Output protection

[jacco vermeulen]

Just budging in to say this is the best thread by far, with all of the right names contributing.
The only thread i am forcing myself to shut-up, read and enjoy.

Just one question:
What speaks against placing relays in the PS rails for drivers and output devices, with the front end on separate rails.
I admit i am a relayholic.
The ones i have been using for a long time do not oxydise, have very low resistance through multiple contacts, take 160 amps peak with 2 in parallel, disconnect in only a few milliseconds.

Jan: i did not have the privilige of observing the home set of Jean Hiraga in France 20 years ago, modest and calm are the words to describe him though.

[jan.didden]

Originally posted by jacco vermeulen
[snip]Just one question:
What speaks against placing relays in the PS rails for drivers and output devices, with the front end on separate rails.
I admit i am a relayholic.
The ones i have been using for a long time do not oxydise, have very low resistance through multiple contacts, take 160 amps peak with 2 in parallel, disconnect in only a few milliseconds.

Jan: i did not have the privilige of observing the home set of Jean Hiraga in France 20 years ago, modest and calm are the words to describe him though.

Jacco,

Which relays are those? I have been using the amplimo silver/wolfram contact relays for my protection circuits. 100A break capacity, don't know the switching speed. Used in the speaker line.
I think that using a couple of small signal transistors are adequate to protect the amp, even if relays are cheap and fast, they are mech devices and most designers are somewhat suspicious about them. It is also a philosophical thing, I can fully understand Lars' decision to use a relay for protection, it's just that I prefer a solid state circuit and accept the possible short audible action over the short complete interruption by a relay. I don't think one is "wrong" and the other "right", just different ideas.

On Hiraga: I once visited him at his offices in a Paris suburb in the building of his publishing company. Since the heydays of L'Audiophile, the successor is now only one of a range of magazines. So his audition room was in that building, this was 1998 or so. This was an audiophile's dream. Wall-to-wall horn speakers, all the famous Voice of the Theather Altecs were there, plus a lot I had never seen. Handwound xover coils looking like fire hose reels. And then these tiny, innocent-looking 1-tube amplifiers. But the sound, man, the sound! There was no need for comments. You just listenend. If I would use one word, it is AUTHORATIVE.

One thing I brought back was a list of records he uses for evaluating equipment. One artist was Mark Curry. Now, there is at this time also a singer called Mark Curry, but he is colored, and the one I mean is caucasian. He only made one CD, early 90'ies, and a couple of singles. He tried again in 1996 or so with the Ten pin trio, but that flopped. He's a kind of modern-day Donovan. The CD is called It's only time. I have scavenged the world and have everything from him now, last one was a demo EP I found somewhere in Australia. demo only, not for sale. Yeah.
This is very cleanly, directly and unprocessed recordings, a pleasure to listen.
If you're interested, mail me off-line.

Jan Didden

[mikeks]

Originally posted by janneman

But, Mike, you wouldn't design the SOA protection against prolonged shorts, would you?
Jan Didden

No...as you noted, this is the province of a thermal latch...............

.....but it may be wise to design for the SOA at the steady state operating case temperature Tc of your transistor...

For instance, the nominal SOA of 200W for the MJL1302A/3281A is refered to Tc=25deg. C...

....while the true Tc during normal operation (viz: with chosen heatsink) is more like Tc~60deg. C.....equivalent roughly to a SOA of 140W....

So...it would be wise to have your protection locus approximate a constant 140W parabola, (linear-linear scale assumed), except of course, where the secondary breakdown parabola intrudes on your nominal ratings...

[mikeks]

Originally posted by Lars Clausen
Isn't 'slope protection' just a fancy word for current limiting?

No...

Originally posted by Lars Clausen

I have simplified the sch to show you what happens in that situation.

Your scheme is deeply flawed for a number of reasons...:

1) It is infact a single-slope scheme with a single breakpoint....very inefficient!

2) The location of the locus in the SOA varies with the rails....Bad!

3) You're using one transistor to detect both polarities....which is a false economy of the most profound variety, as this will result in an inaccurate locus for both polarities at best, and even worse, may result in unnecesary invocation of protection....

