n00b with broken Tectronix FG503

[gadget73]

Greetings folks
I'm attempting to get an old Tek FG-503 function generator going for a friend.  It has two major issues that I can see, however I'm just lost as to what ails it.  I have gone through the calibration steps to confirm the power supply and frequency output.  The voltages come out bang-on for the positive and negative 20, 17 and 5 volt rails.  I even checked them relative to one another, and that all comes out OK.  Voltage is clean, less than 4mv of ripple or noise on any of the test points.

First problem: the sine wave output is triangle wave.  This uses a sine shaper setup with a series of diodes to round over a triangle wave into a sine.  All well and fine, except that it doesn't do this.  I can force it close to a sine wave by jumping R313 and R315 to reduce the amount of bias on the sine shaper diodes.  I could probably get it closer to right by playing with the values, but this just seems like too much of a kludge fix to me. 

Second problem: the dial scale is out to lunch.  When set to 30 on the front, my frequency counter will show anywhere from 55-67, depending on the position of the X30 cal pot inside.  I was able to get it in line by adding a 22k resistor in series with the high side of the pot, but again, kludge fix.

This smacks of power supply problems, but I just can't find it.  If anyone would care to give me a fresh set of eyes on this, I would very much appreciate it. 

Schematic and manual can be found here:
http://bama.edebris.com/manuals/tek/fg503/

Cheers

[jh15]

All your switch contacts are perfect?  Good you checked the power supplies OK. Nice way to start.

[tautech]

Consider that connecting a FG to a live circuit might have hurt the output amp/s.
Check waveforms prior to the output.

[cncjerry]

One problem is that the dial was probably moved or the reticle disk was broken and glue on wrong.  Also, the output can be triangle, sine, square depending on the switch setting so it is again possible it is the switch.  I have three of them for some reason, never had any problems with them other than those I mentioned (e.g. switch and dial misalignment).

[gadget73]

I did check out the switch.  I thought maybe it was missing connection into the sine circuit, but it all ohms out correctly according to the manual.  I even opened it up and ran a sheet of paper between the contacts to make sure it had no garbage in there, but no dice.   The switch setup on this is basically a set of gold fingers soldered to the board above a set of gold contact pads, and a cam on the switch shaft pushes various ones down to make or break things. 


The dial wasn't mis-positioned.  The calibration process has you set it with it fully clockwise to a value, set it to 30 to a value, then run the dial to 1 and make sure it comes out correctly.  I could not make it anywhere close to correct without the resistor.  The output frequency was essentially double what it should have been for no reason I can figure.  It tracked fine, 30 was 60, 15 was 30, etc.  The multiplier worked fine as well and the ranges multiplied as expected.  It didn't appear to be a harmonic issue, output looks clean on the scope, unless there is something going on in the output amplifier that is causing the doubling.  I will check the input to the amplifier stage to see if that gives different results.  Thanks for the suggestion.

One thing I had forgotten about, the original dial lamp was replaced with an LED, and the series resistor to it changed from 200 ohms to 1.5k ohms.  No chance that it relies on current draw through that lamp to make it all work, does it?  The LED was in place when the unit came here. 

[tautech]

Check for the appropriate waveforms before the output amps.

[gadget73]

I lifted R355 at the base of Q355 and sampled the waveform there.  Same result, the sine waves are triangles and it remains off frequency unless I bodge in some parts.

I also tried tacking in a 1k resistor across where the original pilot lamp was to duplicate the load.  No change.

[cncjerry]

I changed one of mine to an led without a problem, then again I use a counter for frequency.  Some of the TM500 modules do vary the light brightness based on output voltage so maybe it does.   I check both of mine though and don't see the brightness change.  One has an led I added and the other the regular bulb.

[gadget73]

I'm really more concerned with the lack of sine wave than the frequency issue.  I didn't think the light had anything to do with it, but it was easy enough to tack in a resistor to check that theory.

I can force it to work, but it really bothers me to change component values to make something work when it originally functioned with the stock parts.  It always ends up covering up a deeper problem.

[tautech]

I'm really more concerned with the lack of sine wave than the frequency issue.  I didn't think the light had anything to do with it, but it was easy enough to tack in a resistor to check that theory.

I can force it to work, but it really bothers me to change component values to make something work when it originally functioned with the stock parts.  It always ends up covering up a deeper problem.

