FG502 with no working FREQUENCY control


 

I'm still waiting for parts to arrive for the 7S14, and I acquired an FG502 for what seemed to be a very low price ($60, including shipping). In fact, the price was low because the instrument is not working (or not entirely working). It produces output signals, but the FREQUENCY control seems to have no effect on the output frequency. The output signal is stuck at a very low frequency (~100 Hz when the MULTIPLIER control is set to 10^6), though the frequency slowly drifts upward as the unit warms up.

At first I thought it had to be either a faulty frequency control pot, or something early in the circuit that leads from the frequency control pot to the main oscillator, but I have checked the two op-amps that drive the positive and negative current sources that drive the oscillator, and they appear to be working. Since I'm getting an oscillation I kind of doubt that there is a failure in the current sources or later. Of course my logic may be wrong, especially since it has not led me, thus far, to the fault.

Here is a list of the symptoms I'm seeing:

1. The FREQUENCY knob appears to have no effect on the output signal's frequency.
2. When started cold the output signal is ~15 Hz (with MULTIPLIER set to 10^6).
3. The frequency climbs by ~1 Hz every 5 seconds as the unit warms up.
4. Setting the MULTIPLIER control to lower values produces correspondingly lower output frequencies, but this is only really measurable for the top three multiplier positions, as the output frequency rapidly descends into the uHz range.
5. The output signal is only a full signal (both rising and falling cycles) when the FREQUENCY control is at the transition point between max (11) and min (0.1). At all other positions only the falling half of the cycle is there.
6. The AMPLITUDE and OFFSET control seem to work as expected.
7. The main triangle and square wave function produce clean signals as expected (when FREQUENCY is in the transition point).
8. The main sine wave function produces a signal that is about twice the amplitude of the main triangle and square wave outputs.
9. The 1/10 duty cycle function don't seem to work, producing either flat lines or 50/50 duty cycle singals like the main functions.

Here is what I have already checked:

1. Checked power rails (+/- 17 V and +/- 20 V). They were slightly high, but I brought them back to spec by adjusting R625.
2. Checked that R110 (FREQUENCY pot) was changing in response to turning the FREQUENCY knob. It runs up to 5 k ohms, as expected.
3. Checked the resistances between pins 2, 3 and 6, and pin 3 and ground, on both U100 and U135 (op amps driven by the FREQUENCY control), and all looked good.
4. Checked that the outputs of U100 and U135 change with the FREQUENCY control (they do).
5. Checked for shorted diodes and capacitors (found none).
6. Check the capacity of the large, silver, axial caps (2 in the PS, 4 in timing, all looked good).
7. Tried putting a signal into the VCF input to see if that would modify the output signal frequency, but it had no effect.

I'm at the limit of my understanding. I feel like this set of symptoms should be telling me EXACTLY where the fault is, but the places I expected to find a fault I didn't find anything (or I didn't recognize a fault, at least).

Any suggestions would be appreciated.

-- Jeff Dutky


Andy Warner
 

I’m thinking three things:

0. Have you tried cleaning the cam switch ?

1. the misbehavior of the frequency control might be telling us something,
can you lift pin3 of U100 and simulate the eff3ct of the freq control with
a known good pot, or scope pin 3 and see 2hat is happening ?

2. I think your logic about the current sources is maybe optimistic, if the
cam switch is healthy, the current source/sinks are apparently totally
incorrect.

On Sun, Sep 5, 2021 at 21:41 Jeff Dutky <jeff.dutky@gmail.com> wrote:

I'm still waiting for parts to arrive for the 7S14, and I acquired an
FG502 for what seemed to be a very low price ($60, including shipping). In
fact, the price was low because the instrument is not working (or not
entirely working). It produces output signals, but the FREQUENCY control
seems to have no effect on the output frequency. The output signal is stuck
at a very low frequency (~100 Hz when the MULTIPLIER control is set to
10^6), though the frequency slowly drifts upward as the unit warms up.

