'Scope 2

The new Telequipment DM63 (19K)As a result of posting my experiences of the DM63 on this site, I've received a surprising amount of e-mail concerning this unit. This was nice because I felt sure there was an audience for this material, however small. I've also received offers from a number of people, and as a result I now have another DM63, and a few interesting goodies such as a V3 Differential Plugin.

In the course of overhauling the new arrival, I've discovered a few new faults, so hopefully this section can continue to be of some use to all the Telequipment users out there.

The picture opposite shows the unit as it arrived on the bench. It had been stored in a garage for the last few years and is in need of a good clean and some TLC. But it basically appeared to be in sound condition.

Just a minor fault...

Screenshot showing the ripple (9K)Upon powering up, a trace appeared, but with a lot of hum! I quickly found that I could trigger on the mains hum, so at least IC1 was OK. This screenshot gives you an idea of the problem. It's pretty bad...

Another problem that happened intermittently - the horizontal sweep would sometimes collapse to just a couple of divisions wide, and display nothing meaningful. But while it was clearly in need of some major power supply work, and the timebase section needed a good looking at, the tube appeared to be in good condition - the trace was nice and sharp and the storage modes seemed to work really well.

The power supply

A quick check showed that the +24V rail had serious amounts of 50Hz ripple. This wasn't quite what I was expecting - dead smoothing capacitors would exhibit 100Hz ripple on the output. Also, the heatsink associated with the two 24V regulators seemed too cool, at least based on how my existing scope behaved. So, I completely removed the PSU module from the unit and removed the PCB from the metal chassis for a closer look.

Faulty PSU (16K)
Faulty PSU (15K)

As you can see, C405 is a bit unwell! But look closely at D406 - yes, that is a crack! So, this explains why there is 50 rather than 100Hz ripple on the output.

Here's the cleaned up PCB, complete with a new set of diodes - I changed them all to eliminate any doubts about their future reliability. The other picture shows the state of the old capacitors, and compares them to a modern replacement. If you look carefully, you can read "7805" on the cans - this is a date code (week 5 of 1978), so it's perhaps not surprising they're a bit past it!

Better PSU (15K)
Yukky capacitors (19K)

Once I'd finished visually checking and rebuilding the PSU, I decided to not reinstall it in the 'scope mainframe just yet. Instead I powered it up on the bench so that I could take some voltage measurements - just in case... I was glad I did! The +24V output was more like +35V!

Upon investigation, it seemed the error amplifier wasn't working. TR408 was dead - base-emitter short - so the pass transistor TR405 was saturated and passing the full unregulated output. This explains why the heatsink was too cold before.

The motherboard

While the power supply was out, it was time to attend to the motherboard. The first task is to change all the electrolytic capacitors - this shouldn't need explaining or justifying.

The motherboard (21K)
Closer view of the motherboard (21K)

These pictures shows the motherboard, free from the brackets and fixings but still attached to the wiring harness. As you can't easily remove it completely from the instrument, it's tricky to work on. You can see the connector for the PSU on the bottom of the PCB, and the pins belonging to the vertical plug-in connectors. This board contains the Y output stages - you can see that the output transistors for channel 1 have been replaced at some point in the past. Unfortunately, the replacements (with their incorrectly fitted heat sinks) aren't up to the job, and the 'scope only manages to reach 7MHz on this channel! I've done a brief survey of a number of these 'scopes and found a variety of different transistors fitted in this role:

Transistor Specifications Comments
FRB750 ? The originally-specified device, in a small plastic package. Often fails.
BD419 100V, 1A, 2W, 75MHz Fitted to serial numbers 739601 and later. Has a larger package (X17) with a heatsink tab and hence is better able to cope with the heat.
BC489 80V, 1A, 0.6W, 100MHz TO92 plastic package
BCX49 80V, 1A, 0.6W, 100MHz TO92 plastic package
BFX84 100V, 1A, 0.8W, 50MHz TO5 metal package. Not fast enough! (See above)

Looking through the various versions of the manuals manuals I can find, it seems that the BD419 and FRB750 devices were the only "official" devices. The BD419 is the safest bet because it is best equipped to deal with the heat dissipated, but I haven't managed to source it yet. The problem is the pin-out, as the base terminal is in the middle - this is unusual, as most power devices have the collector in the middle. If you're willing to bend the leads into the required order, you probably can find a large selection of transistors that will work well - devices used in video output stages would be a good bet.

