Power supply ripple on UA723 based circuit

[RichardM]

I am repairing a spare R&S SMS signal generator and have noticed the ripple on the 5V supply is well above spec. It’s about 60mVrms and should be < 0.4mVrms. I have attached an image of the 5V part of the power supply circuit.

The adjacent +-15v rails have <1mV ripple. There is also a similar small ripple at MP1 on the positive side of C1. The 60mv ripple appears at MP2. I have swapped the UA723 and 741 opamp with new ones but that did not make a difference. Any thoughts on what the most likely culprit is or what to check first is appreciated. Can a noisey T1 or T4 be the reason?

Cheers

Richard

[ondraN]

I would suspect C1 in the first place.

[RichardM]

Ok, the ripple on C1 at MP1 is 0.2mv and C1 measures above 20,000uf ? Could it still be the problem?

[ondraN]

Are some ripple on MP3? It is power suply for 723 and 741 and can cause some problems.

[iMo]

Doublecheck the MP1 voltage as well, it could be too small for proper regulation (you want 5.39V out with 8.2V in)..
..and look at the ripple at R12 too..

[David Hess]

60mVrms is definitely high for a 723.

C1 and C3 should be checked.  Current gain of T1 or T4 could be low.

It is a long shot, but make sure GL3 is not leaky and the output of B2 is low.  Reverse biasing the current sense transistor inside the 723 might cause all kinds of weird behavior.

[Avelino Sampaio]

Check if 60mv ripple is present on pins 5 and 6 of UA723.

[RichardM]

Ok, a few things to check. I have unmounted the pcb from the rear heat sink and TO-3 transistors to get to the back of the pcb. Will report findings.

The voltage at MP1 was 8.8v with < 1mv ripple. Ripple on MP3 was 1.5mv.

Before unmounting, it may be unrelated, but I noticed R39, R40, GL21, GL22 and GL15 get reasonable warm at 70deg C or so.

 Both my LCR meters are limited in uf capacity so all I can really do is check leakage for C1. I have another 22,000uf cap so I may as well replace C1 while it is in parts.

[fzabkar]

What is the ripple frequency? Is it mains frequency or double the mains frequency, or something else?

[RichardM]

Just working on this now. Some observations below.

- R12 less than 3mv ripple
- Ripple frequency is about 2.7 to 2.9khz
- GL3 has -1.48vdc on anode side with 3mv ripple.
- GL3 has the waveform attached on the cathode side. Suspiciously at the same frequency as the ripple.
- less than 3mv ripple on pins 5 and 6 of the 741 opamp
- 741 opamp output is -1.48v as per GL3
- C1 tested fine but replaced it anyway, still 60mv ripple
- C3 tests ok in circuit but may pull out to look at it
- will now pull GL3 and test.

Richard

[iMo]

That is not a ripple from the diode bridge, the stuff oscillates..
Look at C2, C3 (is that a tantalum 470n?).. Is your 5V output loaded?
Btw, the oscillation at the GL3 cathode is 3Vpp when I look at the shots.. That may mean the 741 (the current limiter) senses an overcurrent at the R11 (the shunt) and limits the 723's output (when the 741's output goes up), so it oscillates. It could be the C3 is "shorted" somehow, or a short or overcurrent downstream the 5V line..

[George Edmonds]

Hi

At this point if the 723 is in a  socket, I would replace it.

I have seen strange behaviours from power supplies when the 723’s begin to fail, if my memory is correct the 723 holds the record for the longest length of time that any single IC has been in constant production

G Edmonds

[RichardM]

Replacing the ua723 at B1 has no effect. Replaced GL3 mono effect.

I replaced all the tantalum’s but still see 60mv of ripple. It is present on the emitter of T4 and base of T1.

I put a spare unit on the bench next to this one. The most obvious difference, apart from the lack or ripple on the 5V line on the spare unit, are the voltages across GL3.

On the good unit, -3.95vdc on the anode, 0vdc on the cathode.

On the unit with high ripple, +2.0vdc on the anode, +1.84vdc on the cathode.  I also measure 1.8vac on the cathode🤔

[George Edmonds]

Hi

I assume you have also replaced the 741 B2,

Now you talk about the voltage on GL3, a small signal fast switching diode, being incorrect which is the output of B2 the 741.

Do you realise that the function of  B2 is to provide overload protection for the  +5V supply.  It senses the voltage across R11 a 0.068 Ohm resistor which is the power supply current limit sensor resistor and shuts down the 723 to prevent over load damage to the +5V supply.

Your scope traces clearly show that your problem is NOT power supply ripple.  It looks to me that the power supply is starting to power up and then being shut down by an external overload condition, have you checked for this?

G Edmonds

[RichardM]

Yes, I have replaced B2. If it is not ripple what is it?

The power supply is working and all voltages are correct, +5, +15, -15, +20 and +28. I was just trying to eliminate the ripple, or whatever it is, on the 5V rail per specs.

[RoGeorge]

I'll check the value of the power resistor R11 (the current sensor).  Resistance can increase sometimes for old power resistors.  Check that it has 68 milliohms indeed, and it is well soldered.  Check the voltage drop on it, too (does the voltage drop corresponds with the expected load current?). 

If all OK, I'll check the value of the other passive components around that 741, and the diode, too.

[George Edmonds]

HI

I would suggest reading and fully understanding section 4.14 of the service manual first.

G Edmonds

[RichardM]

Fully understanding section 4.4 is beyond my skill set I’m afraid.😟 I may be able to simulate the circuit is LTSpice as a learning exercise.

