EG&G PAR 5210 Lock-in amplifier repair

[(*steve*)]

Ok, no guessing games this time. It's a Princeton Applied Research model 5210 Lock-in amplifier.

Coming from the same source (a calibration and repair facility) so I'll assume it's BER even though the description didn't say so. (If it's not, I'll sadly move this thread somewhere else).



Again, with a small dog and a far more famous dog for scale. (As an exercise, who is the famous dog?)

And a closer look.



The sticker on the top notes that it was repaired by "RF" on 1-20-03. (What's the 20th month?)

And the rear panel tells me that it is the "property of IBM research instrument pool".



Is that a good thing?

Oh and I guess I had better hope there's nothing wrong with it because (again) there's nothing anyone can repair ;-)

 

[(*steve*)]

Following the DTIOTIA principle, here's a look inside:



Oooh, look, a piece of paper that's been removed and replaced again. I wonder what it says...?



Looks like configuration info. That might be useful!

I can see at least 6 tantalum caps. At the moment I can't read the expressions on their faces though

 

[(*steve*)]

And guess what, when turned on, I get some probably odd readings which change for a while then freeze.

And... The transformer smells a bit. Yep, I did change the line voltage.

A bit more information, this was manufactured in Sept 23, 1988.

The unit draws 350mA and the fuse is a 500mA slow blow. The plate tells me it's 130VA maximum, and it's only drawing 84W, and that doesn't seem too excessive, but I'm still not happy (and I don't think the transformer is either).

 

[kellys_eye]

(As an exercise, who is the famous dog?)

Michelle Obama?

 

[kellys_eye]

I wouldn't be at all surprised to see the almost exact same fault with this one as the other! I wonder if capacitor (tantalum) manufacturers did a MTBF test on their parts and, if so, ever published the results!

 

[(*steve*)]

The mains transformer has 2 windings. An 8VAC and a 48V CT winding.

There is only a little over 1A being drawn from the 8VAC winding.

However, the 48VAC CT winding seems to be providing 1.7A. That seems excessive.

All of the logic I've noticed is CMOS (HC and HCT). The processor is a 68000.

Compared with the SR530, this is a lot harder to gain access to stuff.

However, it looks like I need to find the regulators for the split supply rails. Weirdly, I can't see any regulators except maybe a couple of TO-92 devices. There's quite a few things that the regulators might be under.

The positive 15V rail reads 16.5V (ish) but it varies. The negative 15V rail reads +0.8V.

That may explain the problem.

BTW I'm assuming that the split supply is +/-15V, it may be +/-18V. I guess I'll find out...

There's got to be some regulators around here somewhere!

 

[(*steve*)]

I've removed the board that I think is the second channel (that's the main difference between the models 5209 (single channel) and 5210 (dual channel).

And removing that allowed me to take the cover off the large shield.



Look at all them tantalums!

It looks like I can unplug everything from these boards. Perhaps I can isolate the fault to one of them (or eliminate them...)

Nope. Current draw on the 48VAC CT winding is still almost 2A.

 

[(*steve*)]

Removing that board and the other half of the shield wasn't too hard.



There's only two 18V regulators there, and they're not connected to much of a heatsink.

But that looks odd.



I'm not happy with that resistor.

It measured about 18Ω until it fell apart.



Oops! Well it's an open circuit now.

One end of this resistor leads directly to the 7918. Coincidence? I don't think so.

I wonder what the current draw from the 48VAC CT winding is now?

Ooooh, 220mA now.

And look what's near that regulator!



Those tantrums are on the input side of the regulators.

I'm pretty sure that tantalum on the right is about to join my ever-growing collection of shorted tantalum caps.

That dead resistor had a twin for the positive supply, a 27Ω 1W resistor. I'm sure I can find something similar.

But not until tomorrow.

 

[(*steve*)]

I hope you guys don't think I'm a one trick pony.

 

[(*steve*)]

Damn, my local electronics store has morphed into Tandy electronics!



Anyone want to guess at the cost of labour and packaging vs the cost of the resistors?

 

[(*steve*)]

That seems to fix the +/-18V rails. They now measure +17.6 and -18.2.



Shiny new resistor



And the tantalum has blown up!

I replaced the 1μF 25V with a new 10μF 50V.

Now let's put it back together and see what happens.

 

[(*steve*)]

All our back together.



Pretty!

Now what happens when we apply power?



That looks well behaved.

Previously the two LCD displays showed ever-changing values and the meter swung back and forth until the device locked up and everything froze.

At the moment everything is stable until you move to the 300nV scale. Then things go wild. Given that the inputs are open, and there's no reference input, I'm not sure that's a huge issue (just yet)

 

[(*steve*)]

Woo hoo!

It certainly seems to work.



The functions of the SR530 and the 5210 are very similar. Hey, they were manufactured probably in the same year so it's probably a similar "state of the art".

The 5210 has 2 LCD displays and one analogue meter. I'm pretty sure it can display the same parameters as the SR530 with its 3 LCD displays and 2 analogue meters. However, you need to switch metering mode a lot more on the 5210.

Here are both operating side by side.



There is some difference between the measurement of both the magnitude and angle. It's hard to say which is closer at the moment, and I don't have the procedure for calibrating either of these units. Oh, the measurement of frequency us a little different, 51.0kHz vs 51.2kHz.

As both units warm up, the differences are getting smaller. The magnitude difference is a pretty consistent 6.5% to 7.0%

And another thing, the SR530 has a 1,2,5,10 sequence whereas the 5210 has a 1,3,10 sequence. The practical difference is small, other than the 3 scales needed on the SR5S0 analogue meters.

 

[(*steve*)]

Oh silly me. I had filters turned on. When I turned them off, the difference between the magnitudes fell from 6% to 0.4%.