clamp diodes

[john larkin]

Assume we have a 1-ohm 2512 current shunt resistor that we don't want
to fry. My thought is to put a pair of big diodes across it to limit
the current. The associated ADC will be maybe +-250 mV full scale, so
we don't want the diodes to conduct much current there.

A 3 amp PN power diode might work, maybe starting from 0.6v at 1 mA
and declining 60 mV per decade of current. A pair of TVS zeners, used
in their forward directions might work; we have lots of them in stock.

So, I wonder if some doping magic makes a zener diode have higher
forward drop if it's a higher voltage zener?

Easier to ask than to pull a bunch of parts from stock and test them.
I'm feeling lazy today.

A shorted bridge rectifier would give me two diode drops. The ones in
stock are gigantic.

[Phil Hobbs]

Post by john larkin
Assume we have a 1-ohm 2512 current shunt resistor that we don't want
to fry. My thought is to put a pair of big diodes across it to limit
the current. The associated ADC will be maybe +-250 mV full scale, so
we don't want the diodes to conduct much current there.
A 3 amp PN power diode might work, maybe starting from 0.6v at 1 mA
and declining 60 mV per decade of current. A pair of TVS zeners, used
in their forward directions might work; we have lots of them in stock.
So, I wonder if some doping magic makes a zener diode have higher
forward drop if it's a higher voltage zener?
Easier to ask than to pull a bunch of parts from stock and test them.
I'm feeling lazy today.
A shorted bridge rectifier would give me two diode drops. The ones in
stock are gigantic.

Depending on the accuracy required, you might want either a pair of
antiparallel PN diodes, or else a bridge with a little bit of bias applied
to keep the diodes back-biased.

Cheers

Phil Hobbs

[John Larkin]

On Wed, 28 Feb 2024 02:13:04 -0000 (UTC), Phil Hobbs

Post by Phil Hobbs

Post by john larkin
Assume we have a 1-ohm 2512 current shunt resistor that we don't want
to fry. My thought is to put a pair of big diodes across it to limit
the current. The associated ADC will be maybe +-250 mV full scale, so
we don't want the diodes to conduct much current there.
A 3 amp PN power diode might work, maybe starting from 0.6v at 1 mA
and declining 60 mV per decade of current. A pair of TVS zeners, used
in their forward directions might work; we have lots of them in stock.
So, I wonder if some doping magic makes a zener diode have higher
forward drop if it's a higher voltage zener?
Easier to ask than to pull a bunch of parts from stock and test them.
I'm feeling lazy today.
A shorted bridge rectifier would give me two diode drops. The ones in
stock are gigantic.

Depending on the accuracy required, you might want either a pair of
antiparallel PN diodes, or else a bridge with a little bit of bias applied
to keep the diodes back-biased.
Cheers
Phil Hobbs

The other option is to put a polyfuse in series with the resistor. I'd
have to test that to make sure the poly opens before the resistor
does.

[John Larkin]

Post by John Larkin
On Wed, 28 Feb 2024 02:13:04 -0000 (UTC), Phil Hobbs

Post by Phil Hobbs

Post by john larkin
Assume we have a 1-ohm 2512 current shunt resistor that we don't want
to fry. My thought is to put a pair of big diodes across it to limit
the current. The associated ADC will be maybe +-250 mV full scale, so
we don't want the diodes to conduct much current there.
A 3 amp PN power diode might work, maybe starting from 0.6v at 1 mA
and declining 60 mV per decade of current. A pair of TVS zeners, used
in their forward directions might work; we have lots of them in stock.
So, I wonder if some doping magic makes a zener diode have higher
forward drop if it's a higher voltage zener?
Easier to ask than to pull a bunch of parts from stock and test them.
I'm feeling lazy today.
A shorted bridge rectifier would give me two diode drops. The ones in
stock are gigantic.

Depending on the accuracy required, you might want either a pair of
antiparallel PN diodes, or else a bridge with a little bit of bias applied
to keep the diodes back-biased.
Cheers
Phil Hobbs

The other option is to put a polyfuse in series with the resistor. I'd
have to test that to make sure the poly opens before the resistor
does.

Here's the polyfuse version.

https://www.dropbox.com/scl/fi/iwwe4v4zylzf9j826no8f/23S948A3sh9.pdf?rlkey=7d9qg1jc8gr69gil0zxymvkyc&dl=0

Maybe I could use a wirewound resistor. A thick-film might vaporize
before the poly opens.

