Manufacturer using part over maximum ratings, is it safe?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi, I am working on repairing a Marantz PM17 mk2 and found the fault to be a short in the LC78212 analogue switch. When looking in the datasheet the maximum supply voltage is ± 20V, and on the schematic of the amplifier the power supply voltages is stated to be ± 23V.

So my question is do you think it will be safe to replace the chip with a new from ebay without blowing up? Or do you think I should modify the circuit so the LC78212 gets a lower supply voltages, or maybe you have a good idea of how this could be done?

Thanks for any thoughts.

LC78212  Absolute Maximum Ratings.png LC78212  Marantz PM17 mk2.png
 
Administrator
Joined 2007
Paid Member
I looked at a different data sheet but it to shows 20 volt as a max rating. Curious. I would definitely look at reducing the rails which should be easy in practice.

Swapping these two zeners for a low value will drop the rails. Rail voltage equals the zener voltage less the two Vbe volt drops of the series pass transistors... so 20 volt zeners would give around -/+ 18 volts.
 

Attachments

  • Annotation 2019-08-01 153140.png
    Annotation 2019-08-01 153140.png
    120.7 KB · Views: 347
Administrator
Joined 2007
Paid Member
If you want to lower the supply to just this chip then it's a little more involved. There are a couple of options you could try.

1/ Add two series zeners, one for each rail that feeds the chip. Use something like 4v7 Zeners. The current consumption of the chip is very low and so that really suggests that as well as the zeners you should include two resistors of say 8k2 wired from the output of the zeners to ground. That would add an additional known load of a couple of milliamps per rail and ensure the zeners always give a rail of around -/+18 volts. Without the resistors you could find the rails are higer than expected if the chip has a very low current draw under certain conditions.

So dead easy, two zeners and two resistors.

2/ Same number and type of components but different values and arrangement. Add two 18 volt Zeners across each rail of the chip and add series feed resistors. With 24 volt rails and lets say the chip could use 3 milliamps then you would need a resistor of around 820 ohms per rail. That would drop the excess 6 volts per rail and allow a small Zener current to flow all the time.
 
I wouldn't replace with a Fleabay chip, I would source a Sanyo LC chip.
I don't think Marantz would under engineer any of their designs.
LC78212 Sanyo,DIP-30,04+/05+ | IC Chips | UTSOURCE

UTSOURCE is also an eBay seller that sells counterfeit parts regularly. UTSOURCE is NOT an authorized parts dealer. UTSOURCE is just broker like Littlediode UK. UTSOURCE and Littlediode will gladly sell anything that they can get their hands on.

The older 22xx Marantz series under engineered their electrolytic capacitors. This notion that big names don't make this mistake is just wishful thinking.

Intel screwed up with their Pentium FDIV bug, NASA screwed up with their 1998 Mars orbiter SI units, etc. These are entities that are suppose to be the gold standard in engineering but they still made mistakes despite the rigorous checks they have in their system.

Humans make mistake, we all make mistakes.

Salberg found a Marantz engineering error. Let us not downplay Salberg's findings. Keep an open mind and give Salberg credit for finding this problem for others. These are Salberg's first posts from the last 8 years and I can see why he didn't mention the manufacturer's name on the title.

Kudos Salberg for an informative and quality post!
 
A manufacturer can get parts tested for a particular specification. That looks like what was done here.

I would not change the zener diodes as that may cause the regulator transistors to overheat.

I would use zener diodes in series with the power input pins to drop the voltage to 18 volts. Even a 4.7 volt 1/2 watt unit will work as the current is under 1 mA.
 
Administrator
Joined 2007
Paid Member
These CMOS type switches, and it could even be the same part number, were used extensively in mid price music systems from the likes of Aiwa and many others in the late 80's and 90's and they were a fairly regular failure item. Fwiw, these systems used nothing like -/+20, more like -/+9 from memory.
 

PRR

Member
Joined 2003
Paid Member
...looking in the datasheet the maximum supply voltage is ± 20V...

ABSOLUTE MAX!

Next page, Operating Max, is +/-18.5V.

Get it down to there.

There is an anecdote in the studio console world. A console maker asked for specially selected TL072 chips to stand +/-20V (normally rated +/-18V). They did have a special part number. However after 30++ years many have failed.

True story: I ran 28V chip-amps on 35V. Worked for 9 years and then dead. Replaced, worked another 9 years, dead. By this time they were on Lifetime Buy, so I got 2 more, and then quit that place.
 
Not another anecdote!

Do you have any shred of proof or evidence that supports this unsubstantiated statement of yours?

What is true to one case doesn't necessarily apply to another.

No anecdote having real experience in the field both with the designers and service folks. It is a very common practice as noted by others here.

Do you really think running an 18 volt nominal part at 23 volts would get past a design review? Even make it past the warranty period?

Yes there are actual products out there where parts are used in an off data sheet parameter method. Not surprisingly high warranty failure rates and quickly redesigned. Almost an unknown issue from major manufacturers.
 
Last edited:
I'm fully aware of discrete devices that are cherry-picked, op amps tested for noise, etc. Testing cost money. Bean counters in major manufacturers of mass consumer electronics want to reduce their BOM cost not the other way around yet you suggest the contrary.

The OP is talking about a Sanyo/ON LC78212 CMOS IC which is completely different from what you experienced in your lifetime. If you want to expand your insider knowledge about YOUR own subject, open a new thread instead of thread-jacking this one.

You talk about an imaginary Marantz PM17 II design review that you weren't part of and even have insider knowledge on how it transpired. Seriously?

The OP did his failure analysis and I agree with his assessment.

I think simple, it is elegant. Over analyzing makes you crazy.
 
Wow you are psychic... not. I have used the ON LC78212 and similar units in some of my designs. I also have worked with some of the folks who do designs for Marantz and have a friend who consults with and reviews some of their products.

Using a selected part is usually cheaper than than adding parts. Screening for noise is not. IC processes of often run at 36 volt-ish limit.

I also have mentioned the best solution is to add zeners in series with the voltage rails to drop the voltage, but obviously you didn't read or comprehend that.

BTY today I got the go ahead for a new gizmo I designed. It will be a medium size run of 10,000 pieces.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.