EG&G 5209 lock-in amplifier repair

[MichaelPI]

Hello all,

I bought an used LIA a couple of month ago - unfortunately, the LIA has some problems.

The lock-in had unresponsive buttons and not plausible (not changing) signs on the LC display. After heating up (certainly more than 20 minutes) the error slowly disappeared. It took me a while to figure out, what the root-cause actually was. All supply voltage rails (there are plenty of them in this instruments) had a resonable value - at least measured with the multimeter. Only by checking the supply voltage of the digital part with the scope I could find out, that the +5V rail had repetitive dips. Finally, I could track it down to a partially shorted electrolytic capacitor directly after the full-bridge rectifier.
I replaced all of those caps and the instruments is at least operable.

Two things are still not ok, one is the PLL and the other the calibration.

The first issue is, that at higher frequencies of more than approx. 7 kHz I get always an "PLL unlock" condition. This is also the case, if I use an external function generator.
I have currently difficulties to find the failure, as I only know the rough location of the PLL. Even worse I do not have a service manual and it seems as if someone has already done some "modifications" to the board. I don't really think those "bug fixes" have been installed by the manufacturer - it just looks to bad. 

Maybe someone has the service manual or a similar unit and could support fixing the remaining issues.

Thanks.

Michael

[Kleinstein]

The bodge wires don't look very nice, but they could very well be original. If not a leakage sensitive part, some flux residue is not a problem, if it is no clean flux. Such a change would be odd if done later as a mod - It more like looks a layout error to correct.

It is some 23 years since I used one such LI.  Not directly related to the PLL issue, but I remember a small bug (could be fixed in later versions): the auxiliary DAC outputs uses OPs (TL074 AFAIR) to directly drive the outputs and these don't like capacitive load from a cable.

AFAIR the manual included schematics. At least a know that I looked at the schematics and I don't think we had an extra service manual. I would not be surprised to find an 4046 PLL chip. I would expect the PLL also to be used for the phase shift - so I would first check if the fine phase shift works OK at lower frequency.

[MichaelPI]

This is not the worst modification in the instruments - there is one with removed traces, added diodes and resistors and a couple of bodge wires. Maybe all of this has been fixed in later revisions of the board.

Mine is from 1987 and made in the US. I have seen a couple of these instruments marked as Made in Ireland - which suprises me quiet a bit.
I checked the user manual, but there is no schematic included.

Actually the phase shifting is working fine at lower frequencies. As soon as I increase the frequency, the LIA goes nuts and displays an incorrect signal, before it goes into unlock condition.

The 4046 does not seem to be used, I recognized a MC4044 next to the magic black box. I will do some more tests tomorrow. Today I removed the filter board, which is on top of the main circuit board and the connector to the input amplifier - otherwise the demodulator and oscillator is not accessible.

[Kleinstein]

The 4044 can make sense as part of the PLL. There may actually be more than one PLL, e.g. one for the generator and one for the Lockin part to "generate" the ref. signal and do the phase shift.
However as far as I remember the internal generator frequency is not locked to the main clock used the measure the frequency and than adjust the filter setting accordingly. I remember that the automatic mode for the filter was not that stable: if the frequency drifted, the filter setting could change and this effects the response a little. So it is better to set the filter frequency manually.

For the phase shift I would expect to find some kind of DAC (maybe R2R) and OP in combination with the PLL.
AFAIK the coarse phase shift (90 deg, maby 45 deg. steps is done digitally, starting from 2 or 4 times the frequency. The fine phase shift should be together with the PLL, by adding some analog signal to the output of the phase - frequency converter and than have an integrating filter.

If the PLL does not work at higher frequencies, I see 3 main modes of failure:
1) Excessive noise (reaching the nonlinear range at higher frequency)
    This could be something like power supply ripple.
2) trouble with the VCO that does not work well to higher frequency (e.g. extra offset)
3) if the loop filter is switched depending on the frequency, this could go wrong.

Giving wrong data before the unlock may indicate a noise problem (or an unlock detection that does not work well).

