# Comparing different through hole matched BJT options for exponential voltage to current conversion

I have an interest in analog audio synthesizers and I'm in the process of analyzing and prototyping vintage synth VCOs in order to get a grasp of their inside workings. All schematics I've looked into so far like MFOS's VCO, the Bergfotron advanced VCO, Jim Patchel's No tempco VCO(see page 5) and the ASM-1 VCO use a matched BJT as their exponential converter (a 1V/Octave convention is widely used in analog synthesizers). Its task is to convert a control voltage (CV) exponentially to a current driving the oscillator core.

Now, some of these chips like the intersil CA3046 that I'm currently using are obsolete, the MAT03 is scheduled for obsolescence and who knows what happens in the near future...

That's why I want to compare my options and learn how to substitute one with the other, altering the original circuits if and as needed.

A search on mouser revealed 2 possible series. 3xxP14-U by THAT and MATxx by analog devices, the second being more or less 4 times more expensive than the 1st considering they are dual ICs against quad. The other obvious differences is VCBO which is 9V higher for the MATxx series, CE saturation voltage which is 50mV for the 3xxP14-P and 120mV for the MATxx, DC collector current and gain bandwidth product.

This got rather confusing for me when I started looking into their datasheets because (I need your verification on this) I think they are expressing the same quantities in different ways. Nevertheless, I'm feeling a bit lost and that I don't know exactly what I'm looking for.

So to make this as clear as possible, I would like some help understanding:

1. What are the most important factors when comparing substitute matched BJTs (what to look for in the datasheets)
2. How to deal with differences between parts (scaling input and output Vs and Is) and if this is at all possible
3. Are there any 'out of the box' devices that perform exponential voltage to current conversion at the 0-10V , maybe 0-12V range or maybe +/-5V (I would love to give you a current range but i haven't managed to calculate the expected output current values)?
• Please let me know if you think there is anything I can add to make this question more clear – Schizomorph Mar 30 '17 at 19:39
• I'm not too interested in going after the schematics or getting up to speed on your knowledge here, but there are lots cheaper devices than the MAT** series. Take a look at the BCV61 and BCV62. These are $\beta$-matched. You can also look for ones that are both $\beta$ and $V_{BE}$ matched, which are more expensive (of course.) You can buy a lot of BJTs and match them yourself, as well. And there is literature on adding compensation resistors to help deal with variations -- most of this literature dates into the '60's, though, before ICs made this easy. – jonk Mar 30 '17 at 20:08
• Thanks regardless. I will definitely have a look at the BVC6x series and dig a bit deeper in the literature. You've already given me some key words to use on my google search and an approximate date of the literature I'm looking for. – Schizomorph Mar 30 '17 at 20:16
• Widlar is one author to look for. He added a $\beta$-compensating resistor between the bases of two BJTs, plus emitter resistors for $V_{BE}$ compensation. The $\beta$-compensating resistor value depends upon the values of the $V_{BE}$ compensation resistors, too. You can work out the calcs yourself, if you are familiar enough with the BJT model. BJTs are cheap enough that buying lots of them and matching them could make sense. But you also need to know how your circuits work, well enough to know "how matched" you need and what exactly needs to be matched. – jonk Mar 30 '17 at 20:26
• Taking a look at the MAT03 parts in your "Triangle VCO core" example schematic, I suspect you do NOT need $V_{BE}$ matching nearly as much as $\beta$ matching. So the BCV61 and BCV62 parts may be your friends here. – jonk Mar 30 '17 at 20:29