Reply by Anthony William Sloman●April 1, 20232023-04-01
On Sunday, April 2, 2023 at 1:25:09 AM UTC+11, John Larkin wrote:
> On Fri, 31 Mar 2023 23:43:22 -0700 (PDT), whit3rd <whi...@gmail.com>
> wrote:
> >On Thursday, March 30, 2023 at 2:20:30?PM UTC-7, John Larkin wrote:
> >> On Thu, 30 Mar 2023 22:11:56 +0100, piglet <erichp...@hotmail.com>
> >> wrote:
> >
> >> >Do not try the C-B junction to compensate the E-B breakdown trick on a
> >> >2N3904. It has too much gain. As I wrote up thread on 3 March the
> >> >reverse gain needs to be well below unity otherwise transistor action
> >> >will create interesting effects. If you really want to have a forward
> >> >junction compensate the reverse junction either use a device with low
> >> >reverse beta (as mentioned in the app note) or use a second 2N3904 to
> >> >compensate the first.
> >
> >> Buy a bandgap reference. The transistor zener thing is risky.
> >
> >Another non-risky option is uA723 or LM723 or MC1723; the innards is basically a few
> >general purpose transistors, plus a temperature compensated zener.
> >That was the old-school part, with LOTS of second sources.
>
> 723 has had an amazing lifetime. It's still being designed into power supplies.
A legacy part, used by legacy designers, in much the same way that the NE555 still has it's fans. You can do better, but often you don't have to bother.
--
Bill Sloman, Sydney
Reply by John Larkin●April 1, 20232023-04-01
On Fri, 31 Mar 2023 23:43:22 -0700 (PDT), whit3rd <whit3rd@gmail.com>
wrote:
>On Thursday, March 30, 2023 at 2:20:30?PM UTC-7, John Larkin wrote:
>> On Thu, 30 Mar 2023 22:11:56 +0100, piglet <erichp...@hotmail.com>
>> wrote:
>
>> >Do not try the C-B junction to compensate the E-B breakdown trick on a
>> >2N3904. It has too much gain. As I wrote up thread on 3 March the
>> >reverse gain needs to be well below unity otherwise transistor action
>> >will create interesting effects. If you really want to have a forward
>> >junction compensate the reverse junction either use a device with low
>> >reverse beta (as mentioned in the app note) or use a second 2N3904 to
>> >compensate the first.
>
>> Buy a bandgap reference. The transistor zener thing is risky.
>
>Another non-risky option is uA723 or LM723 or MC1723; the innards is basically a few
>general purpose transistors, plus a temperature compensated zener.
>That was the old-school part, with LOTS of second sources.
723 has had an amazing lifetime. It's still being designed into power
supplies.
Reply by whit3rd●April 1, 20232023-04-01
On Thursday, March 30, 2023 at 2:20:30 PM UTC-7, John Larkin wrote:
> On Thu, 30 Mar 2023 22:11:56 +0100, piglet <erichp...@hotmail.com>
> wrote:
> >Do not try the C-B junction to compensate the E-B breakdown trick on a
> >2N3904. It has too much gain. As I wrote up thread on 3 March the
> >reverse gain needs to be well below unity otherwise transistor action
> >will create interesting effects. If you really want to have a forward
> >junction compensate the reverse junction either use a device with low
> >reverse beta (as mentioned in the app note) or use a second 2N3904 to
> >compensate the first.
> Buy a bandgap reference. The transistor zener thing is risky.
Another non-risky option is uA723 or LM723 or MC1723; the innards is basically a few
general purpose transistors, plus a temperature compensated zener.
That was the old-school part, with LOTS of second sources.
Reply by Anthony William Sloman●March 31, 20232023-03-31
On Friday, March 31, 2023 at 8:20:30 AM UTC+11, John Larkin wrote:
> On Thu, 30 Mar 2023 22:11:56 +0100, piglet <erichp...@hotmail.com>
> wrote:
> >On 30/03/2023 4:20 pm, Anthony William Sloman wrote:
> >> On Wednesday, March 29, 2023 at 11:46:17?PM UTC+11, neo5...@gmail.com wrote:
> >>> On Thursday, March 2, 2023 at 12:07:47?PM UTC-5, whit3rd wrote:
> >>
> >> <snip>
> >>
> >>>> It'll probably work. That doesn't mean it's good design practice to depend on it.
> >>> That is my gut feeling too. I was looking for better corroboration of if it's a good idea to design into something that can go months before anyone comes by to service the flaky analog input section.
