|
Hi all,
Is it parameterized in LTspice (or even in SPICE)?
I know it's not specified in datasheets.
I'm thinking of adding a series combo of diode + zener to the b-e junction but I would prefer a "native" solution.
Regards
|
Datasheets normally specify an abs max for Veb, typically 6V though I have seen 20V parts.
To offer zener-type parameters would presume that the parts can be operated in this region and meet their lifetime specs, they can't.
I have seen the b-e junction used as an especially noisy zener, i.e. a noise source, but the circuit AC-couples the noise and Vzener can be anywhere from 6V to 15V.
Hope this helps
regds
Bonkers
|
+1 agree. Using a BJT this way is going off-piste, so there aren't any guarantees or specs. It *was* the traditional way to make white noise until the mid to late 1970's, but that doesn't mean it's not transistor abuse.
Datasheets normally specify an abs max for Veb, typically 6V though I have seen 20V parts.
To offer zener-type parameters would presume that the parts can be operated in this region and meet their lifetime specs, they can't.
I have seen the b-e junction used as an especially noisy zener, i.e. a noise source, but the circuit AC-couples the noise and Vzener can be anywhere from 6V to 15V.
Hope this helps
regds
Bonkers
|
See file Draft506_.zip in TEMP folder.
|
For some applications the junction degradation doesn't matter.
Check what happens if you replace the diodes in a "perfect"
rectifier with a transistor.
Le 18/11/2024 à 13:18, Bonkers via
groups.io a écrit :
переключити цитоване повідомлення
Показати цитований текст
Datasheets normally specify an abs max for Veb, typically 6V
though I have seen 20V parts.
To offer zener-type parameters would presume that the parts
can be operated in this region and meet their lifetime specs,
they can't.
I have seen the b-e junction used as an especially noisy
zener, i.e. a noise source, but the circuit AC-couples the noise
and Vzener can be anywhere from 6V to 15V.
Hope this helps
regds
Bonkers
|
|
|
On Mon, Nov 18, 2024 at 06:25 AM, Jerry Lee Marcel wrote:
Is it parameterized in LTspice (or even in SPICE)?
You can see in LTspice's Help that there is a model parameter for that:
- BVbe Base-emitter breakdown voltage (Default=)Infinite
- Ibvbe Current at base-emitter breakdown voltage (Default=)1e-10
So it's there, but any given BJT SPICE model may or may not include it.
It was my understanding that many SPICE models did not include it. By omitting it from the .MODEL statement, the breakdown voltage in SPICE would be infinite. I think most models did that because it was a breakdown effect that was best to be avoided.
Looking in LTspice's standard.bjt model library, none of the models there have Bvbe. I think that is typical of what you will find, but some models have it. (This question has come up before in this group.)
I know it's not specified in datasheets.
Actually, almost all datasheets do (somewhat indirectly) include it. Look under Absolute Maximum ratings and you can find a maximum allowed B-E voltage. If the number there is, say, 5 V, they're telling you that the minimum Bvbe is something greater than 5 V. They usually won't give you a typical value, but I think it is not very far off from the minimum.
There are conflicting views about this breakdown. In fact there are some "hobbyist" circuits that exploit it, often to make an oscillator, or a low-voltage Zener diode. Some transistors can withstand it indefinitely, apparently without damage or degradation. But I would not assume that they all can, even if you limit the current.
Andy
|
I am also trying to remember if the SPICE model parameter BVbe was from (Berkeley) SPICE, or if it's something added uniquely to LTspice. Can't remember anymore.
If it wasn't part of Berkeley SPICE, then I think nobody else included B-E breakdown in SPICE models, until Mike Engelhardt added it. But it is a feature, waiting for you to use it.
Andy
|
"For some applications the junction degradation doesn't matter."
All applications will have some limit on the degree of parameter shift that can be accommodated.
Guaranteeing a maximum parameter shift, due to Vbe breakdown, needs a lot of testing - and most/all BJT manufacturers don't bother.
|
Here is an old note from Mike Engelhardt, about BVbe and treating it as a "failure" mode:
Andy
|
It is a common "nuisance" to have to add a vbe antiparallel diode, if there is a risk of the base being driven negative, above Veb.
It's often picked-up in schematic review, for designs that are intended to be robust.
There's little downside, apart from parts cost, most circuits work the same, better even - e.g. for circuits like zero-crossing detectors (10Meg + 1nF from AC mains into the base of a transistor. )
|
FYI - I think Jerry Lee Marcel was wondering if he needed to add a diode + zener to his simulations, to mimic what happens in real hardware.