And you've checked all the resistor strings + & - in the shaper and their GND connections?
Any offset on the output might also provide a clue as to where to look.

Edit
For those watching at home P 24

[cncjerry]

I dug out the schematic.  With the switch in sine shaping, you should see a triangle on the hot side of r349. 

Check to ensure you have the +20 and -20 correct on R310 and r315 respectively.

There are a number of diodes in that circuit, 10 of them to be exact.   If any open, you would see a defect in the sine wave either the positive or negative half of the cycle.  if Q320 is open or shorted, you would see a defect in the positive half of the sine wave and Q340 the negative half though I could have the polarities reversed.

The fact you aren't seeing any change in the output leads me to believe it is the function switch.  DC offset is added to all the waveforms and shouldn't matter.  So when you switch the function switch S380, it should interrupt the triangle path coming from R225 which by the way, feeds the sine shaper all the time through r349.  So put a scope on the bus side of r329, r331 and r333 (the side not connected to the diode) and you should see a sine wave approximation.  If you don't, then something is really hosed because like the schematic teaches, the triangle feeds the shaper constantly thru R349. 

So the diagnostic flow chart would be:

1) Scope the hot side of R349.  The other side of 349 is grounded.  You should see a triangle.  If you don't then the trace from R225 to R349 is open. (unlikely).  Also, if that was open and the function switch was defective then you wouldn't see the triangle.

2) Scope the common connection of R329, R331 and R333 and you should see a sine approximation.  If you do, and it could be really distorted without a load, then the problem is the switch.

I don't see any way for the triangle wave to feed thru the 5 step approximation circuit without seeing some shaping.  Therefore, I suspect the switch.

I have one of my units out with the cover off so if you need to compare signals let me know.


Jerry

edit:  I checked the signal at that junction and it is fine with/without a load.  so you you should see a good signal there.

[gadget73]

And you've checked all the resistor strings + & - in the shaper and their GND connections?
Any offset on the output might also provide a clue as to where to look.

Edit
For those watching at home P 24

No offset either when not set in offset mode.  In offset mode, it responds as expected with no change to the wave shape unless set to maximum offset in either direction and max amplitude.  Reducing either makes it clean up.


Resistors measure reasonably in circuit.  I didn't pull them to verify but nothing seems grossly wrong.  I did re-flow all of the solder connections in that area as well.  Diodes checked OK in circuit, again did not remove them to verify.  It shows absolutely no signs of shaping at all unless I jumper out R313 and R315.  When that is done, the bias on the diodes is reduced and I actually get reasonable sine output.  The bias is a bit too low though, results are cleaner with the trimmer pots set to give maximum voltage.  I could probably sub about a 47 ohm resistor in there and get the output to look good.  I would prefer to find out why it doesn't work with originally engineered values though.

Quick ohm check from the ground TP to the grounded side of those diodes and resistors came in at less than 1/2 ohm resistance, with about 1/3 ohm being test lead resistance.   

I dug out the schematic.  With the switch in sine shaping, you should see a triangle on the hot side of r349. 

Check to ensure you have the +20 and -20 correct on R310 and r315 respectively.

R318 you mean?  That connects to the -20v supply.  Notes I had on the schematic from last night indicate 3.111v on the high side of the + sine pot, and -3.135v on the low side of the - sine pot.   Its not in front of me this second for a volt check on the supply line side of those resistors, but that seems reasonable with some math.

There are a number of diodes in that circuit, 10 of them to be exact.   If any open, you would see a defect in the sine wave either the positive or negative half of the cycle.  if Q320 is open or shorted, you would see a defect in the positive half of the sine wave and Q340 the negative half though I could have the polarities reversed.

No defect at all, its just a nice perfectly clean triangle wave.  It will become a sine if I jump out R315 and R313 to lower the base voltage on Q320 and Q340 though. 

The fact you aren't seeing any change in the output leads me to believe it is the function switch.  DC offset is added to all the waveforms and shouldn't matter.  So when you switch the function switch S380, it should interrupt the triangle path coming from R225 which by the way, feeds the sine shaper all the time through r349.  So put a scope on the bus side of r329, r331 and r333 (the side not connected to the diode) and you should see a sine wave approximation.  If you don't, then something is really hosed because like the schematic teaches, the triangle feeds the shaper constantly thru R349. 

