How to 'separate' dual tracking PSU regulators

[jeffbro]

To assist anyone interested enough to look at the attached schematic, all this text is copied there.

The V+ and V- regulators in this PSU track each other and each will shut down if the other fails. I want to individually troubleshoot and test them but I want to make sure I understand how they track and how I can 'separate' them without the risk of unpleasant surprises.

It appears that when the V- rail goes down, the lack of a voltage at R47 causes Q3 to turn on and rob U5 of its input at pin 5, turning off the V+ rail.

It also appears that the 723 in the -18V regulator gets its pin 5 input signal not from the 723's internal reference voltage (pin 6) as is typical but instead from a voltage divider hanging off the V+ regulator's output rail. So, if the V+ rail goes down there is no input to the 723 and it turns off the V- rail.

If I understand this correctly then...

It seems that I could make the V+ rail operate independently by lifting one end of R46 so the loss of V- can't turn on Q3.

To make the -V rail independent I could add a temporary 4K7 resistor between U7 pins 5+6 and lift the ends of R61 and 62 that join at pin 5.

Am I on the right track here or is there a simpler way of making the +/- regulators operate independently?

 

[jeffbro]

Anyone?

 

[radardoug]

There is a reason they are tracking. If you have faults, you need to disconnect the gear they supply, and then use dummy loads on them.

 

[jeffbro]

That is exactly what I am doing but I want to trouble shoot the regulators one at a time without them 'talking' to each other. Looking for problems in the console will come after the PSU is repaired and tested on the dummy loads.

 

[radardoug]

Well your writeup looks correct without me looking at the schematic, so I would say go for it.

 

[JohnRoberts]

The 723 regulators are old soldiers and pretty straightforward. Changing the tracking feedback may add another variable to the troubleshooting.

JR

 

[jeffbro]

I'm still hoping someone might look at the schematic and let me know if I have overlooked something. This supply (one of two) is for a large 40 year old console and could not be replaced if it was damaged due to some oversight on my part. In other words, a high risk situation where caution is in order.

 

[JohnRoberts]

I would start by reading the data sheet for the 723.... You can probably defeat one of the cross connected feedback path by just removing (disconnecting) the transistor.

What is it doing that has you convinced that you need to change the design?

JR

 

[Hubbub]

For isolating V+ if you lift R46 then Q3 base will try to go to -18V. This exceeds the rated Vbe(reverse) and may damage Q3. A simpler solution is just short Q3 base to emitter, leaving R46 and R47 in place.

I'm thinking the V- isolation solution you propose should work OK. If you fix V+ first then this step is not needed, simplifying your efforts.

 

[jeffbro]

Perhaps I was unclear. I don't want to change the design, I want to temporarily disable rail to rail tracking since both rails are dead. With tracking disabled I will be able to troubleshoot the regulators individually without the other dead regulator turning off the regulator under repair.

Thank you for pointing out the risk to Q3. An oversight like that is exactly why I was hoping for someone else to critique my plan.

 

[JohnRoberts]

You can disable the rail tracking by removing or disconnecting Q3. It may regulate to a different voltage so READ THE DATA SHEET....

Good luck, do no harm.

JR

 

[jeffbro]

Given the cramped component layout and other considerations I think the simplest path in this instance is to lift the legs of R46+47 where they join, leaving the base of Q3 open.

 

[benb]

Regarding the 723 and its operation, go to this site and click on "Download a sample chapter."
https://artofelectronics.net/

 

[Hubbub]

Pulling R46 and R47, thus leaving Q3 base open, allows any base-collector leakage to be multiplied by the transistor's gain and pull down on R41, giving U5 an incorrect value of reference. Can't hurt anything, but might make debugging difficult. Even lifting the Q3 collector allows the base to go too far negative under some conditions, again not recommended. That's why shorting base to emitter is the best choice.

FWIW I not long ago fixed some failed 723 based supplies, the only problem in all 3 was that the setting trimpot wiper had opened up. This supply is vulnerable to the same issue as a safer design would have cost another resistor or two.

 

[radardoug]

I recently mended two 723 based designs by just replacing the 723. Died of old age. That would be my first step.

 

[jeffbro]

Thanks for the additional rational for shorting b-e.

Give me a few days (a week?) and I'll let you know what I find.

Among the consoles of this type that I service there are at least 5 of these supplies and, sadly, I see at least one of them every year, but this is the first time I've seen a supply lose both the + and - rails.

And yes, I've also seen failures of the 723's. They've been working hard for decades and sometimes just give out.