ETI magazine Feb 1979 - power supply (ETI-142)

[(*steve*)]

Someone has dumped an ETI 712 power supply (I think) at my local hackerspace.

Edit: it's an ETI-142

Does anyone have a copy of the magazine article(s) which describe it? Our state reference library may have it, but they note that this particular year is incomplete without giving details.

The power supply itself is interesting. It has a huge transformer followed by a bridge rectifier and filter. This is then pre-regulated by a switch mode regulator before passing through a linear regulator.

As was common for the era, the linear regulator uses an LM723.

The pass transistors (including those in the switching pre regulator) are 2N3055's. You might decide to use one of these in a linear regulator today, but not a switching regulator.

I would really like to bring this back to life, mostly because it was my theoretical introduction to switching regulators (yeah I'm young).

It looks like there was some errata, because there are some mods on the board to add some additional (0.1Ω) resistors in series with the collectors (not emitters, weirdly) of the SMPS pass transistors.

Also, it's clear that the board was designed using a TO-5 LM723, but a 14 pin socket has been wired up on short flying leads for a DIL packaged device.

It looks like a great repair project, but I'd like to have all the documentation before I start.

 

[Bluejets]

Steve,
I found this but it does not list any CB power supply.
Hope it's the right mob.
Cheers ...Jorgo

http://www.americanradiohistory.com/ETI_Magazine.htm

I'm guessing you'll grab them anyway.

 

[(*steve*)]

Don't do things from memory

It's an ETI-142 power supply.



From Feb 1979

See here.

 

[(*steve*)]

My local reference library has copies of the February 1979 ETI (the Australian version), and I have photographed the article.

I might try to clean them up a bit more later, but here they are, hot off the phone (almost)...



There's some interesting design techniques here:

1. Input current measured by the ripple voltage on the filter capacitors (for overload reporting purposes)
2. Transformer temperature measured for thermal shutdown
3. No measuring of heatsink temperature!

 

[(*steve*)]

On the unit I have, there are a couple of resistors bodged onto one of the boards. I expected them to be emitter resistors, but they are in the collectors of the 2N2055's. This is a little confusing, and I didn't manage to find errata for this project so I'm not sure why they're there.

On that board, there were wiggly traces which appeared to be emitter resistors (on the switch mode regulator board). These are also not shown on the schematic. I imagine the traces will have far lower inductance than wirewound resistors.

I should take some photos!

 

[Terry01]

Nice one Steve! I really hope you find all the bits you need to fix this up. I think its awesome when these things find their way to people who have the know how to fix them.
Hope one day I can fix something like that up. That's the kind of thing electronics wise that interests me.
Good luck.

 

[(*steve*)]

OK, Here's some photos...

Here's the unit as I received it:

And with the top off:

Another view of the insides: From left to right there is the linear regulator, the switching pre-regulator, and the power transformer. The wiring is quite nice. This leads me to feel that the messiness elsewhere is probably not related to construction...

A reasonable view of the linear regulator board: Note the IC near the bottom right is a 14pin DIL uA723 mounted to fit the holes for the 10 pin TO-5 original. There's a few visible bodges on this board. It's either been repaired or someone had a very hard time putting it together.

Here's a rear view of a part of the switching pre-regulator: These bodged on resistors are connected in series with the collectors of the chopper transistors. I can't imagine why. The emitter resistors (relatively long, thin traces) are mostly hidden behind those resistors. I can understand why you'd (maybe) want to add a little more emitter resistance, but not this. Any ideas?

Lastly a somewhat dodgy shot of the rectifier: Sorry about the camera shake, it was getting dark. It's not exciting.

I'm not sure exactly how I'm going to test this. The construction article recommends a variac, and while this can be a very bad thing for an SMPS, I'll take their advice.

I hope the transformer is OK (I can't imagine that failing without the fuse blowing, and I hope it's not that because that would be a kiss of death). I'm concerned about the transistors and the diode in the switch mode pre-regulator, but a failure there would (again) probably lead to the main fuse blowing. If it's none of these major things then I hope it's the switching pre-regulator simply not starting up.

Well, that will be the adventure next time... but I'll probably check for any obviously damaged semiconductors first.

 

[Terry01]

Well cool! I can't wait to see it once you've fixed it up. How old is it?

 

[(*steve*)]

How old is it?

I haven't looked at the date codes, but it could be anything up to 38 years old. I'd bet it's more than 30 years old.

 

[Terry01]

Just shows how long these things can last. It'll be awesome if its brought back to life. I had a look inside mine last week. I knew what a few bits were and their purpose but most of it was beyond me....for now!

 

[(*steve*)]

I had a look inside mine last week

You have one of these?

I keep looking at the parts chosen for this and think "I could replace that with a MOSFET", or "a Schottky diode would be great there", but I really want to bring it back to original condition, so I will have to shelve those thoughts.

 

[Terry01]

No,sorry I should have worded that better. I have a modern Tenma PSU. I just ment I looked inside my Tenma last week. Was just having a nosey.

 

[kellys_eye]

Could it be classified as 'vintage' as far as electronics are concerned? The designation seems pretty vague these days but as far as usable power supplies go that model seems to be very well specc'd and worth the effort to fully restore.

I have a hankering to knock up the mini oscilloscope project ETI did - also about 30-odd years ago now, just for nostalgia sake.... it's performance is pizz poor in comparison to todays stuff.

 

[bushtech]

Well, with your knowledge I'm pretty sure you'll end up hotrodding it up to modern performance

Sorry OT

 

[(*steve*)]

Ok, I've finally decided to spend a few moments doing some basic checks.

1. Both windings of the transformer are ok.
2. I get AC at the secondary
3. I get DC after the rectifier
4. With 100V input voltage (240V mains) there apparently no load on the DC rail.