After all my prodding and replacement of capacitors etc the ripple has improved but the 5V rail goes into current limiting mode with anything more than about 250ma. One step forward, 3 steps backwards. 😀

[George Edmonds]

I agree that section 4.14 can be daunting, possibly because it is a translation from German.

Do you have the facilities to run the power supply external to the instrument?

This maybe a long shot, Please  check the voltages and let me know what they are on B2 pins 2,3,4 and 7, also do the same on B5.  If possible do this with a DMM and a scope. B2 pin 7 should be 20.5V and pin 4 should be -14.8V

From the measurements you have done it looks like one of two things is happening (1) The rest of the instrument is putting an excess load on the +5VDC supply causing the +5VDC overload protection to activate (IC B2) or (2) The +5VDC supply protection circuitry is faulty.  Basically the current sensor is R11 which is connected directly across the inverting and none inverting inputs of B2.  The load current at which the supply starts to limit is set by R7, R8, R9 and R10.

A good starting point with any power supply is to check the high voltage and high current parts first so start with T1 the 2N3055, the schematic suggests that it is in a socket so possibly a poor connection exists.  The prime suspect must be R11 so visibly check it and if you can make 4 Wire  (Kelvin) resistance measurements check the resistance from T1 Emitter to output pins 1, 2, 15 and 16, is should be just over 0.068 Ohms.

You are not helping yourself by calling the problem ripple as this usually implies mains frequency.

G Edmonds

[RichardM]

Thanks for hanging in there with a hobbyist.

Yes, I have the power supply separated from the instrument on my bench. It's a pain to remove the TO-3 transistors on the heatsink to properly desolder components properly.  I have done it about 5 times already. I did check the 2N3055 transistors, they varied in gain under low current using a Peak Atlas tester, 23, 63 and 108. I can try a new 2N3055 if need be.

I can definitely make the measurements you suggest and will post the results.

Fyi, I ran a separate test by applying 5v to the instrument without the power supply as I also thought the 5v may have been shorted. I used the schematic and 24pin dip flex cable usually attached to the power supply. I started inserting each card individually and the 5v rail was taking about 750ma in total. Seems reasonable given all the ICs on the boards?

Ok, no more mention of ripple 😀

[George Edmonds]

Hi Again

Congratulations on what you have done so far, I will quite happily work with a person with your positive attitude.

My fist gut reaction to your last posting is that the design of the +5VDC supply is well over the top if only supplying 750mA,

You probably do not need me to say this but the current regulation circuitry for the +5VDC supply is almost identical to that of  the +15VDC supply so simply comparing the voltages on all of the pins of B2 and B5 may well give a clue as to what is happening.  Given that the +15VDC supply uses two current sense resistors in parallel suggests to me that it provides a greater current than the +5VDC supply.

I find it very annoying that service manuals rarely give the current drawn from the various supplies as it is frequently very useful information.

My final point for now is that I forgot to ask you to directly measure the voltage across R11 and R27 with the supply in the instrument, from this we can work out the currents drawn.

Keep smiling

G Edmonds

[RichardM]

Ok, some initial measure,ents.

R11 measures spot on with 4 wire kelvin on m HpP3478A

DC voltages on B2

Pin 1 0v
Pin 2 5.319
Pin 3 5.026
Pin 4 0
Pin 5 0
Pin 6 1.333
Pin 7 22.04
Pin 8 0

DC voltages on B5

Pin 1 0
Pin 2 15.06
Pin 3 14.81
Pin 4 0
Pin 5 0
Pin 6 2.024
Pin 7 25.35
Pin 8 0

DC voltages on B7

Pin 1 0
Pin 2 20.22
Pin 3 20.02
Pin 4 0v
Pin 5 0v
Pin 6 2.005
Pin 7 25.38
Pin 8 0v

One thing I haven’t mentioned is that one of the tracks from the MP headers (MP3) has clearly been overloaded and burnt in the past. The track was lifted so I just snipped it off a few days back. I didn’t think it was critical to maintain the link between the header and the via for MP3 based on the schematic. It may provide some insight though. Photo attached. Headers on the left are MP1 to MP10 bottom to top. I suspiciously measure some DC and AC on MP3 even though it is not connected to the trace anymore. Maybe the burn has contaminated the pcb ? Just an added complication, the board is in pretty good condition otherwise given its age.

[iMo]

I would replace all those red bricks - I assume the tantalums capacitors.. Put there new with a higher voltage rating.
PS: perhaps the PSU is missing the cooling water in the copper pipe there in your shot.. 

[xavier60]

Is there activity on B2 pin 6.

[RichardM]

That is an old photo before I cut the burnt trace that was floating in the air. I have replaced all the tantalums since then.

I also replaced the low gain 2n3055 tonight. It was T2.

I made an observation by chance when testing the 5V under load using a big power resistor like I did previously. Not sure what to make of it.

If I turn on the power supply with the 5V under load from the resistor, I get zero current flow and the 5V is line zero.    If I start the power supply with no load on the 5V line and then add the load, I get current flow. Right up to 800ma no problems. So, I guess I need to swap back the old transistor to see if I observe the same behaviour. Fingers crossed I don’t see the same behaviour suggesting T2 was the problem. Somehow I don’t think it is T2.  Otherwise I will be scratching my head. I also need to get the supply back into the unit with the new transistor to see if it works. All a bit strange to me now. Keeps my grey matter active 😀