[Mike Monett VE3BTI]

Post by Phil Hobbs

Post by john larkin
Assume we have a 1-ohm 2512 current shunt resistor that we don't want
to fry. My thought is to put a pair of big diodes across it to limit
the current. The associated ADC will be maybe +-250 mV full scale, so
we don't want the diodes to conduct much current there.
A 3 amp PN power diode might work, maybe starting from 0.6v at 1 mA
and declining 60 mV per decade of current. A pair of TVS zeners, used
in their forward directions might work; we have lots of them in stock.
So, I wonder if some doping magic makes a zener diode have higher
forward drop if it's a higher voltage zener?
Easier to ask than to pull a bunch of parts from stock and test them.
I'm feeling lazy today.
A shorted bridge rectifier would give me two diode drops. The ones in
stock are gigantic.

Depending on the accuracy required, you might want either a pair of
antiparallel PN diodes, or else a bridge with a little bit of bias applied
to keep the diodes back-biased.
Cheers
Phil Hobbs

Put two antiparallel Schotky diodes in series across the shunt, four
altogether. That also reduces your anode-cathode capacitance and improves
high frequency response.

[piglet]

Post by Phil Hobbs

Post by Phil Hobbs

Post by john larkin
Assume we have a 1-ohm 2512 current shunt resistor that we don't want
to fry. My thought is to put a pair of big diodes across it to limit
the current. The associated ADC will be maybe +-250 mV full scale, so
we don't want the diodes to conduct much current there.
A 3 amp PN power diode might work, maybe starting from 0.6v at 1 mA
and declining 60 mV per decade of current. A pair of TVS zeners, used
in their forward directions might work; we have lots of them in stock.
So, I wonder if some doping magic makes a zener diode have higher
forward drop if it's a higher voltage zener?
Easier to ask than to pull a bunch of parts from stock and test them.
I'm feeling lazy today.
A shorted bridge rectifier would give me two diode drops. The ones in
stock are gigantic.

Depending on the accuracy required, you might want either a pair of
antiparallel PN diodes, or else a bridge with a little bit of bias

applied

Post by Phil Hobbs
to keep the diodes back-biased.
Cheers
Phil Hobbs

Put two antiparallel Schotky diodes in series across the shunt, four
altogether. That also reduces your anode-cathode capacitance and improves
high frequency response.

Or perhaps BJT wired as transdiode (E to C+B) should have low leakage at
low voltage? Two inverse parallel for both directions.

[Mike Monett VE3BTI]

Post by piglet
Or perhaps BJT wired as transdiode (E to C+B) should have low leakage at
low voltage? Two inverse parallel for both directions.
--
piglet

How would that be any different than having two forward-biased diodes in
parallel? Except the current would be limited to the maximum base current.

How would it reduce the leakage at low voltages?

[piglet]

Post by Mike Monett VE3BTI

Post by piglet
Or perhaps BJT wired as transdiode (E to C+B) should have low leakage at
low voltage? Two inverse parallel for both directions.
--
piglet

How would that be any different than having two forward-biased diodes in
parallel? Except the current would be limited to the maximum base current.
How would it reduce the leakage at low voltages?

The current is limited by maximum collector current rating not base
current.

E-B junction is far better than many PN diodes.

[Mike Monett VE3BTI]

Post by piglet

Post by Mike Monett VE3BTI

Post by piglet
Or perhaps BJT wired as transdiode (E to C+B) should have low leakage
at low voltage? Two inverse parallel for both directions.
piglet

How would that be any different than having two forward-biased diodes
in parallel? Except the current would be limited to the maximum base
current.
How would it reduce the leakage at low voltages?

The current is limited by maximum collector current rating not base
current.
E-B junction is far better than many PN diodes.

Your term "transdiode" threw me on a wild goose chase.

The term transdiode was used by Patterson in his 1984 patent:

Patterson Transdiode
US4450414A
Inventor Raymond B. Patterson, III
Current Assignee Intersil Corp

High temperature current mirror amplifier
https://patentimages.storage.googleapis.com/14/d7/e1/096a85e0cc0fba/US44504
14.pdf

It is used in Logarithmic Amplifier feedback networks:

Analog Devices Dual Matched NPN Transistor MAT12
Figure 18. Log Conformance Circuit, Page 9 of 12
https://www.farnell.com/datasheets/2178247.pdf

It is not suitable for a voltage clamp across a 25 ohm resistor.

Your description of an E-B junction as better than many PN diodes threw me
for another loop. The base is very thin and not suited for high current. I
could find no datasheet that specified the maximum base current.

What you really meant to say was a diode-connected transistor, which you
describe as E to C+B.

This conducts very little current at 250 millivolts forward voltage. For
example, a 2N2219 conducts only 240 pa, which is negligible in this
application.

For your enjoyment, I include the LTspice files. I checked to make sure
they work in IV and XII, and that there are no line wrap problems. I cannot
verify they will work in 17 since I am running Win7 32 bit and have no
intention of going to 64 bit or MS 10.

