More current mirror confusion

I think I have a fundamental mis-understanding here.

As stated in an earlier thread, I created a spice sim of the simple amp in chapter 2 of Cordell's book. I am messing around with the input LTP just to try and understand some things. My current incarnation is just the simple amp but with emitter degeneration and a simple mirror load for the LTP.

I have MANY questions, but I noticed one thing that really is confusing me

Why is the current flow in the diode connected half of the mirror out of phase with the collector current of the LTP transistor it is connected to? Like I said, I must be missing something really fundamental here.

Also, other than quiescent current balance, I still don't see what the mirror accomplishes that a current source (sink?) on the collector of the jnon-feedback side of the LTP wouldn't.

Sorry for being dense. I come from a tube guitar amp background, and solid state seems to be putting me back at square one! It appears cultivating distortion is way easier than fighting it!


2003-02-17 7:38 pm
you have to use common sense, sanity checks, be aware that the software may have its own quirky definitions despite how you think of the circuit working

you have to watch out for Spice model, schematic, symbol, probing polarity definitions - this is easy to get "wrong" when probing the degen R - you can't easily see the R device pin #, filpping the R symbol end or end in the cirucit will change the sign of the measured I

there can also be diagreement with current direction in/out of a complex device pin

if you use the LTspice schematic current probe, look closely for the red arrow in the current clamp loop symbol - it shows the direction convention for the 2 terminal part you a hovering over, but not for higher pin count parts
Last edited:
Ok, so I checked and the transistors in the LTP are showing the expected quiescent current, except negative. Components in series with these transistors show the same current but positive. Obviously this shouldn't be, right? I tried many different transistor types including Bob Cordell's. is this just a lame LTSpice thing? It's not like I'm doing anything esoteric here.


2003-02-17 7:38 pm
yes the software is looking at the symbols, pin numbers, conventions for more complex parts to assign a polarity to a current for display - obvioulsy it does "the right thing" internally in the .tran sym matrix,d.dmg&fp=e62569ac204fe7af&biw=1920&bih=1069

try breaking down the circuit to smaller pieces you are confident of the answers for and simiing, probing, building up in complexity or just running quick separate sims on the side as a check of say what polarity definition for Ie(Q1) the software is using

quick hacks on you working schematic can be undo with the back button so sometimes I cut out or add test parts, then undo back to where I started
Last edited:
The OP may be confusing quiescent current and signal current, or persisting in thinking strictly in 'conventional current' terms (i.e. opposite to the electron flow). This mode of thinking is often taught in elementary electrical/electronics classes but it leads to confusion when NPN/PNP comes along, as some of the currents then seem to flow 'backwards'.

I always think of a quiescent current flowing into a collector, and signal current flowing out - whatever polarity the BJT is. Then I separately think about the polarity of that signal current. So an LTP produces two signal currents, but of opposite polarity. This works for me, but if it confuses you please ignore it.


2011-03-13 8:21 pm
I like to don' t care about NPN / PNP.
I care about the ARROW.
Emitter current flows like shown by the arrow,
most of this current flows the same way at the Collector,
a small part of the emitter current flows the same way at the Base.
Beware, this arrow method works for BJT's, it does not work for Mosfet arrows.
In any case current flows FROM some voltage TO a LOWER voltage
There may be two conventions that need to be taken into account here. LTspice allows you to 'orient' the current flow by which way round you connect the component. That way you can have the current arrows showing in the way you want to see it. It's an issue that has come up quite a lot and is quite tricky. At the user group on Yahoo! you will find a number of discussions about it. The second issue comes from the way the Kerchoff based matrix works and relies on the current at any node summing to zero. So, for this to be the case, the collector and base current summing to a positive must imply a negative emitter current. But current flow is as you would normally expect.

Is this the answer you are after or is there something more?
I mostly have this solved I think. It just kind of threw me that clicking on the collector showed one polarity and clicking on the emitter showed the opposite when I know they are the same. Not realizing this qwerk briefly caused me to doubt my fundememtal understanding because it gave strange results in transient simulations if I was unlucky enough to probe the wrong pin.

Knowing what I do now, I have started checking the operating point of a pin to make sure it's the right one to watch.