Looking for the serive manual of an HP 70951 optical spectrum analyzer

[_Wim_]

I know it is a long shot, but does someone here has a service manual for the HP 70950/51 optical spectrum analyzer? Together with the 70004 display module, this combination is also called HP 71450/51. I already have the "installation & verification manual", "quick start guide", "user guide", "reference manual" , "programming manual" and "quick reference guide".

According to the reference guide (see attached screenshot) a service manual has existed. I already looked on the usual places and googled quite a bit, but so far no luck. If the manual is somewhere for sale, that would also be an option for me.

My unit has the following error I am trying to fix: "Cal Failed (2C) ERR 7055". If any-one knows more detail about the "2C" hex code, that would also already be very helpful!

Thanks for looking,

Wim

[zrq]

Although very unlikely with the service manual, we have a similar HP OSA at work, so I hope I can answer some questions.
Does it scan normally? These grating based OSAs have moving parts, so inherently have limited lifetime. The oldest Yokogawas we have at work makes screaming noise when scanning and from time to time aborts with monochromator errors. Good luck.

[_Wim_]

I have been doing various tests this evening, and the monochromator part seems fine I think (my unit has the monochromator out option, so I can use it as an optical "preselector", and this works fine for the optical signals I could test with).

There is at least an issue with the transimpedance amplifier, as the signal comes and goes depending on the sensitivity setting, so probably some gains do not work any longer (could be a faulty COTO relay, will try to check tomorrow).

Attached some pictures on the transimpedance section. If not the COTO relays, the DG441 cmos switches would be my next guess...

[_Wim_]

Forgot to mention, I also replaced 2 opamps on the X-Y control board (U401 and U402) because they were running quite hot (>100°C). After the replacement, U401 dropped to about 85°C, and U402 to 60°C. I did however not have any TL072 on hand, so I replaced them with AD712 as a temporary solution, so this might also explain the temperature difference. Still, this board remains suspect also, but I first want to concentrate on the transimpedance section, because debugging is much easier when a signal is present.

[_Wim_]

This morning I checked both the COTO relays via their driver (74ACT244), and they were fine. The two analog switches (U400 and U402) however both had their port 4 damaged (other ports are working fine). As I do not have any ADG441 cmos-switches in stock, I will have to wait on the parts to continue the repair.

[_Wim_]

The HP journal 44 no. 6 has some very interesting articles about these optical analyzers:

* Optical spectrum analyzers with high dynamic range and excellent input sensitivity
* A double-pass monochromator for wavelength selection in an optical spectrum analyzer
* A high-resolution direct-drive diffraction grating rotation system
* A two-axis micro positioner for optical fiber alignment

[_Wim_]

And the last one

[_Wim_]

This morning I checked both the COTO relays via their driver (74ACT244), and they were fine. The two analog switches (U400 and U402) however both had their port 4 damaged (other ports are working fine). As I do not have any ADG441 cmos-switches in stock, I will have to wait on the parts to continue the repair.

The ADG441 parts arrived, but unfortunately this did not fix the issue. Probably some more parts are damaged in this area. As the HP70951 allows you to connect and external signal to the transimpedance amplifier, this probably damaged this board, so I will check what is in series with those damaged ADG441 ports.

[nctnico]

Opamps running hot may also be caused by a missing power supply rail. I see lots of tantalum capacitors. I'd measure them all to see none have formed a short circuit.
The reed relais on the board may also have become sticky.

[_Wim_]

Opamps running hot may also be caused by a missing power supply rail. I see lots of tantalum capacitors. I'd measure them all to see none have formed a short circuit.
The reed relais on the board may also have become sticky.

Thanks for the tips. The relais were tested before and are ok, and I have verified that the rails are not shorted to ground with a multimeter. However, testing all the rails when the board is powered is harder, due to the way it is installed in the module. I did check all rails of the 70004 main frame, and these were fine (and no overload light blinking).

I also verified that whenever channel 4 is not selected on the DG441, I get an actual signal on the display, so this means whatever is broken on the amplifier board, must be wired to channel 4 only.