Originally posted by Lars Clausen
It monitors the SOA curve......

...not even close....

[jacco vermeulen]

The relays i was referring to are the former SDS S and ST types.
The S model withstands 400 amps for 1 mS on 4 parallel contacts.
Golden contacts, gasfilled and sealed.
With 4 contacts in parallel 20 amps continuous.
Turn-off within 5 mS.
Contact resistance below 10 mOhms.
The ST does even better but has only 2 parallel contacts.

Personally i like a bit of both, and have done so.
I am going to build an amplifier with a number of output devices i have been cherishing a great number of years.
I can not afford to use them for ChateauBriand as they are irreplaceable.
Every protection option mentioned on this thread will be what i intend to use, even Michael's triple slope, i am very gratefull he sent me the article(ordered the 4 pound softcopy too, jan)
I do see the downside of relays on the output though,i wondered if i could move the relays from the output to the rails.


In my goldy locks years my veggie girlfriend and i went long weekends to Paris 2, 3 times a year.
She desired Beaubourg.
Me, La Maison du L'Audiophile, long talks with Monsieur Walther, and café at the bar/café on the corner of La Rue de la Bastille afterwards.
Every year i attended the listening sessions in the backroom.
On one ocassion i had the fortune to be able to write myself on a list for an evening of demonstrations and talk, Jean Hiraga being one of the speakers.

Here is a scrible of a famous French audio person, with a nice sense of humor, on one of my 80s magazines.

[jan.didden]

Originally posted by mikeks


No...as you noted, this is the province of a thermal latch...............

.....but it may be wise to design for the SOA at the steady state operating case temperature Tc of your transistor...

For instance, the nominal SOA of 200W for the MJL1302A/3281A is refered to Tc=25deg. C...

....while the true Tc during normal operation (viz: with chosen heatsink) is more like Tc~60deg. C.....equivalent roughly to a SOA of 140W....

So...it would be wise to have your protection locus approximate a constant 140W parabola, (linear-linear scale assumed), except of course, where the secondary breakdown parabola intrudes on your nominal ratings...

Check. Got you.

Jan Didden

[jan.didden]

Originally posted by jacco vermeulen
[snip]In my goldy locks years my veggie girlfriend and i went long weekends to Paris 2, 3 times a year.
She desired Beaubourg.
Me, La Maison du L'Audiophile, long talks with Monsieur Walther, and café at the bar/café on the corner of La Rue de la Bastille afterwards.
Every year i attended the listening sessions in the backroom.
On one ocassion i had the fortune to be able to write myself on a list for an evening of demonstrations and talk, Jean Hiraga being one of the speakers.

Here is a scrible of a famous French audio person, with a nice sense of humor, on one of my 80s magazines.

Nice memories.....

Jan Didden

[mikeks]

Originally posted by jacco vermeulen
...even Michael's triple slope, i am very gratefull he sent me the article(ordered the 4 pound softcopy too, jan).........

I would suggest you discard the copy you ordered, as it contains a vast number of editorial errors and ommisions...

..which is one reason why i've been giving away the original manuscript...

[mikeks]

Originally posted by jacco vermeulen
I do see the downside of relays on the output though,i wondered if i could move the relays from the output to the rails.

http://www.diyaudio.com/forums/showt...786#post440786

[mikeks]

Originally posted by jacco vermeulen
The ones i have been using for a long time do not oxydise, have very low resistance through multiple contacts, take 160 amps peak with 2 in parallel, disconnect in only a few milliseconds.

...actually contrary to intuition, wiring relay contact in parallel does nothing to increase service life i am afraid....see pg 62 here:

http://relays.tycoelectronics.com/sc.../C0_v4bg_4.pdf

More stuff:


http://relays.tycoelectronics.com/sc...hn/relbook.stm


http://relays.tycoelectronics.com/sc...efinitions.pdf

http://relays.tycoelectronics.com/sc...lay_basics.pdf


http://relays.tycoelectronics.com/application.asp