Cossor CDU150 rescue

[Radio Wrangler]

Oh, from your pulde generator use a good coax cable to a Y-piece on the scope input BNC. Put a good RF grade 50 Ohm termination on the other branch of the T-piece.

It sounds like the service manual is aimed at rough fixing in the field with limited gear... often an approach in manuals for the military. But unlikely to get your scope up to its proper spec.

David

[MelJon66]

Thanks David,

What would be a fast enough rise time?

Mel

[MelJon66]

The CDU150 has a rise time of 10ns and my signal generator is <40ns.

Whatever the shape of the input waveform the gain on Ch2 starts getting larger than Ch1 at any frequency above 20kHz. Even without a fast pulse generator I should be able to locate where in the circuit the differences between channels start to show.

Cheers
Mel

Mel

[MotorBikeLes]

Your signal generator must have a rise time considerably shorter than the equipment being calibrated. my Telequipment C3 calibrator has a fast rise signal with better than 1nS. That it the recommended unit for the 50MHz bandwidth D755 and D83. That suggests you need better than at least 2nS.
Your coax and input 50Ohm attenuator need to be in good condition, or the results will be unsatisfactory.
Les.

[MelJon66]

Hi All,

With the attenuator module disconnected from the pre-amp and a signal generator driving the module alone the outputs from the attenuator vary by a factor of 2 between channels at some sensitivity settings. The channels are equal at the non attenuated 5mV/cm setting. There were signs of overvoltage abuse in the attenuator when I got the CDU150 and I replaced some resistors last year. I suspect some others may be failing now. I won't be able to get onto this until after the Easter week so I won't be posting for a while.

Fast rise time pulse generators are expensive and hard to justify for use once in a blue moon. W2AEW on YouTube built a circuit using a 74AC14 inverter IC some years ago which looks easy and cheap to make. Has anybody used one? Is it good enough for testing 'scope rise time or are there better alternatives these days?

Cheers
Mel

[Radio Wrangler]

The 74AC14 is probably fast enough for that scope. If it takes a fast edge to make an adjustment, it takes a fast edge no matter how infrequently it needs to be done, it just makes the cost harder to justify and makes the search for an alternative way of getting a fast edge all the more attractive. Pulse generators are rather versatile, and you don't need the full range of their capabilities which opens the door to cheaper alternatives.

When good pulse generators do turn up at radio rallies, there is little demand and you can pick one up for pennies - usually the perceived value of the box and PSU section for a homebrew project.

With logic devices, beware of output ringing, ground bounce etc.

David

[Jez1234]

Great thread! I have one of these which although now not working was basically reliable and a fine instrument which served me well for maybe 10 years without any problem... and this was probably 2-3 hours a day for most of that time!

I got mine when loads of them were sold off by the MOD and then re-sold by several test gear suppliers. The first one gave a loud crack from the EHT and died on me after about 1 min use but was replaced by the company I bought it from. Mine gave me the impression it had seen very, very little use before I acquired it.

The heat it gave off was most worrying and I'm amazed it worked with 100% reliability for so long.... One could have been forgiven for assuming it was valved such was the heat from it! I'd guesstimate the outside of the case sat at about 45C! The temp inside which the electrolytics will have been exposed to is not worth thinking about!

It failed eventually due to a FET in the ramp generator giving up the ghost but I found the fault after an hour or so and fitted something like a 2N3819 as a replacement for the original U something or other and this restored correct operation for another year or so.

By the time it died altogether one morning (power light on, no burning smells or anything... but no trace either!) I had another scope ready to take over (a superb little SE Labs SE111) so, expecting every electrolytic in the beast to be in need of replacement after the years of being cooked... it was put into storage in a warm dry place where it has.. ahem... now been for about 20 years!

As my CDU150 is pretty much mint and complete with all accessories and manual, worked very well indeed when going, and as it served me so well for ten years or so, I've always had an idea that someday I'll give it some TLC...

As I repair things like vintage hi fi etc for a living and there are only so many hours a day that I'm willing to be doing electronics I guess it'll have to wait for that batch of ex MOD Round-Tuit's to be released!

This may be of some use to the OP but I vaguely recall that with no input signal whatsoever the trace could indeed flicker on and off pretty randomly. IIRC it could be kind of dealt with by adjusting one of the controls very finely indeed.. as in there was a specific point where it would free run steadily but a degree or so either way on the control and it was back to flickering a bit. Sorry can't recall any more detail than that... Could well be that "they all do that" then? I never found it to be a problem as even a mm or 2 of displayed signal would put everything as it should be.

