Hello!
This is a continuation of
https://www.eevblog.com/forum/testgear/output-capacitance-of-higher-end-bench-power-supplies/ .
I would like to build a high-precision, high-performance power supply. How would I do that?
Some general specs to give direction:
1. Digital control and readback
2. Output voltage range 0..15V (min 12V)
3. Current ranges 0..100mA and 0..3A (min 2A) with 1mA step, and 0.25mA accuracy. Or, alternatively, I can simply build two channels with different current capabilities.
4. Precision current set and readback. Many power supplies have cannot do that because of different parasitic currents, although error can be pretty small
5. Remote sensing. Well, remote up to output terminals, I don't long external connectors.
6. Stability with minimum output capacitance
7. Stability with large output capacitance
8. Performance... Well, how about recovery time from 0 to 100% load and back? Say, in less than 20us. Also fast CV/CC switch without glitches. Not sure how to define that.
Bonus points if PSU can:
1. Sink current
1. Bipolar
1. Multiple voltage ranges, but I don't really need steps more than 1mV. But nice to have
.
1. True remote sensing.
I decided to split task into two. The first one is choosing power supply architecture. The second one is choosing components.
I'm more or less set on parts to be used. And, I think, it's less important than architecture. I need you help there.
May be gurus of analog engineering want to share their designs and ideas? Currently the best I can do is the following. I didn't figure out how to make a beefy, precise and fast power supply. So, I decided that I only need precision and performance for low-power and low-voltage ranges. So, I'd make a low-power channel that would consist of simple emmitter follower with a fast bjt (like 2n2222 or something, point is it has much lower collector-base capacitance than, say, 2STA1943).
To reduce recovery time, I'd add a current sink at the output. This will create an offset current as it goes through the shunt. So, this needs to be taken into account. But even if we resolve this problem, I still haven't figured out how to switch fast between CC and CV mode. I'm currently looking at opamps that have ability to limit output swing, but there are not many around. So, clearly, not the best design.
Now to start answering from that thread:
The TIP2955/TIP3055 are more like normal slow BJTs. For fast response I would consider 2SC5200 or similar transistors made for high end audio. For a smaller version D44H / D45H could be an option.
Thanks for suggestion, I'll try 2SC5200. As of D44H and TIP2955, I think the performance of these part varies a lot from manufacturer to manufacturer. My measurements show that original parts from Toshiba (okay, I have only one sample of 2ta1943) was much slower than 2STA1943 (ST). As of D44H, I think showed similar performance to TIP2955/TIP3055. I can't find curves for D44H8 the moment, but I'll do measurement again and share them here. Here is methodology I'm using for measurements:
https://www.eevblog.com/forum/beginners/evaluating-bjts-for-a-linear-psu-am-i-doing-it-right/msg2950816/#msg2950816 .
As for other questions, I'll answer them next time. It's a lot of topics yet to discuss, and I want to start with something. So, here is the thread for that.
I'm not investigating push-pull stages, but I have zero previous experience with them, so it will take time how to figure out how they work, how to optimize, say, biasing, and how to sense current.