How much input current can the secondary of, say, a Jensen 115KE mic
input transformer, tolerate before it's distortion rises measureably?
What current level will begin magnetizing it appreciably?

Do I have to worry about a few microamps flowing in a microphone's
output transformer
magnetizing it (or raising distortion) if it's direct coupled to an
amplifier with bipolar inputs? I'm worried about hooking up my coles
4038 to this little mic amp i've built.

Thanks.

Sean B

wrote:
How much input current can the secondary of, say, a Jensen 115KE mic
input transformer, tolerate before it's distortion rises measureably?
What current level will begin magnetizing it appreciably?

This is a question that you should probably ask directly to Jensen.
They're responsive to e-mail. Try to give them a number to work with
(how many milliamps DC?) and they might even test one if they don't
already know.

Magnetizing the transformer in a mic used to be a source of worry when
it came to connecting a mic to a phantom powered input, but that seemed
to go away when it was pointed out that many condenser mics of the time
also had transformers and nobody worried about magnetizing them.

Putting DC through a transformer can indeed "bias" the core so that it
gets closer to saturation, but I really don't have a clue as to how
much. Jensen will be able to help you though.

Mike Rivers wrote:

wrote:
How much input current can the secondary of, say, a Jensen 115KE mic
input transformer, tolerate before it's distortion rises measureably?
What current level will begin magnetizing it appreciably?

This is a question that you should probably ask directly to Jensen.
They're responsive to e-mail. Try to give them a number to work with
(how many milliamps DC?) and they might even test one if they don't
already know.

Magnetizing the transformer in a mic used to be a source of worry when
it came to connecting a mic to a phantom powered input, but that seemed
to go away when it was pointed out that many condenser mics of the time
also had transformers and nobody worried about magnetizing them.

Because the current for phantom power passes via the centre tap, there is in
fact no net magnetising force since the current through each half of the winding
'cancels out' the other.

Putting DC through a transformer can indeed "bias" the core so that it
gets closer to saturation, but I really don't have a clue as to how
much. Jensen will be able to help you though.

I doubt that microamps are going to be a problem but why not use a coupling cap
anyway ?

Graham

Pooh Bear wrote:

Because the current for phantom power passes via the centre tap, there is in
fact no net magnetising force since the current through each half of the winding
'cancels out' the other.

I'm talking about the sencondary circuit, and the effects of the
amplifier's input current. Imagine we're recording with a ribbon mic.


SB

wrote:

Pooh Bear wrote:

Because the current for phantom power passes via the centre tap, there is in
fact no net magnetising force since the current through each half of the winding
'cancels out' the other.

I'm talking about the sencondary circuit, and the effects of the
amplifier's input current. Imagine we're recording with a ribbon mic.

Yes I know you are. I was responding to Mike's post.

You'll find my comment about magnetising current of a few microamps also in that post.

Graham

wrote:

Pooh Bear wrote:

Because the current for phantom power passes via the centre tap, there is in
fact no net magnetising force since the current through each half of the winding
'cancels out' the other.

I'm talking about the sencondary circuit, and the effects of the
amplifier's input current. Imagine we're recording with a ribbon mic.

Oh you're using the awful google groups. You probably can't see the thread properly.

Graham

Pooh Bear wrote:

Because the current for phantom power passes via the centre tap, there is in
fact no net magnetising force since the current through each half of the winding
'cancels out' the other.

That's true in the normal, steady state, but the risk (and it's the
same risk with popping the element on a sensitive mic) is that pins 2
and 3 won't make contact at the same time and there will be a brief
time where current can flow through the transformer if it has a
grounded center tap. And of course if you have a cable with a short
between pin 1 and one of the signal pins, you have the full voltage
(admittedly through a resistor that limits the current to about 7 mA)
applied across the element or the transformer.

And as we all know, it's not steady DC that magnetizes things, but a
rapidly rising or falling field caused by a current transient.

I doubt that microamps are going to be a problem but why not use a coupling cap
anyway ?

Because capacitors are evil nasty things that cause distortion and
should be avoided at all costs.

Pooh Bear wrote:

Oh you're using the awful google groups. You probably can't see the thread properly.

I'm usuing Google groups and I can see the thread perfectly. When was
the last time you tried it? And set it up to see the thread?

Mike Rivers wrote:

Pooh Bear wrote:

I doubt that microamps are going to be a problem but why not use a coupling cap
anyway ?

