Repair of Tektronix FG502 11 MHz Function Generator

[xrunner]

Just arrived. A little rough - advertised as "Used". Plugged it into a power module - like the PG501 no light on this one, I think we have another burned out lamp. 

Need to gather the manual from the BAMA (Boat Anchor Manual Archive) -

Boat Anchor Manual Archive

Don't know what awaits inside.

Developing ...

[Johnny10]

I'm going to pull mine out.
Have a broken one also and wanting to watch a pro work out the bugs!

[xrunner]

Have a broken one also and wanting to watch a pro work out the bugs!

Nothing like working without pressure. 

Do you remember what seems to be wrong with yours? I might want to compare symptoms.

[Johnny10]

This is your thread ..

But I did pull my unit out, haven't opened it yet.

Plugged in it shows no response to frequency adjustment.
Shows same waveform  on all settings and only adjusting between 20mv and 40mv amplitude. If amplitude higher or lower.. waveform disappears.
No Trigger output.

But we are following your repair.

Besides I need time to read the theory of operation.

[xrunner]

This is your thread ..

Oh I can share. 

I'm kickin' back with a beer but I did go into the lab and hook it up to my scope. Ugh ... nasty output. Sine - triangle - square - all bad. Badness varies with the amplitude adjustment but the frequency seems to change OK

Will post screen shots in the morning.

[Rerouter]

Ill kick you the schematics this afternoon if you havent found them, i have 3 of these, though i wish i knew what part caused the frequency shift when the output loading varies

[Rerouter]

http://w140.com/smb/FG502_SM.pdf

[xrunner]

http://w140.com/smb/FG502_SM.pdf

Thanks. Here's the screen shots from the output of the FG502. Pretty bad. Shown is sine and triangle selections. Trigger out is better. Looks like the negative half of the waveform is coming out.

Also as I said the lamp is burned out - will convert to LED but I'll tackle that easy job later on.

For now, time to crack it open and see what a visual inspection reveals.

[Kleinstein]

The waveform from xrunners unit looks like a broken output amplifier.

As circuit seems to use no special parts, it should be possible to fix.

If loading the output has an influence on the frequency, the problem is likely with the power supply. The frequency depends slightly on the supply and there seems to be no really good supply stabilization. Anyway I would not expect a very stable (e.g. with temperature) frequency.

[xrunner]

The waveform from xrunners unit looks like a broken output amplifier.

As circuit seems to use no special parts, it should be possible to fix.

Looks very possible. 

Visual inspection reveals two burned resistors. One in the center of the pic and the other one all the way at the back.

Time to heat up the iron and locate them on the schematic ... be back later with some documentation. 

[bitseeker]

xrunner, you're doing a good job making me want to start down this road. 

[xrunner]

xrunner, you're doing a good job making me want to start down this road. 

Come on in the TM500 pool - the water's fine.  (that is of course after you have repaired the entire pool filtering system  )

I pulled the two resistors, as shown on the schematic they are 10 ohm rated @ .25 W. burnt to a crisp. One measures 24 ohms the other is an open. I also pulled the two output transistors - Q540 and Q542. As you can see in the pic they look like they got overheated BUT they test OK in the transistor checker with an hFE of 137 and 145 respectively. I cross referenced them as follows according to the Tektronix part number on the case -

Q540 - 151-0103 --> NPN 2N2219A
Q542 - 151-0134 --> PNP 2N2905A

These are very common transistors so even though they check out I'm going to replace them, they may misbehave at high temperatures. Also will replace the resistors of course but I'm going to go with a .5 W size just for good measure. Have to order these parts. I ordered plenty of these transistors in case the previous stages require replacement. Also will provide socket pins too because I don't want to stress this old board out with any more heat in that area, as you can see they appear to have been replaced at least once already.

I have inspected the entire area; with a mirror I can see underneath Q530 & Q532 and do not see signs of overheating, also junctions read properly so I don't think these are bad. Next will go backwards and make sure the incoming waveforms are being generated properly prior to entering the output stage.

So guys any speculation as to what happened? Remember people do some very dumb things with test equipment up to and including connecting the output of one unit into the output of another ... 

[MarkL]

...
So guys any speculation as to what happened? Remember people do some very dumb things with test equipment up to and including connecting the output of one unit into the output of another ... 

I'm not sure why none of the 33ohm resistors wouldn't be toast too.

I'd make sure the +20V and -20V rails are actually that.  And check the C536 and C546.  And when the 10ohm resistors are replaced, check the bases and emitters of Q540, Q542, Q530, and Q532 to see if they are reasonable voltages.  Note that Q530/Q540 and Q532/542 and paralleled so they should share the current load to the output.

[xrunner]

I'm not sure why none of the 33ohm resistors wouldn't be toast too.

Just checked them per your suggestion; look pristine and check out OK.

I'd make sure the +20V and -20V rails are actually that. 

They are.

And check the C536 and C546. 

Checked in circuit not shorted. I could pull 'em and replace since I have plenty on hand.

And when the 10ohm resistors are replaced, check the bases and emitters of Q540, Q542, Q530, and Q532 to see if they are reasonable voltages.  Note that Q530/Q540 and Q532/542 and paralleled so they should share the current load to the output.

Will do ... 

[bitseeker]

So guys any speculation as to what happened? Remember people do some very dumb things with test equipment up to and including connecting the output of one unit into the output of another ... 

Without looking, my first thought was an overload into the output. However, glancing at the schematic, all those undamaged resistors on the output make it seem unlikely.