[MelJon66]

Hi Jez1234,

It's good to know there is yet another Cossor enthusiast about. I have really got to like the CDU150 in the time I have had it and would love to get it fully operational. It has a classic industrial look and is repairable with standard parts unlike some Tek or HP equipment that have rare as hens teeth special chips. I hope I can repair it and keep it going for years to come. Yes, it does run warm and at first I wondered if it was about to self destruct. After a while I accepted that is just the way they were designed. The combination of the heat and the glow from the CRT display accompanied by some good music is a great combination on a cold winters night.

The flickering trace is just an annoyance but on my 'scope it happens when there is no signal or the signal on one channel is much smaller than the other. I now know which settings stop it happening so I could live with it if I can't fix it. My plan is to fix all the more serious issues and hope that one of the fixes will stop the flickering. The project is a learning exercise for me so the more problems I encounter the better the learning.

The support of the forum members here is invaluable to a beginner like me. I have some classic amp projects to work when I get the chance so your experience might be useful there so please stay in touch.

Cheers
Mel

[MelJon66]

Easter Greetings to All,

I have found some time to investigate the attenuator issue. I have taken out the attenuator module and added a 1M resistor to ground at the end to simulate having R4 (Ch1) or R73 (Ch2) in circuit. I have applied voltages to the input that would give FSD readings at the relevant sensitivity settings and have measured the output voltage across the 1M resistor. This replicates the circuit provided by Buggies in post #29. The results have surprised me.

I expected that the attenuator module would have a flat frequency response with the output staying the same at all frequencies for a given input setting. Indeed it is flat at the straight wire 5mV/cm input setting where there is no attenuation. With 40mV in I get 40mV out at all frequencies, with a slight roll off at 2MHz, which might be because my signal generator is at the end of its range at 2MHz. On other input settings I expected to get similar 40mV output voltages to give FSD. But, at any attenuated setting it is far from flat, mostly with very significant roll offs with increasing frequency. I was expecting 40mV output to give FSD but often measured around only 10mV at 2MHz so there was massive roll off. On the other hand on Ch2 at 5V, 10V and 20V settings I am seeing very large increases with increasing frequency.

It turns out that the adjustable capacitors in the attenuator are having a massive impact on frequency response. For example on Ch2 at 20V/cm vertical sensitivity the twiddler (as Radio Wrangler described them) has to be removed or left almost falling out to get a flat response. At this setting, with the twiddler fully screwed in the frequency response of the attenuator increases dramatically to almost 4X FSD at 2MHz. This is good news as it does explain the anomalies I was finding when trying to go through the calibration procedure but I did not realise that the twiddlers had such an impact on frequency response. Page 4.14 of the handbook suggests that adjustment of these twiddlers is to achieve an optimum square wave shape and doesn't mention their significance to frequency response. To achieve a flat response through the attenuator I would have to have some of them adjusted almost to the point of falling in to the the attenuator and a few almost falling out.

It bothers me that when I tweak the twiddlers to get a good square wave as described by the handbook it might upset the frequency response again, which doesn't seem like a good idea. So, before I go through the laborious process of adjusting these twiddlers (11 per channel) to give a flat frequency response at all input settings would the expert members please confirm that I am now on the right track. I do not want to get further along the calibration procedure and discover that I have misunderstood something gain.

Persistence pays but knowledge pays with interest .

Cheers
Mel

[MelJon66]

Just another thought,

Following the sequence in the handbook:

Page 4.7 - Set the Y amp gain - done
Page 4.8 - Set Ch1 and Ch2 gain - done
Page 4.12 - Set Y Amp HF compensation - done but with no apparent effect
Page 4.14 - Set Attenuator HF response - attempted but with no apparent effect

This sequence makes sense if all the attenuator twiddlers are doing is tweaking waveform shape but when they have such an impact on frequency response any adjustments risk undoing what has been done earlier. This doesn't seem right.

