Reliability of electronics

28 views
Skip to first unread message

Trevor Wilson

unread,
May 30, 2022, 3:24:53 AMMay 30
to
As discussed elsewhere, I commented on the appalling reliability of
certain items purchased from Jaycar and I expressed my opinion that
Jaycar should not be relied on for anything sophisticated. Like
capacitors and transistors.

It got me thinking though. I've been in the service business for quite
awhile and, when I was Marantz service manager I was issued with a
confidential list of reliability figures for various Marantz models. The
list included any reported problem within the 3 year warranty period.
Here's some of the models I recall:

Model 1070. 0.5% failure rate. Which, I might add, includes the early
variant which suffered relay problems.

Model 1060. 4% failure rate.

Model 4230. 63% (yes, SIXTY THREE) failure. I put it down to the use of
very early TO220 pack output devices. These were not reliable back in
the early 1970s.

All the amplifiers fitted with TO3 output devices demonstrated quite
respectable reliability, except....

Model 500. 100% failure rate. In fact, each Model 500 usually required
service several times during it's warranty period. I budgeted 10 hours
to repair each one. As a consequence, of the 3 amplifiers imported by
Marantz, only to were sold to the public. It was cheaper to bury the
last one in landfill. I negotiated a price and I still own the beast. It
is a highly desirable amplifier. The Absolute Sound reported: "The
Marantz Model 500 would be issued with our 'best power amplifier ever'
if only we could manage to get one to operate for more than six weeks
without blowing up. Designed in 1963, it used, unusual for the time,
full complementary symmetry outputs. Unfortunately, the Voltage rating
on the output devices was marginal.

I was a warranty agent for another major brand, when DVD players were
first released. I was instructed by the Australian agent that their
first DVD player would certainly fail within the warranty period. They
were expecting a 100% failure rate. I was told it would be a busy time.
They were correct.

Trevor Wilson

unread,
May 30, 2022, 5:33:40 AMMay 30
to
On 30/05/2022 5:24 pm, Trevor Wilson wrote:
Designed in 1963,

**Oops. Released to market in 1972, so likely designed in 1971-2.

Phil Allison

unread,
May 30, 2022, 7:01:55 PMMay 30
to
Trevor Wilson wrote:
=================
>
>
> Model 500. 100% failure rate. In fact, each Model 500 usually required
> service several times during it's warranty period. I budgeted 10 hours
> to repair each one. As a consequence, of the 3 amplifiers imported by
> Marantz, only to were sold to the public. It was cheaper to bury the
> last one in landfill. I negotiated a price and I still own the beast. It
> is a highly desirable amplifier. The Absolute Sound reported: "The
> Marantz Model 500 would be issued with our 'best power amplifier ever'
> if only we could manage to get one to operate for more than six weeks
> without blowing up. Designed in 1963, it used, unusual for the time,
> full complementary symmetry outputs. Unfortunately, the Voltage rating
> on the output devices was marginal.
>

** The model 500 dates from 1973.
Each channel used 8 x TO3 transistors with SJ prefixes ( SJ2404 and 2405)
So specially selected types made by Motorola.

https://www.hifiengine.com/manual_library/marantz/500.shtml

Rated power was 250W at 8ohms, so 8 devices should be plenty.
Weight was 83 pounds - which is absurd.




..... Phil

Trevor Wilson

unread,
May 30, 2022, 8:35:56 PMMay 30
to
**If you've downloaded the service manual, you can see the complexity of
the beast. SOA adjustments are very comprehensive. The fault with the
amp lay with the VCEO of the output devices. There was very little
margin for mains fluctuations. Marantz finally solved the problem by
specifying 200 Volt, Japanese output devices, rather than the Motorola
ones originally fitted.

BTW: The Model 500 would typically deliver well over 300 Watts @ 8 Ohms.
As I recall, around 320 Watts, both channels driven.

Yeah, it's a heavy sucker. Fan cooled too, so heat sink mass is not
huge. Big power transformer and 4 huge main filter caps.

It was replaced by the Model 510/M, which was far more reliable, more
compact, lower mass and sounded like shit. Series/parallel output
devices. Yuk.



Phil Allison

unread,
May 30, 2022, 8:59:04 PMMay 30
to
Trevor Wilson wrote:
>=================
>> Designed in 1963, it used, unusual for the time,
> >> full complementary symmetry outputs. Unfortunately, the Voltage rating
> >> on the output devices was marginal.
> >>
> >
> > ** The model 500 dates from 1973.
>
> > Each channel used 8 x TO3 transistors with SJ prefixes ( SJ2404 and 2405)
> > So specially selected types made by Motorola.
> >
> > https://www.hifiengine.com/manual_library/marantz/500.shtml
> >
> > Rated power was 250W at 8ohms, so 8 devices should be plenty.
> > Weight was 83 pounds - which is absurd.

> **If you've downloaded the service manual, you can see the complexity of
> the beast. SOA adjustments are very comprehensive. The fault with the
> amp lay with the VCEO of the output devices. There was very little
> margin for mains fluctuations.

