# Beta multiplier reference

In the beta multiplier reference circuit, what is the function of start-up circuit? Can we use the beta multiplier without the start-up circuit?

• For fast startup you need a fast negative feedback bias for n and p. This circuit has a current source with negative feedback to do this . Can you see it? Commented May 31, 2017 at 10:10
• Can you please specify which negative feedback are you referring to?
– Swap
Commented Jun 3, 2017 at 16:21
• The loop around the bottom left corner from MSU3-d to MSU1-gd to MSU3-g with Iref Ref =Vref offset Commented Jun 3, 2017 at 16:26

Zero current in the left-hand branch and zero current in the right-hand branch is a possible steady-state operating point. If you powered up this reference without a startup circuit the reference current would be zero.

MSU2 is a small pull-up, which turns on MSU3, which kickstarts the feedback loop in the beta reference. One the reference hits steady-state, MSU1 overdrives the pull-up and turns of MSU3, leaving the startup circuit high-impedance.

• Can you please clarify what will be the gate voltages of pmos (M3,M4) and nmos (M1,M2) of beta multiplier during the steady state?
– Swap
Commented Jun 4, 2017 at 5:26

I'm now studying this topic, so I'll try to add something more about it; maybe it could be useful for future users..

Beta-Multiplier is a self-biasing circuit and, as in any self-biased circuit, there are two possible operating points: the desired one and the unwanted one, where zero current flows in it.

In the circuit above, the unwanted state occurs when the gates of M1/M2 are at ground while the gates of M3/M4 are at VDD.

In this situation, the gate of MSU1 is at ground and so it is off.

The gate of MSU2 is somewhere between VDD and (VDD - |VTH,P|).

So MSU3, which behaves like an NMOS switch, turns on and leaks current into the gates of M1/M2 from the gates of M3/M4.

This causes the current to snap to the desired state and MSU3 to turn off.

Now Beta-Multiplier works in the desired state and the start-up circuit is off.

PS: This circuit is suitable if we are using long-channel MOSFETs. Indeed, if we use short-channel MOSFETs, the performance of this version are poor, because Iref becomes sensible to the variation of VDD. To overcome this, a different version can be used (with an additional differential amplifier) to have a better power-supply sensitivity.