The amplitude does actually increase relative to the triangle wave output when set to sine.  Squares work perfectly, offsets work perfectly.  Sine + offset on the selector gives me triangle + offset.

So the diagnostic flow chart would be:

1) Scope the hot side of R349.  The other side of 349 is grounded.  You should see a triangle.  If you don't then the trace from R225 to R349 is open. (unlikely).  Also, if that was open and the function switch was defective then you wouldn't see the triangle.

2) Scope the common connection of R329, R331 and R333 and you should see a sine approximation.  If you do, and it could be really distorted without a load, then the problem is the switch.


I don't see any way for the triangle wave to feed thru the 5 step approximation circuit without seeing some shaping.  Therefore, I suspect the switch.

I have one of my units out with the cover off so if you need to compare signals let me know.


Jerry

edit:  I checked the signal at that junction and it is fine with/without a load.  so you you should see a good signal there.

I'll scope the resistors tomorrow.  The part that really just has me befuddled here is why does it turn into a sine wave if I monkey with the bias in that area?  I'd ordinarily agree with you about the switch, but the fact that I can force it to work with crude means is what has me confused.  It really acts like some sort of odd voltage problem in here.  Maybe a broken ground buss somewhere in the mix that I haven't found.

[cncjerry]

You got me on the bias, but I can speculate. You'll have to excuse me as I don't know the technical terms for this circuit.  They feed a triangle into the 5 step shaper.  The sequence of resistors and diodes set the cathode current for the amps, one set for the positive side and the other for the negative.  As the triangle wave voltage hits various levels through its cycle, the diodes short/open the cathode resistors changing the amp.  So basically you are mapping a triangle wave over a level-based grid more or less.  Draw a triangle on a piece of graph paper adjacent to a sine wave. Cut the sine wave and the triangle vertically first by degrees and then map the degrees horizontally extending the cuts into the triangle.  Then map the voltage levels of the incoming triangle at a certain degree to the desired output gain.  This would generate a rough sine wave (once filtered).  So for this incoming triangle voltage, change the gain of the amp to set the mapped output voltage. 

Now instead of using an incoming triangle wave at full amplitude, set the triangle to the lowest level so it only passes under the first cathode diode/resistor grid step.
If the drive is way off, too low would be the worst, then you are basically using a 1 step instead of a 5 step shaper.  The triangle wave isn't swinging high and low enough to hit the upper levels, in my somewhat hack-kneed explanation.  So maybe you have a drive problem coming into the shaper. 

Tektronix uses a shaper in other generators including the most complicated one known to man, the FG5010.

wrt your question on r318: There are two amps in the shaper, pos/neg.  The positive side uses r310 and r313 plus a variable R317 for setting the base voltage.  The negative uses r315 and r318 and variable r311.  R310 and r318 connect to the +20 and -20 respectively. R315 connects to ground on the negative side. C313 and C315 are marked as electro's so you might want to check them.  They stiffen the base bias voltage.

So scope r329 junction with r225 and you should see your triangle.  Then scope r224 both sides and again, triangle feeding the shaper.  Then the junction of r329,r331,r333.  If you see a sine wave there, then the problem is the switch.  If not, then it would only take a few minutes to check the shaper diodes. and Q320 and Q340.  Make note of the p-p voltage on both sides of r224.  If you don't fix it with the above scoping, then I'll check my p-p at r224 and we can compare gain.

I've fixed a number of these modules.  It is almost always the switches.  If not, then it was because I blew something up prior to finding the switch problem.  I had a voltmeter that someone must have run 100 amps through that had melted the solder off, so I guess that doesn't count.  I had another that I pulled and pushed back into the TM5006, that which you don't want to do with power on. That took a while to fix.

[cncjerry]

One other point:  Looking at a couple of FG503 units last night I found that the sine wave shaping varies somewhat by frequency.  The THD calculated up to the 9th harmonic at 3Mhz was 4.2% and at 1Mhz it is 2.8% on one unit.  At 100Khz the THD drops to 1%.  So this unit isn't that far out as the spec at 3Mhz is < 2.5% and 100Khz is < 1%.  Even at this THD the sine wave looks a little more triangle than I would expect.

As a point of reference, the FG5010 THD calculated off my spectrum analyzer to the 9th is 1.35% at 1Mhz.

[gadget73]

If the drive is way off, too low would be the worst, then you are basically using a 1 step instead of a 5 step shaper.  The triangle wave isn't swinging high and low enough to hit the upper levels, in my somewhat hack-kneed explanation.  So maybe you have a drive problem coming into the shaper. 