It's worth saying that the power supply uses two FRB750 devices, so you could probably "rob" these for use as Y-output amps. In the power supply, they are being used as the first half of a Darlington pair for the ±24V rails - TR406 and TR409 - and I'd suggest that this is a reasonably non-critical role for these. Perhaps a BC337 or similar would work well enough in the power supply?

Update: February 2014

A few years back, I trialled a potentially suitable substitute - the 2N3019. This was selected slightly unscientifically - it was based on what I had in stock at the time, having eliminated the obvious non-starters and narrowing down the choice of the remainders. I didn't update this page immediately because I wasn't 100% confident the 2N3019 was a perfect choice as it only has a maximum collector-emitter voltage of 80 volts. But after several years of reliable operation, I'm confident enough to report my findings.

I replaced all four output transistors with the 2N3019 (and clip-on heat sinks), and using my Telequipment setup jig, I was able to go through the high-frequency calibration routine. Once complete, I was pleased to find that the -3dB point was just over 20MHz - comfortably above the original specification.

Again, I'll repeat that there is a residual worry about the VCEO rating, but if you're happy to risk it, the 2N3019 can be purchased from RS Components for £1 each at the time of writing. The stock number is 714-0445.

Capacitors

This scope uses lots of Philips axial capacitors, and it's nice to see that the correct replacements are still available. I ordered the following parts from Farnell:

Value Part number Price each Designation Quantity
1uF, 63V 305-558 £0.217 C1003 1
4.7uF, 63V 345-0831 £0.154 C8, C304, C411, C1004, C1006 5
4.7uF, 160V 345-1665 £0.41 C416 1
47uF, 25V 345-0648 £0.154 C408, C414*, C415*, C417* 4
47uF, 63V 345-0879 £0.226 C305 1
330uF, 200V 966-060 £2.73 C404 1
1000uF, 25V 305-121 £0.73 C818 1
1500uF, 25V 305-133 £0.77 C821 1
2200uF, 63V 106-803 £3.47 C405, C406 2
NB: Capacitors marked with an asterisk (*) are "upgraded" from 22uF. This table was compiled from my scruffy notes, so accuracy is not guaranteed - check parts yourself before ordering! Prices and part numbers correct in 2003. Note that order-multiples apply, so you'll probably have to buy more than you need!

The EHT unit

The insides of the EHT unit (20K)This image shows the insides of the EHT module. This is quite possibly the worst part of the 'scope to work on - there are no connectors; the unit is hardwired into the main wiring harness. Getting behind the larger PCB is very difficult - you have to remove the power transistor and desolder the two high-voltage diodes that connect between the PCB and the transformer. But it's essential to replace the two electrolytic capacitors, as 27KHz pulses caused by the switching action of the unit can leave the unit and contaminate the rest of the 'scope. When this happens, the trace can appear to be a dotted line if you get the right combination of timebase speed and brightness control settings.

Another problem area in this section is all the high value resistors with high voltages across them - these go high-value or open-circuit with age, causing all sorts of strange faults. I've already mentioned R311 on the last page, but there are lots of other candidates. Basically, if there is any fault that affects brightness, focus or storage operations, then check all these resistors. I saw a really obscure fault caused by an open-circuit R325 (3M9), causing a screen that constantly flooded when the scope was in NON-STORE mode.

The resistors originally fitted were standard carbon-film models which aren't suitable for high voltage operation. Make sure that replacements are correctly rated; metal-oxide resistors should be used. The Mullard VR37 types are recommended, and I believe these were used in later models.

That sweep problem

Earlier I mentioned that the sweep would intermittently fall in width. During the restoration, this problem appeared to go away, so naturally I assumed that the problem was caused by the power supply faults, or perhaps it was caused by one of the electrolytic capacitors that had been routinely replaced. To be honest, I had enough to worry about, and didn't give it too much thought.

However, I started using the scope after it had been in storage for a few years, and the fault was back with a vengeance!

This time, the fault was much less intermittent, which forced me to think seriously about it. Using a second scope, it was obvious that the expected sawtooth waveform was replaced with a high frequency oscillation at a few MHz. It didn't take long to decide that the problem was in the area of IC2, the sweep generator, but happily the IC itself was elimated by substitution. Looking at the service manual, each of the pins of this IC are explained, and pin 2 quickly drew attention to itself, being labelled "OSCILLATION SUPPRESSOR". Between this pin and ground are a resisor and capacitor in series (C17 and R65), and there is another capacitor (C18) between pin 2 and 9. All these components seemed to be OK, but after a really close look under an eyeglass, I could see something suspicious about one of the joints on C18...

