Good idea to leave 5 1/2 digit DMM powered on all the time?

[Dave92F1]

I have a ~1975 Systron Donner 7205 bench DMM; 5 1/2 digits (all transistor; no valves).

I modified it so the power switch just turns off the display, while leaving the rest of the DMM powered 24x7, thinking that it'll be at thermal equilibrium then and give accurate readings without a long warm-up period.

It's completely cool to the touch (without a fan), so it doesn't draw much current.

Is this a good idea? Or am I needlessly wearing out 40 year old irreplaceable parts? Do solid-state electronics age quicker when powered, or does that not matter?

[gilbenl]

There's no need to leave it powered up for routine measurements.

[retrolefty]

I agree that it gains little to keep it powered up.

Do solid-state electronics age quicker when powered, or does that not matter?

 However I find the above an interesting question. I know that MIL parts are sometimes bought with a minimum expected lifetime but don't know it powered Vs unpowered would have large variation? Anyone with firsthand experience with solid state lifetime specs?

[electr_peter]

I would not bother and leave it turned off unless I need instant turn on performance for some reason. I seriously doubt that only 5 and 1/2 digit DMM can gain anything from continuous power on...

[Vgkid]

Can you do a teardown of your meter?

[gilbenl]

Can you do a teardown of your meter?

+1

You can't hurt this:

source: http://www.amplifier.cd/Test_Equipment/other/Systron_7205.html

[saturation]

In general, it beneficial so long as the device is well ventilated in standard room temperature, but its wasteful of energy; the quiescent drain of even 1980s style DMMs are in the 20-25W range.  Electrolytic capacitors 'reform' when powered up, and their lifespan will be substantially prolonged in use.  But we find that if you power up rarely used devices 1-2x a year, its enough to keep most electrolytics from blowing.

Bench DMMs will be in ready state in 30min, and at its best in 1 hour at room temperature.  Metrology grade DMMs will be in a stead state too, but they are never shut off to prevent very very small amounts of drift.

Heat is what kills most components, active and passive.  If the operating temperature of a device is nearer to room temperature, it can last 'forever'.  MILSPEC estimates are based on accelerated aging algorithms on MILSPEC grade components though. 

[Vgkid]

With a design like that, you do not need to leave it on continuously. Though after recapping my 3456a I left it on for a week.

[Dave92F1]

There's no need to leave it powered up for routine measurements.

Thanks for all the answers. BTW, mine has a Panaplex display (7 segment neon), not a LED display.

I don't use it for routine measurements, I have a modern Vici VC8145 bench DMM for that.

But the Systron Donner is far more accurate (tho much less convenient to use), so I use it those few times when I need a really precise number (OK, relatively precise - it's still only 5 1/2 digits). And as a 2nd DMM.

Sounds like I'm not doing any harm to it by leaving it powered up. (I'm not worried about the electric bill...)

[Wuerstchenhund]

Is this a good idea? Or am I needlessly wearing out 40 year old irreplaceable parts? Do solid-state electronics age quicker when powered, or does that not matter?

Based on you saying that you don't use the device regularly or very often (i.e. at least daily) I'd say it's not a very good idea.

In general, keeping instruments powered on is a good idea if the device is used regularly (i.e. daily), as it limits the thermal stress from the change between hot and cold cycles. However, continuously powered on instruments also consume their life even when idle, which isn't much of a problem with newer kit that is young and at the beginning of it's useful service life, but for old gear that has more or less reached the end of its life keeping things powered up is pretty wasteful.

Keeping your DMM powered on may not draw much power (or an amount that concerns you), but you're wearing out components that by now are probably already very close to the end of their natural lifetime. Capacitors do "wear out", as do other components like transistors and ICs. Some are easily replaceable, others maybe not. With gear that old, there's a good chance that you consume the remaining life of components for basically nothing.

And I didn't even touch the environmental/ethical impact (i.e. climate change and such) of such wasteful behavior (i.e. creating CO2 for essentially no purpose).

For such an old device and considering your usage profile, you're much better off by keeping it powered down until you really need it, and when you need to use the DMM keep it powered up for a few hours before powering it down again.

[retrolefty]

Capacitors do "wear out", as do other components like transistors and ICs.

 I'm not aware of a "wear out" specification for transistors and ICs?

[Wuerstchenhund]

Capacitors do "wear out", as do other components like transistors and ICs.

 I'm not aware of a "wear out" specification for transistors and ICs?

There isn't a "wear out" specification (aside from the standard MTBF rating when operated in a controlled environment), however there definitely is "weardown" at an atomic level in all areas where there is a flow of electrons. It's pretty minimal (which is why semiconductors in general have a very long service life), but unless there's some external cause like overvoltage/ESD then this "weardown" is usually the reason when a semiconductor eventually fails internally.

Normally this isn't much of a problem, however with a now 40yr old DMM I certainly wouldn't risk it to waste the remaining life margin on useless operation.

[retrolefty]

but unless there's some external cause like overvoltage/ESD then this "weardown" is the reason when a semiconductor eventually fails.

Is this atomic eventuality shorter or longer then the standard lifetime of a human life? 

[Wuerstchenhund]

but unless there's some external cause like overvoltage/ESD then this "weardown" is the reason when a semiconductor eventually fails.

Is this atomic eventuality shorter or longer then the standard lifetime of a human life? 

That depends on various factors. Some components might exceed 100yrs, others often last less than a decade (i.e. flash memory).

[G0HZU]

FWIW I turn off everything at the wall in my work areas whenever I finish working. So nothing is left on overnight or if the house is empty.