> >>
> >> There's nothing flaky about the approach
> >>
> >> National Application note AN-74
> >>
> >> https://www.ti.com/lit/an/snoa654a/snoa654a.pdf?ts=1677789876245&ref_url=https%253A%252F%252Fwww.google.com%252F
> >>
> >> specifies the National Semiconductor process 21 transistor 2N2369. The 2N3904 is process 23, which is much the same but on a slightly bigger chip with more interdigitation. Both are gold-doped.
> >>
> >> It's at Figure 37. It's not going to require anybody to service it.
> >>
> >> It's an unconventional application of the part, so you can't sue anybody if stops working, but it won't. It's essentially the same idea as the 1N829 precision voltage reference diode, but it offers a close to zero temperature coefficient at 1.5mA rather than the 7.5mA you need with the 1N829.
> >>
> >
> >Do not try the C-B junction to compensate the E-B breakdown trick on a
> >2N3904. It has too much gain. As I wrote up thread on 3 March the
> >reverse gain needs to be well below unity otherwise transistor action
> >will create interesting effects. If you really want to have a forward
> >junction compensate the reverse junction either use a device with low
> >reverse beta (as mentioned in the app note) or use a second 2N3904 to
> >compensate the first.
It probably won't be difficult to select for a reverse beta of less that 0.25 (as specified in AN74). Using the base diode in a second 2N3904 as the compensation diode for a reverse biased base-emitted junction in another 2N3904 avoids the problem, but the two diodes aren't guaranteed to be at the same temperature, which was the whole point of the trick
> Buy a bandgap reference. The transistor zener thing is risky.
It's using a part in a way that wasn't intended and exploiting a a feature that isn't specified by the manufacturer, but John Larkin does that all the time and boasts about it. 2N3904s are cheaper than well-specified reference voltage sources, and the performance should be better than a bandgap reference.
--
Bill Sloman Sydney
Reply by John Larkin●March 30, 20232023-03-30
On Thu, 30 Mar 2023 22:11:56 +0100, piglet <erichpwagner@hotmail.com>
wrote:
>On 30/03/2023 4:20 pm, Anthony William Sloman wrote:
>> On Wednesday, March 29, 2023 at 11:46:17?PM UTC+11, neo5...@gmail.com wrote:
>>> On Thursday, March 2, 2023 at 12:07:47?PM UTC-5, whit3rd wrote:
>>
>> <snip>
>>
>>>> It'll probably work. That doesn't mean it's good design practice to depend on it.
>>> That is my gut feeling too. I was looking for better corroboration of if it's a good idea to design into something that can go months before anyone comes by to service the flaky analog input section.
>>
>> There's nothing flaky about the approach
>>
>> National Application note AN-74
>>
>> https://www.ti.com/lit/an/snoa654a/snoa654a.pdf?ts=1677789876245&ref_url=https%253A%252F%252Fwww.google.com%252F
>>
>> specifies the National Semiconductor process 21 transistor 2N2369. The 2N3904 is process 23, which is much the same but on a slightly bigger chip with more interdigitation. Both are gold-doped.
>>
>> It's at Figure 37. It's not going to require anybody to service it.
>>
>> It's an unconventional application of the part, so you can't sue anybody if stops working, but it won't. It's essentially the same idea as the 1N829 precision voltage reference diode, but it offers a close to zero temperature coefficient at 1.5mA rather than the 7.5mA you need with the 1N829.
>>
>
>Do not try the C-B junction to compensate the E-B breakdown trick on a
>2N3904. It has too much gain. As I wrote up thread on 3 March the
>reverse gain needs to be well below unity otherwise transistor action
>will create interesting effects. If you really want to have a forward
>junction compensate the reverse junction either use a device with low
>reverse beta (as mentioned in the app note) or use a second 2N3904 to
>compensate the first.
>
>piglet
Buy a bandgap reference. The transistor zener thing is risky.
Reply by piglet●March 30, 20232023-03-30
On 30/03/2023 4:20 pm, Anthony William Sloman wrote:
> On Wednesday, March 29, 2023 at 11:46:17 PM UTC+11, neo5...@gmail.com wrote:
>> On Thursday, March 2, 2023 at 12:07:47 PM UTC-5, whit3rd wrote:
>
> <snip>
>
>>> It'll probably work. That doesn't mean it's good design practice to depend on it.
>> That is my gut feeling too. I was looking for better corroboration of if it's a good idea to design into something that can go months before anyone comes by to service the flaky analog input section.