The solder joint itself was perfect, but there was oxidisation between the tin plating and the underlying metal used on the leg of the capacitor. This was a 200pF axial polystyrene part - I don't know how critical the value actually was, but the nearest I had was 220p, and it was a ceramic disc. I was able to clean up the lead of the original component and solder it back in place. If it happens again, I'll know where to look.

Cosmetic restoration

These old workhorses can get quite dirty with age and use. Personally, I regard a full cleanup as an essential part of the overhaul process, and here are some pictures to show what's involved:

Close-up of a plug-in (13K)
Plug-in in pieces (17K)

This picture shows how grubby these things can get. Dirt and grime gets stuck in the control knobs and on the front panel. The second picture should leave you in no doubt as to the extremes you need to go to! It's time consuming, but ultimately very rewarding.

The empty chassis (15K)
Showing the bent chassis (12K)

The rest of the chassis and all the other controls get the same treatment. As you can see, the front sub-panel is slightly bent - this was presumably caused by dropping the instrument at some point in the past. From what I've seen and heard, this is a reasonably common occurrence with these - the tube is clamped at the rear of the instrument and is intended to push against this front panel. It's easy enough to straighten out the panel, but as you can see there is a serious amount of dismantling to be done before you can do the job.

Update: January 2015. More problems!

I'm beginning to think that this example is cursed! Afer another period of non-use - perhaps just a couple of years - it decided to sulk. Here we go again...

Failed X output stage

Upon being powered, there was no display. Pressing "Trace Locate" resulted in a pair of dots on the far-right of the display, and a quick inspection with a digital temperature probe (my finger!) revealed that TR102 - one of the horizontal output transistors - was faulty.

These are listed as SPS5286, which is a type of transistor that no-one seems to know about. A quick search reveals nothing but people in exactly my position!

I looked inside a later example of this model, just in case different types were used. I found a pair of TO92 transistors with no heatsink, and they are a different type - a type I can't readily interpret. The markings on the transistor are as follows:

    F S4
    3308
    814
    

Means nothing to me!

Then I unclipped the faulty transistor from its heat sink and found that it was the same type. Looking for clues, I dug out my box file full of Telequipment service manuals, and found that the SPS5286 is extensively used in the D83 (X-amp, Bright-up, PSU). In the DM63, it is also used in the power supply (TR412). But the manuals contained no other part numbers that might give me some clues about the characteristics of this mystery part.

I did consider taking the one from the power supply - thinking along similar lines as I mentioned above when discussing Y output transistors (the HF performance of this transistor in the PSU is probably not a concern, so would be an easier substitution), but instead, I decided to subsitute the transistors in the X output stage.

Approximately, these transistors are subject to a maximum collector-emitter voltage of around 100V, and could pass a maximum current of about 25mA. The power dissipation is somewhere around 0.6 watts on average. They are in a TO92 package with the base in the centre.

I have a decent range of TO92 devices that would have done the job in theory - mostly video output devices (e.g. BF422) - but they all had the collector lead in the middle. Also, I wasn't altogether happy about using a TO92 device in this role, as they only seem to last 40 years or so(!). With that in mind, I transferred the clip-on heat sinks to the other 'scope mentioned above, and reached for my drawer of TO5 devices. I noted that I had plenty of BF259s, and these comfortably fit the requirements. The only downside is that the TO5 package doesn't fit the TO92 transistor sockets very well.

With the Y-amps, I decided not to worry about this - it is really hard to work on the motherboard, so let's be pragmatic. But the timebase PCB is easily removed, so perhaps we can be a little more professional?

I have some TO5 transistor sockets, but they don't fit in the available space. Instead, I removed the TO92 sockets and mounted the transistors directly on the PCB, using lengths of PTFE sleeving to stand the devices away from the PCB. Finally, small clip-on heat sinks were attached, and the performance of the unit checked. No problems were found - the BF259 works perfectly here.

But we're not done yet...

Loss of trigger on one channel

Applying a signal to each of the four channels in turn, all seemed well apart from one, which didn't trigger at all when the trigger was set to CH1 or CH2. When set to ALT trigger, it worked fine, but in this mode, the trigger signal is "picked off" the Y signal on the motherboard, as opposed to being supplied by the plugin (via pins 7 and 10).

We've been here before with another one of these, and sure enough, the problem was caused by an MPS6518 on the trigger control PCB (PC153). There are four of these transistors - TR756, TR757, TR758 and TR759 - and any one of these should be instantly suspected. As previously mentioned, the ubiquitous BC327-25 works perfectly here.