** That assertion makes no sense.
Facts:
1. Output transistors do not operate in Vceo mode.
2. They are only exposed to * half* the DC supply at idle or moderate volumes.


> Marantz finally solved the problem by
> specifying 200 Volt, Japanese output devices, rather than the Motorola
> ones originally fitted.

** MJ numbered power BJTs all exceed their Vceo ratings on test.
They way exceeded the ( higher) Vcer rating too - as it almost matches the Vcb rating.

SJ numbers have no published ratings.


..... Phil

Trevor Wilson

unread,
May 30, 2022, 9:31:55 PMMay 30
to
**Correct. All the US built Marantz models used Motorola devices with an
SJ prefix. Even some of the early Japanese ones used Motorolas as well.
Very unusual. I measured a few back in the day. As I recall none of the
Model 500 output devices exceeded 160 Volts. I may still have a few
originals lying around. If I locate them, I'll measure the breakdown
Voltages. Don't forget: It was 1973. High voltage, high power PNP
devices were very scarce.

Phil Allison

unread,
May 30, 2022, 9:53:00 PMMay 30
to
** Measured with open base ? Not how they are used.

> I may still have a few originals lying around.
> If I locate them, I'll measure the breakdown
> Voltages.

** If you do, put 100ohms between B and E.
Makes at least a 10% increase .

> Don't forget: It was 1973. High voltage, high power PNP
> devices were very scarce.

** SJ types were factory selected from stock, mostly for Vbe matching.
Crown, SAE and Peavey used them all the time.

Total PITA for repairers, who had zero clue what the types really were.

BTW

if the Marantz SJs lacked adequate Vce, examples would have commonly failed bench testing.
So I don't buy the idea.



...... Phil






Trevor Wilson

unread,
May 30, 2022, 10:01:00 PMMay 30
to
**Will do.

>
>> Don't forget: It was 1973. High voltage, high power PNP
>> devices were very scarce.
>
> ** SJ types were factory selected from stock, mostly for Vbe matching.
> Crown, SAE and Peavey used them all the time.
>
> Total PITA for repairers, who had zero clue what the types really were.

**Yep.

>
> BTW
>
> if the Marantz SJs lacked adequate Vce, examples would have commonly failed bench testing.
> So I don't buy the idea.

**Point taken. When the Japanese devices were installed, the Model 500
became reasonably reliable. If I recall correctly, most amps failed at
switch on, rather than under use. I could be wrong.

Phil Allison

unread,
May 30, 2022, 11:03:06 PMMay 30
to
Trevor Wilson wrote:
=================
> >>>
> >>> SJ numbers have no published ratings.
> >>
> >> **Correct. All the US built Marantz models used Motorola devices with an
> >> SJ prefix. Even some of the early Japanese ones used Motorolas as well.
> >> Very unusual. I measured a few back in the day. As I recall none of the
> >> Model 500 output devices exceeded 160 Volts.
> >
> > ** Measured with open base ? Not how they are used.
> >
> >> I may still have a few originals lying around.
> >> If I locate them, I'll measure the breakdown
> >> Voltages.
> >
> > ** If you do, put 100ohms between B and E.
> > Makes at least a 10% increase .
>
> **Will do.
> >
> >> Don't forget: It was 1973. High voltage, high power PNP
> >> devices were very scarce.
> >
> > ** SJ types were factory selected from stock, mostly for Vbe matching.
> > Crown, SAE and Peavey used them all the time.
> >
> > Total PITA for repairers, who had zero clue what the types really were.
>
> **Yep.
> >
> > BTW
> >
> > if the Marantz SJs lacked adequate Vce, examples would have commonly failed bench testing.
> > So I don't buy the idea.
>
> **Point taken. When the Japanese devices were installed, the Model 500
> became reasonably reliable. If I recall correctly, most amps failed at
> switch on, rather than under use. I could be wrong.


** I once used to see a lot of Phase Linear 400 mk2 amps.
A revised version of the famous PL400 hi-fi model.
Complementary MJ output devices and input op-amps on the pcb.
These were all used in PA systems - not a good idea.

Saw a lot of blown output stages and many that simply went DC taking speakers with them.
Nothing to do with Vce or even SAO limits.

The first problem was due to overheating of output devices - since owners had to devise fan cool themselves.
The installed temp cut out was useless, since all TO3s were mounted using thick, pink silicone thermal pads.
I called the "thermal insulators" as they allowed devices to get 50 C hotter than when mica and grease was used.

Going DC was due to 5W ww resistors that fed +/- 16V zener regulated rails for the op-amps.
The resistors were all faulty, going bright green inside and hence open cct.

Did a nice trade in fitting relay speaker protectors in many power amps.
( wired the RIGHT way so DC arcs were eliminated)


...... Phil

Trevor Wilson

unread,
May 30, 2022, 11:45:17 PMMay 30
to
**Good plan. FWIW: The Model 500 never overheated. Thermostat
controlled, two speed fan cooled. Output devices were all mica/thermal
paste mounted. Even after repair (using original output devices) the
amps failed. Only after the output devices were substituted with
Japanese types did reliability significantly improve. The amp could run
all day at 40% max power. Easy. Even into 4 Ohm loads.

