1
\$\begingroup\$

I have seen in several schematics where a precision resistor, or matched resistors are called for, but every one of them talk about you going to purchase special resistors for this purpose.

I am building the Go, No-go transistor matching circuit shown here. It stated the just about all the resistors have to be matched to 1% at the most for it to work. I have cheapo "1% resistors", but don't trust them to be in spec. For this particular circuit it would seem that the closer I can get to precise resistances, the more accurate the vbe measurements will be.

Question: do I really need to find resistor "pairs" for this or can I simply put resistors in series until the total resistance is exactly what I want (my 20000 count multimeter would let me get to 0.1% or even 0.01% precision this way)? So for example, to get a perfect 20k resistor, I might use a 15k, 4.7k, 270 and maybe something small between 1-100 ohms all in series so that the total real resistance measured is 19.999-20,001 ohms. It seems like a simple no brainer, but every time I see schematics talking about precision resistors, none of them offer this approach, which got me to wonder if there was a reason.

\$\endgroup\$
4
\$\begingroup\$

This approach works well for 'one-off' instruments designed to work at room temperature. I use it all the time. It has two drawbacks.

1) It's labour intensive, which is why it's not used for mass produced equipment, where it's cheaper to buy 0.1% resistors and just fit them, rather than muck about measuring and matching.

2) 1% resistors won't necessarily have as good a tempco of resistance as 'precision' 0.1% or 0.01% resistors. This will be mitigated to some extent if you are only after tracking between multiple resistors, but even then there is no guarrantee cheaper resistors will track well. A set of precision resistors in a pack will be specified to track to a certain extent.

\$\endgroup\$
  • \$\begingroup\$ I think I understand. Though the transistor matching circuit is certainly temp sensitive, this is only for the parts where for example the inverting unity gain amplifier that produces -10v from +10v. Both the input and feedback resistors of that op amp need to be as close in value to eachother as possible (in this case the circuit calls for 20k). So it sounds like there is no reason why you need to have two physical resistors that match as opposed to two resistances (regardless of the number of resistors you use in series) that match to the precision you want. \$\endgroup\$ – Dave Guenther Feb 21 '16 at 14:56
1
\$\begingroup\$

Not suitable for high frequency or high speed circuit because the modified resistor may exhibit high stray inductance and capacitance.

But for your application it may work fine.

\$\endgroup\$
0
\$\begingroup\$

p.s. that circuit is complete overkill to find pairs of transistors with equal VBEs.

There is no reason to set the test current to any precise value -- if you want to match VBES at (say) 100 uA, then something 'close' to 100 uA would be OK -- if the 'matching' changes depending on whether you are at 100 uA or 110 uA, your application is likely not using transistors correctly.

So, if take (say) a 9 V battery, and a resistor of about (9-0.6)/100uA = 84k -- so use 82k, and connect the transistor as a diode (C&B = 1 terminal, E = other) in series, and measure the VBE with a DMM. Swap in your other devices and find matched ones.

Alternatively, use 2 82k resistors, 1 for each transistor and use the DMM to compare directly between the two collector/base connections. You don't need precision resistors -- 1 % mismatch will cause about 0.26 mV difference between the transistors -- about the same as 0.13 deg. C temperature difference.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.