SHEET 1 1140 1108
WIRE -736 -704 -864 -704
WIRE -736 -672 -736 -704
WIRE -1008 -624 -1040 -624
WIRE -976 -624 -1008 -624
WIRE -864 -624 -864 -704
WIRE -864 -624 -896 -624
WIRE -800 -624 -864 -624
WIRE -1040 -608 -1040 -624
WIRE -736 -560 -736 -576
WIRE -1040 -512 -1040 -528
FLAG -1008 -624 Vin
FLAG -1040 -512 0
FLAG -736 -560 0
SYMBOL voltage -1040 -624 R0
WINDOW 0 49 39 VRight 2
WINDOW 3 9 56 Right 2
SYMATTR InstName V1
SYMATTR Value 1
SYMATTR Value2 AC 1
SYMATTR SpiceLine Rser=2
SYMBOL npn -800 -672 R0
WINDOW 0 63 29 Left 2
WINDOW 3 56 59 Left 2
SYMATTR InstName Q1
SYMATTR Value 2N2219A
SYMBOL res -992 -608 R270
WINDOW 0 32 56 VTop 2
WINDOW 3 0 56 VBottom 2
SYMATTR InstName R1
SYMATTR Value 500µ
TEXT -1048 -784 Left 2 ;'2N2219 Forward Voltage
TEXT -1024 -744 Left 2 !.dc V1 0 0.25 1m

[DC transfer characteristic]
{
Npanes: 3
Active Pane: 1
{
traces: 1 {303038468,0,"Ic(Q1)"}
X: ('m',0,0,0.03,0.25)
Y[0]: ('p',0,0,2e-011,2.4e-010)
Y[1]: ('_',0,1e+308,0,-1e+308)
Amps: ('p',0,0,0,0,2e-011,2.4e-010)
Log: 0 0 0
GridStyle: 1
},
{
traces: 1 {34603011,0,"Ib(Q1)"}
X: ('m',0,0,0.03,0.25)
Y[0]: ('p',0,0,2e-012,2.6e-011)
Y[1]: ('_',0,1e+308,0,-1e+308)
Amps: ('p',0,0,0,0,2e-012,2.6e-011)
Log: 0 0 0
GridStyle: 1
},
{
traces: 1 {303038466,0,"I(R1)"}
X: ('m',0,0,0.03,0.25)
Y[0]: ('p',0,0,3e-011,2.7e-010)
Y[1]: ('_',0,1e+308,0,-1e+308)
Amps: ('p',0,0,0,0,3e-011,2.7e-010)
Log: 0 0 0
GridStyle: 1
}
}

You are right about diode-connected transistors. An ordinary PN diode such
as the MURS120 conducts 9.5 ua at 250 mv, and would not be suitable in this
application.

[piglet]

Post by Mike Monett VE3BTI

Post by piglet

Post by Mike Monett VE3BTI

Post by piglet
Or perhaps BJT wired as transdiode (E to C+B) should have low leakage
at low voltage? Two inverse parallel for both directions.
piglet

How would that be any different than having two forward-biased diodes
in parallel? Except the current would be limited to the maximum base
current.
How would it reduce the leakage at low voltages?

The current is limited by maximum collector current rating not base
current.
E-B junction is far better than many PN diodes.