Use the undocumented "AKO" syntax to add BVbe to any model that does not have it already:
.model 2N2222mod AKO AKO: 2N2222 BVbe=6.5 Ibvbe=1m
Andy
|
AKO ! - what a fantastic feature to find, I never knew it existed.
(for the tldr folks, it allows you to change a model parameter without editing/saving a new model)
As for the OP's purpose, to mimic real hardware, outside of SOA, would need a complex integral of reverse current vs time - a bit like characterising a memristor. That integral would then modulate other DC parameters, like gain. Short heavy overloads are not the same as long gentle overloads, see the paper I referred to earlier - the integral would need to be polynomial in "i_reverse", at least. Real hardware under these circumstances is a complex problem, if you include the known degradation. It is outside the scope of standard BJT models.
I'd make the hardware robust, with the Vbe diode, and simulate from there - for any sort of production-intent design.
|
On 18/11/2024 12:25, Jerry Lee Marcel
via groups.io wrote:
Is it
parameterized in LTspice (or even in SPICE)?
I know it's not specified in datasheets.
I'm thinking of adding a series combo of diode + zener to the b-e
junction but I would prefer a "native" solution.
Emitter-base breakdown is modelled in the enhanced BJT SPICE model
by the BVbe, Ibvbe and Nbvbe parameters. By default, Bvbe=∞.
--
Regards,
Tony
|
On 18/11/2024 15:02, Andy I via
groups.io wrote:
FYI - I think Jerry Lee Marcel was wondering if he needed to
add a diode + zener to his simulations, to mimic what happens in
real hardware.
Use the undocumented
"AKO" syntax to add BVbe to any model that does not have it
already:
.model 2N2222mod AKO 2N2222 BVbe=6.5 Ibvbe=1m
The "AKO" syntax requires the colon:
.model 2N2222mod AKO: 2N2222 BVbe=6.5 Ibvbe=1m
--
Regards,
Tony
|
On Mon, Nov 18, 2024 at 09:38 AM, Tony Casey wrote:
The "AKO" syntax requires the colon:
Yes - and this person requires better glasses. :-)
Andy
|
don't beat yourself up - publicising the little-known AKO function is the real value here.
A typo in the "serving suggestion" is no problem, since you link to the actual required protocol.
Credit to Mr. Casey for his immediate correction.
|
My aim is not to protect anything. I just want to see the effects
of the b-e junction zenering when used in a perfect rectifier.
I have uploaded file "peak detector discharge". The rightmost
version simulates well but doesn't in real life because b-e
junctions zenering discharges capacitor.
I would like the sim to show this behaviour.
Le 18/11/2024 à 14:48, Bonkers via
groups.io a écrit :
переключити цитоване повідомлення
Показати цитований текст
It is a common "nuisance" to have to add a vbe antiparallel
diode, if there is a risk of the base being driven negative,
above Veb.
It's often picked-up in schematic review, for designs that
are intended to be robust.
There's little downside, apart from parts cost, most circuits
work the same, better even - e.g. for circuits like
zero-crossing detectors (10Meg + 1nF from AC mains into the
base of a transistor. )
|
Good update, thanks :)
It's never easy to decode the actual purpose and intention of the questions here, the upload helps a lot.
Adding a separate Vbe zener would make the 3rd circuit simulate closer to reality. This extra part keeps it easy to understand.
Or maybe use the AKO function mentioned earlier to force a non-infinite Vbe breakdown, maybe "purer" but less direct.
To make a real circuit for production, which is my main expertise - you have a few options:
1) lower the voltage rails to avoid Vbe breakdown altogether. The RRO voltage range of your opamp needs to be constrained to Vbe_max. like Voh-Vol < 6V .
2) find BJT's with a high Veb breakdown voltage, there are better parts (usually PNP) that reach 20V
3) use nFETs not BJT's - it's a "perfect rectifier" so the Vgs_th parameter should be absorbed within the loop feedback. 20Vgs or 30Vgs are available.
All the above remedies don't fix the fact that the opamp FB loop is open for input voltages below the peak voltage stored on the output capacitor.
The opamp wil "rail" most of the time and this invokes de-saturation time in the opamp.
I'll have a closer look for you, a second NPN should fix all this...
|
have a look at peak detector 2 or 3.asm - recently uploaded
this adds another transistor to keep Veb below 0.6V.
It also gives a load for the op-amp so it might de-saturate quicker.
To really keep the op-amp out of saturation will need further work, and more parts.
For a small inaccuracy, this could just be a resistor, a gain-limit on the opamp.
let me know if it helps
regds
|
Bonkers
Tony Casey