Mel

[MotorBikeLes]

You say setting up for a square waves seems to affect frequency response.
Yes, of course. Think of a square wave as made up of an infinite series of sine waves of increasing frequency. If you have a good square wave, you have a good high frequency response. The steep rise and corner shape reflect the highest frequencies, the slope (if any) of the horizontal part reflects the low frequency response. If the peak of the rising part overshoots too much, your frequency response will likely have a few "ups and downs" at the higher end.
I don't have the CDU150 info here, but all these settings must be done in the correct order. i cut my teeth on this starting with a Telequipment S31, something out of the arc compared with the CDU150. My test gear was VERY limited back then, but I was only working with a scope of 3MHz bandwidth, less than 10% of what you have there. The TQ manuals laid out exactly how to go about the settings, from stating with an internal fast rise signal for your 10Xs probe compensation. Everything must be done in the correct order or you will never succeed. David explained the order a few posts ago, but probably in a way few novices would understand. I can improve on his explanation by completely bypassing it and advising follow the manual's instructions to the letter in the order they are given. Of course if you have no manual, then it is time to understand his advice.
If a twiddler will not do what it should, find out why. The little silvered capacitors can go open circuit, and the little ones with plastic dielectric spacing can get damaged and go short circuit. Both of these can be intermittent. If a "fully open" capacitative twiddler has no effect, is there some parallel capacitance there that should not be there.
Finally, for final high frequency compensation, you must use a "proper" signal generator designed for the purpose, which gives its output into a 50ohm load.
Follow those rules and you will identify each fault (or fault area) one at a time, finishing with a properly calibrated unit.
Les

[MelJon66]

Greetings All,

Despite a lot of time spent checking things out I am still puzzled by the behaviour of the attenuator in the CDU150. I have the attenuator disconnected from the pre-amp and terminated with a 1M resistor to ground as per the diagram provided by Buggies in post #29. Applying input voltages to give FSD (where possible) I have measured the output voltages across the 1M resistor at all sensitivity settings. I have tabled the results in the attached spreadsheet. My observations are noted in the spreadsheet.

As you can see the results show inconsistency between channels and significant frequency dependency. At the 5mV sensitivity setting, which has no attenuation, everything looks linear so I would expect the attenuated ranges to also be linear. They are not. Everything is reasonable at 1kHz, which is the frequency at which the amplifier gain is set, but at higher frequencies the readings are all over the place, possibly indicating problems on nearly all ranges.

When probing around inside the attenuator the frequency dependency starts immediately after the large 1M attenuator resistors. I wondered if the resistors were becoming frequency dependent due to aging so I removed R14 on Ch2 (5,10,20V range) and substituted it (It’s the easiest to access). The frequency dependency after R14 remained. I also clipped one of the leads to the 1200pf fixed capacitor C18 but that made no difference. The variable capacitors appear to be ceramic tubes with adjustable steel cores so they look very robust and I doubt that many, if any, can have drifted very badly. C16 is in parallel with R14 and adjusting it significantly alters the signal amplitude after R14 but the signal is still very frequency dependent. A flat profile cannot be achieved. Adjusting C17 doesn’t alter the signal at all. It should alter the input capacitance but I can get no reading at all at the input BNC’s.

At the moment I can’t work out what the problem is but it appears to affect all attenuated ranges which suggests that many components may be faulty. If I can get one attenuator range working I can apply the fix to the others.

I have ordered the parts to make a simple fast rise time pulse generator (W2AEW type) to test the amplifier. I am also about to order a new signal generator as mine are both so old they both have strange behaviours that have to be allowed for, which is frustrating. I was using a 50 ohm termination while carrying out these tests but it turns out that was reading 300K ohms so I have ordered one of those too. Old test gear is a major pain.

Any ideas or suggestions on what I could try next while I wait for parts and new equipment would be welcome.

Cheers
Mel

[MelJon66]

.... and here is the attenuators circuit diagram.

Cheers
Mel

[Radio Wrangler]

Note that some of those twiddlers aren't there to flatten the scope's response , but to set it so the input capacitance presented to say a probe, stays the same on all attenuator settings.

Otherwise you'd have to readjust the comp cap on your probe every time you changed range.

You set these with an RF capacitance meter on the input BNC. They won't affect the scope response itself at all.

David

[MelJon66]

Greetings All,

I haven’t posted for a while but I have been working on the CDU150 between gardening projects and getting the campervan ready for the summer.

To pick up from my last post the tests I carried out on the attenuator got me nowhere. In the end I just replicated the attenuator capacitor twiddler settings, i.e. depth of adjustment screw and rotational angle, from Ch1 (untouched by me) over to Ch2. It worked and Ch2 behaves just like Ch1. Some attenuator resistors on Ch2 were burnt out when I got the Cossor so I can only conclude that when I replaced them last year I messed up all the settings afterwards. Luckily I left Ch1 alone so it was good as a reference.