Your term "transdiode" threw me on a wild goose chase.
Patterson Transdiode
US4450414A
Inventor Raymond B. Patterson, III
Current Assignee Intersil Corp
High temperature current mirror amplifier
https://patentimages.storage.googleapis.com/14/d7/e1/096a85e0cc0fba/US44504
14.pdf
Analog Devices Dual Matched NPN Transistor MAT12
Figure 18. Log Conformance Circuit, Page 9 of 12
https://www.farnell.com/datasheets/2178247.pdf
It is not suitable for a voltage clamp across a 25 ohm resistor.
Your description of an E-B junction as better than many PN diodes threw me
for another loop. The base is very thin and not suited for high current. I
could find no datasheet that specified the maximum base current.
What you really meant to say was a diode-connected transistor, which you
describe as E to C+B.
This conducts very little current at 250 millivolts forward voltage. For
example, a 2N2219 conducts only 240 pa, which is negligible in this
application.
For your enjoyment, I include the LTspice files. I checked to make sure
they work in IV and XII, and that there are no line wrap problems. I cannot
verify they will work in 17 since I am running Win7 32 bit and have no
intention of going to 64 bit or MS 10.
SHEET 1 1140 1108
WIRE -736 -704 -864 -704
WIRE -736 -672 -736 -704
WIRE -1008 -624 -1040 -624
WIRE -976 -624 -1008 -624
WIRE -864 -624 -864 -704
WIRE -864 -624 -896 -624
WIRE -800 -624 -864 -624
WIRE -1040 -608 -1040 -624
WIRE -736 -560 -736 -576
WIRE -1040 -512 -1040 -528
FLAG -1008 -624 Vin
FLAG -1040 -512 0
FLAG -736 -560 0
SYMBOL voltage -1040 -624 R0
WINDOW 0 49 39 VRight 2
WINDOW 3 9 56 Right 2
SYMATTR InstName V1
SYMATTR Value 1
SYMATTR Value2 AC 1
SYMATTR SpiceLine Rser=2
SYMBOL npn -800 -672 R0
WINDOW 0 63 29 Left 2
WINDOW 3 56 59 Left 2
SYMATTR InstName Q1
SYMATTR Value 2N2219A
SYMBOL res -992 -608 R270
WINDOW 0 32 56 VTop 2
WINDOW 3 0 56 VBottom 2
SYMATTR InstName R1
SYMATTR Value 500ต
TEXT -1048 -784 Left 2 ;'2N2219 Forward Voltage
TEXT -1024 -744 Left 2 !.dc V1 0 0.25 1m
[DC transfer characteristic]
{
Npanes: 3
Active Pane: 1
{
traces: 1 {303038468,0,"Ic(Q1)"}
X: ('m',0,0,0.03,0.25)
Y[0]: ('p',0,0,2e-011,2.4e-010)
Y[1]: ('_',0,1e+308,0,-1e+308)
Amps: ('p',0,0,0,0,2e-011,2.4e-010)
Log: 0 0 0
GridStyle: 1
},
{
traces: 1 {34603011,0,"Ib(Q1)"}
X: ('m',0,0,0.03,0.25)
Y[0]: ('p',0,0,2e-012,2.6e-011)
Y[1]: ('_',0,1e+308,0,-1e+308)
Amps: ('p',0,0,0,0,2e-012,2.6e-011)
Log: 0 0 0
GridStyle: 1
},
{
traces: 1 {303038466,0,"I(R1)"}
X: ('m',0,0,0.03,0.25)
Y[0]: ('p',0,0,3e-011,2.7e-010)
Y[1]: ('_',0,1e+308,0,-1e+308)
Amps: ('p',0,0,0,0,3e-011,2.7e-010)
Log: 0 0 0
GridStyle: 1
}
}
You are right about diode-connected transistors. An ordinary PN diode such
as the MURS120 conducts 9.5 ua at 250 mv, and would not be suitable in this
application.

Sorry if I created confusion, I first heard the term “transdiode “ for the
E to CB connection back in the 1960s and assumed it was well known. There
are other possible permutations of using a bipolar transistor as a diode
each with different characteristics.

[Mike Monett VE3BTI]

Sorry if I created confusion, I first heard the term ƒ otransdiode ƒ o
for the E to CB connection back in the 1960s and assumed it was well
known. There are other possible permutations of using a bipolar
transistor as a diode each with different characteristics.
--
piglet

Nontheless, a diode-connected 2N2219 would be an ideal solution for Larkin.
240 pa is a billion times lower than 250 ma, so it is negligible.

And your post is the reason we found out.

[Phil Hobbs]

Post by Mike Monett VE3BTI

Sorry if I created confusion, I first heard the term â otransdiode â o
for the E to CB connection back in the 1960s and assumed it was well
known. There are other possible permutations of using a bipolar
transistor as a diode each with different characteristics.
--
piglet

Nontheless, a diode-connected 2N2219 would be an ideal solution for Larkin.
240 pa is a billion times lower than 250 ma, so it is negligible.
And your post is the reason we found out.

Too wimpy. Fault currents of 2A were mentioned.

I hadn’t thought about transdiode forward current being super low—that’s a
good candidate for the bag of tricks, thanks.

Of course the beta tanks at very low V_CE, so you’d basically just get the
forward conduction of the B-E diode.

Cheers

Phil Hobbs

[Mike Monett VE3BTI]

Post by Phil Hobbs

Post by Mike Monett VE3BTI

Sorry if I created confusion, I first heard the term â otransdiode â
o for the E to CB connection back in the 1960s and assumed it was well
known. There are other possible permutations of using a bipolar
transistor as a diode each with different characteristics.
--
piglet

Nontheless, a diode-connected 2N2219 would be an ideal solution for
Larkin. 240 pa is a billion times lower than 250 ma, so it is
negligible.
And your post is the reason we found out.

Too wimpy. Fault currents of 2A were mentioned.

Model is constrained by what is available in LTspice IV and XVII. You are
free to choose another NPN.

Post by Phil Hobbs
I hadn’t thought about transdiode forward current being super
low—that’s a good candidate for the bag of tricks, thanks.

Transdiode configuration is for Log Amplifiers. You are interested in diode
connection.

Post by Phil Hobbs
Of course the beta tanks at very low V_CE, so you’d basically just get
the forward conduction of the B-E diode.

The goal is minimum forward current at low voltage. Low beta helps meet
this.

Now the question is why is the B-E conduction of a transistor so different
from a P-N diode?

