Tektronix 49xx Spectrum Analyzer series power supply failure.

[Gazucha]

Hi all,

 I recently met someone with a shed full of old kit he wants to sell. In amongst everything were 4 1980's Tektronix 492 Spectrum Analyzers which all have a failed power supply (PN 337-2553-03).

  I've brought one PSU home to open it up and found approximately 14 bad caps, which have now been changed. I am suspecting that there are other problems as there are 5 internal fuses, none of which are reading any voltage.

What I am looking for, is someone who has the repair schematic or otherwise experience in repairing these PS units. The 49x manual doesn't go into much detail regarding these PSU's, such as which voltages should be found where. 

Any help would be great, as I have 2 more days before I meet up with the fella and test out any repair.

[Gazucha]

Here's more photo's to help get a better picture of the PSU module, which as can be seen is plug and play type.

 The first small pcb pictured I am assuming to be the fan controller.

[JoeO]

Schematics and info:

http://www.ko4bb.com/manuals/index.php?dir=Tektronix/Tektronix_-_492_Spectrum_Analyzer

[Gazucha]

Hi,
 but as I said, there is nothing too detailed regarding the 49xx PSU.

 Been doing some further testing and one of the 2 large NPN transistors is blown( the one nearest the fuses). The transistors are bolted through the board, between two large caps in the middle, on the same side as the fuse bank. (See post 1 picture 2)

 As mentioned here ...

According to a Usenet post (sci.electronics.equipment, 2-Sep-01), the large stud-mounted
transistors with Tektronix part number 151-0703-00 cross-reference to the industry-standard part
number 2N6586 (10A/450V, 12.5 MHz fT, TO61 package). With the appropriate mechanical
modification, 2SC4237 transistors have also been successfully used


 Would the bad caps have blown this transistor or should i check something else nearby?

[madsci]

Those transistors appear to be the primary side drivers.

They die because they get hot under operation and then cool when off. Over time this causes flaws in the semiconductor and the device fails. It's often that simple. If there are any back EMF clamp diodes, check those too. You may or may not have a blown mains fuse in this failure mode.

A shorted transformer or rectifier can also blow them. Usually this will be accompanied by a blown mains fuse. There is a current transformer in the primary side switch supply. I presume this is for overcurrent protection and if so, it *should* be protected from this failure mode. (I've seen enough SMPS failures to know that it's not foolproof!)

Check the control circuitry and the base drive waveforms. If it's push pull, there MUST be dead time between the waveforms or you will have a short circuit when both transistors turn on. Again, you'll likely see a blown mains fuse. This also usually kills all the power transistors feeding the load.

Check the output of U6069 and the two little transistors it drives. These feed a small transformer that feeds the drive signals to the main switching transistors. At the output of this transformer are some filters. Check those 47uF caps and 4.7 ohm resistors. This is probably an oscillation prevention circuit and if it dies, the gate drive waveforms will be severely disturbed.

Their output seems to be fed through C1063. There's also some extra damping across the primary with R3064 and C3074. Check these.

[madsci]

Are your output fuses OK?
Is your mains input fuse OK?

[Gazucha]

Hi madsci,

 thanks for your detailed reply. 

 All fuses ARE still ok.
 The transformer is NOT shorted and also there IS dead time between waveforms.

 Will check U6069 and its transistors. The 47uF caps tested fine but will also check the resistors you mentioned.
 The plan for today was to pull and test the diodes and long white capacitor next to the failed transistor.

 One thing that may need modification is the transistor itself, as last night it looked impossible to find one here in Brazil. Perhaps will buy something cheap and compatible - just to test - before I start importing pricey components.

 I'll let you know if anything turns up fried.

 thanks again.

[Gazucha]

Hi madsci,

 rewind on that last post.
 
 My mistake, I hadn't checked those 47 uF caps ~ and they were both bad as was the 10uF nearby. The 4.7 ohm resistors read 5.7 - 5.8 ohm, so will change them too.

