# How precise resistor value needed in feedback?

I'm trying to design an isolated feedback loop for a Flyback converter using TLV431 and an optocoupler. One of the TLV431 pins is the reference voltage, where the output's resistor divider (R1 and R2) should be connected so the reference voltage there is 1,24V. However I saw some Flyback converter designs and the resistor values aren't the most precise (I mean they don't give 1,24V reference in case of Vout=5.0V), calculating the output needed for 1,24V reference they give about 4,8V instead of the 5.0V nominal output. Does that mean that those designs aren't giving 5.0V output, only 4.8V? If not then how is that regulated to 5.0V and how precise the resistor divider has to be?

• It has to be as precise as you need the output to be – PlasmaHH May 9 '18 at 11:39
• You need to show an example circuit but, bear in mind that the TLV431 is +/- 3%. – Andy aka May 9 '18 at 11:39
• The TLV431 is ±1.5%, according to TI's page for it. – Hearth May 9 '18 at 11:43
• @Felthry re-read the datasheet NOT the sales pitch. That is ONLY at 25C. Full range it is just over +-3% – JonRB May 9 '18 at 11:47
• You really should wait a while before accepting any answer. 24 hours is a good time, since that gives everyone around the world a chance to chime in. – Olin Lathrop May 9 '18 at 12:04

To really answer this, you have to look at all sources of error:

1. The accuracy of the TLV431 itself. This includes initial value error and any additional offsets due to the temperature range.

2. The accuracies of R1 and R2. The way I usually deal with a resistor divider is to go thru the calculation twice. Find the divide ratio with max R1 and min R2, then also at min R1 and max R2. Again, you may have to increase the error range of the resistors according to the temperature range the circuit will operate in. As always, read the datasheet carefully.

3. How close to the regulation threshold the feedback system can keep the output voltage. This is what the error in the output voltage is under different loads and input voltages.

4. The worst case output ripple. This doesn't matter if you are looking for the average, but does matter if you need to guarantee a min/max voltage range to parts powered by the supply.

After adding all these up and projecting them to the output voltage, you'll usually find that there is a lot of slop in common off the shelf and low cost "5 V" supplies. ±200 mV would actually be quite tight after taking into account all possible errors from all possible conditions. In most cases, you'd be lucky to get ±500 mV over the full ranges of input voltage, output load, and temperature.

So to more directly answer your question, yes, it's quite possible that any one design actually has a nominal output of 4.8 V instead of 5.0 V. This is especially true for circuits you find on the 'net someplace.