Replacing old SCR's

[desertrout]

In the Macintosh restoration community, we're finding that some components are getting difficult to replace. One in particular is a 0.8A/30V/200uA fast SCR that has several old part #'s: E0102YA, E0122, CR400Y... but none of these are in production. The modern equivalent is the 2N5060 - it looks the same on paper - but it doesn't work. (In this case, it's part of a 'kickstarting' circuit of a switching power supply, the 2N5060 results in crowbarring, the E0102/E0122 don't)

My question is why would that be?

Here's the datasheet for the E1020YA: https://datasheet.datasheetarchive.com/originals/scans/Scans-007/Scans-00148351.pdf
Here's the datasheet for the 2N5060: https://my.centralsemi.com/datasheets/2N5060_SERIES.PDF

One theory is that the gate current trigger test conditions are different for each... that the E0102YA tested at 10Ohms, the 2N5060 was tested at 100Ohms, a magnitude difference... but I can't find a datasheet to confirm that.

What am I missing?

[bdunham7]

Can you post  a schematic of the circuit that it is in?

[desertrout]

Unfortunately there's no official schematic for this circuit... it's for a Macintosh 128k / 512k/ Plus analog board, which includes a switching power supply and the video sweep circuit all on the same board. Some folks have attempted to reverse-engineer schematics... I've attached a screenshot from repair guide (https://vintageapple.org/gamba2/images/plus_analog.PDF) that isolates the part of the circuit where the E0122 (at Q10) resides...

[akimpowerscr]

Try to select 2N3050 for the lowest Tq possible.
tq typical is 10µs.

Tq of E0102YA is not specified on the datasheet but must be lower than that.

[David Hess]

Try to select 2N3050 for the lowest Tq possible.
tq typical is 10µs.

Exactly, going by the datasheets, the 2N3050 does not turn off as quickly as the original "fast" SCR.  The E0102 is 10 microseconds maximum while the 2N3050 is 10 microseconds typical, so could be much slower.

If I was desperate, I might replace the SCR with a pair of fast high voltage bipolar transistors and some passives in the SCR configuration.

[Doctorandus_P]

If you can't find a faster SCR, then maybe de sensitizing the crowbar circuit a bit is a viable option?

But instead of changing random stuff, modifications like this beg for some serious measurements with a scope to diagnose what is really happening.

[David Hess]

If you can't find a faster SCR, then maybe de sensitizing the crowbar circuit a bit is a viable option?

It is not a crowbar circuit exactly.  Other switching converters based on the blocking oscillator use a transistor switch in place of that SCR to do the same thing.  That is duty cycle modulate the blocking oscillator.

[akimpowerscr]

It is a self-oscillating flyback converter that works on the same principle as some Philips oscilloscope power supplies.

When the current in transistor Q11 reaches around 1A, transistor Q9 conducts and triggers thyristor Q10 which blocks Q11.

Thyristor Q10 must have a very short extinction time (tq) so as not to remain in conduction at the end of the cycle when Q11 must again enter into conduction.

Although the BRY39 used as a fast thyristor in Philips power supplies is also difficult to find, but less so than the E0122, it could replace the E0122 with a simple modification.

https://www.eevblog.com/forum/repair/replacement-for-an-old-bry39-programmable-unijunction-transistor/msg3292608/#msg3292608

[desertrout]

Thanks everyone. What you're all saying makes sense.

There is in fact an official schematic - I just found it this afternoon, so I've attached here in case there's more to say. I'll continue to look for a faster SCR with the right values.

Of note is also this post (https://hackaday.io/project/168100-fat-macintosh-repair/details) where the author made some observations of the system with a scope and made some alterations that seem to hold - for curiosity's sake.

[akimpowerscr]

I'll continue to look for a faster SCR with the right values.

Fast thyristor technology is absolete and even for power inverters it has become difficult to find spare parts.

I believe that you have no chance to find any other alternative than those given to you in this topic.

But if you feel like wasting your time, it's your right. 

Buy 10 x 2N3050, try them in the power supply, some will probably work, some will not.