Post by Phil Hobbs
Cheers
Phil Hobbs

[John Larkin]

On Sun, 3 Mar 2024 20:56:52 -0000 (UTC), Mike Monett VE3BTI

Post by Mike Monett VE3BTI

Post by Phil Hobbs

Post by Mike Monett VE3BTI

Sorry if I created confusion, I first heard the term â otransdiode â
o for the E to CB connection back in the 1960s and assumed it was well
known. There are other possible permutations of using a bipolar
transistor as a diode each with different characteristics.
--
piglet

Nontheless, a diode-connected 2N2219 would be an ideal solution for
Larkin. 240 pa is a billion times lower than 250 ma, so it is
negligible.
And your post is the reason we found out.

Too wimpy. Fault currents of 2A were mentioned.

Model is constrained by what is available in LTspice IV and XVII. You are
free to choose another NPN.

Post by Phil Hobbs
I hadn’t thought about transdiode forward current being super
low—that’s a good candidate for the bag of tricks, thanks.

Transdiode configuration is for Log Amplifiers. You are interested in diode
connection.

Post by Phil Hobbs
Of course the beta tanks at very low V_CE, so you’d basically just get
the forward conduction of the B-E diode.

The goal is minimum forward current at low voltage. Low beta helps meet
this.

Any beta increases current.

Post by Mike Monett VE3BTI
Now the question is why is the B-E conduction of a transistor so different
from a P-N diode?

Reverse zener voltage, for one. That won't matter in my case, with
back-to-back clamp diodes.

[Mike Monett VE3BTI]

Now the question is why is the BC-E conduction of a transistor so
different from a P-N diode?

Something is fishy. I extended the sim to 1 volt, and the current remained
very low, well below a plain diode. I suspect the simulation for a
diode-connected NPN may be faulty, and it's time for bench testing.
Unfortunately, I'm in the process of remodeling and my lab is temporarily
shut down.

Version 4
SHEET 1 1140 1108
WIRE -736 -704 -864 -704
WIRE -736 -672 -736 -704
WIRE -1136 -624 -1168 -624
WIRE -1040 -624 -1136 -624
WIRE -864 -624 -864 -704
WIRE -864 -624 -1040 -624
WIRE -800 -624 -864 -624
WIRE -1168 -608 -1168 -624
WIRE -736 -560 -736 -576
WIRE -1168 -512 -1168 -528
WIRE -1040 -432 -1040 -624
WIRE -736 -432 -1040 -432
WIRE -736 -400 -736 -432
WIRE -736 -320 -736 -336
FLAG -1136 -624 Vin
FLAG -1168 -512 0
FLAG -736 -560 0
FLAG -736 -320 0
SYMBOL voltage -1168 -624 R0
WINDOW 0 63 53 Right 2
WINDOW 3 9 56 Right 2
SYMATTR InstName V1
SYMATTR Value 1
SYMATTR Value2 AC 1
SYMATTR SpiceLine Rser=0
SYMBOL npn -800 -672 R0
WINDOW 0 63 29 Left 2
WINDOW 3 56 59 Left 2
SYMATTR InstName Q1
SYMATTR Value 2N2219A
SYMBOL diode -752 -400 R0
SYMATTR InstName D1
SYMATTR Value MURS120
TEXT -1048 -784 Left 2 ;'2N2219 Forward Voltage
TEXT -1024 -744 Left 2 !.dc V1 0 1 1m

[DC transfer characteristic]
{
Npanes: 3
Active Pane: 1
{
traces: 1 {34603012,0,"Ic(Q1)"}
X: (' ',1,0,0.1,1)
Y[0]: ('m',0,0,0.08,0.88)
Y[1]: ('_',0,1e+308,0,-1e+308)
Amps: ('m',0,0,1,0,0.08,0.88)
Log: 0 0 0
GridStyle: 1
},
{
traces: 1 {34603011,0,"Ib(Q1)"}
X: (' ',1,0,0.1,1)
Y[0]: ('m',0,0,0.001,0.015)
Y[1]: ('_',0,1e+308,0,-1e+308)
Amps: ('m',0,0,0,0,0.001,0.015)
Log: 0 0 0
GridStyle: 1
},
{
traces: 1 {34603010,0,"I(D1)"}
X: (' ',1,0,0.1,1)
Y[0]: (' ',1,0,0.6,6.6)
Y[1]: ('_',0,1e+308,0,-1e+308)
Amps: (' ',0,0,0,0,0.6,6.6)
Log: 0 0 0
GridStyle: 1
}
}

[john larkin]

On Mon, 4 Mar 2024 03:27:35 -0000 (UTC), Mike Monett VE3BTI

Post by Mike Monett VE3BTI

Now the question is why is the BC-E conduction of a transistor so
different from a P-N diode?

