Good day, i'm designing a logarithmic amplifier in a sensing mechanism and i need matched diodes to keep the sensing mechanism the same in every product.

Has anyone some idea of how well SMD diodes of the same reel are matched. I need the diode parameter Is (reverse leakage current) to be essentially the same.
My experience is that SMD transistors of the same reel are very well matched.

 

You'll probably find by statistical sampling that 75%-85% fall in a very narrow range (typical Vf, typical Vbr, etc.) but the remainder will fall outside of that range in varying degrees, much like the standard bell curve.

Consult the manufacturer's datasheet for the parts in question to determine their specifications.

Keep temperature range in mind, too. As temperature increases, Vf and Vbr will decrease.

If you want matched diodes, then look for matched diode arrays. If no luck there, consider matched transistor arrays; you could use the BE junctions as diodes.

 

thanks for the reply. I would look for arrays if i was looking for matched diodes for use within the same circuit.
I'm however talking about independent circuits. I need the sensing mechanism to be the same so that i can produce numerous circuits without needing to calibrate every one of them.

 

The manufacturer may have some 'in-house' population results as part of its quality control. That could possibly give you the deviation characteristic, which may then give you confidence that effectively "all" your product will have acceptable specs - or to the contrary, that you are actually stuck with testing each part!

Or you could do you own statistical sample of completed product.

 

Is is a very strong function of temperature. So even if you have perfectly matched diodes, any difference in temperature would destroy any hope of matched circuits. The only way to make this a reliable design choice is to change the laws of physics.

As for the question of matching... devices from the same Silicon slug are generally very closely matched, those from the same slice even more so. However, you can not be assured that those on the same reel are from the same slice, or even the same slug. But, again, matched Is is not likely to get you what you are looking for anyway.

Generally, what you do for a log amp is to have matched transistors, in a single package. One is referenced to the other, guaranteeing a consistent circuit regardless of Is. In this case, not only are the 'diodes' from the same slug, and the same slice, but they are from the same area of the slice since they are adjacent to each other on the same piece of silicon. Being this close, they are also closely thermally coupled so that any temperature dependence of Is is also compensated for. Better design choice all around.

 

Are you using the diode as a reference?

Is the system a uC/DSP that would allow an auto-zero to occur on startup, and/or compensate for temperature?

 

DonQ, thanks for the reply. I had already switched from using a single diode to transistors in my logaritmic amplifier (for temperature compensation). Had intended to use smd transistors from the same reel but forgot the option of using a dual transistor on a single chip.

thatoneguy, yes it is temperature compensated and there is a DSP in the product but it does not feature auto-zero as of yet.

Keep in mind though that i am trying to match these for different circuits. I'm making a product (digital feedback control system) which i intend to sell. I want by careful design to get past having to calibrate the sensing mechanism after i build each circuit so that its essentially the same in every one.