Hopefully tomorrow I have some time in the lab to further test what is wrong.

[nctnico]

I'd be carefull with saying those reed relays are OK. I had an HP function generator and after a while (with the internal temperature rising) the relays started to act up causing all kinds of weird behaviour. It took me quite a while to figure out the cause was in the relays because they tested just fine.

[_Wim_]

I'd be carefull with saying those reed relays are OK. I had an HP function generator and after a while (with the internal temperature rising) the relays started to act up causing all kinds of weird behaviour. It took me quite a while to figure out the cause was in the relays because they tested just fine.

I agree that there could also be and inconsistency issue with them, but when I tested them (cycling each approx. 5 times) they worked flawlessly, and the issue the 70951 has is very consistent (does not change when just started up to hours later).

I get only a signal on high sensitivity settings, and that signal can be partly blanked-out when "auto-ranging" is "on" (so when it tries to switch the gain in the middle of the sweep). This auto-ranging is presumably done with the solid state switches, although I still need to verify if that is the case.

[_Wim_]

On lower sensitivity settings I get something that looks like this (so essentially equal to the blanked out section above, but then for the entire spectrum). So it really seems the unit is unhappy with "strong" signals, indicated it might have been severely overloaded in the past

[zrq]

I'd be a bit surprised if it was an overload that caused the problem. The photodiode will likely die before the electronics.

[nctnico]

Can you burn out an optical attenuator? I suppose there are some mechanically operated optical attenuators somewhere in the signal path. And with the device working for low level signals, it could be these are switched in / out of the path. I'd check whether these attenuators still work.

[_Wim_]

I'd be a bit surprised if it was an overload that caused the problem. The photodiode will likely die before the electronics.

One of the read relays selects between the internal photodiode (including initial amplification) and the external photodiode input. I suspect some previous owner overloaded the external photo diode input (max 10V and max -10mA).The internal photodiode would not be damaged is this case, as the reed relay provides sufficient isolation. This would also explain why the very sensitive amplification was not damaged (external signal is only injected after this part) but some parts of the following sections are damaged. Off course, this is just a theory, and a schematic of this amplifier section would be very handy!

Worst case I could brute force and replace all active components in the transimpedance section, but for now I still want to find out what was exactly damaged.

[_Wim_]

Can you burn out an optical attenuator? I suppose there are some mechanically operated optical attenuators somewhere in the signal path. And with the device working for low level signals, it could be these are switched in / out of the path. I'd check whether these attenuators still work.

As far as I can see (and also browsing through the manual) I do not think this OSA has an optical attenuator fitted.

[nctnico]

I just checked the specs and the maximum input level is +20dBm which is far more than I expected. So yes, the problem could be in transimpedance section. I'd try to retrace the schematic and make more measurements before going for a brute force replacement of random parts. IF you can't find the service manual. Isn't there are similar machine from HP from the same era for which the service manual exists? It is likely the electronics are similar enough to get an idea of how ranges are controlled and how the amplifier works.

[zrq]

I have never seen an OSA with optical attenuator.
Is it possible to manually inject a current and see how the TIA respond in different ranges? (maybe you have already tried this)

[_Wim_]

I have never seen an OSA with optical attenuator.

They do exist, my other OSA (Anritsu MS9710C) has one built in, it allows to increase the range from +10dBm to +23dBm.

Is it possible to manually inject a current and see how the TIA respond in different ranges? (maybe you have already tried this)

That is indeed my next plan (currently not in the lab). I plan to use the keithley 220 in program mode continuously stepping between 2 currents. This way is should be easier to trace where the actual signal is. Ideally however I would have a board extender, as currently probing is not easily possible (apart from the IC towards the top, it means pulling the board out, making the connections with micro grabbers, reinstalling the board (and hoping the grabbers to not disconnect), measuring, repeat all over again...

[_Wim_]

Isn't there are similar machine from HP from the same era for which the service manual exists? It is likely the electronics are similar enough to get an idea of how ranges are controlled and how the amplifier works.