Something is fishy. I extended the sim to 1 volt, and the current remained
very low, well below a plain diode. I suspect the simulation for a
diode-connected NPN may be faulty, and it's time for bench testing.
Unfortunately, I'm in the process of remodeling and my lab is temporarily
shut down.

The big diode is hogging the current.

Post by Mike Monett VE3BTI
Version 4
SHEET 1 1140 1108
WIRE -736 -704 -864 -704
WIRE -736 -672 -736 -704
WIRE -1136 -624 -1168 -624
WIRE -1040 -624 -1136 -624
WIRE -864 -624 -864 -704
WIRE -864 -624 -1040 -624
WIRE -800 -624 -864 -624
WIRE -1168 -608 -1168 -624
WIRE -736 -560 -736 -576
WIRE -1168 -512 -1168 -528
WIRE -1040 -432 -1040 -624
WIRE -736 -432 -1040 -432
WIRE -736 -400 -736 -432
WIRE -736 -320 -736 -336
FLAG -1136 -624 Vin
FLAG -1168 -512 0
FLAG -736 -560 0
FLAG -736 -320 0
SYMBOL voltage -1168 -624 R0
WINDOW 0 63 53 Right 2
WINDOW 3 9 56 Right 2
SYMATTR InstName V1
SYMATTR Value 1
SYMATTR Value2 AC 1
SYMATTR SpiceLine Rser=0
SYMBOL npn -800 -672 R0
WINDOW 0 63 29 Left 2
WINDOW 3 56 59 Left 2
SYMATTR InstName Q1
SYMATTR Value 2N2219A
SYMBOL diode -752 -400 R0
SYMATTR InstName D1
SYMATTR Value MURS120
TEXT -1048 -784 Left 2 ;'2N2219 Forward Voltage
TEXT -1024 -744 Left 2 !.dc V1 0 1 1m
[DC transfer characteristic]
{
Npanes: 3
Active Pane: 1
{
traces: 1 {34603012,0,"Ic(Q1)"}
X: (' ',1,0,0.1,1)
Y[0]: ('m',0,0,0.08,0.88)
Y[1]: ('_',0,1e+308,0,-1e+308)
Amps: ('m',0,0,1,0,0.08,0.88)
Log: 0 0 0
GridStyle: 1
},
{
traces: 1 {34603011,0,"Ib(Q1)"}
X: (' ',1,0,0.1,1)
Y[0]: ('m',0,0,0.001,0.015)
Y[1]: ('_',0,1e+308,0,-1e+308)
Amps: ('m',0,0,0,0,0.001,0.015)
Log: 0 0 0
GridStyle: 1
},
{
traces: 1 {34603010,0,"I(D1)"}
X: (' ',1,0,0.1,1)
Y[0]: (' ',1,0,0.6,6.6)
Y[1]: ('_',0,1e+308,0,-1e+308)
Amps: (' ',0,0,0,0,0.6,6.6)
Log: 0 0 0
GridStyle: 1
}
}

[John Larkin]

On Sun, 3 Mar 2024 13:50:03 -0000 (UTC), Mike Monett VE3BTI

Post by Mike Monett VE3BTI

Sorry if I created confusion, I first heard the term ƒ otransdiode ƒ o
for the E to CB connection back in the 1960s and assumed it was well
known. There are other possible permutations of using a bipolar
transistor as a diode each with different characteristics.
--
piglet

Nontheless, a diode-connected 2N2219 would be an ideal solution for Larkin.
240 pa is a billion times lower than 250 ma, so it is negligible.
And your post is the reason we found out.

I need to tolerate a customer applying a brute source to my ammeter. I
will have 2 amp polyfuses in both legs, but they can conduct a lot of
current for a while, so I need hunky diodes to protect my 1 ohm shunt
resistor.

It looks like a 3 amp PN diode will work. I've tested some to be sure
that their non-ideal behavior, at both high and low currents, doesn't
mess up the theory.

I wonder if a giant wirewound resistor would be a better current
shunt, something that tolerates the current surge through the
polyfuses. I'd have to test that too.

[Lasse Langwadt]

Post by piglet

Post by Phil Hobbs

Post by Phil Hobbs

Post by john larkin
Assume we have a 1-ohm 2512 current shunt resistor that we don't want
to fry. My thought is to put a pair of big diodes across it to limit
the current. The associated ADC will be maybe +-250 mV full scale, so
we don't want the diodes to conduct much current there.
A 3 amp PN power diode might work, maybe starting from 0.6v at 1 mA
and declining 60 mV per decade of current. A pair of TVS zeners, used
in their forward directions might work; we have lots of them in stock.
So, I wonder if some doping magic makes a zener diode have higher
forward drop if it's a higher voltage zener?
Easier to ask than to pull a bunch of parts from stock and test them.
I'm feeling lazy today.
A shorted bridge rectifier would give me two diode drops. The ones in
stock are gigantic.

