Tektronix 2232 scope.... in need of a doctor..

[David Hess]

Procedure says you need a x10 attenuator which I don't have, a 50 ohm terminator which I don't have either, and a Function Generator with minimum 99% linearity, which I probably don't have since my Generator is an old 1970's basic analog Philips unit ! Never had a very stable amplitude nor Frequency. As for linearity well....

The display is 25 counts per division so the digitizer has a range of about 10.2 divisions.  The setup procedure is to start with a centered 6 division triangle wave which you can see and then double the amplitude to 12 divisions so it covers the entire digitizer range.

So for an informal test, the attenuators and terminations and 10mV/div and 20 mV/div ranges are not really needed.

BTW Dave, I gather from your previous posts here, that you have several 2XXX scopes ?

I only have 2232s and 2230s but I've worked one some of the others.

My 2232 is my only one of this era so I can't compare, but I am wondering about the big time base knob : I find it extremely "stiff"/hard to turn.. especially compared to that of my old Hameg scope which is absolutely feather light in comparison, like your typical detented rotary encoder, say, to give an idea.  So the Tek being so stiff, makes the knob feel a bit "spongy", every time I turn it, I am scared that it will (again) slide on its shaft and the dial loose track of things again ! Admittedly that knob has not one but TWO screws, 90° apart, to secure it to the shaft... but still, even though I tighten them quite firmly, it still doesn't feel "right".  Is this stiffness normal on these scopes ? Or is it just the switch getting stiffer due to age, and I could make it a bit smoother with some lubricant/contact cleaner spread liberally in it ?

It is not the knob but the cam switch which is causing the problem.  When the timebases are locked together, the horizontal control should be stiffer than the vertical controls but not spongy.  Individually the timebases on mine seem about the same stiffness as the vertical controls.

The cam switch bushings are just plastic and can be oiled but getting to them is not trivial.  They might have split in which case they can be sanded down a little bit and then lubricated but that requires removing the cam.

[Vince]

So for an informal test, the attenuators and terminations and 10mV/div and 20 mV/div ranges are not really needed.

Hi Dave,

Terminators definitely not needed indeed, since I don't have any, but the amplifier range most definitely needs to be at teh requested 10mV/div. I took the picture at 20mV/div so the waveforms fits on the screen, but running the tests like this results invariably in 100 missing codes or so. See screen capture below. Looking at the pattern the test returns (with flat lines at the start and finish), I guess these correspond to the top and bottom of the screen where no  signal is present at 20mV/div.  Switching to 10mV makes sure that the signal covers the entire height of the screen, and some more, so that the test doesn't risk missing codes because a signal that might not span completely the viewing area.

It is not the knob but the cam switch which is causing the problem.  When the time bases are locked together, the horizontal control should be stiffer than the vertical controls

Yes you are right, if I pull the knob to unlock the second time base, the knob feels perfectly fine, same as the vertical amplifiers.
OK so the stiffness in single time base mode is mechanically normal/explainable, I will stop worrying then, it's not going to fall apart any time soon...

but not spongy

Ah thanks, so I do indeed have problem there... see the video clip I just made a minute ago to make myself clear (hopefully) :

https://youtu.be/Pi-mr1mysAo

There appears to be some play in the knob, like 2 or 3 of the splines that are molded into the button. When you grab it and go rotate it, the shaft does not move immediately, it does not feel connected to the button... even though I have tightened the two screws really well.
So when you turn the knob it doesn't inspire a lot of confidence, it feels like every time you go to turn it, it might suddenly spin round ! 
So maybe the button is actually well secured to the shaft, and it's the shaft itself which some play inside the switch mechanism. In other word the mechanical assembly is worn out and it's not repairable.... or with difficulty.


Oh and the scope did it again : when I turned it on it was already in store mode and... the triangle was again not centered properly, again like yesterday, it started 2 small divisions below zero/ground level... and just as yesterday, after a couple minutes it progressively climbed to end up spot on in the middle, where I set it to with the trimmer before putting the cover back on.
So it must indeed be a warm up "issue", in quotes... A practical real world example that yes, when test gear manufacturers  tell you all the time to let the instrument warm-up before attempting any measurements.... it's not all bullshit, they designed the thing so they know a thing or two ;-)

Anyway, was nice to witness a practical example of this, something you can actually easily see. 