After applying 240V to the unit I noticed that the voltage meter was pinned to the right.

Further tests show that the voltage at the output terminals is controllable using the voltage pot. Turning the current pot to near minimum turns the output off.

Is this actually working?

Time to test the voltage differential across the linear regulator, and if that is ok, proceed to a load test...

 

[(*steve*)]

Ok, starting with my 100 ohm resistor.

1. 100Ω PSU goes from 162mV up to 33.2V. about 0.5mV ripple
2. 10Ω at 22V 1.27mV ripple
3. 1Ω PSU goes from 2.7mV. At 14V output there is 0.45V of 100Hz ripple and I'm drawing 2A from the mains. And yes, my dummy load actually got warm for the first time
4. 0.1Ω. pay goes up to 1.85V before max current limit is reached. Linear power supply heatsink is noticably quite warm (possibly dissipating about 100W. AC ripple maxes out at 70mV at 100Hz.

It's very interesting that the 100Hz ripple makes it's way to the output. My thought is that it may well be related to the ESR of the main filter caps.

So I'm beginning to doubt the "No Power (not fuse)" label.

It would be nice to as power light, a load switch, and indication/shutdown for high heatsink temperature, but I think filter caps come first.

 

[(*steve*)]

OK, so here's the results very quickly obtained from my analog scope.

Firstly, 14V 14A output:



Next 6V 6A output (and with the trace brightness turned down):



And the same with a better vertical and horizontal setting



This is 100Hz ripple, and I guess, if I take a look, it's the voltage sagging on the capacitors (that will be the next test).

What you may be able to see is how fuzzy the trace is. It has approx 20kHz ripple impressed on the output. Because it's not locked to the 100Hz, it just appears as fuzz. I'll try to do a single shot capture of it later.

As a power supply, it may "work" but the output leaves a lot to be desired. As @bushtech suggested, I may have to pimp it up a bit...

 

[(*steve*)]

OK, there's something wrong with the switching pre-regulator. It's not putting out a voltage that is tracking the output voltage well. It seems to track something else instead...

I have displayed here the voltage across the main regulator. There is supposed to be about 7V across here. I've used line sync to make life easier.

Here is the difference at almost no load:

That's 2V per division, with 0V set to the bottom graticule. The overhead for the regulator is 7V.

With a low current (500mA):

Horizontal timebase is 2ms/division, so with a period of 10ms, that's a frequency of 100Hz. About 1V peak to peak. This is well within the linear regulator's ability to handle.

Now at 1.5A:

Weirdly, the dominant frequency is not 50Hz. Note that the minimum overhead for the linear regulator is now 4V. Probably OK, but it must be pushing it.

At 2.5A:

The frequency is 100Hz again, but note that you have a smaller then a larger negative going cycle.

And at 5A:

The overhead is less than 2V. It actually doesn't get worse than this because I'm pretty sure this is the minimum overhead of the linear regulator. Now is when we start to see periodic dips in the output voltage.

This is a real problem because the output of the switching pre-regulator shouldn't have all this 100Hz rubbish on it.

Is it slow in responding? Do I have a problem in the diode bridge?

Also... with this amount of ripple, I'd expect the overload indicator to turn on. Was this a problem from the beginning?

 

[(*steve*)]

Things to look at tomorrow:

1. Ripple on unregulated DC rails WRT centre tap (50Hz or 100Hz?)
2. -12V rail
3. 20kHz sawtooth
   
4. Error signal
   
5. Ripple voltage at pin 2 of IC3
   
6. Setting of overload

Have I overlooked something obvious?

 

[(*steve*)]

I couldn't wait...

Here's the voltage between the centre tap of the transformer (referred to as +20V) and each of the supply rails

These traces are AC coupled, and without thinking I inverted the second channel to make it look right. I shouldn't have needed to do that. There is an LC filter here, and I suspect that the ripple is being inductively coupled from one rail to the other (due to a greater load on one half of the supply???)

As I increase the load from nothing (above) to 15A (225W) the ripple actually decreases, and the phase relationship changes:

Both halves don't look the same, and that's kinda interesting. But the amount of 100Hz ripple actually decreases! I didn't expect that. (The amount of ripple across the 0V-40V increases, but the magnitude as seen from the 20V rail reduces)

So it seems pretty clear to me that the rectifier is doing what it should (these are both 100Hz signals, and the ripple is pretty well controlled).

Note that the traces above are both at 1V per division. The ripple is under 2V at full load.

So I again look at the difference between the input and output of the linear regulator. This time with 2 traces, with both being measured WRT the negative output. This does not show the input ripple correctly. The grounds are NOT common. And I left the B channel inverted :-(

First with no load:

Both AC coupled, the output voltage at 0.5V/div, the input at 2V/division. Pretty much flat traces at zero current as you'd expect.

At 1A (I think):

There's 8V of ripple!! However the output is not affected.

At 5A:

Now we see the linear regulator falling out of regulation

At 10A:

It's even more horrible (as is the camera shake)

And finally at 15A:

Yuk! 1.5V of ripple on the output.

I'm pretty sure this removes blame from the rectifier and the filter caps.

The one thing I'm worried about is that the rail for the op-amp in the pre-regulator is referenced to the +40ve supply rail, where the voltages in the linear regulator are referenced to the +20V rail. I wonder if this affects the error voltage which is transferred between them (I don't think it will)?

I don't fully understand how the control output is generated. My naive approach would be to use a zener diode connected to the +ve output to generate a voltage Vz higher than the output, and make the pre-regulator track that. I'm sure there's a good reason that's not done, but what is that?

I think the control output is going to be interesting...