Depending on the accuracy required, you might want either a pair of
antiparallel PN diodes, or else a bridge with a little bit of bias

applied

Post by Phil Hobbs
to keep the diodes back-biased.
Cheers
Phil Hobbs

Put two antiparallel Schotky diodes in series across the shunt, four
altogether. That also reduces your anode-cathode capacitance and improves
high frequency response.

Or perhaps BJT wired as transdiode (E to C+B) should have low leakage at
low voltage? Two inverse parallel for both directions.

low leakage in reverse, but in forward, maybe

https://www.eevblog.com/forum/projects/forward-leakage-of-a-diode/msg1287509/#msg1287509

[john larkin]

Post by Lasse Langwadt

Post by piglet

Post by Phil Hobbs

Post by Phil Hobbs

Post by john larkin
Assume we have a 1-ohm 2512 current shunt resistor that we don't want
to fry. My thought is to put a pair of big diodes across it to limit
the current. The associated ADC will be maybe +-250 mV full scale, so
we don't want the diodes to conduct much current there.
A 3 amp PN power diode might work, maybe starting from 0.6v at 1 mA
and declining 60 mV per decade of current. A pair of TVS zeners, used
in their forward directions might work; we have lots of them in stock.
So, I wonder if some doping magic makes a zener diode have higher
forward drop if it's a higher voltage zener?
Easier to ask than to pull a bunch of parts from stock and test them.
I'm feeling lazy today.
A shorted bridge rectifier would give me two diode drops. The ones in
stock are gigantic.

Depending on the accuracy required, you might want either a pair of
antiparallel PN diodes, or else a bridge with a little bit of bias

applied

Post by Phil Hobbs
to keep the diodes back-biased.
Cheers
Phil Hobbs

Put two antiparallel Schotky diodes in series across the shunt, four
altogether. That also reduces your anode-cathode capacitance and improves
high frequency response.

Or perhaps BJT wired as transdiode (E to C+B) should have low leakage at
low voltage? Two inverse parallel for both directions.

low leakage in reverse, but in forward, maybe
https://www.eevblog.com/forum/projects/forward-leakage-of-a-diode/msg1287509/#msg1287509

I guess the question is, given a hunky diode, rated 2 amps or so, does
it behave, forward-direction, exponentially all the way down to
nanoamps?

[Lasse Langwadt]

Post by john larkin

Post by Lasse Langwadt

Post by piglet

Post by Phil Hobbs

Post by Phil Hobbs

Post by john larkin
Assume we have a 1-ohm 2512 current shunt resistor that we don't want
to fry. My thought is to put a pair of big diodes across it to limit
the current. The associated ADC will be maybe +-250 mV full scale, so
we don't want the diodes to conduct much current there.
A 3 amp PN power diode might work, maybe starting from 0.6v at 1 mA
and declining 60 mV per decade of current. A pair of TVS zeners, used
in their forward directions might work; we have lots of them in stock.
So, I wonder if some doping magic makes a zener diode have higher
forward drop if it's a higher voltage zener?
Easier to ask than to pull a bunch of parts from stock and test them.
I'm feeling lazy today.
A shorted bridge rectifier would give me two diode drops. The ones in
stock are gigantic.

Depending on the accuracy required, you might want either a pair of
antiparallel PN diodes, or else a bridge with a little bit of bias

applied

Post by Phil Hobbs
to keep the diodes back-biased.
Cheers
Phil Hobbs

Put two antiparallel Schotky diodes in series across the shunt, four
altogether. That also reduces your anode-cathode capacitance and improves
high frequency response.

Or perhaps BJT wired as transdiode (E to C+B) should have low leakage at
low voltage? Two inverse parallel for both directions.

low leakage in reverse, but in forward, maybe
https://www.eevblog.com/forum/projects/forward-leakage-of-a-diode/msg1287509/#msg1287509

I guess the question is, given a hunky diode, rated 2 amps or so, does
it behave, forward-direction, exponentially all the way down to
nanoamps?

the curves show a few 1A diodes down to 1nA

[john larkin]

Post by Lasse Langwadt

Post by john larkin

Post by Lasse Langwadt

Post by piglet

Post by Phil Hobbs

Post by Phil Hobbs

Post by john larkin
Assume we have a 1-ohm 2512 current shunt resistor that we don't want
to fry. My thought is to put a pair of big diodes across it to limit
the current. The associated ADC will be maybe +-250 mV full scale, so
we don't want the diodes to conduct much current there.
A 3 amp PN power diode might work, maybe starting from 0.6v at 1 mA
and declining 60 mV per decade of current. A pair of TVS zeners, used
in their forward directions might work; we have lots of them in stock.
So, I wonder if some doping magic makes a zener diode have higher
forward drop if it's a higher voltage zener?
Easier to ask than to pull a bunch of parts from stock and test them.
I'm feeling lazy today.
A shorted bridge rectifier would give me two diode drops. The ones in
stock are gigantic.