Hmm, thats something I hadn't considered.  Fudging the bias level would make up for a reduced triangle drive level. 

[/quote]


wrt your question on r318: There are two amps in the shaper, pos/neg.  The positive side uses r310 and r313 plus a variable R317 for setting the base voltage.  The negative uses r315 and r318 and variable r311.  R310 and r318 connect to the +20 and -20 respectively. R315 connects to ground on the negative side. C313 and C315 are marked as electro's so you might want to check them.  They stiffen the base bias voltage.
[/quote]

Caps look like dipped tantalum. I have no real means of checking low voltage caps.  I do have a crappy meter that will read value, but my tester for leakage was built in the 1930s and scoffs at the idea that anything can operate at less than 50 volts. 

R310 sees +20.006 on the rail side and +3.111 at the junction of R310 and R317.
R318 sees -20.005 on the rail side and --3.132 at the junction of R381 and R317.
I would expect those to be a lot lower if the cap was shorted.  I do see some high frequency ringing at that point on the scope when the triangle switches slope, but its about 6mv p-p.

So scope r329 junction with r225 and you should see your triangle. 

R349 you mean?  Doesn't appear that R329 and R225 meet anywhere.  R225 and R349 do meet. 
At that point,  Looks about like 2.4v p-p with the function selector on triangle if I count the hash marks right.
If you meant R329 and R224, those do connect via the switch with the selector in sine mode.  I get 3.6v p-p at that point with the selector in sine.

Then scope r224 both sides and again, triangle feeding the shaper.

3.8v p-p at the junction of R224 and R225.  Smidge less on the switch side, say 3.7v p-p.

Then the junction of r329,r331,r333.  If you see a sine wave there, then the problem is the switch.

triangle wave, 3.7v p-p.

All scope measurements done with the frequency dial at 15, multiplier on 10^3 using an x10 probe.  Scope is a B&K 1540.

If not, then it would only take a few minutes to check the shaper diodes. and Q320 and Q340.  Make note of the p-p voltage on both sides of r224.  If you don't fix it with the above scoping, then I'll check my p-p at r224 and we can compare gain.

Diodes check OK, ranging from 0.585v and 0.611v.  Most were 0.585-0.595, one was at 0.602 and one at 0.611.  Transistors check OK using the diode test function.  Beyond that I'd need to spend some time replicating a long-obsolete battery for my 1960s vintage Hickock 890 transistor tester.  I really don't do much solid state repair and I haven't needed that thing enough to spend the time on it.  In theory my tube tester will do them too, but it might be germanium only.  Not really sure.

I've fixed a number of these modules.  It is almost always the switches.  If not, then it was because I blew something up prior to finding the switch problem.  I had a voltmeter that someone must have run 100 amps through that had melted the solder off, so I guess that doesn't count.  I had another that I pulled and pushed back into the TM5006, that which you don't want to do with power on. That took a while to fix.

I can guarantee nothing about how this was treated prior to it showing up here.  My friend said when he got it the output was basically gone on one side of the waveform, and upon testing the voltages came out wrong at the test points so he replaced U512 and U566.  At that point he said something else didn't work right, but he couldn't remember what it was.  He apparently put this aside a couple years ago.  It ended up here, and you know the rest of the story now.

Side question, does R260 on yours show any evidence of having been warm?  This one is a bit darkened and the board under it is also discolored.  Its dropping 7.14v across it, which is less than 1/4w on a 1/2w part.

Thanks for your help with this.  Solid state gear really isn't my thing, so I sometimes need help working my way through it.  I mostly mess with tube gear, but tube test gear for the most part is awful or just has so many functional and usability problems that I'm not interested in dealing with it.  Nothing like trying to align a tuner with a signal generator that drifts off frequency every time a bird flies overhead or the planetary alignment shifts more than it deems appropriate.

[cncjerry]

My mistake on r329.  I meant r349.  I'll check the voltages but the only thing I can think of now is that the switch isn't opening and it is feeding a triangle at the same time as the sine.  I can't think of anything else.  I'll get to the voltages today.  I also look at that resistor.  This is a very simple circuit at this point, we'll fix it.