Because capacitors are evil nasty things that cause distortion and
should be avoided at all costs.

**** LMAO ! **** I assume that _was_ a joke ? !!! ;~}

Graham

Mike Rivers wrote:

Pooh Bear wrote:

Because the current for phantom power passes via the centre tap, there is in
fact no net magnetising force since the current through each half of the winding
'cancels out' the other.

That's true in the normal, steady state, but the risk (and it's the
same risk with popping the element on a sensitive mic) is that pins 2
and 3 won't make contact at the same time and there will be a brief
time where current can flow through the transformer if it has a
grounded center tap.

Good point.

And of course if you have a cable with a short
between pin 1 and one of the signal pins, you have the full voltage
(admittedly through a resistor that limits the current to about 7 mA)
applied across the element or the transformer.

And as we all know, it's not steady DC that magnetizes things, but a
rapidly rising or falling field caused by a current transient.

On the subject of transformers and phantom power, years ago at Studiomaster we used to
get the occasional mic input fried by problems with phantom power. A former colleague
of mine had put in place some of the typical clamp diodes to deal with this but
occasionally it wouls still happen. We eventually found out that AKG C451s were often
implicated and they have a transformer in them. I concluded that the problem was due
to inductive back emf from exactly the kind of 'hot pin' situation you describe.
Beefing up the protection fixed things.

Graham

Mike Rivers wrote:

Pooh Bear wrote:

Oh you're using the awful google groups. You probably can't see the thread properly.

I'm usuing Google groups and I can see the thread perfectly. When was
the last time you tried it?

At least a month ago - probably more.

And set it up to see the thread?

Barely worth the trouble. I know what I'm looking for, so won't be confused by any lack of
quoting and I just use it as a fall back if I get problems with my normal newsserver.

Now, what did you reckon to the references I gave you for mic impedance ?

Graham

Pooh Bear wrote:

Now, what did you reckon to the references I gave you for mic impedance ?

I was overwhelmed, but see my response to Phil, who gave me some
similar examples.

To tell you the truth, I don't believe most specifications that are
published for audio gear, and about the only ones that mean much to me
are the size, weight, and things that aren't really specifications,
like number of inputs and outputs, number of bands of EQ, stuff like
that. You can't even tell if a "balanced" output is differential or
single-ended nowadays by reading most spec sheets.

So I may have overlooked the change toward more information (even
though it isn't always useful) in today's spec sheets simply because I
haven't looked at one for ages.

Pooh Bear wrote:

On the subject of transformers and phantom power, years ago at Studiomaster we used to
get the occasional mic input fried by problems with phantom power. A former colleague
of mine had put in place some of the typical clamp diodes to deal with this but
occasionally it wouls still happen. We eventually found out that AKG C451s were often
implicated and they have a transformer in them. I concluded that the problem was due
to inductive back emf from exactly the kind of 'hot pin' situation you describe.
Beefing up the protection fixed things.

Interesting. The problem with "hot plugging" mics was rampant in the
mid '80s. The Biamp (a Mackie relative) had to replace a lot of input
stages, and the original Mackie CR1604 followed suit. By the time they
came out with the VLZ series, they had figured out the clamping diodes.
A lot of manufacturers use them now, and you'll find them recommended
in the application notes for mic preamp chips.

Some people say you can hear them, but that might be preferable to not
hearing anything because you've fried the preamp input.

Pooh Bear wrote:
Mike Rivers wrote:

Because capacitors are evil nasty things that cause distortion and
should be avoided at all costs.

**** LMAO ! **** I assume that _was_ a joke ? !!! ;~}

Only partially.

Mike Rivers wrote:

Pooh Bear wrote:
Mike Rivers wrote:

Because capacitors are evil nasty things that cause distortion and
should be avoided at all costs.

**** LMAO ! **** I assume that _was_ a joke ? !!! ;~}

Only partially.

Inasmuch as some are indeed non-linear the really bad ones are Hi-K ceramics and
tantalums with zero bias. From my own measurements I've established that zero
bias aluminium electrolytics used sensibly ( use a large value ) are blameless
despite popular belief.

Graham

Mike Rivers wrote:
By the time they
came out with the VLZ series, they had figured out the clamping diodes.
A lot of manufacturers use them now, and you'll find them recommended
in the application notes for mic preamp chips.