For the large frequency range I would expect to have the actual VCO to run at a relatively high frequency and than have a switchable divider for the lower frequency ranges. The loop filter may also need switching to get faster response at higher frequency.

[LCR]

Hi, I do not know if a service manual for 5209 exists somewhere. There is an old document “MODEL 5209 LOCK-IN AMPLIFIER INSTRUCTION MANUAL 219567-A-MNL-C” with schematics. Do you have it?

[MichaelPI]

I really have to say sorry for the late reply. I have been busy with other topics, so I was not really able to dig deeper into the repair.

I recognized powering on the device, while the filter box is unscrewed and connected via the cables is not a good idea. There is connection screw, which is not only designed for mechanical purposes. It seems to serve in addition as star ground connection. The readings went all over the place, but except that the LIA is not damaged.

The lock-in amplifier has at least two oscillators, I can locate. One oscillator should be an analog VCO (for the internal reference oscillator) and the other one a crystal oscillator for the MCU. The frequency of the internal reference oscillator is digitally adjusted with buttons on the front panel. Based on this adjustments, the MCU seems to change the output value of a DAC. The "actual" frequency (read back via the MCU) can be displayed on the LCD. The difference between both frequency was rather high, so you are absolutely right, they are not locked.
The analog VCO seems to work properly - the OSC output has shown a plausible signal up to the upper frequency limit and the unlock condition occures with the internal and an external source.

I have seen an ADI part close to the area, where I assume the PLL, VCO and the phase shifter for the reference input is located - I guess, this could be the DAC for the phase-shift. Maybe this could be a good starting point for further tests. 

@LCR: Hi, I have read this document, but there is no schematic included - just a block diagram. I have seen a service manual for the 5210 (should be pretty much the same) on ebay a couple of weeks ago. The seller has shown a pictures which included a schematic.

[LCR]

Hi, I have the document for 5209 with schematics but their quality could be better. Overall size is about 32MB, so I can upload page by page. A list of schematics first.

[LCR]

The Reference Channel Schematics, 2 pages.

[MichaelPI]

Hi,

that is exactly, what I was looking for. Thank you!

Would it be possible to send me the whole schematics via email?

Regards

[EricCA]

Thanks for sharing this detailed post!  I have the same amp in our lab.  Can send more of the manual if you need it!

I wondered if you could offer any advice after your experience.  Our amp is "working", at least the Analog Meter and BNC output give the expected values for any input signal.  The LCD Output , however, displays a much lower value. (e.g. with no input, SIGNAL %FS = -150, SIGNAL = -4.5V on 3V sensititivity, or -1.5V on 1V Sens) With input, the LCD display scales with the Analog Meter and BNC output, but offset by -15V.

I think I've traced the problem to somewhere around chips U427 - U430 on the main board.  I see the "good" (albeit inverted) value at pin 2 on U427, appearing as short 1ms pulses.  But at pin 6 on U429, I see those good pulses and pulses with the -15V offset corresponding to the display value.  Both make it to U430 which I think is the digitizer.

I'm uncertain what to try next.  Supply voltages to all chips look nominally OK.  Don't know if a wrong value is being written into the pulse train, or maybe if the digitizer is somehow reading its value at the wrong time.  Last recourse is to cover the LCD with black tape!

[Kleinstein]

The circuit looks like a sucsessive approximation ADC build from a DAC and comparator controlled by the CPU. U 430 is a DAC and U427 is a comparator to compare the signal to the DAC value.
U429 is inverting am amplifying the DAC output to some +-15 V range.  U429 pin 6 is the ouput and should normally show a sequence of up to 16 steps converging towards the input voltage. U427 Pin 7 is the comparator output and should show a digital sequence that will than cause the CPU to adjust the DAC sequence.
I would check U427 to see if the output gives both digital high and low signals.

[EricCA]

Wow thanks!  With just a few keywords you made the logic of the circuit a lot clearer.

The output from the comparator output U427 pin 7 shows Hi and Lo output, but also some in-between value (noise?):

[Kleinstein]

The comparator output does not look good. This would be a lot of high frequency "noise". The circuit shows some hysteresis (feedback to the non inverting input) that should normally suppress these intermediate states - though the capacitor between the inputs kind of defeats it.