>
> There's nothing flaky about the approach
>
> National Application note AN-74
>
> https://www.ti.com/lit/an/snoa654a/snoa654a.pdf?ts=1677789876245&ref_url=https%253A%252F%252Fwww.google.com%252F
>
> specifies the National Semiconductor process 21 transistor 2N2369. The 2N3904 is process 23, which is much the same but on a slightly bigger chip with more interdigitation. Both are gold-doped.
>
> It's at Figure 37. It's not going to require anybody to service it.
>
> It's an unconventional application of the part, so you can't sue anybody if stops working, but it won't. It's essentially the same idea as the 1N829 precision voltage reference diode, but it offers a close to zero temperature coefficient at 1.5mA rather than the 7.5mA you need with the 1N829.
>
Do not try the C-B junction to compensate the E-B breakdown trick on a
2N3904. It has too much gain. As I wrote up thread on 3 March the
reverse gain needs to be well below unity otherwise transistor action
will create interesting effects. If you really want to have a forward
junction compensate the reverse junction either use a device with low
reverse beta (as mentioned in the app note) or use a second 2N3904 to
compensate the first.
piglet
Reply by Anthony William Sloman●March 30, 20232023-03-30
On Wednesday, March 29, 2023 at 11:46:17 PM UTC+11, neo5...@gmail.com wrote:
> On Thursday, March 2, 2023 at 12:07:47 PM UTC-5, whit3rd wrote:
<snip>
> > It'll probably work. That doesn't mean it's good design practice to depend on it.
> That is my gut feeling too. I was looking for better corroboration of if it's a good idea to design into something that can go months before anyone comes by to service the flaky analog input section.
There's nothing flaky about the approach
National Application note AN-74
https://www.ti.com/lit/an/snoa654a/snoa654a.pdf?ts=1677789876245&ref_url=https%253A%252F%252Fwww.google.com%252F
specifies the National Semiconductor process 21 transistor 2N2369. The 2N3904 is process 23, which is much the same but on a slightly bigger chip with more interdigitation. Both are gold-doped.
It's at Figure 37. It's not going to require anybody to service it.
It's an unconventional application of the part, so you can't sue anybody if stops working, but it won't. It's essentially the same idea as the 1N829 precision voltage reference diode, but it offers a close to zero temperature coefficient at 1.5mA rather than the 7.5mA you need with the 1N829.
--
Bill Sloman, Sydney
Reply by neo5...@gmail.com●March 29, 20232023-03-29
On Thursday, March 2, 2023 at 12:07:47 PM UTC-5, whit3rd wrote:
>>Snip>>
> It'll probably work. That doesn't mean it's good design practice to depend on it.
That is my gut feeling too. I was looking for better corroboration of if it's a good idea to design into something that can go months before anyone comes by to service the flaky analog input section.
Don
Reply by sibolis jahanamnaburju●March 8, 20232023-03-08
Reply by Anthony William Sloman●March 4, 20232023-03-04
On Saturday, March 4, 2023 at 7:36:05 AM UTC+11, piglet wrote:
> On 01/03/2023 14:54, Anthony William Sloman wrote:
> > There's a grab-bag National Semiconductor application note for the LM339 that tells much the same story on page 30.
> >
> > https://www.ti.com/lit/an/snoa654a/snoa654a.pdf
> >
> > "Experimental data has shown that any of National's process 21 transistors
> > which have been selected for low reverse beta (βR <.25) can be used quite satisfactorily as a zero T.C.
> > Zener. When connected as shown in Figure 37, the T.C. of the base-emitter Zener voltage is exactly
> > cancelled by the T.C. of the forward biased base-collector junction if biased at 1.5 mA. The diode can be
> > properly biased from any supply by adjusting RS to set lq equal to 1.5 mA."
> >
> > Process 21 was a gold-doped NPN fast switch - 2N2369 is the JEDEC nunber.
> >
> Using the C-B diode to temp compensate the E-B breakdown looks
> superficially neat but the devil is in the detail - they specify below
> unity reverse beta and that is vital.
It isn't being recommended for every part. The National Semiconductor application note was quite specific about what would work. Others have recommended the 2N3904 part which is process 23, a slight;ly bigger chip with a bit more interdigitation.
> Otherwise once E-B breakdown
> occurs transistor action kicks in and you have a negistor or 4-layer
> diode kind of effect (imagine an SCR with zener gate to anode),
> conceptually like a uniploar low voltage diac. It is a hobbyists trick
> low parts count oscillator.
You probably couldn't see that in LTSpice - all the transistor models are Gummel -Poon which doesn't model inverted gain all that accurately.
LTSpice will run the more accurate VBIC model, but the parameters are commercial in confidence, and hobbyists can't get at them'
--
Bill Sloman, Sydney