[gadget73]

I just ohmed out the switch again, and it checks out fine.  I get 0.3 ohms across #5 and #4 in sine, and 0.3 ohms from pin 4 to 5.  I get 825 ohms from 4 or 5 to pin 3, and effectively infinite to pins 1 and 2. There is an 820 ohm resistor (R355) that would measure between pin 3 and 4, so that makes sense.

[cncjerry]

At the connection of R225 and R349 I get 6.8V P2P.  This is a triangle with my scope in 1M impedance. 

The sine wave at r329, r331, r333 junction is 6.0V P2P.  So it looks like the triangle voltage feeding your shaper is about 45% less than expected.  Based on how I think the shaper works, this could be the problem.  I'll look at that resistor after the holiday.  Wife is yelling for me right now as I am cooking the roast.

[gadget73]

Half the voltage into the shaper would make good sense.  That would also explain why changing the bias makes it work too.  Much appreciated for that info, now I know where to start looking.

OK, I guess the next step is to back up into the triangle output section and see whats going on there.  I did notice that the slopes are just a smidge asymmetrical.  Same in triangle and sine wave, but the rise is quicker than the fall.  I didn't calculate exactly how off it was, but playing with knobs until I had one triangle covering the entire face of the scope, the peak was a tick or two to the left of center.  No idea if thats a clue or not.

[cncjerry]

If the triangle is somewhat symmetrical, then you would think everything prior to q201 would be ok. Anything broken in there would hose up the symmetry, I would think.   This leaves you with the three stage preamp consisting of q201, q211, and q221.  Q201 is a dual FET and they are prone to leakage over time.  I've pulled a dozen of them out of old HP meters.  I'll take some readings in there later today. 

If you celebrate it, have a Merry Christmas today.

[cncjerry]

By the way, you can get a square wave out of it, correct? How is the symmetry of the square wave?

[gadget73]

Square wave works fine.  Symmetry is good too as far as I can see.  Nice and flat with sharp rise and fall times.  No ringing or any of that either after it runs for about a minute.  Initial turn-on has the frequency somewhat unstable and the leading side of the square over-shoots.  After a minute or so it snaps into stability and the squares sort themselves out.   Amplitude of the square signal is higher than the triangle too, but I did not measure how much.

I do sort of wonder if the problem isn't  related to the oscillator section.  It still has that odd double frequency thing going on that I had to fudge in a 22k resistor in series with the frequency control pot in order to fix.   Perhaps that might also explain the poor triangle output, the unstable behavior initially, and the serious off-frequency habits.



Happy belated Christmas to you as well, and a joyous non-specific winter holiday to all  

[cncjerry]

I set the freq to 1Khz.  I have a triangle wave of 9.3V P2P at the following points that drive the triangle into the shaper:

The three stage amp voltages that feed the shaper are:

Junction of CR130/Cr140 gate of Q201 (double FET): 9.3V P2P;

Base of Q221 taken from R211: 9.3V P2P;

Junction of R223/R224 Emitter of Q221: 9.3V P2P;

Just thought of something else when I was looking at your frequency issue.  Your frequency problem is probably related to the VCF control voltage.   That would also support the problem you 'fixed' by adding that resistor. It is not unusual for people to drive too much DC or sweep into the VCF connector.  You will notice that those opamps are socketed.  One is a 741 and the other a LM1458.  I now suspect one or the other is bad.  They cost about a nickel.  I would swap the LM1458 first as the 'B' section is closest to the VCF port.  Make sure you note the pin one location on both when replacing them.  At one time you could get both at Radioshack.  I wouldn't even bother checking the other voltages above until you swap them out.

[gadget73]

I did actually consider that perhaps the VCF was "stuck", but it responded when I fed 1.5v to it.  It swung the frequency both directions depending on polarity.  Might still be the chip is dinked but its doing something.  Not going to pretend I paid any attention as to how much it moved.  I just hooked a handy AAA battery to it to see that it did something, and it did.   I can toss a pair of them on my next Mouser order easy enough though.  I have some other things on the list so I can justify the shipping. 

Will check the other voltages to see what its doing once I get those on hand, assuming I can't scare up the parts from my stock of various junk.

Thanks again for checking what yours is doing.  Having a working example to reference certainly does make troubleshooting easier.

[cncjerry]

Let, me know how you progress and if you need any other voltages checked. A bad op amp in that area could be doing something strange.  I have the same problem with mouser as it seems like the shipping costs went way up since they closed the more local depot.