So, to sum it up, I would say this repair is done for now, but that there are still some things I might do later on, to polish it up even further :

- Take a deep breath and get to the time base switch assembly to see where the play comes from exactly, and if there is any way/hope of fixing it, so that it feels "just right" and become even more of a pleasure to use.

- Replace that bloody power supply connector on the digital board, if one day I come across that type of connector

- Hissing from the regulator coil : didn't do the measurements Dave asked for to see if the hissing was inherent to the coil, or if it could be a symptom of something not working quite right... so might come back to the SMPS later on to do that

- Have a spare coil now (and still the project of winding my own !) so could try it to see if it makes any difference.

- Then I could replace each and every electrolytic cap in this SMPS, not just the big ones on the secondaries, but all the the smaller ones in the regulator and inverter section. One of them might be bad and contribute to the hissing by some mechanism yet to be determined.  The smaller value ones I could replace with tantalum capacitors maybe, for better performance and reliability, to even further reduce that hissing noise maybe.

- I could record that hissing noise and run an FFT on it with the TDS 544A, once I have fixed it, and fixed the firmware issue which causes it NOT to have FFT enabled even though Tek's catalogs of the day  clearly say that the 544A had FFT as standard.  Could be cool to see the spectrum of that hissing noise, might give a clue as to what is happening in the regulator, and we could maybe correlate it to some ripple voltages probed here and there. I don't know... could be fun and educational.
Of course the "soft" FFT on scopes suck big time compared to a proper 'spectrum analyzer, but since we are in the audible range here/low frequency range, and we don't expect or require ultra high performance in any way shape or form.... well maybe the soft FFT of the scope will be good enough to actually see something useful. If not, I might repeat the experiment later on, once I have enough money to buy a real/proper spectrum analyzer. I have gathered by now that the "big" ones, that go up to 1 or 2 or 22 GHz, suck at low frequency, and that for very low frequencies they made dedicated "audio" spectrum analyzers. I guess they are bound to be cheaper.. so maybe I could hunt for one of those to start with, then get a "big one" later, to cover the higher frequencies, when I have saved some more money.

So in short... there is ALWAYS something you can do, always something to fiddle with and learn from, I like that, brain never ceases to be stimulated !

[David Hess]

So for an informal test, the attenuators and terminations and 10mV/div and 20 mV/div ranges are not really needed.

Terminators definitely not needed indeed, since I don't have any, but teh amplifier range most definitely needs to be at teh requested 10mV/div. I took the picture at 20mV/div so the waveforms fits on the screen, but running the tests like this results invariably in 100 missing codes or so. See screen capture below. Looking at the pattern the test returns (with flat lines at the start and finish), I guess these correspond to the top and bottom of the screen where no  signal is present at 20mV/div.  Switching to 10mV makes sure that the signal covers the entire height of the screen, and some more, so that the test doesn't risk missing codes because a signal that might not span completely the viewing area.

My point is that you could do the test at any vertical deflection sensitivity as long as you can produce 6 divisions of amplitude and then double it.  Switching from 200mV/div to 100mV/div or 2V/div to 1V/div would work.  Or start with a properly terminated x2 coaxial attenuator and then remove it.

Tektronix could have recommended using a 5 division peak-to-peak signal and then switching from 50mV/div to 20mV/div or removing a x2.5 coaxial attenuator.  What really matters is that the waveform exceed the entire ADC range of 10.2 divisions of which only 8 are visible by some amount.

but not spongy

Ah thanks, so I do indeed have problem there... see the video clip I just made a minute ago to make myself clear (hopefully) :

https://youtu.be/Pi-mr1mysAo

There appears to be some play in the knob, like 2 or 3 of the splines that are molded into the button. When you grab it and go rotate it, the shaft does not move immediately, it does not feel connected to the button... even thought I have tightened the two screws really well.
So when you turn the knob it doesn't inspire a lot of confidence, it feels like every time you go to turn it, it might suddenly spin round ! 

That is way more play than it should have.  On all of mine, the horizontal timebase controls are stiffer when locked together but each have the same play whether locked or unlocked.

Note that when the controls are unlocked, you can still turn the A sweep control by reaching around the B control knob and using the transparent skirt which conveniently has that raised thingy to push.

