the voltage rails is not recommended

The good news  is that the asic is definitly work because they drive the reference voltages as the dacs (on the source the othere is the for the reference voltages).

I think it's too early to tell. What you can tell is that the voltage is set once, but not for sure if the value is correct or not. The fact that the polarity on the test pins is inverted, and also 0.3 V off suggests that the reference DAC *might* be giving a false value. The signal DAC is giving a lot of incorrect values. Maybe the reference DAC is giving one such incorrect value, but is never updated again so you can't see it as clearly as with the signal DAC.

[vaualbus]

Or maybe there is another opamp before that are blown.
Yah if the source dac use the same reference voltage it could be that. But If the fpga is blown at all the reference voltage isn't there. The voltage is produced by a dac controled by the fpga. Look at the block diagram. So my guess is to track that ref voltage and see if after the dac pass through a opamp.
It could be the case that the dac ouput is ok but that then signal pass through some active filter to make the sine wave.

[Fagear]

You have good oscilloscope (not one of them), PCB is not too complex. Output BNC is at 15 volts, that is leading thoughts to one opamp pulling its output to positive rail. Because it has failed or component before it has. Why just not probe all signal pins of all opamps (only three of them per amp) in that area ? It's not so big! Try to trace path from ouput BNC to DAC at first.
Then you can deal with DAC if it still will be outputting some garbage.

[EEVblog]

Why just not probe all signal pins of all opamps (only three of them per amp) in that area ? It's not so big!

Because it's stuck inside the box!
I really need an extender card or cable before I can properly work on this any further.

[vaualbus]

you culd to download all the datasheet for the opamps and see which have the lower range to +-15, couple have +-18 and theoricly they are ok (+19 + 15 = almost 34).
I've reseen the video and at least I've see another opamp with the same hp number so also them will be wrong.
At this point the best way is to fix the source. For my point of view the output dac is ok, the signal is than passed through a active high pass fileter to make the sine wave, if the op fail of course you will see that signal. At list I have see a test point call adc ini, put a signa in the instrument and see If the signal arrive at that pin If not see If the signa arrive at the two coaxial cable and than try to track the signal.
I think I've found the main ADC is an MC1405 but I'm sure. it is near the voltage reference circuitary. 
I suppose that the attenuator board are ok indeed, there isn't regulator problem on it.
Why you not make an extender connector?
Best regards, Alberto

[SeanB]

Why just not probe all signal pins of all opamps (only three of them per amp) in that area ? It's not so big!

Because it's stuck inside the box!
I really need an extender card or cable before I can properly work on this any further.

Stick a tone on the connector to the board and look for it on the board with a tracer . Works well to find the pin, then simply check the opamp with diode test and see if the output is stuck. Simple enough to do to all the opamps to see if any have dead output stages.

[Rutger]

Stick a tone on the connector to the board and look for it on the board with a tracer . Works well to find the pin, then simply check the opamp with diode test and see if the output is stuck. Simple enough to do to all the opamps to see if any have dead output stages.

Yes, can't you take the board out and power it with an external supply? Then you can follow the signals with a scope.

[nikifena]

Hi, Dave
Power the board with external power suppply. Power just these +-15VDC for the opamps

Than using a multimeter read the voltages of each opamp - positive and negative inputs and the output. You know how the opamp works, so maybe you will find something unusual BEFORE desoldering all of the chips there.


Niki

[aaronsnoswell]

Hi all. Sorry if this question is obvious to the non-beginners out there. Around 5:04 (and later on) in the video, you can see a close up of the point where one of the coax connectors is. There is some odd looking silver disk behind the coax connector. What is the purpose of this component - it looks a bit like a small battery?

Really really enjoying this video series, by the way - keep it up, Dave!

[bktemp]

As I had said in the other thread: I don't think the opamp died from overvoltage, more likely it was a reverse voltage from the negative supply missing. Then the inputs (at GND level) are the most negative voltage which can easily damage the input stage because the input voltage of most opamps is limited to the supply voltage rail (or a little bit beyond).
The same applies to most other ICs connected to the -15V rail. I would also check the ADGxxx ICs. Some of them don't like if not all supply rails are powered.

It looks not very promising, but it is a great opportunity to make a series of repair videos. Especially showing how to repair stuff without having the exact schematic.
I would start with the DAC, because ist is partially working: You know how the sinewave should look like at the output and input, go through the digital input signals and check where they come from. Either the DAC or some logic driving the DAC seems to be damaged.

[SeanB]

Hi all. Sorry if this question is obvious to the non-beginners out there. Around 5:04 (and later on) in the video, you can see a close up of the point where one of the coax connectors is. There is some odd looking silver disk behind the coax connector. What is the purpose of this component - it looks a bit like a small battery?

Really really enjoying this video series, by the way - keep it up, Dave!

That is a PTC fuse to protect the outputs from being accidentally fed a large signal. Hopefully the PTC will heat up fast enough and go high resistance to limit the current flowing into the outputs. there are large clamp diodes there to enable this to happen without killing the opamp driving the output. Same as the PTC in your meter ( hopefully) protecting the inputs from overvoltage.

[Nermash]

In previous dsa Dave repaired that was the 50 ohm terminator IIRC.

[Napalm2002]

If the cost of the opamps are cheap enough it may be worth throwing a handful in and shot gunning it so to speak.

[pickle9000]

Swapping out a pile of chips is a valid method but is fairly heavy handed. A board extender would be very nice to help for testing (the op amps). I'd probably swap out 2-3 questionable chips at a time then test and hope I get lucky, still pretty heavy handed.

A schematic makes life so much easier, but unless you plan on repairing these for a living hardly worth drawing it out.

[LaurenceW]

... I think the real cost issue here is not a handful of op-amps (and the odd unobtainable esoteric IC), but Dave's time!

Yes, I concur that that "sine wave" (looked more like a porcupine wave!) could be stuck bits. Are these available, between the ADC chip and a counter/latch/buffer, or all internal to the ADC?

[EEVblog]

A board extender would be very nice to help for testing (the op amps).

A board extender is pretty much essential at this point.

[Rasz]

you talk about sucking chips up - that is time consuming
setup your hot air to be under the board heating solder joints, then you can swap chips for sockets in a matter of seconds without playing with individual holes and solder wick

[poorchava]

The bright side of the whole situation is that only FPGA and the Raytheon chip (what does it do anyway?)  seem to be unobtanium.  I'd try to stick a logic analyzer on the bus between FPGA and DAC to see if the data sent to DAC resembles a sine.  If not,  then most likely the FPGA is blown and you're screwed.  Otherwise it seems reperable.

[Noize]

A board extender would be very nice to help for testing (the op amps).

A board extender is pretty much essential at this point.

If you do get/make board extender maybe you can use that AIM TTI Current probe that you reviewed to quickly find out what op-amps etc are buggered?
I assume you still have that probe.

[alm]

A thermal camera would probably be easier if the opamp is drawing excessive power.

[daddario]

A thermal camera would probably be easier if the opamp is drawing excessive power.

I don't think Noize was suggesting excessive power consumption, but rather finding an opamp that's an 'odd-one-out'.
i.e. no supply current at all, an unequal current between rails or some kind of a third option that stands out from all of the other opas.