I do not think the was a predecessor to the 70950/51 series. Later models also do not have a service manual. I will try to retrace part of the circuit.   

[_Wim_]

Today I did the following:
- Made a short board extender out of some IDC pin headers. At least now a big part of the circuit is accessible to probe
- I tried to use the transimpedance amplifier external input with a current source, could not get it to work (this was before I started the reverse engineering)
- I noticed the analog output is also broken (is stuck continuously around 8.9V). Currently I did not investigate further.
- Tested the voltage rails in circuit, +15 and -15V are fine
- In power meter mode, optical signals below -44dBm work fine, signals above to not work. I suspects this is similar in OSA mode, but it is a bit more complex as it depends on RBW and some other settings
- Started to reverse engineer the board (see attached drawing were I am so far)

Whenever K401 is in the "1" position (so using the "low signal" gain stage), the OSA works fine. When K401 is in the "0" position, no signal is seen. On U407 I see around 0.7V between + and - inputs, and the output of the opamp (AD797) is around 13.7 V . So I tried to replace U407, but got exactly the same result (was replaced by an AD829 I got on hand).

I do not understand the Q403 part of the circuit. There is another one of these on the board, and it measures differently in that position, so this part could be broken. I however do not understand how this could be a plain transistor, although the other one does seems to measure like one. If anyone recognizes Q403, please let me know.

[nctnico]

Q403 could be a 2N7002 mosfet. It would be a good choice as a gain selector and a mosfet can be driven from a logic output directly (without a series resistor).

[_Wim_]

Q403 could be a 2N7002 mosfet. It would be a good choice as a gain selector and a mosfet can be driven from a logic output directly (without a series resistor).

I agree a FET would make more sense, but I measure diode junction voltages like typically seen for a transistor (from base to emitter and base to collector). On Q403 I also measure junction voltages in reverse (and only 0.17V) which made me think that it is possibly broken.

I did remove Q403 (I should have measured it out of circuit also, but did not yet do so) but this still did not fix U407 continuously hitting its rail.

[_Wim_]

One more piece of information, K402 is most of the time in its "0" position (connector to GND) unless the reference level is set very low (below -50dbm if I remember correctly). At that point K401 also switches to its "0" position.

[_Wim_]

Q403 could be a 2N7002 mosfet. It would be a good choice as a gain selector and a mosfet can be driven from a logic output directly (without a series resistor).

I think you are correct. I now start to doubt myself about the measured junction voltages, I will hopefully have some more time tomorrow evening. I will short the pads for low gain, this should then at least show a signal, albeit too low.

[nctnico]

Removing Q403 would be a good idea. If it is broken, it could inject enough DC to overdrive the opamp to get it stuck to a rail.

[_Wim_]

Removing Q403 would be a good idea. If it is broken, it could inject enough DC to overdrive the opamp to get it stuck to a rail.

That was indeed what was happening. However U407 was also broken. After I replaced U407 and got the same result (railed to the positive rail) I wrongly assumed nothing was wrong with U407 and reinstalled the original part  Afterwards I removed Q403, this did not fix it also.

However both Q403 and U407 were broken! By removing Q403  and shorting Q403 to the lowest gain setting I now have a signal on lower sensitivity settings also. Off course some gain switching is not correct, I have some 2N7002 on order, as I did not have any suitable FET in stock.

But certainly good progress! Will off course also reinstall an AD797 for U407.