It may be best not to mess with the cam switches but I would undo the set screws and remove the knob to find out if something is going on there.  Perhaps look for a cheap 2232 timebase board online to mess with and get an idea of how the mechanics work.

Oh and the scope did it again : when I turned it on it was already in store mode and... the triangle was again not centered properly, again like yesterday, it started 2 small divisions below zero/ground level... and just as yesterday, after a couple minutes it progressively climbed to end up spot on in the middle, where I set it to with the trimmer before putting the cover back on.
So it must indeed be a warm up "issue", in quotes... A practical real world example that yes, when test gear manufacturers  tell you all the time to let the instrument warm-up before attempting any measurements.... it's not all bullshit, they designed the thing so they know a thing or two ;-)

Is the warm up offset different in analog and storage modes?  The worst of my 22xx oscilloscopes has a warm up drift of about 1 or 2 minor divisions in the vertical signal path and for the others it is smaller.  Also the entire CRT display shifts by a similar amount during warm up but that is not unexpected.

[Vince]

My point is that...

Sorry for the misunderstanding, we were thinking alike in the end

That is way more play than it should have

Ah thanks, so there is definitely room for improvement there ! 

Perhaps look for a cheap 2232 timebase board online to mess with and get an idea of how the mechanics work.

Hey that's an idea ! Sure would be nice to do that, will try and keep an eye open for such a thing on Ebay... if I can find something cheap enough, will go for it.

Note that when the controls are unlocked, you can still turn the A sweep control by reaching around the B control knob and using the transparent skirt which conveniently has that raised thingy to push.

Yep noticed that. it did save my beacon at the beginning, when the timebase was behaving erratically... I somehow didn't even think of readjusting the knob and tightening its screws, and jumped straight to the "the mechanism itself must be to blame" conclusion... grabbing the skirt was then the only way I had to actually use the scope at all !  Eventually got tired of it and decided to open the scope to have a close look at the switch assembly, then realized only the screws needed tightening, then put the cover back on, turned ti on and.... the SMPS blew and here I was starting this thread !   

Is the warm up offset different in analog and storage modes?]Is the warm up offset different in analog and storage modes?

I have not noticed any offset in analog mode. At power up the trace is centered and stay there. It's only in digital storage where it starts one to tow small division below the center line, before settling to correct level after 2 or 3 minutes.  I guess the thermals of the scope maybe at play. I mean, the analog front-end on the digital board is in a place which I guess must be much hotter than analog circuitry at the bottom front of the scope. So I guess it makes sens that any thermal induced offset would show up quicker and more prominently on the digital board than on the analog board ?!

Also the entire CRT display shifts by a similar amount during warm up but that is not unexpected.

Have not noticed that so either it must be minor on my scope and/or, most probably... I just have not been using the scope often enough and for long enough for me to realize this !   .... built the lab only 9 months ago, still building it so haven't really had time/opportunity to start using any of my gear for any amount of time, really. For now I am constantly hunting for old gear that might be of interest to me, then I dismantle them to seee how they are put together, learn from that, clean them thoroughly, refurbish all the little things that need attention, and trouble-shoot/repair them when need be. It's time consuming (and I have other things to do as well). I am getting there progressively, though ! I think in a year or so, the lab should be in a decent shape and I could start actually doing something with it, plenty of mod or design projects on my to do list.

[David Hess]

Is the warm up offset different in analog and storage modes?]Is the warm up offset different in analog and storage modes?

I have not noticed any offset in analog mode. At power up the trace is centered and stay there. It's only in digital storage where it starts one to tow small division below the center line, before settling to correct level after 2 or 3 minutes.  I guess the thermals of the scope maybe at play. I mean, the analog front-end on the digital board is in a place which I guess must be much hotter than analog circuitry at the bottom front of the scope. So I guess it makes sens that any thermal induced offset would show up quicker and more prominently on the digital board than on the analog board ?!

1 or 2 minor divisions is not enough to worry about.

Also the entire CRT display shifts by a similar amount during warm up but that is not unexpected.

Have not noticed that so either it must be minor on my scope and/or, most probably... I just have not been using the scope often enough and for long enough for me to realize this !

I noticed it from the shift in the readout position which is more apparent in storage mode because there are more readout elements.  Sometimes the text lines up perfectly with the graticule and sometimes it does not.