I have also built a fast rise time pulse generator using a 74AC14 Schmitt trigger IC. The Mk1 replicated Marcus’s version of W2AEW’s build on YouTube. See W2AEW’s video “#203 Schmitt Trigger Oscillator revisited”. The layout looked good and it produced a nice 5V output with minimal overshoot. I needed a 25mV output for the Cossor rise time tests so in the Mk2 I added a resistor divider to reduce the output voltage. Unfortunately, the amplitude of the overshoot and ringing was greater than the amplitude of the signal and the rising edge was impossible to see. I now know that long ground connections result in ground bounce which was producing massive oscillations relative to the small 25mV output signal. I had followed Marcus’s layout not realising that he had used a PCB with a ground plane and my prototyping board version had ground wires running around the periphery.

The Mk 3 is a completely new layout and uses only one inverter nearest to the ground pin and dispenses with the multiple 220R resistors as I didn’t need to drive a 50 ohm load. I still have the 50 ohm version if I ever need it. All the components are directly soldered as close to the IC socket pins as possible. It takes its input from the TTL output of my signal generator and outputs 25mV to a BNC connector directly into the Cossor. It isn’t pretty but it works.

I also had problems with the 74AC14 IC failing when switching on. The 5V supply was provided by a good quality Metrix power supply but it has no switches on the output and the switch on spike was taking out the IC. I lost 3 IC’s before I changed over to using a Thandar PSU which has output switches. No problems since but I have now added a 7805 voltage regulator on the Schmitt trigger PCB to prevent this happening again if I use a different PSU.

So, it took a few attempts to get acceptable performance from the Schmitt trigger but the result is excellent. I have now worked through all the calibration procedure given in the Cossor handbook and the CDU150 is working really well. Consistency between channels is excellent, the rise time is as specified, the crosstalk between channels has reduced by around 50% and the flickering trace is no more. Photo attached of the rise time test at 1MHz with 25mV input from the Schmitt trigger with the timebase at 20ns/cm (X5 mag is on). I read 12ns for the 10 – 90% rise time on screen but the Schmitt trigger probably contributes about 2ns of this so the Cossor is achieving its specified 10ns.

The only calibration step that I failed on was reading the input capacitance. I have many pieces of equipment that can measure capacitance but none of them can get a reading across the input BNC’s. If I calibrate a 10X probe on Ch1 and then reconnect it to Ch2 I get the same flat top square wave so that will have to be good enough.

The Cossor is now my oscilloscope of choice for everyday use and it looks just right next to all my other vintage equipment. I may encounter other problems with it in future but all the work done on the power supply and pre-amp module must have done a lot to improve reliability. It’s a great outcome to a long project and I am very happy with the result.

I have a number of other projects that I will be working on but progress will be very slow over the summer. I will start new threads when I get stuck, which is highly still likely, despite the knowledge I have gained from this project.

Thanks to everyone for their contributions to the thread and in particular to Chris Williams for supplying a replacement transformer without which the project would have ended before it had really started.

Cheers
Mel

[Radio Wrangler]

A great outcome, and you've learned a fair bit along the way that will stand you in good stead on later projects. A real win-win.

David

[Chris55000]

Afternoon Mel!

Congratulations on getting your CDU150 up to spec and your perseverance with it!

As WME_Bill and others have said, they're well worth the time and trouble to restore properly and as long as you have a pulse generator with sufficiently fast rise–time, you'll get there in the end!

A large Parcel of all the remaining spares you'll need to maintain it, including attenuator assemblies, spare boards & CRT, is all packed up ready to send as soon as my friend runs me to Aldridge Post Office with it!

Chris Williams

[MelJon66]

Thanks David and Chris,

Chris, that is unbelievably generous of you.

To save you any hassle I can easily arrange for UPS to collect at my cost. Just let me know the size of the package and I will arrange it. UPS don't require the weight as long as it's under 70 Kg. I have your address.

If a collection is inconvenient I can send a label and it could be dropped at a UPS Access Point.

The nearest one appears to be very close:

D.E.I. CONVENIENCE LOCAL
80 FRIEZLAND LANE, WALSALL WOOD, WALSALL, WS8 7AR

Please don't go to any more trouble, just let me know what suits you best.

Cheers
Mel