simulate this circuit – Schematic created using CircuitLab

I1 is an wilson modified current source by 4 pnp transistors. SW1 and SW2 are doubts. When SW1 is open, SW2 must be closed and vice versa.

I have only one control that is the variable source V1. When V1 = 3.3 V, then I1 = 15mA... and when V1 = 0V, I1= 0mA. For example, when V1 > 0.7V, SW1 is closed and SW2 is open... when V1 < 0.7V, then my load is grounded.

My question is, which are the best components to use in SW1 and SW2? BJT or MOSFETS? I need to implement this with discrete components. I'm using 2N3906 on my current mirror and a 2N3904 to control the current on it.

If I use BJT's for SW1 and SW2, will change the current form I1 to my load?

  • \$\begingroup\$ So, put in different words: You want to either apply a current source to the grounded load or else you want the load actively shorted? \$\endgroup\$ – jonk Apr 1 '19 at 17:45
  • 1
    \$\begingroup\$ My question is, which are the best components to use in SW1 and SW2? BJT or MOSFETS? ..... I don't know. You haven't specified why you need to make a choice. Either of them could work but they will behave differently. What type of behavior are you looking for? \$\endgroup\$ – user103380 Apr 1 '19 at 17:50
  • \$\begingroup\$ If you want to use mosfets, you want a low threshold N-channel MOSFET. Check: electronics.stackexchange.com/questions/36529/… \$\endgroup\$ – Huisman Apr 1 '19 at 17:51
  • \$\begingroup\$ Jonk, yes, basically is this. \$\endgroup\$ – Asfribas Apr 1 '19 at 17:51
  • \$\begingroup\$ KingDuken, I need is something that doesn't affect the basic behavior of the current source, the problem is that some component between I1 and the load affect the level of the current expected. \$\endgroup\$ – Asfribas Apr 1 '19 at 17:57

My question is, which are the best components to use in SW1 and SW2? BJT or MOSFETS?

Use logic level mosfets with a low rdson, make sure you select mosfets that have a gate drive voltage (that works with 3.3V). If you use BJT's, you'll have rougly a 0.7V drop from the BJT which would be like using diodes and contribute to loss.

Make sure mosfets are fully off or fully on (the gate voltage is either 0V or 3.3V) otherwise the mosfet could dissipate power.

| improve this answer | |
  • \$\begingroup\$ I think I have to use one pmos and one nmos, correctly? Well, in this case maybe I can use the 2n7000 nmos, can you indicate some pmos? \$\endgroup\$ – Asfribas Apr 1 '19 at 19:42

If you have only one control, perhaps you may as well just short the current source to ground rather than opening it up.

If that's acceptable, a single N-channel MOSFET will do an okay job there. For example, an AO6404 has Rds(on) of < 24m\$\Omega\$ with 2.5V drive (less with 3.3V drive), so the load will have less than 350uV across it if the layout is good. It can withstand up to 20V when off.

If you don't care so much about the last little bit of voltage, a BJT will work well (eg. 2N4401) and will have < perhaps 50mV across it when saturated with 1mA or so base current.

Generally opening up a current source causes saturation of something and it may take a while to recover so it's better to short it even if you do disconnect it completely from the load.

You could use analog switches to do this too (disconnect the current source, short it to ground, and short the load to ground, depending on the voltage rating required etc.

| improve this answer | |
  • \$\begingroup\$ I can't use analog switches because I need to do this automatically. The idea to open the connection to source and close the SW2 is to pull down the voltage in the load... So, I think just using one switch to short the current source to ground wouldn't solve my problem. \$\endgroup\$ – Asfribas Apr 1 '19 at 19:40
  • \$\begingroup\$ Neither of those sentences makes sense to me, but maybe it's because you're not really explaining. Without more numbers (current source details, maximum load resistance etc.) we're just guessing. Analog switches can contain level shifters which are going to be an issue if you just try to use transistors, as a guess. And yes, they have control inputs, many times usable from 3.3V directly. \$\endgroup\$ – Spehro Pefhany Apr 1 '19 at 19:42
  • \$\begingroup\$ Sorry, I got it wrong (english is not my language). You say something like MAX4623? \$\endgroup\$ – Asfribas Apr 1 '19 at 20:11
  • \$\begingroup\$ Many options here. DG509, for example. \$\endgroup\$ – Spehro Pefhany Apr 1 '19 at 20:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.