I forgot about it until just messing with it but the secret menu has an option to clear all of memory and restore defaults.  To get to the secret menu:

Press the “ADV FUNCT” button once.
Press the “SAVE REF 3? two times.
Use the cursor control.

One thing I miss on the 2232 is having the little "Tek" logo in the lower right hand corner of the readout like the 2230 has.  I thought the secret menu might have an option to turn it on but it does not.

[Vince]

1 or 2 minor divisions is not enough to worry about. 

OK so the scope is perfectly fine then, I am just being a bit too critical I guess ! LOL Still, after adjusting the trim pot, it's now spot on once warmed-up, so quite happy 

Also the entire CRT display shifts by a similar amount during warm up but that is not unexpected.

I noticed it from the shift in the readout position which is more apparent in storage mode because there are more readout elements.  Sometimes the text lines up perfectly with the graticule and sometimes it does not.

Hmm... yes, this I might have noticed indeed... I thought it was being temperamental, but well if your examples do it too then I will consider normal and not loose sleep over it.

I forgot about it until just messing with it but the secret menu has an option to clear all of memory and restore defaults.  To get to the secret menu:

Oh yeah I know about this menu, entered it 3 days ago when I posted here a picture of the Tek bug and wizard 

... but I did not pay attention to the other two options avaialble in thie menu !  Clear all MEM as tyou say, and also some option to set a straing/label for a "plot" something... not sure what it refers to. Maybe some text add to the waveform, when you ask for an XY plotter output. I am very find of XY plotter and it was one of the reason why I ghot attracted to the 2232 model, am I not ashamed of admitting !   So I plan on hunting a little/cheap XY plotter just so I can hook it to the 2232 !    I saw videos of this on youtube, works reasonably well if you don't ask the 2232 to plot the waveform too fast. Luckily you one can adjust the speed at will, ISTR.

One thing I miss on the 2232 is having the little "Tek" logo in the lower right hand corner of the readout like the 2230 has.  I thought the secret menu might have an option to turn it on but it does not.

Well that's one project for you then : read the contents of the two EPROM containing the firmware, disassemble it, reverse engineer it, and I guess with sufficient time (10 years  of hard work ?) it might be possible to modify the firmware to add what you desire, without ruining the scope. What do you mean, it's not worth the time and effort ?! 
In these days of open software and hardware, how great would that be, if Tek and other major brands, would be so kind as to make all their documented source code available on their website, for all their out of warranty products ! 

[Vince]

Bench was a bit of a mess after nearly 2 months working on that scope ! Now that the repair is over, did some cleaning/house keeping.

Put all the remains from the scope repair in a box (fans, news components for the regulator and caps for the storage board, remains of the old coil, plus the 2 new ones), along with all the notes I wrote during the repair, some data sheets. So, everything is being archived... will come handy next time I have to put my nose in there !

Obviously this very topic is an archive in its own right ! Along with all the pics and notes I wrote on the computers, and lots of datasheets.

The "Tek 2232" Chapter is now closed, until further notice ! 


Was quite a ride, my first repair of a complex/professional piece of test gear was not without its ups and downs !  But I sure learned a lot. Most important thing is likely to be : PERSEVERANCE ! Good things/outcomes rarely come easy... you have to earn them one way or another...

[David Hess]

Maybe some text add to the waveform, when you ask for an XY plotter output. I am very find of XY plotter and it was one of the reason why I ghot attracted to the 2232 model, am I not ashamed of admitting !   So I plan on hunting a little/cheap XY plotter just so I can hook it to the 2232 !    I saw videos of this on youtube, works reasonably well if you don't ask the 2232 to plot the waveform too fast. Luckily you one can adjust the speed at will, ISTR.

The hidden plot label function adds a custom line of text to the plotter output.

At least twice online, I have run across hardware HP plotter emulators which accepted RS-232 and produced graphics files with a specified resolution on a memory card which could then loaded into a computer.  A software application could do this very easily as well and I used to use one which accepted HPGL.

One thing I miss on the 2232 is having the little "Tek" logo in the lower right hand corner of the readout like the 2230 has.  I thought the secret menu might have an option to turn it on but it does not.