Some people say you can hear them

The cheapskate way to do it is to put a back-to-back pair of zeners
between each input and ground. Zeners leak quite a lot (and doubtless
not very linearly) below threshold voltage so could affect the sound.

Better to use ordinary Si diodes clamping to a zener that's already
biased on. The reverse leakage of an ordinary diode is negligible and
shouldn't have any audible effect.

--
Anahata
-+- http://www.treewind.co.uk
Home: 01638 720444 Mob: 07976 263827

I just got an email from Jensen, apparently a few microamps through the
secondary is OK, up to a limit of 50 they say. They say the low end
goes first with increasing DC current. FYI.

SB

anahata wrote:

The cheapskate way to do it is to put a back-to-back pair of zeners
between each input and ground. Zeners leak quite a lot (and doubtless
not very linearly) below threshold voltage so could affect the sound.

Better to use ordinary Si diodes clamping to a zener that's already
biased on. The reverse leakage of an ordinary diode is negligible and
shouldn't have any audible effect.

They use the diodes to clamp to the power supply rails. Still, some
people say you can hear the difference with them in or out.

wrote:
I just got an email from Jensen, apparently a few microamps through the
secondary is OK, up to a limit of 50 they say. They say the low end
goes first with increasing DC current.

Good that you went to the folks who have the right answers. A few uA
might be the kind of leakage you're getting out of your preamp, but be
aware that phantom power, if shorted, can give you up to about 7 mA.

anahata wrote:

Mike Rivers wrote:
By the time they
came out with the VLZ series, they had figured out the clamping diodes.
A lot of manufacturers use them now, and you'll find them recommended
in the application notes for mic preamp chips.

Some people say you can hear them

The cheapskate way to do it is to put a back-to-back pair of zeners
between each input and ground. Zeners leak quite a lot (and doubtless
not very linearly) below threshold voltage so could affect the sound.

Better to use ordinary Si diodes clamping to a zener that's already
biased on. The reverse leakage of an ordinary diode is negligible and
shouldn't have any audible effect.

I agree zeners are too 'leaky'. Most designs I've seen simply clamp to the
supply rails actually ( mine included ).

Graham

Mike Rivers wrote:

anahata wrote:

The cheapskate way to do it is to put a back-to-back pair of zeners
between each input and ground. Zeners leak quite a lot (and doubtless
not very linearly) below threshold voltage so could affect the sound.

Better to use ordinary Si diodes clamping to a zener that's already
biased on. The reverse leakage of an ordinary diode is negligible and
shouldn't have any audible effect.

They use the diodes to clamp to the power supply rails. Still, some
people say you can hear the difference with them in or out.

I reckon they're imagining it.

Graham

Pooh Bear wrote:

I agree zeners are too 'leaky'. Most designs I've seen simply clamp to the
supply rails actually ( mine included ).

Isn't there a small danger of an internal junction in the IC conducting
before the external diode does if you do that?

Anahata

anahata wrote:
The cheapskate way to do it is to put a back-to-back pair of zeners
between each input and ground. Zeners leak quite a lot (and doubtless
not very linearly) below threshold voltage so could affect the sound.

I have tried this and it's pretty much always audible. It soured me on
the use of diodes for a long time.

Better to use ordinary Si diodes clamping to a zener that's already
biased on. The reverse leakage of an ordinary diode is negligible and
shouldn't have any audible effect.

There is a third way that is even cheaper and even better. Use ordinary
Si diodes to clamp to the supply rails. Worst case your signal exceeds
one diode drop over the supply rail. When the thing is turned off, they
are clamped to ground reasonably well. I can't hear it working and it
costs two 1N4004s.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Anahata wrote:

Pooh Bear wrote:

I agree zeners are too 'leaky'. Most designs I've seen simply clamp to the
supply rails actually ( mine included ).

Isn't there a small danger of an internal junction in the IC conducting
before the external diode does if you do that?

Not really. I use a decent size diode that has low internal resistance. You
could use schottkys but I think they have more C and are leaky too.

Graham

Scott Dorsey wrote:

anahata wrote:
The cheapskate way to do it is to put a back-to-back pair of zeners
between each input and ground. Zeners leak quite a lot (and doubtless
not very linearly) below threshold voltage so could affect the sound.

