Calibrating and repairing a function generator versus safety

[McMonster]

I've just watched video #217 and I've decided it's time to repair/calibrate my Wavetek function generator, but I'm concerned about safety of this operation. There are informations about this scattered through different videos and a separate one on isolation transformers and I'm going to watch those again, but are there any other practical guides to basic rules of safety when repairing or calibrating bench equipment? My health first, not destroying anything if possible.

More specifically, I have Wavetek 172B function generator which is working, but on higher frequencies (say above 1 MHz and it can go up to 13 MHz) all waveforms start to smooth out until they look almost like a sine wave at 13 MHz. I have the service manual and I intend to start with checking the voltages. The most dangerous place is 250 V in the display section. The manual includes all the schematics, location of pots and test points, detailed service procedures etc. My gear include two scopes, two meters (one decent and one cheap, 50$ one), common sense and a plastic calibration tool.

Any hints abouth both safety and possible ways to repair the generator?

[sub]

It sounds like you are seeing the effects of the output filter.  Are you certain that this is indeed a fault?  If you have been making 13MHz square waves for the last decade on the device then something would certainly be amiss, but the frequency response in the link below suggests that the harmonic content at higher frequencies might be rather diminished:

   http://www.teknetelectronics.com/DataSheet/WAVETEK/WAVET_172B105856.pdf

[rf-loop]

1. Output is specified to 50ohm load.
2. Square wave rise and fall time is 15 - 20ns. (specs value <20ns)
(it also means, read first user manual and specifications before you thinking it need repair or adjustment. Most time peoples first use equipment and reading manual is last point - or more worse: sometimes  peoples make some "this is bad" rewievs before they know how stuffs work in principle and what is specified) . but     also this is natural )

One cycle have 2 edges. Worst case 40ns + this time what need between 0% to 10% and 90% to 100%..
If you have sine where cycle is 40ns what is frequency.
1/.00000004
(25MHz)

But normally risetime is specified so that it is time between 10% and 90% points. (sometimes but more rare 20% to 80%)
If you have square there is more time just becouse it is measured as 10 and  90%.
It can think something like 2 tr + 0.85 tr. (not exactly but just as practical imagine) It means 40ns + 17ns = 57ns. 57ns cycle time sinewave is around 17.5MHz.

With this risetime maximum frequency is.  17.5MHz 20ns risetime square looks nearly as some amount distorted sine in practice and level also may be dropped around 3dB. (just also with gaussian BW shape oscoilloscopes there is equation 0.35/tr = freq.)   If with this risetime still rise frequency signal level drops more than 3dB and there can not really see any square.


So you can easy imagine that 13MHz "square" looks not square at all. It looks very soft corner and with eyes it looks more sinewave as square wave.

So, this is nominal for this generator.
And it all reads very clearly in specifications.

[McMonster]

Thanks for the info, I'll analyze it later and do some more checking with DS1052E (only tested it with analog scope before). I read the manual before, but I was not completely sure if it really needs calibration because I have nearly no experience in the matter of signal generators and high frequencies, I've only worked with less than 500 kHz before. All I knew for sure is that this unit spent a few years in poor storage conditions, was sold as untested (got it for less than half the price of a good one) and waveforms looked very bad for my liking.

I guess I have a lot to learn.

[alm]

I'm impressed that rf-loop can determine that the generator is within specifications without even looking at it. That would save a bundle on calibration costs!

Based on the specs, it sound likely, but it's hard to tell if that's all that's going on based on just a description. It should be easy to measure rise time with a DSO. Ideal connection would indeed be a coax cable with 50 ohm terminator on the scope side, but given the slow rise time, I don't think it's that critical. Keep in mind that the amplitude will be double the indicated amplitude without termination. Just sticking a probe in the output with a BNC-to-probe tip adapter, or even just a regular ground lead, should give a fair representation of the signal. Then set up the scope to measure rise and fall times (10% to 90% if you have to specify the levels), and it should give something close to the 20ns. Note that aberrations like overshoot can screw up the rise time measurements.

