Frequency generator keeps burning out. 50 Ohm load required?

[Skyfox]

To be more specific, it's an opamp on the board.  I have a Wavetek 184 5Mhz function generator.  Last summer I found it not working, so I'm just now getting around to fixing it.  I narrowed the problem down to the generated voltage control loop, and eventually swapped out the LM301 opamp in that part of the circuit, and the thing started working.  But after playing with the unit and my oscilloscope for about 10 minutes, the thing quit again.  All the symptoms are the same so I'm assuming it's the opamp again.

Could there be something I'm doing wrong with this thing to burn out that opamp?  The output connection is labeled 50 Ohm out, and the manual describes connecting a 50 Ohm load to it and getting the frequency output off of that, but is that load mandatory?  Is it harming the opamp to not have that load?  Or would that not be the case since the signal has to go through so much other circuitry before even reaching the output?  While I have several more LM301s, I don't want to have to keep replacing them.  I really don't understand 50 Ohm terminators or when or why they should be used.

[hagster]

Without a load all the power you siggen creates is reflected back and dumped in the output amplifier.

However most sig gens I have used have been designed to tollerate any passive load you can attach(from open to short).

One common way to protect an output stage is to add a 2 or 3dB attenuator in line. The reflected signal gets a double helping of that attenuation (i.e. 4 or 6dB) before hitting the output amplifier.

[Skyfox]

I don't know what that is.  Is that something I'd have to buy or make?

[DaJMasta]

Either would work at these frequencies, an attenuator just reduces the output signal, that said, I think there probably is something with that function in the design, otherwise running it without anything connected would just burn it out - doesn't seem likely test equipment design.

Did you check the rails on the output amp you replaced?  Maybe they're running a little over and it's gradually cooking the chip - could be the cause of the last failure as well.  Or maybe the input level for the amp is over/under spec and is damaging the input stages?


Using the right load on the output will keep your distortion down, but while reflected power can be an issue, I'd assume using a load that was too low would be a bigger issue, if there isn't a current limit on the output amp you could cook it by trying to draw too much.  Still, not really what I'd expect from test equipment designed for use.

[Stray Electron]

   I think you need to take a CLOSE look at where your replacement parts are coming from.  There are a LOT of cheap copied parts out there that sort of work but not for long.

[David Hess]

The output termination has no effect on the voltage control loop and properly designed function generators do not care about output terminations.

Look for some other cause of the problem.  Exactly which LM301A (1) is failing?  Check that the plus and minus 15 volt regulated supplies are correct.

(1) There is no difference between the LM301 and LM301A in this application.  The LM301A has improved input bias current characteristics and was released just a year after the LM301.  The LM301A should actually improve some of the specifications.

[Skyfox]

Thanks all for your input on this.  I'm attaching a clip of the schematic with this opamp in it.  To answer questions:

The voltage rails on the opamp are what they should be; +15V on the positive and -15V on the negative.  The input+ and input- are both 0V ±0.3 mV, and the output varies from -2.8V to -11.3V when adjusting the main frequency pot (R8 at the top of the schematic).  Pin 1 (balance/compensation) is -13.7V (so is pin 5, Balance, but that's not connected to anything).  Pin 8, Compensation, is -7.9V.  The LM301AN can take a maximum supply voltage of ±18V, so the rails are within limits.  I do have to point out that these tests are done on the now-bad chip, and while the first test showed about -.25V on the input-, a second test showed it sitting at 0V.

As far as the replacement part source, it's from a huge cache of brand new chips a friend of mine got at an auction from the electronics maintenance department of a large company, I think a pharmaceutical if I remember right.  I'd tend to think they would buy from known good suppliers.  It's a National Semiconductor chip, just like the one it replaced, and the date code says 849, which I take to mean sometime in 1984 (please correct me if I'm wrong in that).

It is in fact an LM301AN that I'm using, and which the old chip was.  All the system supply voltages check out as perfect, within ±.02V.  However, in running through the voltage checks in the calibration section of the manual, I found something that was curiously out of spec in a different part of the circuit.  In the trigger zeroing circuit there's a variable resistor for setting zero (I'll attach the schematic).  From either side of the resistor R192, it runs to the emitters of the PNP transistors Q42 and Q43.  When turning the variable resistor back and forth, the emitter voltage on Q42 goes up and down just as expected, but the emitter voltage on Q43 never moves.  It sits right at -1.8V.  Coming off the base of Q43 is N2, and N1 after the resistor.  N1 runs to another transistor somewhere else on the board which is where I found the faulty voltage point showing -2.5V when it's supposed to be 0V.  (N2 runs off to the input+ of an LM301AN opamp in the cap multiplier, which is showing -2.5V on that input+ pin and about -2.3V on the input-.)  I don't know if, or how, this voltage discrepancy might affect the LM301 that keeps burning out, but what I can't figure out is why the emitter side of Q43 doesn't budge when adjusting R192.