I have tried this and it's pretty much always audible. It soured me on
the use of diodes for a long time.

Better to use ordinary Si diodes clamping to a zener that's already
biased on. The reverse leakage of an ordinary diode is negligible and
shouldn't have any audible effect.

There is a third way that is even cheaper and even better. Use ordinary
Si diodes to clamp to the supply rails. Worst case your signal exceeds
one diode drop over the supply rail. When the thing is turned off, they
are clamped to ground reasonably well. I can't hear it working and it
costs two 1N4004s.

Well, four actually ( two on each leg ). I use the 1N4004 there now too.

Graham

Pooh Bear wrote:
Anahata wrote:

Pooh Bear wrote:

I agree zeners are too 'leaky'. Most designs I've seen simply clamp to the
supply rails actually ( mine included ).

Isn't there a small danger of an internal junction in the IC conducting
before the external diode does if you do that?

Not really. I use a decent size diode that has low internal resistance. You
could use schottkys but I think they have more C and are leaky too.

You don't care if the internal junction conducts or not, just if it breaks
down. If you use, say, an INA103 that is rated to run on 20V rails, and you
run it on 18v rails with 1N4004 input clamps, then the input won't exceed
18.6V which is well below the rated breakdown voltage for the part.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Scott Dorsey wrote:

Pooh Bear wrote:
Anahata wrote:

Pooh Bear wrote:

I agree zeners are too 'leaky'. Most designs I've seen simply clamp to the
supply rails actually ( mine included ).

Isn't there a small danger of an internal junction in the IC conducting
before the external diode does if you do that?

Not really. I use a decent size diode that has low internal resistance. You
could use schottkys but I think they have more C and are leaky too.

You don't care if the internal junction conducts or not, just if it breaks
down. If you use, say, an INA103 that is rated to run on 20V rails, and you
run it on 18v rails with 1N4004 input clamps, then the input won't exceed
18.6V which is well below the rated breakdown voltage for the part.

That's not how it works though Scott. If you look carefully, you'll see that the
maximum allowable voltages on any pin are related to the power supply voltage.

Typically most pins have a 'parasitic' diode structure internally that conducts
when that voltage is exceeded. In bad cases it may cause 'scr latchup' when the
chip structure is more complex. I've seen it happen. Hence low diode resistance is
indeed a factor.

Graham

Pooh Bear wrote:

You don't care if the internal junction conducts or not, just if it breaks
down. If you use, say, an INA103 that is rated to run on 20V rails, and you
run it on 18v rails with 1N4004 input clamps, then the input won't exceed
18.6V which is well below the rated breakdown voltage for the part.

That's not how it works though Scott. If you look carefully, you'll see that the
maximum allowable voltages on any pin are related to the power supply voltage.

Typically most pins have a 'parasitic' diode structure internally that conducts
when that voltage is exceeded. In bad cases it may cause 'scr latchup' when the
chip structure is more complex. I've seen it happen. Hence low diode resistance is
indeed a factor.

Okay, so it already HAS internal protection diodes in that case, which
means the maximum allowable voltage rating isn't the same thing as the
breakdown voltage, it's instead the rating of the protection diode system.
Is that correct?

This is very different than discretes, where you just worry about the
breakdown voltage on the data sheet, and you always have the option of
picking a higher breakdown voltage (like those great Motorola CRT drive
transistors) at the expense of noise and hFE.

What would the actual breakdown voltages of the input transistors on
one of those chips actually be?
--scott


--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Pooh Bear wrote:

Scott Dorsey wrote:

anahata wrote:
The cheapskate way to do it is to put a back-to-back pair of zeners
between each input and ground. Zeners leak quite a lot (and doubtless
not very linearly) below threshold voltage so could affect the sound.

I have tried this and it's pretty much always audible. It soured me on
the use of diodes for a long time.

Better to use ordinary Si diodes clamping to a zener that's already
biased on. The reverse leakage of an ordinary diode is negligible and
shouldn't have any audible effect.

There is a third way that is even cheaper and even better. Use ordinary
Si diodes to clamp to the supply rails. Worst case your signal exceeds
one diode drop over the supply rail. When the thing is turned off, they
are clamped to ground reasonably well. I can't hear it working and it
costs two 1N4004s.

Well, four actually ( two on each leg ). I use the 1N4004 there now too.

