Second source horror stories

Way down the 2SK3018 thread,

On 8/6/19 5:24 AM, Winfield Hill wrote:> Robert Baer wrote... >> WARNING. Be advised that every manufacturer of a given part (the >> 2SK3018 in this case) runs their fab differently, and you may find that >> those particular specs may be WIDELY different - enough to make the part >> virtually useless. > > Or not. Of course, one must fully evaluate a > second-source manufacturer's parts, before > committing it to manufacturing.

Too right--and worth a separate thread of its own, I think.

My favourite horrible example is the TLV431. The Diodes Inc version is super well behaved and stable, as shown in the stability plot on P.7 of .

Ooooonnn the other hand, see ON Semi's (in)stability plot at Figure 18 of

and TI's not-quite-as-bad-but-still-pretty-awful TI version, Figure 18 of .

The Diodes part is stable for all load capacitances and all output voltages, provided the anode current is over a milliamp or so.

The other two have horribly complicated behaviour, with the ON Semi part taking the biscuit with four separate regions of instability.

What are your horror stories?

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs
Loading thread data ...

Wasn't TI the TLV431 originator? I have a 1996 datasheet.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

I think you win the campfire horror story contest with that one.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

It is just about possible to pick shunt capacitor values that ensure stability of all makes of TLV431 (and Tl431).

John

Reply to
jrwalliker

I think so--it's sort of the low power version of their hoary old TL431, the one JT used to use for everything. ;)

The TI one has a nasty stripe of instability right in the range of capacitors you might want to use.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Sure, assuming that your circuit is okay with 10 uF.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

180 uF polymer, just to be safe.
--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Having some 22 or 47 uF ceramics in stock is handy lately. Some switchers want 100 uF or even more.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

I often use one of those in parallel with, like, 150 uF AlPo with some acceptable minimum ESR. You get the lead-lag effect that way.

Since my latest laser board design uses 2.14 MHz switchers without any spurs being detectable above shot noise even at a 10-Hz IF bandwidth, I'll probably start using those a lot more. (The spec had a noise limit from 30 Hz to 2 MHz, so 2.14 was a good number just in case.)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

As an aside, most adjustable shunt regulators, including the TLV431, work like NPN V_BE multipliers--the cathode current is whatever it takes to make the FB-anode voltage 1.23 V. Thus to make a 1.23V reference, you attach FB to cathode.

A couple, the LM4041-ADJ and the LM385, are like PNPs--they keep the FB-cathode voltage constant. Having both helps a lot with stuff like floating current limiters.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

If you are concerned about frequency-domain noise (but not real noise) the spread-spectrum switchers, like TPS54302, are great.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

My sister told me about her terrible trip to Hawaii, Oh wait, different kind of second source!

Mikek

Reply to
amdx

The BJT-based TL431 does seem pretty consistent, even when bought in bags of 50 no-name from dubious eBay supplier.

The issue with the CMOS ones probably consequence of the cheaper ones being fabbed with the well-regarded "eh, whatever"-micron process to exacting tolerances of, ah, y'know. whatever they are.

but if you need 1.2 volts reference instead of 2.5 and lower cathode current than the 1mA the hoary TL341 likes to have then that's where you're at I guess

Reply to
bitrex

Is the spread-spectrum optional? Because it's super useful to be able to concentrate all the spur power in one spot so I can see it. Sometimes I do things that require knowing the amplitude statistics of the noise pretty accurately, and being able to estimate the switcher's contribution is super helpful.

There are other ways to do that, of course, e.g. trigger a scope off the switcher and average like mad.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

For fairness, the original TI TLV271 is a great deal quieter than the Diodes version, especially in the 1/f region.

Then there's the original LM324 with the single bias network for four sections (fixed in the Fairchild LM324 and the NS LM324A but not the Motorola one). Having the bias collapse for all sections when one rails is not nice, but the independently-biased ones are fine, for very price-sensitive values of 'fine'. ;)

In my business, the Osram BPW34 photodiode is several times faster (as well as quieter) than the Vishay one. You wouldn't learn that from their datasheets.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

given part (the

sion is

of

en one

r

TTL had the same problems. Some repair specs required one to use a particular manufacturers 74(LS)XXX part (TI in most cases) due to 2nd source irregularities from the original specs.

And I too have seen op-amp circuits that were very fussy about the manufacturer. TL081 rings a bell for fussy - had to try several (8 to

10) from 2nd source to get one where the circuit would work properly.

John :-#)#

--
(Please post followups or tech inquiries to the USENET newsgroup) 
                      John's Jukes Ltd. 
 Click to see the full signature
Reply to
John Robertson

Curiuos - how do you measure the noise in a photodiode? This would seem to be a problem with the op amp and feedback resistor as part of the picture.

Reply to
Steve Wilson

It's only a SOT-23/6, so it doesn't have a lot of options. Fixed freq, always ss.

formatting link

But it's very nice.

Reply to
John Larkin

The original National data sheet had a footnote in 2-point type that warned against pulling any input below ground. It should have been more obvious.

Reply to
John Larkin

There's a bunch of series resistance in the very thin epi layer, like

50-200 ohms. Osram seems to dope theirs more, which helps. With a really good bootstrap (see e.g. , the additive Johnson noise dominates the dark noise. That's the main reason the QL01B is quieter than the QL01A.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.