[_Wim_]

A short update:
- Q403 has now been replaced by a 2N7002, this seems to work correctly, although I do not see much difference between the temporary short and with the 2N7002 in place (which is normal I guess, as it only adapts the gain slightly)
- broken analog output: is now fixed, this was also caused by a damaged AD797BR opamp (U105)
- non working transimpedance input: this was a user error, I did not configure the OSA correctly to use the external transimpedance input
- replaced U401 and U402 again by the original TL072, temperature went again above 100°C, which I guess is than normal (absolute max is 150°C according to the datasheet)

TODO's:
- verify all functionality
- find a keypad (not urgent as currently I am using the rubber buttons and a screenshot of the keypad)

[_Wim_]

Some further updates that might be useful for others in the future:

- KE5FX 7470 plotter emulated works mostly correct to take screenshots (using an agilent 82357 GPIB usb adaptor)

Some information is the plotter screenshot is not correct or missing:
- Start,center and stop wavelengths are missing
- RBW is missing
- Video averaging number is correct, but is also shown when video averaging is off
- screenshot reference level is at top of display, while on the 70951 this is at the 9th graticule.

I have also done wavelength and power calibration using a 1550nm single mode laser. Wavelength was off by 0.6nm, power was off by 1.5dbm, but for power I am not really sure about the accuracy of my calibration. For wavelength I verified the result using an Acetylene reference cell (installed in the Anritsu MS9710C), and the result was bang on.

[_Wim_]

TODO's:
- find a keypad (not urgent as currently I am using the rubber buttons and a screenshot of the keypad)

In the spirit of DIY, I deceived to make a 3D printed one myself. Not as nice as an original, but it works fine and is much easier to use than the rubber buttons. Attached the 3D design files should any-one be in a similar situation in the future.

[_Wim_]

I have made some improvements to the 3D panel file to ensure the panel is more rigid, and to allow some additional travel for the buttons. The updated files are added in the post above.

[_Wim_]

Another update:
- I tested all functionality as it seems everything is working now 
- As suggested in this blog (https://medium.com/@HP70K/70004a-display-screen-capture-43f7954d32b), I bought an Epiphan VGA2USB LR to allow me to easily take screenshots with all data (as the KE5FX 7470 plotter is missing some info). Attached an example when I was characterizing 2 Thorlabs  F240FC-980 collimators
- I routed the external input and outputs to the front for easily accessibility (see shelf picture)
- The OSA is now installed on my shelf with all the other optical test gear

This will be a very useful additional characterizing unknown optical parts bought on Ebay or when reverse engineering industrial optical analyzers.

If somebody would still come across the service manual, still interested to have a read through it. Also very interested in any software applications that work together with this OSA or functional that can be installed on the OSA via "measurement personalities".

[smaultre]

Hello _wim_

Just received HP 71450B  -002 option White Light source module (separately).

I have OSA Agilent 86146B (not tested) and would like to test it with light source.

Additionally i have an Anritsu MT9810A optical test set with one power sensor module.

I would like to check the power of mine White Light source module.
And check Agilent 86146B with them.

Can you provide the spectrum of your built-in light source? I would like to compare with my.

[_Wim_]

Hello _wim_

Just received HP 71450B  -002 option White Light source module (separately).

I have OSA Agilent 86146B (not tested) and would like to test it with light source.

Additionally i have an Anritsu MT9810A optical test set with one power sensor module.

I would like to check the power of mine White Light source module.
And check Agilent 86146B with them.

Can you provide the spectrum of your built-in light source? I would like to compare with my.

Here you go. Please also post yours, always good to have a reference.

[_Wim_]

It does seem mine does not meet its spec any longer. According to the datasheet it should produce -50dbm/nm, but is only achieves -55dbm/nm...

[smaultre]

I have now only on dB measurments.

[_Wim_]

Hi Smaultre, when measuring light with an optical power meter, you always measure the "broadband" power. The only thing the wavelength setting does is adjust the correction factor of the photo diode to convert to power for that specific wavelength, but it is not possible to measure the power of a specific wavelength from a broadband source. 

When I connect the HP 70951 light source to a VIS-power meter, I get -20.4dBm, when I connect it to a NIR power meter, I get -8.2dBm (both set to a wavelength of 850nm). This means it also makes it hard to compare results, as the sensitivity spectrum of the photo diode is different for different power meters. As you measure around -9dBm, I expect you are also using a NIR power meter (InGaAs sensor or Ge), but this does certainly not mean your source has a slightly lower output.