Phil Allison

unread,
May 31, 2022, 12:02:13 AMMay 31
to
Trevor Wilson wrote:
=================
> >
> **Good plan. FWIW: The Model 500 never overheated. Thermostat
> controlled, two speed fan cooled. Output devices were all mica/thermal
> paste mounted. Even after repair (using original output devices) the
> amps failed. Only after the output devices were substituted with
> Japanese types did reliability significantly improve.
>

** Were the SJs in steel or Aluminium paks?

Early 70s Motorola TO3s had a big problem with thermal expansion of chip headers.
The silicon and the header had differing tempcos of expansion.
The end result was the chip developed micro cracks.

Such devices were speced at a mere 5000 thermal cycles.
OK for some apps but not class AB audio.
Sudden, unaccountable failure was the norm.

Aluminium paks got dumped and were never seen again.


..... Phil

Trevor Wilson

unread,
May 31, 2022, 12:05:38 AMMay 31
to
On 31/05/2022 2:02 pm, Phil Allison wrote:
> Trevor Wilson wrote:
> =================
>>>
>> **Good plan. FWIW: The Model 500 never overheated. Thermostat
>> controlled, two speed fan cooled. Output devices were all mica/thermal
>> paste mounted. Even after repair (using original output devices) the
>> amps failed. Only after the output devices were substituted with
>> Japanese types did reliability significantly improve.
>>
>
> ** Were the SJs in steel or Aluminium paks?

**Like all Motorola TO3 devices of the early 1970s, they were in those
horrible aluminium packs.

>
> Early 70s Motorola TO3s had a big problem with thermal expansion of chip headers.
> The silicon and the header had differing tempcos of expansion.
> The end result was the chip developed micro cracks.

**Yep. I recall reading the RCA white paper on the issue.

>
> Such devices were speced at a mere 5000 thermal cycles.
> OK for some apps but not class AB audio.
> Sudden, unaccountable failure was the norm.
>
> Aluminium paks got dumped and were never seen again.

**Yep. Steel or copper is the way to go for TO3 devices. RCA won the
argument.

Trevor Wilson

unread,
May 31, 2022, 12:08:05 AMMay 31
to
On 31/05/2022 2:02 pm, Phil Allison wrote:
**BTW: The Marantz Models: 240, 250, 250M, 1120, 1200/B, 140, 2270, 2275
all used Motorola aluminium cased To3 devices. All demonstrated
reasonable reliability. None were fan cooled like the Model 500. The
Model 500 ran much cooler than all those models under normal operation.

Trevor Wilson

unread,
May 31, 2022, 12:18:20 AMMay 31
to
**Oh and with all those models, a visual inspection of the output
devices would frequently and quickly reveal the problem - A tiny pinhole
in the case, where, presumably, a piece of white hot silicon ejected
itself from the header.

Phil Allison

unread,
May 31, 2022, 12:26:47 AMMay 31
to
Trevor Wilson wrote:
---------------------------------
>
> > Such devices were speced at a mere 5000 thermal cycles.
> > OK for some apps but not class AB audio.
> > Sudden, unaccountable failure was the norm.
> >
> > Aluminium paks got dumped and were never seen again.
> **BTW: The Marantz Models: 240, 250, 250M, 1120, 1200/B, 140, 2270, 2275
> all used Motorola aluminium cased To3 devices. All demonstrated
> reasonable reliability. None were fan cooled like the Model 500. The
> Model 500 ran much cooler than all those models under normal operation.
>

** The chip in a TO3 can heat in milliseconds - or at least a small part of it can.
This is why excursions beyond SOA limits are often fatal.
High Vces are the killer - aka "second breakdown".

Motorola 2N3055s and MJ802/4502 in Al paks were OK - I used heaps of them.
Big chip devices like the MJ15003/4 and MJ15024/5 were not so lucky.

...... Phil


Phil Allison

unread,
May 31, 2022, 12:32:09 AMMay 31
to
Trevor Wilson wrote:
=================
> >>>
> >>> **Good plan. FWIW: The Model 500 never overheated. Thermostat
> >>> controlled, two speed fan cooled. Output devices were all mica/thermal
> >>> paste mounted. Even after repair (using original output devices) the
> >>> amps failed. Only after the output devices were substituted with
> >>> Japanese types did reliability significantly improve.
> >>>
> >>
> >> ** Were the SJs in steel or Aluminium paks?
> >>
> >> Early 70s Motorola TO3s had a big problem with thermal expansion of
> >> chip headers.
>
> **Oh and with all those models, a visual inspection of the output
> devices would frequently and quickly reveal the problem - A tiny pinhole
> in the case, where, presumably, a piece of white hot silicon ejected
> itself from the header.
>

** Ha ha, no way is that true.

What really happens is the Al feed wire ( aka fuse) from chip to TO3 emitter pin vaporises.
This gives a path for a DC arc to jump from the pin to the inside top of the pak.
A neat hole then gets drilled until the arc self quenches.

Hitachi TO3 lateral mosfets were famous for doing that !!!


..... Phil



Reply all
Reply to author
Forward
0 new messages