At first I thought it had to be either a faulty frequency control pot, or
something early in the circuit that leads from the frequency control pot to
the main oscillator, but I have checked the two op-amps that drive the
positive and negative current sources that drive the oscillator, and they
appear to be working. Since I'm getting an oscillation I kind of doubt that
there is a failure in the current sources or later. Of course my logic may
be wrong, especially since it has not led me, thus far, to the fault.

Here is a list of the symptoms I'm seeing:

1. The FREQUENCY knob appears to have no effect on the output signal's
frequency.
2. When started cold the output signal is ~15 Hz (with MULTIPLIER set to
10^6).
3. The frequency climbs by ~1 Hz every 5 seconds as the unit warms up.
4. Setting the MULTIPLIER control to lower values produces
correspondingly lower output frequencies, but this is only really
measurable for the top three multiplier positions, as the output frequency
rapidly descends into the uHz range.
5. The output signal is only a full signal (both rising and falling
cycles) when the FREQUENCY control is at the transition point between max
(11) and min (0.1). At all other positions only the falling half of the
cycle is there.
6. The AMPLITUDE and OFFSET control seem to work as expected.
7. The main triangle and square wave function produce clean signals as
expected (when FREQUENCY is in the transition point).
8. The main sine wave function produces a signal that is about twice the
amplitude of the main triangle and square wave outputs.
9. The 1/10 duty cycle function don't seem to work, producing either
flat lines or 50/50 duty cycle singals like the main functions.

Here is what I have already checked:

1. Checked power rails (+/- 17 V and +/- 20 V). They were slightly high,
but I brought them back to spec by adjusting R625.
2. Checked that R110 (FREQUENCY pot) was changing in response to turning
the FREQUENCY knob. It runs up to 5 k ohms, as expected.
3. Checked the resistances between pins 2, 3 and 6, and pin 3 and
ground, on both U100 and U135 (op amps driven by the FREQUENCY control),
and all looked good.
4. Checked that the outputs of U100 and U135 change with the FREQUENCY
control (they do).
5. Checked for shorted diodes and capacitors (found none).
6. Check the capacity of the large, silver, axial caps (2 in the PS, 4
in timing, all looked good).
7. Tried putting a signal into the VCF input to see if that would modify
the output signal frequency, but it had no effect.

I'm at the limit of my understanding. I feel like this set of symptoms
should be telling me EXACTLY where the fault is, but the places I expected
to find a fault I didn't find anything (or I didn't recognize a fault, at
least).

Any suggestions would be appreciated.

-- Jeff Dutky





--
Andy


 

Andy,

I have NOT tried cleaning the cam switches yet, but I was ruling them out because both the MULTIPLIER and the FUNCTION switches seem to have the intended effect. Also, my experience with dirty/oxidized cam switch contacts is that you can get the to work momentarily by vigorous exercise or by "edging" them into a position; neither of those tactics has produced noticeable results here.

I'm not dismissing your suggestion, just saying why I haven't done the obvious thing yet.

I completely agree that I am probably being overly optimistic about the current sources. My logic, however, is that in order to have both the working triangle (and other) output signals, but also to have no frequency control, you would have to have both positive and negative current sources failed in the same way, and in a way that did not affect their operation as current sources, only their response to input. That seemed unlikely, but it's the best bet now that I've ruled out U100 and U135.

The good news, regarding the current sources, is that they are just as easy to check as U100 and U135 were, and will be also be easy to fix: all the op amps in the unit are LM741s (except for two, in the PS, which are also LM741s, but are "selected").

-- Jeff Dutky


 

Okay,

New theory, suppose that both transistors Q140 and Q175 are failed and passing only minimal current regardless of whatever is happening at their bases? That would mean that it would take a very long time for the timing capacitor to charge, in both the rising and the falling phases.

Also, another symptom that I just noticed after running the instrument for over 20 minutes: the main triangle and square waves are no longer completely symmetric: the rising portion is slightly longer than the falling portion. Allowing the instrument to run even longer slightly decreases the duty cycle difference, but it's still there. This would be explained by slightly different E-C leakage currents in the PNP vs. NPN transistors that compose the voltage source and sink.