Loss of trace in ALT mode

This was a complete pain! What made this worse is that I didn't spot the problem until the oscillo­scope was at work (as I was lending it to a friend), so I didn't have access to other examples for quick substitution checks.

All four traces worked individually. All four traces could be displayed at once using CHOP mode. But when switched to ALT mode - which of course is the best mode to use wherever the sweep speed allows - one of the traces disappeared. The trace in question is the second channel on the right hand plugin. The fault moves with the plugin.

Looking at the schematics, there are two transistors on PC153 that control the traces. The outputs of the two amplifiers in the V4 plugin are selected using diode switching, and the control voltages for these switches are generated by TR751 and TR753. These transistors are nothing exotic - just a BSX20 in a TO18 can - and are reliable types. Swapping them with their peers in the other plugin didn't change the fault.

Time to apply logic. The two transistors work as an astable multivibrator when in CHOP mode. When just one trace is required, the appropriate transistor is switched off. When in ALT mode, the two transistors work as a T-type latch, taking the GATE signal from the timebase module as the clock signal.

As the unit worked in all modes apart from ALT, that automatically proves that the transistors and most of the passive components are probably OK. With another 'scope, I followed the GATE signal via the switches and to the 100pF capacitors (C750 and C754). All seemed OK coming in, but I noticed that the spikes seen after the capacitors looked different, so that made me suspect the steering diodes (D751 and D753). And sure enough, D751 was open-circuit.

I rushed down to the workshop with the plugin and confidentally fitted replacement diodes (decided to change both). Everything made sense, and surely that would fix it? Sadly, no. My confidence was short-lived, as the symptoms were exactly the same. Including how the pulses looked after the differentiating capacitors.

Back to the workshop - the 'scope was in my office, and it's too heavy to carry all the way down to the workshop - and start checking things carefully. As I had some, I decided to change the capacitors - these are good quality polystrene types, but I remembered that I'd had problems with one of these on the timebase board, so it was worth doing just in case. But I decided that this wasn't the answer, and that there must be something else amiss. But what? The transistors are OK, the diodes have been replaced, and we'd already checked for problems with the switches. The resistors should be fine, as there are no high value types, and there are no high voltages to speak of.

But, knowing that the pulse didn't look quite right on the base of TR751, I went straight for R752 (39k), and it was completely open-circuit! OK, that must be it, at last! This makes total sense, and although I can't explain why it has failed - nor the associated diode (D751) - that doesn't change the fact that it has failed. Good.

As it's a long walk between the office and the workshop, I did take the time to check a few more resistors, but they were all blameless (and I was not surprised). Full of confidence once more, I braved the cold and made the long walk back to the office. Several colleagues had been following the saga, and were keen to see the outcome - so imagine my sense of total deflation when it proved to make no difference whatsoever :-(

Actually, that's not quite true. With the left-hand plugin in any mode other than ALT, the right-hand plugin was now working correctly. But selecting ALT on the left-hand plugin removed Channel 2 of the right-hand plugin. Very strange.

Following back, the GATE signal comes from pin 14 of IC1 in the timebase module. This signal goes to IC2 via R67, and tells it to start to sweep. It also goes via R40 (330Ω) to the motherboard (pin 2). And it also goes via R30 (1k) to the GATE OUT socket on the front panel.

On the motherboard, it is connected directly to the two vertical plugin connectors (pin 15). There are no components relating to this signal on the motherboard beyond PCB traces. Simple.

Using another 'scope, we could see a healthy pulse on pin 14 of about 1.8V in amplitude. This seemed a bit small, but I have since checked it against a working model and this is correct. This was also present at the front panel output as well. We noted that a x1 'scope probe presented enough loading to affect Channel 2 of the affected plugin, so something is clearly very much on the edge here.

As a crude test, I temporarily connected a 1k resistor between pin 14 and +5V, and this increased the amplitude of the pulse, and made things much more reliable. But I wasn't happy about this, so removed it.

Subsequently, I was able to try another IC from another 'scope, and this improved matters. Yet the original IC worked perfectly in this other unit. So perhaps we were seeing the effects of tolerances? But I'm really not happy about this - it's unusual to have multiple faults, and I really don't like coincidences - but there was definitely a faulty diode and a faulty resistor, for which I can provide no explanation beyond "natural causes". Hmm....

Conclusion

To conclude, this unit has been a useful addition to the workshop. The overhaul detailed above, coupled with lots of time and switch-cleaner, plus a full re-calibration has brought this old-timer back to life. The CRT is in much better condition than my original unit, and the storage functions work perfectly.

  • Go to page:
  • 1
  • 2
  • 3