true VCE = 0V

[Tcheko]

Hello,

I am designing some circuit and I felt on a basic problem : I need a real 0V (read ground) after a BJT NPN used as a switch.

I did some search and found nothing revelent. Most pages says you can neglect it but in my use case, I can't.

How to overcome the VCE(sat) current leak and reach a real VCE = 0V ?

See attached picture of current design. Any tips welcome.

Regards,

Czk

[davidallancole]

You seem to not have posted the picture.

[Mr Evil]

Have you considered using MOSFETs? If you use ones with very low on resistance then you can get very close to ground.

[sawreyrw]

I agree, consider a MOSFET. A couple of questions. Why do you think you need 0 volts? At what current?

[Elvee]

As said above, jFETs or MOSfets have no saturation voltage; bjts always have some saturation voltage.
Tens of years ago, some bipolars were designed for chopper application and only had mV's of Vce under the rignt conditions, but they are not necessary anymore.
The best substitute you can find nowadays are low Vsat transistors: I just tested a ZTX849 in the sim, and it gives 3.9mV @1mA Ic and forced beta of 10. You won't get very much better than that.
If it's not enough, turn to the FEts, or use another similar transistor in series to cancel the voltage.

[sawreyrw]

I forgot to mention that old fashion bipolar chopper transistors were developed for this purpose.

[AndrewT]

why can't you consider a mechanical switch or even a relay?

[Tcheko]

Originally posted by sawreyrw
I agree, consider a MOSFET. A couple of questions. Why do you think you need 0 volts? At what current?

I am designing a DAC for educational purpose. The r2r ladder's bottom end is at ground. The Vce at saturation burry the LSB by a magnitude of 10 at least... (8mV Vce for 600uV LSB in simulation).

I think about applying a negative voltage at the emiter to correct the Vce at saturation. I uses an opamp in substraction mode. It worked nice in simulation : about 50uV. I guess it could be better with another opamp.

The opamp solution is a bit overkill to correct the Vce. Is there another way to solve this ?

[Tcheko]

Originally posted by AndrewT
why can't you consider a mechanical switch or even a relay?

Swith speed ?

48Khz for a mechanical swith or relay might be a bit ineffective.

[Elvee]

You can use a compensating transistor in series; with dual transistors, this will achieve a good accuracy.
But in my opinion, the circuit should be redesigned to make this problem irrelevant.