I can examine the voltages at the bases of Q140 and Q175 and see if they are changing with different FREQUENCY knob positions. If they are, but the current through the transistors isn't changing, then I will have to check the transistors themselves. If the voltage at the bases is NOT changing, then I must have a failed op-amp that feeds them both (that would be U135, which I already checked, and it looked good, but maybe I don't know what "good" means).

-- Jeff Dutky


Tom Lee
 

Jeff,

Although it's theoretically possible, leakages just don't match like that. For now, I would move that theory to near the bottom of the list, perhaps just slightly above "mischievous gnomes". Only after eliminating pretty much everything else would I move it up.

Check the current sources directly. The op-amp outputs shouldn't be pegged to a rail, and the transistors they're driving shouldn't be cutoff or in saturation. The control voltage from the frequency pot should be varying appropriately as you rotate the control. Follow that voltage all the way through to the current source stuff.

The 502 is a very straightforward function gen, if I recall correctly, so you'll be able to find and fix the problem fairly quickly, I am sure. It's a nice little unit -- the second fastest Tek ever made.

-- Cheers,
Tom


--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
http://www-smirc.stanford.edu

On 9/5/2021 22:20, Jeff Dutky wrote:
Okay,

New theory, suppose that both transistors Q140 and Q175 are failed and passing only minimal current regardless of whatever is happening at their bases? That would mean that it would take a very long time for the timing capacitor to charge, in both the rising and the falling phases.

Also, another symptom that I just noticed after running the instrument for over 20 minutes: the main triangle and square waves are no longer completely symmetric: the rising portion is slightly longer than the falling portion. Allowing the instrument to run even longer slightly decreases the duty cycle difference, but it's still there. This would be explained by slightly different E-C leakage currents in the PNP vs. NPN transistors that compose the voltage source and sink.

I can examine the voltages at the bases of Q140 and Q175 and see if they are changing with different FREQUENCY knob positions. If they are, but the current through the transistors isn't changing, then I will have to check the transistors themselves. If the voltage at the bases is NOT changing, then I must have a failed op-amp that feeds them both (that would be U135, which I already checked, and it looked good, but maybe I don't know what "good" means).

-- Jeff Dutky




 

Tom,

Yes, it's a nice little unit. It was a toss up between this and a CFG280, and this came up at an unbeatable price. I may still get a CFG280, but the only feature it seems to have, that this lacks, is a counter/timer and AM modulation. I've got an HP 8116A that can do the AM modulation, and the FG502 was acquired so I would have something that could drive the modulation on the HP.

This does seem to be a very simple design: I'm actually able to follow the theory of operation! It's also a spartan little PCB, compared to some of the equipment I'm working on (or trying to work on).

I've just been through the circuit leading up to Q140 and Q175 and I haven't found anything that looks obviously failed. Q140 and Q175 look perfectly good, at least according to the diode test function of my multimeter. So I'm forced to agree that my theory must move to the bottom of the list.

It looks like I won't find this problem by theorizing, but rather by brute force checking of the unit (and I'm clearly better at the latter than the former). I'm feeling a little too tired be poking around in a powered PCB tonight. I'll look at live voltages tomorrow.

-- Jeff Dutky


Roger Evans
 

Jeff,

I had a look at my working FG502 to see what happens in the dead space between 0.1 and 1.1 on the frequency control. On the x10^6 setting I see a slow (around 100Hz) signal which has a duty cycle of around 3:1 and which switches between square, ramp and 'sine' as you would expect. Its frequency changes appropriately as I go from 10^6 - 10^5 - 10^4. So its not unlike what you see.

Looking at your original point 5, is it possible that the rising edge is there, at the correct rate but you don't see it at the slow sweep you need to capture the complete waveform?

The voltages on pins 3 of U140 and U175 should just be buffered versions of the voltage on the frequency setting pot slider (on U175 it is inverted).

Regards,

Roger


SCMenasian
 

Jeff,
I recently restored 4 FG502s to calibratible operation and learned a few things in the process. Some may be useful to you.