The 2 little transistors are good but I cannot locate U6069.

Off to hunt pieces...

[Gazucha]

Back from town and new caps and resistors installed...

Could not find a 2N6587 - nor anyone who thinks they even exist in Brazil!

In one shop where I asked for a 450+V 10A NPN transistor, I was given a C3182N and told it was good for 500V. I have just looked online and see that it is 140V...

Bummer!

But this is Brazil..... And we make do with people talking rubbish. 

Did find the 2SC4237 in town but got talked into the 3187 as it was half the price. 

 Please someone tell me I read it wrong and that the C3187 is good to use...

(edit to add that C3187 is a typo... meant to write C3182)


 

[madsci]

The original transistors are "air wired" anyway so have you considered retrofitting a TO247 plastic package device? The selection in plastic packages is vast! To install one, simply wrap tinned wire around the leads, solder and insulate with heat shrink tubing. You may need to drill a hole to mount it. Isolation with mica, silpad or Kapton may be required if the tab of the new transistor is connected to a different pin than the case of the original transistor.

U6069 is a CD4012 dual 4 input NAND located on the opposite side of the board from the main switching transistors next to the main transformer. Supply voltage should be around 10V.

At this point due to age and past history of this unit, all electrolytic caps are suspect. Do check the critical film caps though...the one in series with the transformer is critical to safe operation of the SMPS.

2SC3182? Don't even think about it....140V. It's for audio amps...it has a PNP complement and a pair is good for ~70W or so.

C1387 is a tiny TO-92 device. 300V 300mA

[Gazucha]

So, went back and changed for a 2SC4237. Have drilled the heatsink and now have all things nicely lined up for soldering.

 My only question is what to do with the ring washer which is soldered to the 1M Ohm resistor? To me, this had presumably been to ground the metal chassis of the original transistor, which had been totally isolated from all 3 pins, although it doesn't seem to be connected to ground?
 
 The 2SC4237 backplate is connected to the Collector pin.

Below are a few pics without the transistor in place (and one from before it's removal).

 
 

[madsci]

I assume that resistor is connected to chassis ground via that ring terminal. Consult the schematic to be certain.

[Gazucha]

Hi Madsci,

 unfortunately don't have the schematic.... But I was assuming the same thing. In fact, this little circuit connected the transistor chassis to emitter (via the 1Meg resistor).
 I was imagining the washer was to drain any noise from the transistors metal chassis, although the ring itself was insulated from the large heatsink by a nylon(?) washer and the heatsink was definitely connected to the 492 chassis ground.

 So if it was a 'noise' thing, would I need to connect the 1Meg resistor to the 4237? But where? The old transistor had a separate chassis, this 4237 has the chassis connected to Collector. Is it wise to connect Emitter and Collector direct, via a 1Meg resistor?
 Or, seeing how the 4237 has a mainly plastic casing and the rear metal plate is insulated from the heatsink by the nylon(?) washer, to disregard the 1M resistor?

 Other than this detail I am now ready to fire her up again....

[Gazucha]

 Back from a long weekend.

I assume that resistor is connected to chassis ground via that ring terminal. Consult the schematic to be certain.

Seeing as there was no chassis on the new transistor (2SC4237), I decided to simply disconnect and remove the 1M Ohm resistor.

 When I fired the PSU up, there is now the correct voltages on the probe pins, as well as full potential across all the internal fuses. So without getting too excited, we have definite progress, as these points were all dead before!
 
 We now have to wait until tomorrow to plug the module back in, as the actual TEK 492 is on the other side of the city.

 Big thank you madsci, for your input.

 

[Gazucha]

Hi all,

here's an update - which hopefully may help someone else in the future.

Re-connected the Power Supply module today and the Tektronix 492 fired straight up  . (although no display yet due to what I assume is a fault in the Deflection Amplifier).

