I've being working with constant current sources in designing a voltage reference for valve based power supplies, and I started out using a single MOSFET(DN2540- TO-92). Im only using TO-92 packages to save board space, because this will be a plug in module for the power supply.
Below I have an example of what I have arrived at. the VR stablizer is also variable. I would welcome any opinions from those that have worked with current sources. All active BJTs are MPAS42 divices. With this CCS Im able to deliver 2X the amount of current to the zener and temp comp diode.
Below I have an example of what I have arrived at. the VR stablizer is also variable. I would welcome any opinions from those that have worked with current sources. All active BJTs are MPAS42 divices. With this CCS Im able to deliver 2X the amount of current to the zener and temp comp diode.
Attachments
The current through Q1 should be fairly independent of the unregulated supply, but the current through R2 and Q2 spoils it. Why not use a PNP current source that doesn't have that issue?
???
The sum of the currents through Q1 and R2 must equal the sum of the currents through R4 and LED1, as it has nowhere else to go to.
When you change it into the PNP equivalent (everything upside down), you can connect R2 to ground and only the collector of Q1 to the Zener + diode.
The sum of the currents through Q1 and R2 must equal the sum of the currents through R4 and LED1, as it has nowhere else to go to.
When you change it into the PNP equivalent (everything upside down), you can connect R2 to ground and only the collector of Q1 to the Zener + diode.
Last edited:
Here is a possible fix for the bias current:
https://www.diyaudio.com/community/threads/improved-current-source-sink.173005/post-2294196
And here is another possibility:
https://www.diyaudio.com/community/threads/improved-2w-current-sources-ii.197104/post-2720209
https://www.diyaudio.com/community/threads/improved-current-source-sink.173005/post-2294196
And here is another possibility:
https://www.diyaudio.com/community/threads/improved-2w-current-sources-ii.197104/post-2720209
I came up with a better solution for my VR module, I basically converted it into an op amp, and added a LM329 instead of using a simple diode, R2 programs the current draw for the mirrors and differential stage. So far spice simulations look good. I havent run a noise test yet, but Im sure its excellent.
Attachments
Undegenerated current mirrors are pretty noisy, whether this matters in your circuit I can't tell.
The reference voltage will actually be RC filtered by an additional buffer curcuit, so any noise generated will be filtered out.