It all gets tuned off at the wall when I finish work. So there is no live mains fed to any of my equipment or test gear and nothing can run in any kind of standby mode.

I'm not in the least concerned if this lengthens or shortens the life of my test gear or if it has any impact on calibration ageing etc. I am much more concerned about fire safety and also not having numerous items drawing (pointless and wasteful) standby current 24/7.

So in my situation I would not consider it a good idea to leave an old DMM running 24/7 especially if it also means leaving a live feed to several other items of test gear even if they are switched off at the front panel.

[gilbenl]

FWIW I turn off everything at the wall in my work areas whenever I finish working. So nothing is left on overnight or if the house is empty.

It all gets tuned off at the wall when I finish work. So there is no live mains fed to any of my equipment or test gear and nothing can run in any kind of standby mode.

I'm not in the least concerned if this lengthens or shortens the life of my test gear or if it has any impact on calibration ageing etc. I am much more concerned about fire safety and also not having numerous items drawing (pointless and wasteful) standby current 24/7.

So in my situation I would not consider it a good idea to leave an old DMM running 24/7 especially if it also means leaving a live feed to several other items of test gear even if they are switched off at the front panel.

Do you mind sharing what type of work/equipment? Do you turn each item off individually first or just flip the mains?

This conversation does raise a good point about cycling. Lets take a semi-precision DMM like a 3456, 3457 or 34401. Assume you work an 8 hour day, 5 days a week. Over the course of a year, will you put greater wear/stress by:

A) Turning the meter on only during the work day. As a result, you reduce the operating hours by ~85%, but power cycle ~250 times/year
B) Turning on the meter on Monday and off on Friday evening. You only reduce the operating hours by ~15%, but only power cycle ~50 times/year
C) Leaving the meter on 24/7, never power cycling

It's my understanding that failures most commonly occur at power up. So, it would seem that keeping the instrument at a "steady state" is ideal.

[G0HZU]

It's just typical electronics gear. Mostly test gear plus a PC and some powered tools.
I generally turn each item off at its regular front panel control and then turn off everything at the wall.

It's my understanding that failures most commonly occur at power up. So, it would seem that keeping the instrument at a "steady state" is ideal.

Very possibly, but this type of failure mode isn't much of a concern to me. I'd rather fix the odd failure in my test gear than leave it running 24/7. The expensive 10MHz OCXO ovens in my old (but incredibly expensive when new) spectrum analysers are designed to be left running 24/7 to improve ageing etc but I don't leave them running in standby overnight here at home and they weren't left in standby overnight when they were still in use at my place of work. So that means over 25 years of daily power cycling when they should have been either fully turned on or in standby all of their working life. But nobody at work was ever concerned about this because you can't realistically leave a spectrum analyser permanently powered in one place day after day in a busy R&D lab full of engineers.

My old HP8568B analyser got moved a lot between work benches and from lab to lab (and even between buildings) all its working life before I bought it from the company about 10 years ago.

[LaurentR]

In VLSI, there are plenty of on-die mechanisms that reduce MTBF like electromigration and hot-carrier injection. There are all kept in check by running checkers during the design process, tweaking the flagged circuit/wire when necessary. Many consumer-grade chips aim for around a 10yr MTTF for their use case. It is mostly "by construction" (hard to test), so most chips will survive much longer in practice.

Powering on/off has nasty internal effect on die, but the real issue is thermal cycling at the package level that can create all kinds of cracking on bumps or balls. I've seen large ASICs with esoteric packages for server/networking environments rated at 100 lifetime cycles. Better not switch those off too much.

[lowimpedance]

Thanks for all the answers. BTW, mine has a Panaplex display (7 segment neon), not a LED display.

I don't use it for routine measurements, I have a modern Vici VC8145 bench DMM for that.

The panaplex displays do have limited life times so definitely best to use only when needed and given the small usage there is no good reason to leave it on .
(even with the mod to turn the display only off).

I have one of these old timers in a non functional state (no manual), and also a working 7004 and 6052 counter with nixie displays (+manuals). All the same form factor.
Just for the collection though, not used for any measurements as such.

[Wuerstchenhund]

This conversation does raise a good point about cycling. Lets take a semi-precision DMM like a 3456, 3457 or 34401. Assume you work an 8 hour day, 5 days a week. Over the course of a year, will you put greater wear/stress by:

A) Turning the meter on only during the work day. As a result, you reduce the operating hours by ~85%, but power cycle ~250 times/year
B) Turning on the meter on Monday and off on Friday evening. You only reduce the operating hours by ~15%, but only power cycle ~50 times/year
C) Leaving the meter on 24/7, never power cycling

I've worked (and still do) in a lot of labs, and most of them switch gear off (kit with OCXOs are left in standby) at the end of the working day.

I've only seen test kit powered on continuously in labs which basically run tests 24/7.

It's my understanding that failures most commonly occur at power up.

Yes, because the power-on cycle is usually the most stressful operating mode. However, there still needs to be a defect somewhere for a device to fail at power on, which means even if it was powered constantly it very likely would have failed anyways.

Powering on/off has nasty internal effect on die, but the real issue is thermal cycling at the package level that can create all kinds of cracking on bumps or balls. I've seen large ASICs with esoteric packages for server/networking environments rated at 100 lifetime cycles. Better not switch those off too much.

Anyone remembers SiliconGraphics? The PSUs they used in their servers and most of their workstations had a very low power cycle lifetime, and died quickly if switched on/off regularly.

However, test kit is usually designed for a high number of power cycles, so switching test instruments off at the end of the day shouldn't really be an issue.