I'm still contemplating what would've caused the transistors to open up so far and long enough to fry. Could the open resistor have (unusually) failed as a short before blowing open?

[MarkL]

And check the C536 and C546. 

Checked in circuit not shorted. I could pull 'em and replace since I have plenty on hand.

Quickie in-circuit check is fine.  If a cap was shorted I guess it doesn't explain why both 10ohm R's are toasted.

There's always a possibility that a shorted pass transistor (for the +/-20V) in a former TM500 cooked it.

[bitseeker]

Ah, yes, I forgot about the original chassis that it had come from. It could very well have had over-voltage rails.

[xrunner]

There's always a possibility that a shorted pass transistor (for the +/-20V) in a former TM500 cooked it.

Ah, yes, I forgot about the original chassis that it had come from. It could very well have had over-voltage rails.

Yep, the power module is the random variable for sure. I've personally used these systems in my career, (as the lab user not the repair/cal person), and I can tell you for a fact the equipment gets abused and slammed in the chassis etc. It's very possible this FG502 got tangled up with a bad MoJo chassis. 

[David Hess]

The FG502 is my favorite function generator although it is also my fastest which probably explains that.  The other function generators in this series are slower (except the double wide FG504) but have some very useful features.

So guys any speculation as to what happened? Remember people do some very dumb things with test equipment up to and including connecting the output of one unit into the output of another ... 

It looks like a typical output short to me; I have repaired several function generators with similar class-AB output stages and similar symptoms.

Usually if the output stage transistors have not outright failed they are damaged so that should be checked.  They might have been damaged from thermal runaway resulting in high leakage and/or low Vce and the resulting high collector current damaged the resistors.  If you have a curve tracer, then this is a good place to use it.  Or test Icbo and Iebo.

Unfortunately there is a dearth of modern medium power complementary VHF transistors to replace TO-39 packaged parts.  I am cheap so I often Vbe match pairs of 2N4401s and 2N4403s and put them in parallel for each TO-39 part.

I'm still contemplating what would've caused the transistors to open up so far and long enough to fry. Could the open resistor have (unusually) failed as a short before blowing open?

Thin film resistors usually fail short before failing open so this is likely.

The puzzle is why the 33 ohm emitter resistors show no signs of damage or overheating.  My guess is that high transistor temperatures allowed enough collector-base leakage so there were *two* current paths; one through the emitter resistors and one through the 10 ohm resistors and biasing diodes.

[Kleinstein]

The one 10 Ohms resistor reading 24 Ohms now might indicate that those resistors could drift upwards when they get hot and old. As these resistors are seeing a more or less constant maximum current (in case of a short at the output), it could be just continued aging with a final run away once the resistance go too high. So the rest of the circuit might be still OK.

A slightly higher temperature (poor ventilation) a higher supply, or a short towards an external voltage could have also be a reason.

The transistors look slightly strange with the dark color. But this should be the collector pin, thus must be something from the outside - maybe flux of something during manufacturing the transistors. If they are actually 2N2905 / 2N2219 one could replace them of cause.

[tautech]

Check those three series'd diodes, if one is open the output stages may have suffered from too much drive.

[xrunner]

It looks like a typical output short to me; I have repaired several function generators with similar class-AB output stages and similar symptoms.

Usually if the output stage transistors have not outright failed they are damaged so that should be checked.  They might have been damaged from thermal runaway resulting in high leakage and/or low Vce and the resulting high collector current damaged the resistors.  If you have a curve tracer, then this is a good place to use it.  Or test Icbo and Iebo.

Yea I'll be replacing at least two of the transistors, but the way it's going I might have quite a pile of components to replace.

The transistors look slightly strange with the dark color. But this should be the collector pin, thus must be something from the outside - maybe flux of something during manufacturing the transistors. If they are actually 2N2905 / 2N2219 one could replace them of cause.

I don't think it's flux - it doesn't come off with alcohol, plus I would think it would be on the other pins as well, and the other two transistors I looked under, but it's only on these two and only the collector pins. I dunno ... 

Check those three series'd diodes, if one is open the output stages may have suffered from too much drive.

Checked and OK.

Update:

I decided to pull C536 & C546 - 10 uF caps to check, good thing. In the "transistor checker" the report on both is "damaged". With the Rigol DMM one reads no capacitance and 54k ohms, the other reads 0.7 nF. Will replace with new parts. Now I'm on the hunt for bad caps.

The pile is growing. 

[Radio Tech]

Looks like another great mystery novel. Will be watching this one.

[MarkL]

Update:

I decided to pull C536 & C546 - 10 uF caps to check, good thing. In the "transistor checker" the report on both is "damaged". With the Rigol DMM one reads no capacitance and 54k ohms, the other reads 0.7 nF. Will replace with new parts. Now I'm on the hunt for bad caps.

The pile is growing. 

Interesting find.  You can connect the caps up to a power supply and see if they will withstand at least 20V with no appreciable current.  Maybe they're breaking down at a low voltage and that's what killed the 10R's?

If they do start drawing a lot of current, look out for exploding tantalums!

[xrunner]

Interesting find.  You can connect the caps up to a power supply and see if they will withstand at least 20V with no appreciable current.  Maybe they're breaking down at a low voltage and that's what killed the 10R's?

If they do start drawing a lot of current, look out for exploding tantalums!

Well I'm going to replace with aluminum electrolytics - anybody disagree?