There's no harm in doing a performance verification (calibration procedure without doing any adjustments). Best is to use an isolation transformer between mains and the DUT (function generator), especially when working on the primary side of its transformer. Keep your fingers away from the high voltage bits. Be careful that in the past it was common to route mains connections over the main PCB near the low voltage section. The supply to the display is likely fairly low current and much less dangerous than mains. If you feel nervous about reaching a test point, clipping the probe on while the power is off and turning the power on with your hands away from the circuit is another option. Your interaction with the high voltage parts is likely limited to measuring the voltage. Be careful not to let a probe slip, accidentally shorting something can damage the DUT.

[rf-loop]

I'm impressed that rf-loop can determine that the generator is within specifications without even looking at it. That would save a bundle on calibration costs!

Based on the specs, it sound likely, but it's hard to tell if that's all that's going on based on just a description. It should be easy to measure rise time with a DSO. Ideal connection would indeed be a coax cable with 50 ohm terminator on the scope side, but given the slow rise time, I don't think it's that critical. Keep in mind that the amplitude will be double the indicated amplitude without termination. Just sticking a probe in the output with a BNC-to-probe tip adapter, or even just a regular ground lead, should give a fair representation of the signal. Then set up the scope to measure rise and fall times (10% to 90% if you have to specify the levels), and it should give something close to the 20ns. Note that aberrations like overshoot can screw up the rise time measurements.

There's no harm in doing a performance verification (calibration procedure without doing any adjustments). Best is to use an isolation transformer between mains and the DUT (function generator), especially when working on the primary side of its transformer. Keep your fingers away from the high voltage bits. Be careful that in the past it was common to route mains connections over the main PCB near the low voltage section. The supply to the display is likely fairly low current and much less dangerous than mains. If you feel nervous about reaching a test point, clipping the probe on while the power is off and turning the power on with your hands away from the circuit is another option. Your interaction with the high voltage parts is likely limited to measuring the voltage. Be careful not to let a probe slip, accidentally shorting something can damage the DUT.

My answer was related to this what McMonster wonder  and yes it is typical for this model of wavetek:

McMonster: "I have Wavetek 172B function generator which is working, but on higher frequencies (say above 1 MHz and it can go up to 13 MHz) all waveforms start to smooth out until they look almost like a sine wave at 13 MHz."

rf-loop:

littlebit about what specs meaning and then:

"So, this is nominal for this generator.
And it all reads very clearly in specifications."

I did not tell it is inside specifications if meaning real calibration check.
but I do not know what all there reads invisible in my text, but some peoples may read these also - I can not.

[McMonster]

I guess I'm out of luck. I tried turning the generator on and no display or beeper response and it was working when I powered it last time like o month ago. I've taken it apart and checked voltages as manual said, turned out I have a shorted digital +5 V supply somewhere, bad thing is that all the internal boards use this voltage so I'll have to disassemble it completely and check all of them. I hope it's just some passive component or 7400 series chip and I'll be able to fix it, I don't think I'll be able to get any spare boards if it's a broader failure.

And I must say I really love this classic design. All through hole components, tons of 7400 series chips, some old style, violet/gold DIP chips, and I haven't seen any bogdy things as far, but I've only checked a few boards. Simply sex on a stick.

[alm]

My money is on a shorted tantalum capacitor. Measuring voltage drop over traces with a sensitive multimeter while shoving some current through there may help you locate it.

[McMonster]

No luck again. I've taken the thing apart and measured voltages as manual said (again the manual is brilliant, no guesswork required). I've discovered digital +5 V line is probably shorted to ground (reads about -0.38 V). Manual said to take each board using this supply (that means all of them) and check the power rails with ohmmeter for shorts. And that's where the problem is, I've checked all the boards and didn't find any shorts, they read from kiloohms to hundreds of kiloohms.

@alm, thanks, I'll check this, there aren't many tantalums there fortunately.

I couldn't visually locate any failed component, the only bad thing I've noticed is rust on pins of some of the 7400 chips in the front panel. Do you have any ideas where to look next? Did I miss something or did wrong? And is there a way to check 7400 chips for failures without desoldering them from the circuit?

