Kosmo 1222 source a reasonable LM4040AIZ 4.1?

Define “high precision”. You cannot buy ±0.00% references, and once you get below 0.1% (*) things can get expensive quickly (e.g. AD586 at 0.05% is $10 ea or more, and for even more precision you may need more complicated circuitry that’s harder to build and source).

So reasonably cheap and widely available reference + a bit of trimming seems like a reasonable engineering tradeoff, and keeps the budget within bounds.

*) I’ll leave it to @antoine.pasde2 to calculate the lowest tolerance you’d need to skip the trimmer :slight_smile:


I have my 4 oscillators built with a TL431 and so far I can tell you that it works :wink:
Three of them go out of tune for the higher octaves, but I suspect that it is because I did not calibrate them correctly… I used the octave switch for those and the one I tuned with external V/oct (beatstep) is better in that regard!


You’ve been visiting the North Coast again, haven’t you? :wink:

That’s where I got mine too.

Oh, look at that. But no, the TL431 came up in a different context and I had your ppm calculations fresh in memory so the 50 ppm bit stood out more than usual.

If you go back to the Patreon posts when Sam talked about VCO tuning you may find me babbling about the TL431 in the comments :smiley: (I’d link but have no idea how to find old comments on Patreon…)

EDIT: Found it. The zener Sam mentions there might be the LM4040, not sure.


I found this calculator if it helps…


I am assuming you would want high precision resistors in the circuit. If this is a solution I will draw up a little daughter board that will fit in the original regulator hole.

Looks promising…


The TI data sheet I found for the TL431 doesn’t specify the temperature deviation in ppm, it just gives a maximum for the whole temperature range.
The smallest range is commercial: 0°C to 70°C and has a maximum deviation of 25mV.
When you scale the 2.495V to 4V, the deviation scales proportionally to 40mV which seems to be worse than the 18mV calculated above for the LM4040AIZ.
Admittedly that is the worst case, and the typical case of 4mV (scaled to 6.4mV) looks much better, but are we willing to take the risk, or must we always design for the worst case?

There’s a bunch of manufacturers, and it seems they all report things in slightly different ways :grimacing: (the 50 ppm I saw was from OnSemi, but when I click through to their site they say “typical”, while TI has a “do it yourself” formula in the datasheet but spell out max on the site (ATL431LI seems to be their best 431 option at max 66 ppm, and that’s still cheaper than the LM4040)).

But optimistically speaking, I’d expect the distribution to be somewhere close to normal, so on average you “should” be ok. But that may be too optimistic.

I got mine from Farnell but check its the right one as I’m still building my units and may have cocked up!


Let me know if I’ve got the wrong one! Thanks

LM4040AIZ-4.1 is the right one, and the best LM4040 option (A for best tolerance, 4.1 for “just over 4 V” – nominal voltage is 4.096 V to match 12-bit integers, for use with A/D conversion, but that’s trimmed down to 4.000 V for the VCO).

1 Like

Well that’s me confused, spent half the morning searching the usual but did not find that on Farnell and seems darn cheap…

Best get me some :slight_smile:

Good Find

This looks like the one!

Minimum order 10, and a re-reeling charge. Unless you have a placement machine I suspect you want the one linked above :slight_smile:

1 Like

Yeh I think your right, and further more they are not delivering to Germany and Farnell Germany does not have them in stock. :frowning:

Not seeing any restrictions at


and adding them to the cart worked just fine (but ok, I haven’t tried to place the order).


Item charge
handeling charge
Re-Reeling chage???

so £2.62 each …

I can find out how much letter post is and send them on at cost if i have interet

…or you could buy the “cut tape” one linked earlier that doesn’t have a minimum order or extra fees, instead of the “reel” one that’s intended for industrial users with automated assembly machines. Here’s how TI describes these options:

A custom reel is a continuous length of cut tape from one reel to maintain lot and date code traceability built to the exact quantity requested on a plastic reel. One 18-inch (45cm) flat black leader tape and one 18-inch (45cm) flat black trailer tape are connected on both sides of the cut tape via a brass shim (length 22mm) so that the sprocket holes are aligned. The final tape is then rewound onto a plastic reel in compliance with Electronics Industries Alliance (EIA) standards. This splicing operation enables direct feeding into automated board assembly equipment.

Cut tape is a length of tape, cut from a reel. Cut tape orders may be fulfilled utilizing multiple strips of cut tape or as individual parts for your requested quantity and shipped in an ESD or MBB bag.

Ah yeah, don’t use this version of the link…

Cheers :slight_smile:

There are some lm4040 at reichelt but no 4.1 currently :-/ maybe you could try with a 5.0 and hope that it will not make a big difference? As I have said, I am using a tl431, which has worse specs, but so far I am happy with my oscillators :wink:

Haven’t tested it, but it seems to me that the only difference would be where you find 4.000 V (or whatever you can trim it to) on the pot. As long as you’re above 4 V and the cathode current is within spec so it regulates properly, it should work.

(it’s also not entirely impossible to solder wires to SOT-23 legs, btw)

1 Like