 So as a re-cap, after much removal and testing of components, I changed all the capacitors outlined in photo 1, as well as the two 47uF and 4.7 Ohm resistors located just above the smaller transformer, along with the 10uF capacitor located between the 2 transformers. Finally, pulled the blown large bolt-through transistor (151-0703-00) and replaced that with a 2S C4237 which was fixed by drilling a new hole in the heat-sink.

 The 1 MegaOhm resistor and the washer via which it had been connected to the 151-0703-00 chassis, were removed (photo 2).

 All the diodes and other caps tested OK whilst any suspect-looking solder joints were touched up.


The text in Post #4 came from a 49x_notes.pdf which was found online. The rest of the Power Supply text is ...

General notes on power-supply service
Regulation tolerances for the low-voltage supply buses are specified in service volume 1, and are
not usually a problem. However, the +100V and +300V supplies are frequently below their
specified values. If any 66-kHz ripple or sagging is observed, all of the miniature 2.2 uF/200V
axial-lead electrolytic capacitors in the power supply should be replaced. These are inexpensive
high-failure-rate parts, so they should be replaced in any event if the power supply is otherwise
disassembled for service.
I recommend checking the ESR on any replacement high-voltage electrolytics as well. Several
"new" capacitors in this voltage range have proven defective when obtained from surplus
sources.
High ESR in the filter capacitors on the lower-voltage rails (+17V, +5V) has been reported as a
cause of excessive power-supply temperature, so when servicing the power supply, you should
check ESR on all electrolytics as a matter of habit. Check heat-sink fasteners for tightness, and
renew the heat-sink compound while you're at it. Avoid losing track of the mica washers and
aluminum spacers that may fall out when you remove the PCB.
Use 105 degree C-rated electrolytics where possible, e.g., 3.3 uF 350V 105C, Digi-Key part #
493-2046-ND. These can replace all of the smaller high-voltage capacitors in the power supply
(6x 2.2uF/200V, 1x 1uF/350V).


 As I said, hope this saves someone some head-ache.

[madsci]

I'm happy to see that 83 hours of university credit and a nearly life long obsession with taking things apart has helped you!

The following are some tidbits on cap selection in repairs:

I recommend not using surplus electrolytic caps as they may have sat for two decades or more in some cases and are just waiting to blow up in your repair projects. I bought a "new" cap from a small electronics shop and I was not amused when I saw a date code that meant 26th week of 1986.

Buy caps from a distributor to ensure that you receive fresh ones. You'll find that the price difference is negligible, especially on the smaller caps where you may actually *save* money. Experience has also told me that the smaller leaded cap seals are inferior to the seals on the screw terminal, snap in and soldering star equipped caps.

ALWAYS use low ESR caps in an SMPS filter. General purpose caps will fail (sometimes very rapidly!) and will give poor ripple control before they fail. Depending on the ripple current or frequency, you may need ultra low ESR caps. Ultra low ESR caps are uncommon in old designs or for places where large caps fit (the larger capacitance units are very low ESR inherently) and so a Nippon ChemiCon KY, Rubycon ZL or Nichicon PW series usually does the trick in all cases.

CPU VRMs are the most common item where ULTRA low ESR is required. The new ones have switching frequencies in the low RF range; 500kHz is not unheard of. Polymer, tantalum and MLCC rule these designs. The older 100kHz range stuff used ultra low ESR electrolytics like the Nippon ChemiCon KZJ, Nichicon HZ and Rubycon MBZ. Now, those are the minimum to filter the *input* to these regulators because we cant get suitable ESR from anything else in the size constraint. My newer PCs don't have a single aqueous electrolytic cap on their logic boards, it's all polymer and MLCC  with the occasional tantalum. The only electrolytics are in things like the disk drives, the card reader and of course, the PSU.