Firstly, Tektronix made a big design error in the placement of tantalum capacitors, C536 and C546. Also, the voltage rating seemed a bit low to me. These capacitors are adjacent to Q530, Q532, Q540 and Q542 heat sinks. I replaced mine with 35 Volt Sprague 150Ds (mounted vertically as far from the heat sinks as I could get them). Of the 8 capacitors in 4 units, 4 were shorted, taking out 3 of R536 and R546.

Modpots, in 3 of my units, were assembled with screws and easily cleaned by loosening the screws enough to allow injection of some IPA. In the remaining unit, I replaced the rivited 2k pots with 2.5k pots (I had on hand), paralleled with 4,9k resistors.

On one unit, there was a cam switch problem caused by a previous owner's repair attempt. The moveable contact is, effectively, soldered to the board at the pivot point and, if the connection point is overheated, it can get out of alignment. On mine, this was obvious under visual examination. I got it back into place.

Transistors, Q140 and Q175, were socketed (connection points soldered to the board). They did not make good contact and caused problems which might be related to what you see.

On one unit, charging current problems were solved by replacing U140 and U175 with TL081 op amps.

Finally, more than one 1uF capacitor was bad, causing extraneous ringing. Since I had a pile of 1uF, 35V Sprague 199Ds, I replaced C192, C390, C392, C405, C418 and C442 on all 4 units - even the good ones.
In one case, one of these capacitors was shorted and its associated resistor was affected. The ringing was cured.

Stephen Menasian


 

I have verified that the frequency control is in fact working for one half of the cycle (falling half) so it appears that the positive current source is busted. This significantly narrows the focus of my investigation.

I did notice an odd reading on one of the four Schottky diodes in the switching bridge, an apparently much higher forward voltage drop than the other three (~1.2 V instead of ~0.3 V). I am also going to try reseating Q140 and Q175, and check the live voltages on positive current source op amp (U140).

— Jeff Dutky


Tom Lee
 

If you confirm that high forward reading, that diode is dead.

Cheers
Tom

Sent from my iThing, so please forgive brevity and typos

On Sep 6, 2021, at 13:27, Jeff Dutky <jeff.dutky@gmail.com> wrote:

I have verified that the frequency control is in fact working for one half of the cycle (falling half) so it appears that the positive current source is busted. This significantly narrows the focus of my investigation.

I did notice an odd reading on one of the four Schottky diodes in the switching bridge, an apparently much higher forward voltage drop than the other three (~1.2 V instead of ~0.3 V). I am also going to try reseating Q140 and Q175, and check the live voltages on positive current source op amp (U140).

— Jeff Dutky





 

Tom,

I’ve got a pack of BAT41 on order. When those arrive I will unsolder the switching bridge diodes and measure them out of circuit.

I’m don’t understand why they built the current sources and switching bridge the way they did. The transistors are socketed, but only the base and emitter pins. The collector pin is soldered directly to a resistor, which is then soldered directly to two of the diodes in the bride, all flying up over the PCB. At that point they could have just soldered the B and E pins of the transistors down, rather than docketing them, or they could have socketed the other corners on the bridge, so we could have removed the transistors and bridge as a single unit.

I’m assuming that there’s some electrical advantage to the flying lead arrangement, but I don’t really understand what it is. We talked about this last year with respect to a pair of resistors in the Z-axis amp of a 475/A I was working on, but I don’t recall what the conclusion was (either it was helpful to reduce parasitic capacitance, or it was better for heat dissipation).

— Jeff Dutky


Harvey White
 

You're not considering "oops, it needs to be modified"? are you?

Tek could have done that for very good engineering reasons, with pages and pages of calculations.  Occam's Razor suggests "oops...." instead.

Harvey

On 9/6/2021 5:37 PM, Jeff Dutky wrote:
Tom,

I’ve got a pack of BAT41 on order. When those arrive I will unsolder the switching bridge diodes and measure them out of circuit.

I’m don’t understand why they built the current sources and switching bridge the way they did. The transistors are socketed, but only the base and emitter pins. The collector pin is soldered directly to a resistor, which is then soldered directly to two of the diodes in the bride, all flying up over the PCB. At that point they could have just soldered the B and E pins of the transistors down, rather than docketing them, or they could have socketed the other corners on the bridge, so we could have removed the transistors and bridge as a single unit.