I've always had a nagging worry with that system. If an input overload
exceeds the total current being drawn from the supply rail, it will take
over from the regulator, pull the rail higher and put excessive voltage
on all the chips.

The resulting damage will be much more than just one blown input chip.


I have never seen it happen, though.


--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk

Adrian Tuddenham wrote:

Pooh Bear wrote:

Scott Dorsey wrote:

anahata wrote:
The cheapskate way to do it is to put a back-to-back pair of zeners
between each input and ground. Zeners leak quite a lot (and doubtless
not very linearly) below threshold voltage so could affect the sound.

I have tried this and it's pretty much always audible. It soured me on
the use of diodes for a long time.

Better to use ordinary Si diodes clamping to a zener that's already
biased on. The reverse leakage of an ordinary diode is negligible and
shouldn't have any audible effect.

There is a third way that is even cheaper and even better. Use ordinary
Si diodes to clamp to the supply rails. Worst case your signal exceeds
one diode drop over the supply rail. When the thing is turned off, they
are clamped to ground reasonably well. I can't hear it working and it
costs two 1N4004s.

Well, four actually ( two on each leg ). I use the 1N4004 there now too.

I've always had a nagging worry with that system. If an input overload
exceeds the total current being drawn from the supply rail, it will take
over from the regulator, pull the rail higher and put excessive voltage
on all the chips.

That would be one heck of a large belt of energy required to do that.

The resulting damage will be much more than just one blown input chip.

*If it happened * then yes. In the meantime it's the best way to protect what's
there.

I have never seen it happen, though.

Likewise. I doubt it could happen unless someone plugged an amplifier into the
mic input. In any event I also fit smal value series Rs and they'll fry if too
much current goes that way, providing another level of protection.

Graham

Pooh Bear wrote:

Adrian Tuddenham wrote:

Pooh Bear wrote:

Scott Dorsey wrote:

anahata wrote:
The cheapskate way to do it is to put a back-to-back pair of zeners
between each input and ground. Zeners leak quite a lot (and doubtless
not very linearly) below threshold voltage so could affect the sound.

I have tried this and it's pretty much always audible. It soured me on
the use of diodes for a long time.

Better to use ordinary Si diodes clamping to a zener that's already
biased on. The reverse leakage of an ordinary diode is negligible and
shouldn't have any audible effect.

There is a third way that is even cheaper and even better. Use ordinary
Si diodes to clamp to the supply rails. Worst case your signal exceeds
one diode drop over the supply rail. When the thing is turned off, they
are clamped to ground reasonably well. I can't hear it working and it
costs two 1N4004s.

Well, four actually ( two on each leg ). I use the 1N4004 there now too.

I've always had a nagging worry with that system. If an input overload
exceeds the total current being drawn from the supply rail, it will take
over from the regulator, pull the rail higher and put excessive voltage
on all the chips.

That would be one heck of a large belt of energy required to do that.

The resulting damage will be much more than just one blown input chip.

*If it happened * then yes. In the meantime it's the best way to protect
what's there.

I have never seen it happen, though.

Likewise. I doubt it could happen unless someone plugged an amplifier into the
mic input. In any event I also fit smal value series Rs and they'll fry if too
much current goes that way, providing another level of protection.

That's probably the best answer, a sacrificial resistor. The sort of
situation I had in mind was if someone found a way of plugging a power
supply into a mic socket.

A really elaborate way to protect against it would be to crowbar the
supply rails - but that is a bit over the top for most applications.


Many years ago I was asked to design an amplifier that would be
undamaged by mains (U.K. = 240 v) on the mic inputs. I sacrificed the
S/N ratio and used a virtual earth input stage with 1N4000 clamp diodes
to earth from the inverting input. As there was no voltage signal at
that point under normal conditions, there was no increase in distortion.

The input impedance had to be about 47K, so I used a pair of 1 watt
resistors in series. They were sufficient to withstand the mains
indefinitely without overheating.


--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk

Adrian Tuddenham wrote:

I've always had a nagging worry with that system. If an input overload
exceeds the total current being drawn from the supply rail, it will take
over from the regulator, pull the rail higher and put excessive voltage
on all the chips.

If it's a short pulse, it should go right through all the decoupling
caps. The decoupling caps should appear like a dead short across the
power line to AC.

If it's a lot of DC for a long time, I could see maybe it being an issue,
but it would have to be pretty catastrophic.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."