If this is all clear as mud I can attach the complete schematic with the spots I'm referring to highlighted for clarity.

[DaJMasta]

it may require lifting a pin, but it may be worth checking the output current of the LM301 - it could be that the thing downstream is pulling more current than it should and is causing them to fail early.  I sort of had it in my head when reflections were mentioned that this was the output amp, but if the LM301 isn't the output stage, it's definitely something internal to the circuit that's making it burn out.  If the voltages are good, then it could be inputs overrange (sounds like it is not), the output current is too high, or there's some something being applied to the output that it doesn't like.

An opamp can't get cooked quickly in too many ways if you've ruled out the power supply.


An unlikely one that may worth checking - is C4 dead?  Wouldn't expect a problem, but who knows.

[duak]

Are there any board to board connectors in your unit?

I have a similar Wavetek FG with a digital display that uses some pin & leaf connectors that after time got intermittant.   While I was diagnosing this I saw that one of DC supply voltages was way high.  Maybe one of the ground connections was blinky.  Anyway, cleaning all the pins & leaves solved the problem.

I noticed on the schematic that there's a 15 pF cap on the LM301.  Is it one of those little blue or amber 0.1" size guys?  I've encounted a couple of them over the years becoming intermittant or even breaking.  If this part failed the LM301 could oscillate.  I doubt it would fail but it would act strangely.

Cheers,

[Skyfox]

C4 is odd...with my L/C meter it measures at about 16.5 pf (10% high), but my component tester doesn't even recognize that anything's in it ("No, unknown, or damaged part").  That cap shows no cracking or other damage at all and appears perfectly fine. 

If something downstream is pulling more current then something downstream must have failed because this function generator worked just fine for a few years after I got it, but then over the course of a long span that I wasn't using it, it went from working to not working.  I don't know what could have gone bad while it was unplugged other than maybe the electrolytic caps, but they all look good.  And since the thing worked for about 10 minutes after I replaced the opamp the other day, it seems nothing downstream from the opamp is causing the lack of signal.  I'll see what I can do about measuring the current coming off the output.  I'm installing a socket to make things easier, so it shouldn't be too difficult to bend a pin out of the way of the hole.

The only board connectors are on a small daughter board with the sweep circuitry.  There's a long multi-pin connector with the main voltages, and two single post pins.  They all connect securely and are clean and shiny with no corrosion or anything.  All other connections are directly soldered, or a couple bolted down with lock washers and nuts.

In the past I've had a knack for finding just the right burned out part even though I knew nothing about the device I was working on (for example, my TV was salvaged from the trash after it was hit by lightning, and I happened to find exactly what was burned out even though nothing showed damage), but so far this function generator has me stumped.

Here's the two pages of full schematic from the manual.  The light green area is the spot with the trouble opamp.  The light orange area is where the trigger zero adjust is that's changing voltage on only one side of the voltage divider.  The voltages N1 and N2 coming off of Q43 are color coded for where they end up going elsewhere in the schematic.

[duak]

IC2 seems to bring a bunch of varied frequency control inputs together to drive the VCO.  Pots and trimpots can get blinky.  I'd consider performing a calibration of this part of the circuit to see if they do the right things or if they need cleaning or replacement.  Replacing the LM301 could have jogged things enough to make it work for a few minutes before it reverted to acting NFG.

Cheers,

[thedoc298]

Maybe the unit is just intermittent and the changing of the ic did not really fix the problem and the intermittent just came back. I would try chip #2 and monitor the temperature of it with my finger.   

[Skyfox]

Well, I'm just lost with this thing.  Nothing is getting excessively hot, I've tried swapping out the trim pot I thought was going bad and it made no difference, I've swapped out some of the transistors I thought could be going bad but that made no difference, and I can't find any place on the board that's even attempting to generate a frequency.  I can't even figure out in the schematics where the frequency is supposed to be getting generated so I don't know what to focus on.  It seems that LM301 opamp I thought was going bad might actually have been fine, since the output voltage does vary proportionally with the turning of the main frequency knob.  That gets fed into the current source circuit along with the generated control voltage, which outputs a voltage to the emitters of some NPN transistors in the trigger & gate logic circuit, which goes to both the squaring circuit and the timing capacitors and the cap multiplier circuit, and so on.

I forgot to mention, I put in another brand new LM301 briefly and the output was still flatlining, which kind of makes me think the opamp isn't the problem.

Duak mentioned the VCO, which I'm guessing means voltage controlled oscillator, but where in this schematic is the VCO?  I really want to get this thing working, but the frustration is making me tempted to get one of those cheap function generator kits from Banggood (or even go as far as building my own, because in that aforementioned cache of ICs from an electronics repair shop there are some function generator ICs).