I’m assuming that there’s some electrical advantage to the flying lead arrangement, but I don’t really understand what it is. We talked about this last year with respect to a pair of resistors in the Z-axis amp of a 475/A I was working on, but I don’t recall what the conclusion was (either it was helpful to reduce parasitic capacitance, or it was better for heat dissipation).

— Jeff Dutky





Tom Lee
 

Jeff,

That bridge is sensitive to capacitive loading at those nodes, so having them in air is good practice. The still-faster FG504 does the same, and even threads diodes through a grounded shield plate to minimize blow-by. It's not an oops.

As for the 475 series-resistor example, Ed's analysis convinced me that minimizing capacitance there would not have been the motivation. But neither was it for remediation (i.e., again, no oops). Putting them in series allowed achieving a higher dissipation rating without having to add a new part to inventory, thus avoiding the massive ripple effect that such a seemingly trivial act has.

-- Cheers
Tom

--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
http://www-smirc.stanford.edu

On 9/6/2021 14:37, Jeff Dutky wrote:
Tom,

I’ve got a pack of BAT41 on order. When those arrive I will unsolder the switching bridge diodes and measure them out of circuit.

I’m don’t understand why they built the current sources and switching bridge the way they did. The transistors are socketed, but only the base and emitter pins. The collector pin is soldered directly to a resistor, which is then soldered directly to two of the diodes in the bride, all flying up over the PCB. At that point they could have just soldered the B and E pins of the transistors down, rather than docketing them, or they could have socketed the other corners on the bridge, so we could have removed the transistors and bridge as a single unit.

I’m assuming that there’s some electrical advantage to the flying lead arrangement, but I don’t really understand what it is. We talked about this last year with respect to a pair of resistors in the Z-axis amp of a 475/A I was working on, but I don’t recall what the conclusion was (either it was helpful to reduce parasitic capacitance, or it was better for heat dissipation).

— Jeff Dutky




 

Tom,

There were other cases of similar construction in the 475/A, mostly in the vertical pre amp, where it would make sense to be using that pyramid configuration to reduce capacitance. I understand why that would be needed at 200 MHz. I’m just a little surprised that it’s needed at 11 MHz (which doubtlessly reflects on my ignorance more than anything else; case in point: I have no idea what the magnitude of the parasitic capacitance is for a diode or resistor laying flat against the board versus raised up by a cm or so).

— Jeff Dutky


Ozan
 

I’m don’t understand why they built the current sources and switching
bridge the way they did. The transistors are socketed, but only the base and
emitter pins. The collector pin is soldered directly to a resistor, which is
then soldered directly to two of the diodes in the bride, all flying up over
the PCB.
As Tom said it could be to minimize the capacitance. Other possibility is if the minimum current at the lowest frequency setting is very small they may be worried about leakage in the PCB. I have seen flying leads connections in integrator summing junctions of HP voltmeters.
Ozan


 

Tom, Ozan,

Yes, the flying leads were probably done to reduce capacitance, but why not solder the E and B pins down to the board?

They were very inconsistent about docketing things on this FG502: many of the other transistors are soldered down. Three of the op amps are soldered down, but the rest are socketed (I’ve seen pictures of other FG502s where all the op amps are in sockets). Maybe these specific transistors were socketed because their pads are obscured on the other side of the board by a cam switch housing, but other socketed items are completely accessible on both sides of the PCB.

I understand that Tek was not operating under the kind of constraints that I am used to thinking about with mass market consumer electronics, and that a large amount of manual labor in construction was justified, but some of their choices just seem random.

— Jeff Dutky


 

Tom,

It turns out that I was NOT able to verify my measurement of the high forward voltage on the Schottky diode; a more deliberate set of measurements showed that all four diodes were working correctly, as were the two transistors. That only left the two op amps that drove the negative voltage source, U170 and U175. A quick check of their output voltages with the unit powered up while turning the FREQUENCY knob showed that U175's output was pegged at the negative rail (or close to it, -18 V and change, where the negative rail is -20 V). I replaced U175 with an LM741 I had lying around (bought because I thought that my 7CT1N was broken, when it was just my own ignorance of how to operate the instrument: happy accidents!) and now the FREQUENCY control is working correctly on all ranges!