Depending on the accuracy required, you might want either a pair of
antiparallel PN diodes, or else a bridge with a little bit of bias

applied

Post by Phil Hobbs
to keep the diodes back-biased.
Cheers
Phil Hobbs

Put two antiparallel Schotky diodes in series across the shunt, four
altogether. That also reduces your anode-cathode capacitance and improves
high frequency response.

Or perhaps BJT wired as transdiode (E to C+B) should have low leakage at
low voltage? Two inverse parallel for both directions.

low leakage in reverse, but in forward, maybe
https://www.eevblog.com/forum/projects/forward-leakage-of-a-diode/msg1287509/#msg1287509

I guess the question is, given a hunky diode, rated 2 amps or so, does
it behave, forward-direction, exponentially all the way down to
nanoamps?

the curves show a few 1A diodes down to 1nA

I'm setting up to test some power diodes and TVS diodes, at 2 amps and
10 uA. 10 uA at 0.3v and about 1 volt at 2 amps would be good enough
to protect my 1R shunt resistor and ADC if some yahoo applies 2 amps
to my gadget.

Temperature matters too, of course, but I can adjust for that.

The exponential thing doesn't work at high currents, when diodes get
ohmic.

[Chris Jones]

Post by john larkin
Assume we have a 1-ohm 2512 current shunt resistor that we don't want
to fry. My thought is to put a pair of big diodes across it to limit
the current. The associated ADC will be maybe +-250 mV full scale, so
we don't want the diodes to conduct much current there.
A 3 amp PN power diode might work, maybe starting from 0.6v at 1 mA
and declining 60 mV per decade of current. A pair of TVS zeners, used
in their forward directions might work; we have lots of them in stock.
So, I wonder if some doping magic makes a zener diode have higher
forward drop if it's a higher voltage zener?
Easier to ask than to pull a bunch of parts from stock and test them.
I'm feeling lazy today.
A shorted bridge rectifier would give me two diode drops. The ones in
stock are gigantic.

If you want really quite low leakage, you could put two pairs of
antiparallel diodes in series, and use an op-amp to drive the voltage
across one of the pairs of diodes to about zero, by driving the midpoint
of the diode string to the same voltage as one of the ends.

[piglet]

Post by Chris Jones

Post by john larkin
Assume we have a 1-ohm 2512 current shunt resistor that we don't want
to fry. My thought is to put a pair of big diodes across it to limit
the current. The associated ADC will be maybe +-250 mV full scale, so
we don't want the diodes to conduct much current there.
A 3 amp PN power diode might work, maybe starting from 0.6v at 1 mA
and declining 60 mV per decade of current. A pair of TVS zeners, used
in their forward directions might work; we have lots of them in stock.
So, I wonder if some doping magic makes a zener diode have higher
forward drop if it's a higher voltage zener?
Easier to ask than to pull a bunch of parts from stock and test them.
I'm feeling lazy today.
A shorted bridge rectifier would give me two diode drops. The ones in
stock are gigantic.

If you want really quite low leakage, you could put two pairs of
antiparallel diodes in series, and use an op-amp to drive the voltage
across one of the pairs of diodes to about zero, by driving the midpoint
of the diode string to the same voltage as one of the ends.

Neat idea! Thanks

[john larkin]

Post by john larkin
Assume we have a 1-ohm 2512 current shunt resistor that we don't want
to fry. My thought is to put a pair of big diodes across it to limit
the current. The associated ADC will be maybe +-250 mV full scale, so
we don't want the diodes to conduct much current there.
A 3 amp PN power diode might work, maybe starting from 0.6v at 1 mA
and declining 60 mV per decade of current. A pair of TVS zeners, used
in their forward directions might work; we have lots of them in stock.
So, I wonder if some doping magic makes a zener diode have higher
forward drop if it's a higher voltage zener?
Easier to ask than to pull a bunch of parts from stock and test them.
I'm feeling lazy today.
A shorted bridge rectifier would give me two diode drops. The ones in
stock are gigantic.

I tested a bunch of diodes.

A big conventional PN rectifier diode, SMB package, rated 3 amps, 200
volts, measures 0.76v at 2 amps and 0.37 at 10 uA.

The 600v version is 0.94 and 0.37 respectively.

A 28 volt TVS is 0.91v at 2 amps and 0.45 at 10 uA. The TVS is either
a very small junction, or doped such as to conduct less. It does get
pretty hot at 2 amps.

I should X-ray them.

A pair of antiparallel diodes, any of these, will protect my 1-ohm
shunt and have no serious affect on my 250 mV ADC.