  I found one: BR103 from Siemens.

https://www.ebay.de/itm/5x-30V-0-8A-Thyristor-BR103-Siemens-Halske/281011512080?epid=1169291962&hash=item416d976310:g:-AUAAOxyZqhRiQic

[desertrout]

I've also found on ebay NOS ECG5589's and NTE5400's that considered replacements. But this of course is a short term win. You are correct, ultimately, once this old stock has dried up, other solutions will need to be found.

[akimpowerscr]

No way......

ECG5589 is a thyristor of 350A 

NTE5400 is not a fast scr, it will never work. (Tq is not even specified in the data sheet!)

[desertrout]

Mis-type - I meant SK3950, but it is billed as a NTE5400 replacement. How do you know for sure the NTE5400 will not work - is the fact that tq is not on the datasheet proof?

[akimpowerscr]

Tq (or toff) is an essential parameter of a fast thyristor.

It is obvious that if this parameter is not even informed in the datasheet, it is not a fast thyristor.

Fast thyristors (including power thyristors) have a tq (or toff) generally between 5 and 50µs
Main's thyristors generally have a tq between 50 and 150µS.

Fast thyristors had been developed for inverters.
In this application, they were first replaced by GTOs (gate turn off thyristors, which was not a good technology) and they are now replaced by mosfets or IGBTs

Fast thyristors are absolete technology, which is why they are hard to find because they have not been manufactured for a long time.

[Alex Eisenhut]

Fast thyristors are absolute technology, which is why they are hard to find because they have not been manufactured for a long time.

I think you mean "obsolete".

[akimpowerscr]

Yes indeed, it is a google translation error

I am correcting in my 2 previous posts.

Thanks for this correction.

[mag_therm]

2N1595 : TO39 1.6 A 50 V  tq=10 usec typ
I saw some available on surplus in USA and UK.

[desertrout]

It is obvious that if this parameter is not even informed in the datasheet, it is not a fast thyristor.

OK - thank you. I'm a new hobbyist, and your clarifications are very helpful 

[akimpowerscr]

For more informations about turn off time of scr, see page 30, 31 and 32 : 3.4.2.4 turn off time tq

https://www.infineon.com/dgdl/Infineon-AN20012_01_Bipolar_Technical_Information-ApplicationNotes-v01_00-EN.pdf?fileId=db3a304412b407950112b40ec42b126a

[desertrout]

2N1595 : TO39 1.6 A 50 V  tq=10 usec typ
I saw some available on surplus in USA and UK.

Yeah, the 2N5060 has the same tq... apparently not fast enough

[mag_therm]

Yeah, the 2N5060 has the same tq... apparently not fast enough

I bet you will not find, ever made,  a faster small scr with published tq <= 10 usec

[akimpowerscr]

Yeah, the 2N5060 has the same tq... apparently not fast enough

I bet you will not find, ever made,  a faster small scr with published tq <= 10 usec

You loose: BR103 from Siemens has a published tq<6µs

Datasheet in one of my previous posts.

[desertrout]

Yeah, the 2N5060 has the same tq... apparently not fast enough

I bet you will not find, ever made,  a faster small scr with published tq <= 10 usec

And the OE part E0122YA has a tq MAX of 10us

[David Hess]

It is a self-oscillating flyback converter that works on the same principle as some Philips oscilloscope power supplies.

Tektronix used it also and not just for power conversion.  This might look familiar:

https://en.wikipedia.org/wiki/Blocking_oscillator

[desertrout]

For more informations about turn off time of scr, see page 30, 31 and 32 : 3.4.2.4 turn off time tq

https://www.infineon.com/dgdl/Infineon-AN20012_01_Bipolar_Technical_Information-ApplicationNotes-v01_00-EN.pdf?fileId=db3a304412b407950112b40ec42b126a

Thank you. I also was looking at this document, which suggests the turn off capability of an SCR can be improved in various ways, including adding a heatsink (not possibly in this case), or a low R_GK resistor (see page 19). Are any of these possibilities in the circuit I'm working with?