Small caps (under 100uF) with orange or yellow jackets next to caps of the same brand and size that are a more typical color are probably low leakage (low self discharge) types. If they are performing coupling duties, they are almost certainly LL types and should be replace with another LL electrolytic or film type cap. Advancements in film caps in regards to size have allowed me to replace almost all sub 2uF electrolytic caps in audio gear and SMPSes with film for a substantial boost in reliability.

[Gazucha]

Thanks madsci for your helpful words and for sharing your experience.

As you know, it's always a good feeling when something 'dead' comes back to life.

Now we're on to the Deflection Amp, where I found a burned LM318. Hopefully that is the last problem (don't suppose you've ever repaired one of those, no?). I started a new thread especially for it...

We'll see.

Anyway, thanks again fella, you saved the day!

 Gazucha.

[madsci]

Dead LM318....that's unusual and tells me the original fault lies elsewhere. It's probably driving a larger amp, possibly a larger IC or a discrete transistor based unit. Check that. Failures in a direct coupled amplifier often take out the previous stages. For example, it is common in audio amps for the drive stage to fail when the output stage shorts.

The deflection amp is in many respects, quite similar to an audio amp.

[Gazucha]

Hi madsci,

 it was mentioned that there may be a prob on the output, so after first testing 'in circuit' I pulled one of the output transistors which then tested the same on the bench. All the others were then tested in circuit. Certainly no 'short' on any of the larger transistors.

So that you know ~ the other five LM318's all have very clean and solid soldering, whilst the two 16 pin LF13333N's on the board have quite weak and dirty solder connections. Perhaps one of these opened the circuit? The problem is that there are internal traces within the pcb, which without a schematic leaves me in entering the realm of guesswork.

Another member, Orange, mentioned that he normally just changed these boards but it would obviously be much better (here in Brazil) to fix it. It's just a bit of a ball-ache that the 492 is on the other side of Rio, which means a good 4-5 hours return travel just to test any repairs I may do.

 One last thing before things get any more involved... Am I correct in assuming that this burnt Deflection Amp is what is causing the no-display symptoms?
 Or is it not possible that something else, between the Defl Amp and CRT, caused this failure and therefore the blank CRT? Or could I have missed something in the PSU and that is where the failure is coming from? (this is the problem with no service manual)

 Maybe it's time to join the Tekscopes yahoo group?
 

[Dan Gibbs]

Hello All
I have one of these Tek 492 SA's here at work that was dead when I started this job. I'm trying to get it working again and have poured over this post and followed the recommended parts replacements.

Our unit had a blown Q2061, diode CR2065, and mains-fuse along with some aged-looking capacitors. I ordered replacement 2SC4237 transistors from a trusted vendor and mounted them securely and isolated to the heatsink for the originals. I replaced all electrolytics except for the two large 750uF/200V units. I replaced the two small 47u/6v tantalum C5084 and C5085 caps. I have a bag of NOS 6A100 6-amp 1000V diodes and used one to replace the shorted CR2065 diode.

I reassembled and powered the unit on after double-checking everything I could think of. It ran for 3 or 4 minutes and seemed fine before blowing the mains fuse again. It blew the exact same parts as when I found it - Q2061, CR2065, and the mains fuse - no more and no less.

I'm considering trying to power the supply on it's own and temporarily fusing the collector and the base of the replacement transistor for testing purposes - I hate the idea of replacing transistors like fuses if they keep blowing before I can get to anything to test for the remaining problem.

Does anyone have any ideas of what else might be wrong? Again, Q2071 and the other diodes never burn out, only Q2061 and CR2065.

Thanks,
DG

Update five hours later... I wired in new Q2061 and CR2065. I fused the collector with 8A fast-blow fuse and the base with a 4A fast-blow fuse. I also fused the CR2065 diode with a 5A fast-blow fuse. Replaced the main 4A fast-blow fuse. I powered on the power supply module by itself disconnected from the main chassis. It is unloaded of course but has been running fine. It tells me a little but not much that it runs by itself - unloaded of course