Well that's one project for you then : read the contents of the two EPROM containing the firmware, disassemble it, reverse engineer it, and I guess with sufficient time (10 years  of hard work ?) it might be possible to modify the firmware to add what you desire, without ruining the scope. What do you mean, it's not worth the time and effort ?! 
In these days of open software and hardware, how great would that be, if Tek and other major brands, would be so kind as to make all their documented source code available on their website, for all their out of warranty products ! 

I wonder though why they removed it.  Maybe they ran out of ROM space?

[Vince]

At least twice online, I have run across hardware HP plotter emulators which accepted RS-232 and produced graphics files with a specified resolution on a memory card which could then loaded into a computer.

That sounds quite clever !   I won"t be needing this on the 2232 though, seeing as it has a GPIB port and waveforms can be retrieved this way.
I am presently trying to setup the GPIB stuff in the lab, hoping to get something working while I am in vacation. Zero prior experience so the learning curve takes some time.

One thing I miss on the 2232 is having the little "Tek" logo in the lower right hand corner of the readout like the 2230 has. [..] I wonder though why they removed it.  Maybe they ran out of ROM space?

EPROM space sure would have been at a premium ! Now that said, by teh sound of it the bitmap picture was probably very small so maybe they could managed to cram it in to the EPROM(s) if really tehy cared to.  Never seen this logo in action, care to power up one of your 2230 and post a picture for me to see it ? 

[Vince]

Come to think of it... GPIB sure allows to read stored waveforms but ... the "plot" function is different since it also includes the readouts. I am not sure GPIB can get that... would have to read the manual closely to figure out what GPIB can do exactly....

[David Hess]

EPROM space sure would have been at a premium ! Now that said, by teh sound of it the bitmap picture was probably very small so maybe they could managed to cram it in to the EPROM(s) if really tehy cared to.  Never seen this logo in action, care to power up one of your 2230 and post a picture for me to see it ? 

The text is written using vectors although I have no idea how they are encoded on the 2230/2232.  With some cleverness, each vector can be implemented with 3 bits.  (1) I think the late 7000 series digital vector readouts and the 2230/2232 readouts have the best look.

I just replaced the batteries in my camera so here are 2230 screen photographs of analog and digital mode.

Differences between the 2230 and 2232 include:

1. The 2230 has no graticule illumination.
2. The 2230 readout does not display the trigger level.
3. The 2230 has the Tek logo in the lower right corner.
4. The 2230 does not implement 20 MHz bandwidth limiting in DSO mode. (2)
5. The 2230 does not store its configuration in NVRAM.  Settings are always restored from defaults on power up.
6. The 2230 lacks the rather clever "smooth" mode available on the 2232 when using peak detection.  Tektronix might not have thought of it at the time but also the 2230 has a slower processor than the 2232 likely making it infeasible.  I usually leave smooth mode on the 2232 disabled for performance reasons.
7. The 2230 has a very different DSO interface.  The 2232 interface is recognizably like more modern DSOs.  Part of this is simply because the 2230 lacks the 5 buttons under its CRT.

(1) See the change notice for the readout on the 7904A oscilloscope for details on 3 bit vector encoding.

(2) Like most analog oscilloscopes, the 2230 20 MHz bandwidth limit is implemented after the vertical channel switch and this is after the trigger and digital pickoffs.  Tektronix recognized this as a problem in a combined analog and digital storage oscilloscope so the 2232 fixed this by implementing the 20 MHz bandwidth limit in the vertical preamplifiers before the trigger and digital pickoffs.  After studying the various ways to implement switchable bandwidth limiting, I get the feeling this this was considered a hard problem up until the 2235 series including the 2230 and 2232 which make it look trivial and that the secret was later lost.  The modern designs I have seen are worse.

[Vince]

Thanks for the piccies Dave 

... and for thje good into as usual !

I always wondered what exactly differenciated the 2230 form the 223é, aside from the missing bezel buttons which was the only obvious thing I could see on pictures. Now I know ! 

Strange about the missing Tek logo, maybe it's just some random marketing decision they wanted to try out, then every body complained and they put it back...
Maybe they thought it was too intrusive, was interfering with the trace/signal too much, so they removed it. Then later marketing said No, we need to have a logo present at all times, so later in the TDS series they put it back, but this time in the upper left corner where it doesn't get in the way so to speak.
I don't know, there must be reason, but we will never know !