[alm]

Can you check the +5V with the load disconnected? The very low voltage (-1 diode drop?) suggests a very low resistance somewhere, so the fact that none of the boards have a low resistance suggests to me that the power supply might be at fault. Maybe a dead regulator? Can you measure the current from the +5V supply? Are the other voltages correct?

A dead tant tends to be either low resistance or shorted.

[McMonster]

All other voltages are within specs. I've located the regulator, it would require some desoldering and hacking in a wire to measure current, so I'll do it tomorrow. Can resistance between pins be a hint for dead regulator? I grabbed a fresh 7805 from my parts bin and it has megaohms between pins 2 and 3 while the one on the regulator board has a little over 4.7 kiloohms between the same pins (no other connections on the board, tab insulated from heatsink).

[alm]

5k sounds low to me, but testing the regulator with no load should be a good test. It should produce a stable 5V then. Did you also check if the input to the regulator is correct, i.e. a 7V+ DC signal with a moderate amount of ripple? Best to use a scope for this. It could also be something like a rectifier with a dead diode, bridge rectifiers also seem to die from time to time. I would expect the trouble shooting procedure in the manual to suggest this.

Is the -0.38V you measure on the 5V rail actually DC or is there AC on the 5V rail? That might indicate a dead smoothing cap.

[McMonster]

I'll do more tests tomorrow, but I analyzed the schematics for the board and found out that there's a bridge rectifier (1N4002 diodes) supplying the 7805, but label and the rest of schematics and diagrams say there's 5 VAC going into the rectifier directly from one of the transformers, isn't it too low?

I only tested this bridge rectifier for now, all the diodes are good.

[McMonster]

I've finally found some time and motivation to take this thing apart again just to encounter some Murphy's laws being in action.

I've desoldered the 7805 regulator and checked it separately, it works fine, then put it back onto the supply/transistor pass board and soldered some wires to measure input and output voltage and current. Kill me, but those were perfectly fine, 9-10 volts in and 4.95 out, about 0.38 amps at the output, but the digital supply line test point on the regulator board was still -0.3 volts!

It sounds to me that either I screwed up reading the schematics and digital +5V gets its input from elswhere or there's a broken connection somewhere in the regulator board and there's no current reaching the rest of the device, but something's still drawing some current. Either way it'll be a pain in the ass to probe for it because of the base interconnection boards and daughter cards construction, not enough space to stick a probe and check it while it's running, I would have to turn it off, remove the board and change the probe connection point each time.

A long night before me, any help and suggestions are greatly appreciated. I can paste a link to the service manual if you want.

[don.r]

I feel your pain. I have a Wavetek 144 with a busted 24V supply. Luckily it reads low even when disconnected from all the other boards so I have a new 7824 on the way. They are luckily excellent units to work on and beautifully put together: a real learning experience to debug one of these for the novices as they are usually fairly simple circuits made from through-hole discretes (save for the pulse logic).

[McMonster]

I absolutely agree. Though there's one more ugly thing with this kind of construction, mutiple schematics covering multiple boards, so you can see one schematic showing parts from three boards and vice versa. Add multiple offboard power signals and a headache guaranteed. Sometimes the signals are hard to trace over this mess, but most of the time they do a good job with labeling and marking board boundaries.

[McMonster]

I think I've found it after halfway disassembling of the generator. It's is in fact a classic short on the motherboard. I have 4 suspicious 220 uF 16 V  electrolytics between +5V and GND rails reading stable 6 ohms in both directions between them. I'd apreciate if someone could confirm if this sounds valid.  After this the daunting task of disassembling the generator almost completely to lift the caps and check them...

The trick was that they were hard to notice under the daughter boards (axial leads) and the motherboard does not have a full schematic (too complex) but just the signal listing for the slots and a note indicating the existence of those caps. I simply didn't expect so plain and boring board to have a faulty component.

I've watched Daves latest video just a few hours ago. Murphy would surely laugh at me. I did.