The noise may be something like an oscillation, either at the input signal or from a problem (lost decopupling)  with the supply at the LM311 comparator itself.
Normal noise would hardly keep the comparator from reacing a full high signal.

Besides the "noise" the sequence at the comparator looks plausible. A 15 V shift would correspond to 1 MSB error. As the fine part seems to work, the error may also be with the digital side.  Than I would however expect that half the range should not work at all, as the ADC would jump to a fixed value if the MSB is wrong. 

Are the +-5 V reference voltages to the DAC correct ?

[7jp4-guy]

I know this is an old thread, but I would very much like to have the rest of the schematics (and the adjustment procedure...) for my EG&G 5210s. I manged to fix my immediate problem with the schematics in this thread (R48 was not soldered in - I think this was a factory build error - and this was causing the reference never to lock.) However, I would really like to have the complete schematics to address other / future problems as they arise.

-Matthew D'Asaro
(www.dasarodesigns.com)

[PvdT]

Hello,

I'm new to the forum.
Maybe you can help me out. I've a EG&G 5209 lock-in and have some 'burnt' resistors e.g.
In the foto i've indicated them and number them.

Could someone tell me which components numbers it are?

with kind regards, Peter

[Kleinstein]

The numbers 7 and 8 look more like tantalum caps, not resistors. These can be the root cause of resistors burning by failing with a short. There are more of these on the board and it is a good idea to at least check them in circuit and possible replace them.

The other resistors don't look so bad on the picture. The value may be very well readable.  There is a chance that these are fusible resistors - that is resistors that fail open relatively gracefully after overload.

[PvdT]

Actually the picture posted shows the components still intact.

This picture is how it is in mine 5209:


Does any of you have a good diagram of the actual part of this PCB?

[Kleinstein]

The black resistors seem to be 20 Ohms. At least with most of them one can read the value on one or the other of the pictures.
Chances are these resistors are for the power supply filtering and a tantalum cap may have failed short to cause the damage.

[MichaelPI]

The 4 black resistors are 20 Ohms (3 W). The smaller ones are 27 Ohms (1/2 W). They are most probably for filtering and current limitation as Kleinstein mentioned. They are connected on one side to +/- 24 V provided by 7924, 7824 regulators and distribute the voltage to the analog circuit boards. There are plenty of tantalum capacitors on the different boards. In your case, I suspect that more than one failed short.
Even without a short, those resistors run relatively hot at least with my unit. Do you need a schematic of the power supply section?

[PvdT]

Hello Michael,

Thanks for your info this will definitely help finding the problem.

If you could please post the schematics of the power supply section.

[MichaelPI]

Here is the one, that I have. Unfortunately the image quality is rather bad and it does not seem to be 100% accurate for my unit.

[MichaelPI]

The tantalum capacitors after the +/- 24 V regulators and series resistors are only rated up to +25V. I got the same problem after powering on the LIA. One of the tantalum capacitor failed short and smoke came out of one of the 2W resistors. Fortunately the cover was not mounted, so I could react after a couple of seconds and pulled the plug. The resistor does not look good anymore, but the resistance value is still ok, so I left it as it is and only replaced the shorted capacitors. I attached a couple of pictures to show where to look at if this resistor fails. This is certainly not complete, but can give some hints.
 

[PvdT]

I've changed the resistors and cap's but the one indicated on the photo still is smoking if i start up the LIA.
Any idea where to look for else?

[Kleinstein]

There can be more of the drop form tantalum caps that can be dead, e.g. near some amplifiers.
Too much current may also come from a broken chip supplied via that resistor.

[guenthert]

Sorry for being late to the party, but my PAR 5209 had similar issues: here the tantalum at the output of the LM78L18 shorted and killed the LM78L18, which then smoked one of those 20Ohm 3W resistors (which I guess serve also as fuse).  I replaced all three and then it worked.

At that time I had only an aluminium cap around and intended now to fix it properly, but in the schematic I cannot read the value of the capacitor.  The LM78L18 data sheet recommends 0.01uF, I guess, I go with that unless someone has a better scan of the schematic.