Now I have a pack of BAT41s that will be lying around, but there's plenty of other broken instruments that might need a part that I thought I was buying for something else. No problem.

Now I've got three lesser faults to iron out (one or two of which may not be a fault):

1. The sine wave output is about twice the amplitude of the triangle and square wave outputs for the same setting of the AMPLITUDE control.
2. The sine wave is distorted (clipped?) at maximum amplitude.
3. All signals start to roll off significantly as the frequency approaches (and exceeds) 1 MHz.

I figure that all of these symptoms sound like they might be related, so I'm going to start tracking down the distortion and see where that leads. If I find a bad output amplification stage then I might fix all three in one blow.

Thanks for the great advice and support!

-- Jeff Dutky


Tom Lee
 

Jeff,

Glad to hear of your progress, and especially to hear that a new 741 is all it took to restore frequency control. The remaining problems you've identified should be straightforward to track down. I have every confidence that you'll find and fix the fault(s) soon.

Looking forward to hearing of your ultimate triumph.

-- Cheers
Tom

--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
http://www-smirc.stanford.edu

On 9/7/2021 22:32, Jeff Dutky wrote:
Tom,

It turns out that I was NOT able to verify my measurement of the high forward voltage on the Schottky diode; a more deliberate set of measurements showed that all four diodes were working correctly, as were the two transistors. That only left the two op amps that drove the negative voltage source, U170 and U175. A quick check of their output voltages with the unit powered up while turning the FREQUENCY knob showed that U175's output was pegged at the negative rail (or close to it, -18 V and change, where the negative rail is -20 V). I replaced U175 with an LM741 I had lying around (bought because I thought that my 7CT1N was broken, when it was just my own ignorance of how to operate the instrument: happy accidents!) and now the FREQUENCY control is working correctly on all ranges!

Now I have a pack of BAT41s that will be lying around, but there's plenty of other broken instruments that might need a part that I thought I was buying for something else. No problem.

Now I've got three lesser faults to iron out (one or two of which may not be a fault):

1. The sine wave output is about twice the amplitude of the triangle and square wave outputs for the same setting of the AMPLITUDE control.
2. The sine wave is distorted (clipped?) at maximum amplitude.
3. All signals start to roll off significantly as the frequency approaches (and exceeds) 1 MHz.

I figure that all of these symptoms sound like they might be related, so I'm going to start tracking down the distortion and see where that leads. If I find a bad output amplification stage then I might fix all three in one blow.

Thanks for the great advice and support!

-- Jeff Dutky




 

I took a moment to try adjusting the sine wave amplitude using the calibration trimmers (1 kHz trim pot, and 11 MHz trim cap) and got the high frequency output down to a level that is comparable to the levels of the triangle and square wave output at the same frequency. The 1 kHz trim pot is epoxied in place, so I haven’t been able to tweak that yet.

However, I went back and looked at the output, and the distortion of the sine wave has changed. Now the clipping is only on the negative sweep of the sine wave, and it is independent of the AMPLITUDE control. This tells me that the clipping is happening before the main output amplifier, in the sine wave amplifier.

So I went in and started checking the transistors in the sine wave amplifier, working backwards from the output, and I pretty quickly found two transistors (Q420 and Q440) that look bad: one has forward voltage for one of the junctions is only 0.3 V, and the other has forward voltages for both junctions that are way too high (1.3 V and 1.8 V). Fortunately both transistors are “jelly bean” parts (a 2N2222A and a 2N2709), which I’ve got in stock.

My question is this: the parts that Tek used are TO-18 (metal can) parts, but what I have in stock are TO-92 (plastic encapsulated) parts. Is there a good reason to prefer TO-18 to TO-92 in this application? I’m going to replace them with what I have on hand, but I’m wondering if I should bother to order some TO-18 parts.

— Jeff Dutky


 

I meant 2N2907, not 2N2709.

— Jeff Dutky